Electrical and Electronic Engineering Series

Frederick Emmons Terman, Consulting Editor

McGraw-Hill

W. W. Harman and

J.

G. Truxal, Associate Consulting Editors

RADIO ELECTRONICS

McGraw-Hill

Electrical and Electronic Engineering Series

Frederick Emmons Terman, Consulting Editor
W. W. Harman and J. G. Truxal, Associate Consulting Editors

Bailey and Gault Alternating-current Machinery

Beranek Acoustics
Bruns and Saunders Analysis of Feedback Control Systems
Cage Theory and Application of Industrial Electronics

Catjer

Synthesis of Linear Communication Networks, Vol. 1

Harmonics, Sidebands, and Transients in Communication
Engineering

Cuccia

Davenport and Root Random Signals and Noise

Eastman Fundamentals of Vacuum Tubes
Evans Control-system Dynamics
Fitzgerald and Higginbotham Basic Electrical Engineering
Fitzgerald and Kingsley Electric Machinery
Geppert Basic Electron Tubes
Glasford Fundamentals of Television Engineering
Happell and Hesselberth Engineering Electronics

Harman Fundamentals

of Electronic

Harrington Introduction

Hayt

Motion

to Electromagnetic Engineering

Engineering Electromagnetics

Hessler and Carey Fundamentals of Electrical Engineering

Hill Electronics in Engineering
Johnson Transmission Lines and Networks

Kraus Antennas
Kraus Electromagnetics
LePage Analysis of Alternating-current Circuits
LePage and Seeley General Network Analysis
Millman and Seely Electronics
Millman and Taub Pulse and Digital Circuits
Rogers Introduction to Electric Fields
Rudenberg Transient Performance of Electric Power Systems
Ryder Engineering Electronics

Seely Electromagnetic Engineering

Seely Electron-tube Circuits
Seely Electronic Engineering
Seely Radio Electronics
Siskind Direct-current Machinery

Skilling
Skilling

Electric Transmission Lines

Transient Electric Currents

Spangenberg Fundamentals of Electron Devices

Spangenberg Vacuum Tubes
Stevenson Elements of Power System Analysis
Storer Passive Network Synthesis

Terman Electronic and Radio Engineering

Terman and Pettit Electronic Measurements
Thaler Elements of Servomechanism Theory
Thaler and Brown Servomechanism Analysis
Thompson Alternating-current and Transient Circuit Analysis
Truxal Automatic Feedback Control System Synthesis

Radio Electronics
SAMUEL SEELY,

Ph.D.

Department of Electrical Engineering

Case Institute of Technology

McGRAW-HILL BOOK COMPANY, INC

New York

Toronto
1956

London

RADIO ELECTRONICS

1956 by the McGraw-Hill Book Company, Inc. Printed in the

Copyright
United States of America. All rights reserved. This book, or parts thereof,
may not be reproduced in any form without permission of the publishers.
Library of Congress Catalog Card

Number 55-5696

III

THE MAPLE PRESS COMPANY, YORK,

PA.

PREFACE
"Radio Electronics" and "Electronic Engineering" have been written
as companion volumes, though written as independent textbooks.
Together they represent a revision and extension of the author's "Electron-tube Circuits." It has been necessary to include a certain amount
of material that is

common

to both books in order to ensure completeness

and continuity of text material. Moreover, the common material is the

same in each book, in the interests of economy of production. However,
of duplicated material has been kept to a minimum, conwith
the desire to have these books independent, and also to
sistent
provide a complete and continuous development.
While one of the main objectives of this book, as its title would indicate,
is a study of the important electronic aspects of radio, a much broader

the

amount

contemplated. That is, more than just an analysis of the elesystems is undertaken. For this reason general system
ments
Many
block diagrams are discussed before any analysis is undertaken.
scope

is

of radio

of the

important aspects of communication-systems engineering are also

some of the major factors that must

be considered in system synthesis. This accounts for the inclusion of
such topics as noise, signal/noise ratios, gain-bandwidth products, and
an introduction to information theory, with consideration of the rate of
transmission of information. Some of the systems discussed superdiscussed in order to bring into focus

ficially

require specialized techniques in their operation,

of these is

deemed to be outside

of the scope of this book.

and a discussion
Therefore only

very limited discussion will be found of pulse-modulation systems.

Wherever possible, the analysis proceeds in two stages. An effort
is made first to present an explanation of the operation of the circuits
from a physical point of view. This is followed by fairly rigorous mathematical analyses. Such mathematical analyses have a threefold objecOften, in fact, altertive: (1) To illustrate the techniques of analysis.
native developments have been included to demonstrate different methods
of analysis.

(2)

To deduce a

operation of the circuit.

(3)

solution which yields a description of the

To examine

the effects of the various

parameters on the operation of the circuit.

In all analyses considerable care has been taken to include the requisite
reference conditions for

potential

polarities,

current directions,

and

PREFACE

vi

transformer-winding sense. These are an essential part of any circuit

diagram, and without them the ultimate choice of a positive or negative
sign would require a major decision.
Much of the material in this book has been used in two courses in
radio at Syracuse University.

The introductory chapters

are essential

in the first course, as this provides the students with their first introduc-

tion to electronic devices

and

their circuit applications.

retical course in physical electronics follows rather

more theo-

than precedes

this

course.

The author wishes

to acknowledge the helpful discussions with many of

former colleagues at Syracuse University. He is particularly indebted
to Dr. Herbert Hellerman for his many helpful suggestions and for his
assistance in proofreading the entire text.
his

Samuel Seely

CONTENTS
v

Preface

Chapter

1.

Introduction to Communication Systems

2.

Characteristics of Electron

3.

Rectifiers

4.

Vacuum

5.

Basic Amplifier Principles

100

6.

Untuned Potential Amplifiers

119

7.

Special Amplifier Considerations

143

8.

Untuned Power Amplifiers

187

9.

Tuned

212

and

Tubes

15

53

Filters

Triodes as Circuit Elements

Potential Amplifiers

83

10.

Tuned Power

11.

Oscillators

283

12.

Amplitude Modulation

320

13.

Demodulation

348

14.

Frequency Modulation and Detection

377

15.

Information Theory

429

Notes on General Network Analysis

443

Appendix A.
B.

241

Amplifiers

Plate Characteristics of

Vacuum Tubes
Tubes

C.

Characteristics of Transmitting

D.

Table of Bessel Functions of the First Kind

454
471

473
475

Index

vii

CHAPTER

INTRODUCTION TO COMMUNICATION SYSTEMS

Elements

Communication Systems. Communication systems

and reception of intelligence. The
form and features of the systems depend upon the type of intelligence
being communicated. For example, telegraph transmission requires a
frequency band of only several cycles per second, whereas the frequency
band required for transmitting television intelligence is several mega1-1.

of

are concerned with the transmission

cycles per second; consequently, the details of the systems will differ

markedly.

However, despite the different waveforms and frequencies

may demand for its transmission, all communication

that the intelligence

Intelligence

Intelligence in
electrical form

Energy

Transducer

Fig. 1-1.

Energy modulated
by intelligence

Receiver

source

The elements required

Intelligence in
suitable form to
actuate output device

of a complete

Intelligence

Output

communication system.

systems may be crudely represented by essentially the same blocks in their

schematic representations. These are illustrated graphically in Fig. 1-1.
In such a system, it is the function of the transducer to convert the
intelligence, whether it is voice or music as in radio, whether it is a
message for a telegraph system, whether it is voice for a telephone system,
or whether it is both aural and visual as in television, into such a form
as to be suitable for modulating an energy source.
The modulated
energy source is in a form which will allow the energy to pass through the
transmitting medium. The modulated energy reaches a second transducer the receiver which extracts the intelligence therefrom and provides this intelligence in a suitable form for actuating the output device.
In the specific case of radiobroadcast transmission, the transducer is
the microphone, which converts the sound energy impinging on it into
an appropriate electrical form at low power level. The energy source is
a complicated device which includes amplifiers for raising the power level
of the audio signal at the microphone to a sufficient value to modulate a

RADIO ELECTRONICS

[Chap. 1

high-power high-frequency signal, which is generated in an oscillator.

This modulated h-f "carrier" is fed to an antenna, from which radiation
of the energy into free space occurs.
At the receiver some of the radiated energy is absorbed by an antenna.
This energy is at very low power level, which is then amplified, and the
intelligence is extracted from the modulated signal.
The extracted signal
is then applied to a loud-speaker, which is to reproduce the original signal.

From

the foregoing discussion, the essentials of a communication

system are the following:

1.

means

or

medium

for transferring energy

from a transmitter to a

receiver.
2. A means for modulating the energy which
from the transmitter to the receiver.
3.

will carry the intelligence

A means for preparing the intelligence in a form suitable for perform-

ing the function under (2).

4.
5.

A
A

means
means

for extracting the intelligence

from the transmitted energy.

for presenting the intelligence in proper form.

Certain aspects under these topics will be considered in this chapter.

This book will be concerned with details of many of the circuits required

number

of communication systems.
Systems for Communication. In the example given above of a
communication system, it is stated that an h-f carrier is used. Actually,
however, three basic systems of communication exist. These are:

in a

1-2.

1.

Noncarrier systems.

2.

Carrier systems.

3.

Suppressed-carrier systems.

noncarrier system

signal.

is one in which energy appears only with the

In a carrier system, energy flows between the transmitter and

the receiver even

fied

by the

when no signal appears. The carrier current is modiway which will permit the ultimate extraction of

signal in a

In the suppressed-carrier systems, no carrier

energy flows between the transmitter and the receiver. However, the
carrier originally existed in the system and was suppressed prior to
transmission.
The carrier must be reintroduced at the receiver in order
the signal information.

to extract the signal information.

An example of a noncarrier system is found

which a dot is a short-duration square pulse,

in simple telegraphy, in
first

with one polarity,

by the same duration square pulse, with reversed polarity.

That is, a dot is a single-cycle square wave of relatively short duration.
The dash is a single-cycle square wave of relatively long duration. A
space is a region of no current. In this type of telegraph system, the
complete signal consists of a series of square a-c waveform elements, the
followed

average current being zero.

COMMUNICATION SYSTEMS

Sec. 1-3]

Systems. A second method of telegraph communication

one in which a dot is a current pulse of short duration, a dash is a
current pulse of long duration, and a region of no current is a space.
In this method, the average current is not zero, and the dot and the dash
may be considered to modulate the current. The essential aspects of the
1-3. Carrier

is

noncarrier and carrier methods are illustrated graphically in Fig. 1-2.

A modulated a-c carrier system is used also, as well as the amplitudemodulated d-c carrier system, for telegraphy. If the amplitude of the
a-c carrier is altered by the signal, the system is known as an amplitudemodulated (a-m) system. If the frequency of the a-c carrier is altered
by the signal, the system is known as a frequency-modulated (f-m)
system. If the phase of the a-c carrier is altered by the signal, the system
is known as a phase-modulated (p-m) system.

Dot
Dot

Dash

D-c

Dash

Space

Fig. 1-2.

noncarrier and carrier telegraph signal.

and this may be

communication,
Mc (megacycles per second) for voice communication
for television signals, is to be found in the frequency

The need

for a relatively h-f carrier,

(kilocycles per second)

necessary for the

for telegraph

amount

as low as 15 kc
as high as 200

and 1,000 Mc
band which is

of information that is being transmitted per

second, and also in the fact that an h-f carrier

mission by radiation of electromagnetic waves.

is

essential for the trans-

number

of practical

considerations are also involved, an important one being the fact that the

antenna problem for both transmitter and receiver

is

simplified at the

higher frequencies.

Consider a waveform which

may

be represented analytically by the

expression
e

= A

cos (<

0)

(1-1)

where t is the time. In amplitude modulation, the amplitude A is varied

accordance with the modulating or signal information, while o> and
remain constant. In frequency modulation, the frequency w is varied,
while both A and 8 remain constant.
In phase modulation, the phase 6
is varied relative to some arbitrary datum, while A and oo remain constant.
The essential character of a modulated carrier may be illustrated
in

RADIO ELECTRONICS

4
graphically.

This

is

done in Fig.

1-3,

[Chap. 1

which shows an

1-f

square-wave

and the form of the a-m, f-m, and p-m waves.

1-4. Comparison of Modulation Systems.
Amplitude modulation is
the oldest method of carrier transmission and is used almost exclusively
in radiotelephony and -telegraphy.
Commercial radiobroadcast transsignal

mission is predominately amplitude-modulated, although considerable

f-m broadcast activity exists. The sound channel of TV transmission is
of the f-m type.
Also, frequency-shift keying is important for telegraphic work. This is an f-m system, with dots being pulses of equal

Signal

Fig. 1-3.

The

essential character of

modulated

carrier waves.

amplitude and duration which have been shifted in frequency by one

amount, and dashes being pulses of the same amplitude and duration
but which have been shifted in frequency by a different amount.
An important feature of a-m transmission is that the total frequency
spread around the carrier frequency, which arises because of the modulation which is applied to the carrier, is twice the highest frequency in the
modulating signal, the spread extending from o)m mai the maximum
modulating frequency, above the carrier to an equal spread below the
carrier frequency.
The frequency band above the carrier is called the
upper sidebands, and the frequency band below the carrier is called the
lower sidebands.
For most commercial broadcasting, a total band spread
of 10 kc is allowed.
Inherently, therefore, a-m transmission might be
,

COMMUNICATION SYSTEMS

Sec. 1-4]

narrow-band transmission. Because of this, a-m transmission is

possible over a very wide range of carrier frequencies, and such transmission is carried out commercially from several hundred kilocycles to
30 Mc.
In so far as the receiver is concerned, a consideration of fundamental
importance is the ratio of signal to noise at the output. The nature and
called

sources of noise will receive detailed consideration later in the text.

will suffice here to note that
is

may

continuous and

sinusoidal

random

noise

is

be resolved into an

components at

all

frequencies.

of

two

classes.

infinite

The

number

One

It

class

of small

other class consists of

occasional pulses of relatively short duration but of large amplitude.

These, too, can be resolved into sinusoidal components covering a range

Those components of noise from both sources which lie
of frequencies.
within the pass band of the reciever contribute to noise output.
In an a-m receiver, the noise output results from the interaction of the

thereby producing an 1-f varisame character as the a-m signal.

cannot be separated from the signal and remains as an undesired

noise

components with the desired

ation of
It

While

signal.

noise,

The

amplitude.

it is

the random

carrier,

noise has the

possible to limit the effect of large pulse bursts of

noise serves to limit the total allowable gain or sensi-

tivity of the receiver,

and so the minimum desired

signal that

might be

received.

Frequency-modulated transmission has found considerable application,

both for radiobroadcasting and as the sound channel of TV receivers.
Phase modulation has found very little application, except as an intermediate step in certain systems for producing f-m waves.
Frequency modulation differs from amplitude modulation, in addition
to the fundamental differences of character of modulation as illustrated
in Fig. 1-3, in the fact that, in order to include all the sideband energy due
to a complicated signal spectrum, such as that, for example, from an
orchestra, a frequency band of approximately 100 kc on each side of the
Because of this fact, frequency modulation is usually
carrier is required.
confined to the high frequencies, from perhaps 50 Mc and higher, in order
Such frequencies are also
to provide room for a number of f-m channels.
dictated by the receiver problem, since it would be extremely difficult to
design band-pass circuits of 200 kc bandwidth except at the higher
frequencies.
It is instructive to

mitter.

examine certain

of the features of

an f-m trans-

Since the amplitude of the carrier remains constant during

modulation, the transmitter can be adjusted to give its rated output

power, at which point the transmitter is operating at its optimum efficiency.
In an a-m transmitter, the peak power with full modulation
(100 per cent modulation) is four times the average value.
Because of the

RADIO ELECTRONICS

[Chap.

essential differences in operation, a given transmitter will produce twice

when

the modulation power at the receiver

it

is

operated as an f-m

system.

In the f-m receiver, since the intelligence is carried in the frequency

deviation and not in amplitude variations, a limiter is used to remove
amplitude variations. Noise, which is amplitude-modulated, will similarly

be limited when the noise

in fact, that those

will

produce

When

less

components

is

near

its

peak.

It will

be shown later,
frequency

of noise nearer to the carrier

frequency modulation, and so

less noise in the output.

is about
30 db higher than with an a-m transmitter of the same peak power, when
the deviation frequency is about 75 kc and the input signal/ noise ratio is

all

factors are taken into account, the signal/noise ratio

fairly good.
If

the signal/noise ratio at the receiver is fairly poor, then, because

wide receiver bandwidth of the f-m receiver, the extra noise potenIn such cases, a relatively
will contribute to noise in the output.

of the
tials

narrow-band f-m system is desirable and will yield a better signal/noise

an a-m receiver in the same situation.
Another fact which will be discussed later is that two f-m signals of
differing amplitude at the same carrier frequency will interfere less with
each other than will comparable a-m signals. In fact, if the desired
signal is about twice the amplitude of the interfering signal, the interference at the receiver output is very small.
In an a-m system, even though most of
1-5. Carrier Suppression.
ratio than

the radiated power is contained in the carrier, the intelligence is contained

only in the upper- and lower-frequency sidebands. In fact, the intelligence is contained in each sideband. Clearly, therefore, if the carrier

were suppressed before amplification, the efficiency of the system could

be improved materially.
The problem of the receiver design becomes more complex for suppressed-carrier reception than for ordinary amplitude modulation,
because, as will later be seen, it is necessary to reintroduce the carrier
While it is not difficult to
before the intelligence can be extracted.
generate locally an h-f signal, the reintroduced carrier must maintain
the proper phase relation with respect to the sidebands if distortion is
While it would be possible to transmit a pilot carrier to
to be avoided.
serve as a reference for controlling the locally produced carrier, this
would introduce considerable complications. Also, since the phase of the
locally generated carrier could not be controlled accurately enough for
such operation, suppressed-carrier transmission is not very practical.
1-6. Single -sideband Systems.
A satisfactory method in which the
carrier is suppressed is possible
is

also suppressed.

if,

at the

Such a system

is

same

time, one of the sidebands

theoretically feasible, since, as

COMMUNICATION SYSTEMS

Sec. 1-7]

mentioned in Sec.
sidebands, and,

by

may be extracted.

1-5,

the signal information

is

contained in each of the

later reintroducing the carrier, the signal information

Such single-sideband operation is very desirable, since

is half that when both sidebands are

the total required band spread

transmitted.

The phase

of the reintroduced carrier in the single-sideband suppressed-

system must be fixed with respect to the transmitter datum;

otherwise, phase distortion of the output results.
However, for telephony work, no difficulty arises, as the ear is not sensitive to phase changes
but does respond to frequency changes. As a result, if the frequency
of the reintroduced carrier is different from that of the suppressed carrier,
serious distortion may arise.
However, it is possible to set the carrier
carrier

and satisfactory operation is feasible. Such single-sideband

is employed for long-distance (transoceanic)
transmission by commercial organizations, the added complexity of the
receiver being justified in view of the efficiency of the system.
1-7. Pulse Modulation.
A number of p-m methods of transmission
have been developed during recent years. Such methods depend on the
fact that it is possible to reconstruct a given signal by sampling the wave
properly,

suppressed-carrier telephony

at periodic intervals, provided that the frequency of sampling

is

made

sufficiently high.

"Signal

/
Pulses

Time

*-

Fig. 1-4. Pulse-amplitude modulation.

In the pulse-amplitude-modulation (PAM) method, short pulses of

power are transmitted at regular intervals. The amplitude of each

r-f

pulse

is

proportional to the instantaneous amplitude of the signal.

The

For speech confined to a

frequency spread of 5 kc, the sampling frequency must be not less than
If the pulse duration at each sampling point is, say, 5 /*sec, this
10 kc.
means that the duty cycle is only 5 per cent, permitting a theoretical
situation

is

illustrated graphically in Fig. 1-4.

20 simultaneous messages at a given carrier frequency without overlap or

An adequate commutating means must be provided for

interference.

switching both transmitter and receiver to the different channels.

number

If

messages is small, then, for a given average power, the

peak power during transmission may be quite high.
the

of

RADIO ELECTRONICS

[Chap.

In pulse-duration, or pulse-width, modulation, the amplitude of each

pulse is constant, but the width, or duration, of the pulse is made proportional to the instantaneous amplitude of the signal. This scheme is
In such a method, the pulse duration might
illustrated in Fig. 1-5.
of 5 usee, the widths varying from 1 to 9 usee.
the deviation of the pulse from its datum (no signal) position is made
proportional to the instantaneous amplitude of the signal, the system

have a mean value

If

Fig. 1-5. Pulse-duration, or pulse-width, modulation.

is

the pulse-time, pulse-phase, or pulse-position modulation.

ation

is

This

situ-

illustrated in Fig. 1-6.

In pulse-code modulation (PCM), somewhat different factors must be

considered in a discussion of the method. In pulse-code modulation the
process is characterized by three steps: (1) the signal is sampled, as in all
pulse-modulation methods;

(2)

the pulse heights are "quantized" and

;

(3)

Fig. 1-6. Pulse-position modulation.

the quantized signal is converted into a sequence of coded pulses. The

term quantization as here used implies the selection of a particular
amplitude level to represent a range of amplitude variations. The
quantizing process thus selects the nearest amplitude level, instead of the
continuity of amplitudes that actually exists.
Suppose, for example, that a four-pulse group

the amplitude level.

The

is

available for coding

signal value at a given single point will be

COMMUNICATION SYSTEMS

Sec. 1-7]

TABLE 1-1
BINARY CODE FOR FOUR-PULSE GROUPS IN PULSE-CODE

MODULATION (PCM)
Amplitude

Amplitude
Code

represented

0000
0001
0010
0011
0100
0101
0110
0111

3
4

5
6
7

Code

represented

1000
1001
1010
1011
1100
1101
1110
1111

8
9
10
11

12
13
14
15

coded as in Table 1-1. In this table the presence of a pulse is indicated

Clearly the four-pulse
1, and its absence is indicated by a 0.
groups can express all integral values from
to 15. Owing to this limitation in amplitude levels available, when the pulse groups are reconverted into a signal there will be a discrepancy between it and the original.
This discrepancy is called "quantizing noise." Evidently, the larger
the number of quantization levels available, the smaller will be the
relative distortion produced by the system.

by a

Code

TimeFig. 1-7. Pulse-code modulation (a typical code for several signal amplitudes only is
and these do not have a realizable code pattern for the amplitudes shown)

illustrated,

As indicated above, the

tems

is

essential

advantage

that time-division multiplexing

that the duty cycle

is

are equally favorable.

relatively small.

is

of

pulse-modulation sys-

made possible, owing to the fact

However, not all pulse methods
will be examined in the light

The various systems

of receiver signal/noise ratio.

For the pulse-amplitude system, the signal/noise ratio is worse than

with ordinary amplitude modulation, owing to the fact that the receiver

bandwidth must be considerably wider in the pulse-amplitude-modulation

receiver than for ordinary amplitude modulation.
This broader bandwidth is necessary in order to permit a reproduction of the pulses.
In the pulse-duration and pulse-position methods, since the amplitudes

RADIO ELECTRONICS

10

[Chap.

remain constant, limiters can be used. Although

loss of energy, the noise, which is amplitudemodulated, will be removed. However, noise which appears on the
leading and trailing edges of the pulses will produce an effect on the overConsequently, by increasing the bandwidth of
all signal/noise ratio.
of the pulses are to

this

may

result in

some

the receiver, thereby permitting steeper sides to the pulses, the noise
output is reduced. This is offset to some extent by the fact that,

in the

with a broader bandwidth, the input noise is increased. The net effect
that the signal/noise ratio decreases with increasing bandwidths,
although an optimum bandwidth is dictated in such applications. If

is

impulse noise occurs, this might increase the effective duration, in pulseduration modulation, or it might alter the position in pulse-position
modulation, with a resultant production of noise.
An important feature of pulse-code modulation is that distortion and
noise must be very large before the presence or absence of a pulse cannot

be recognized.

In

fact,

if it is

pulse combined with noise,

original

makes

pulse.

it is

possible to establish the existence of a

theoretically possible to regenerate the

This characteristic of pulse-code-modulation systems

their use in radio relay links particularly desirable.

for the regeneration process to

permitting a

new

It is possible

be carried out at each relay station, thus

signal to be available for retransmission.

System Block Diagram. As has probably been surmised, many

the fundamental circuits and techniques are common to all communica-

1-8.
of

tion systems, although certain of the system elements will be special to the
To serve as an introduction to the analytical disparticular system.

and also to indicate the general features of

more important systems of communication, block diagrams

cussion of circuits to follow,

certain of the

It
of the essential elements incorporated in such systems will be given.
should be emphasized that certain of the blocks which are illustrated
as separate entities might, in fact, be combined in achieving a given

operation.
a.

A-M

Systems.

The

essential elements of a simple broadcast trans-

mitter and of appropriate receivers are given below.

The block diagram

Nothing

is here said
an a-m transmitting system is given
elements
or
stages.
These
concerning the power capacities of the various
will vary according to the particular circuits or systems of operation used
and will be considered at the appropriate points in the text. The
frequencies which appear at various points in the circuit are indicated
on the diagram.
It will be observed that the a-m transmitter is a relatively simple
device, consisting essentially of an audio amplifier for raising the level
of the intelligence signal, a stable oscillator which establishes the carrier
frequency of the transmitter, a modulator in which the signal and the

of

in Fig. 1-8.

COMMUNICATION SYSTEMS

Sec. 1-8]

11

combined to produce a modulated carrier, and power amplifiers

power level of the modulated carrier before transmission.
Observe that the modulator serves to translate the signal information at
frequency co m as measured from the zero level to the information given
carrier are

for raising the

with reference to the carrier

The

level,

a-m wave

receiver for an

In the receiver, the

Fig. 1-9.

amplified from

its

cc c .

shown

is

r-f

form

in

potential induced in the antenna

is

in its simplest block

usual low level of 5 /xvolts to

mv

to perhaps

to 30

r~ Signal in
electrical -form

Radiator

Modulating
signal Wm,

Audio

amplifier

Modulated carrier

Transducer

Modulator

m ,coc -a>m

coc ,ai c +a)

High freq.

power
amplifier

Carrier
Oscillator

Fig. 1-8.

coc

The elements

of

an a-m transmitting system.

may all occur at the center frequency co

may be changed to some intermediate value

This amplification

volts.

the transmitter, or

it

The

amplified modulated

of
coj

then passed
through the demodulator, which extracts the signal from the modulated
wave. In essence, therefore, the demodulator, or detector, serves in the
within the receiver.

r-f

signal

is

The demodulator may be considered

from the carrier level ut to the zero level,
so that the signal frequencies w are specified with respect to the zeroinverse

manner

of the

modulator.

to translate the frequency level

frequency

level.

The audio

ate the loud-speaker

signal

is

then amplified sufficiently to oper-

and usually provides about 3 to 30 watts output.

-With frequency changing u>i,coiw

Wi+houf frequency changing ojc ,ajc a>m
High freq.
u>c , coc

wm potential

<>>m

Demodulator

amplifier

amplifier

Fig. 1-9.

Audio
frequency

The elements

of

Mm.

Loud

speaker

an a-m receiver.

Two important different types of receivers are in widespread use, the

tuned radio-frequency (trf) and the superheterodyne types. The tuned
r-f receiver is illustrated in block form in Fig. 1-10, and the superheterodyne receiver is illustrated in block form in Fig. 1-11.
The tuned r-f receiver shown in Fig. 1-10 performs in precisely the
manner of the general block diagram of Fig. 1-9. Tuning is accomplished
by changing the resonant frequency of the parallel LC circuits of each
stage, either

by changing the capacitance

or

by changing the inductance.

RADIO ELECTRONICS

12

The superheterodyne

[Chap. 1

receiver illustrated in Fig. 1-11

is

more common

a single stage of tuned r-f

amplification, although this is often omitted in many receivers, the output of which feeds into a mixer (and this is often referred to as the "first
for

normal broadcast reception.

detector").

The

lator.

It comprises

Also feeding into the mixer

is

the output from a local

oscil-

frequencies which are centered about the carrier w c combine

with the local oscillator frequency co to produce a modulated carrier

coo or around
is now centered either around the frequency co c
,

which

V
Tuned
r-f
amplifier

coc

Audio
frequency

Demodulator

wm

amplifier

Fig. 1-10. Block diagram of the tuned

r-f receiver.

depending on choice, which is called the intermediate frequency

The modulated i-f signal then passes through a fixed tuned amplifier,
coi.
the output from which feeds a detector, or demodulator, as before.
The essential differences between the tuned r-f and the superheterodyne receivers are important. In the tuned r-f receiver, tuning and
amplification are effected at the incoming carrier-frequency level co c
Because of this requirement, the circuits used are best from tuning conIn the
siderations but are not optimum from selectivity considerations.
frereceiver,
tuning
is
accomplished
by
adjusting
the
superheterodyne
the
incoming
signal
is
converted
to
quency of the local oscillator. Each
co c

coo,

V
COc

Tuned

r-f
amp.

<oc
ojc

wm

Fixed

Mixer

U>l*U

u>i

u>iw m
m frequency
amp.

Detector

Audio
amplifier

Local
oscillator

Fig. 1-11.

The elements

same intermediate frequency

to,-,

of a superheterodyne receiver.

and the

i-f

signal

is

amplified in high-

gain fixed tuned stages which have been chosen for optimum selectivity.
This is possible since the requirement of tunability has been removed as a

requirement of the amplifiers. The detector and audio amplifiers are

substantially the same in both receivers.
An f-m transmitter in its most elementary form is
b. F-M Systems.
given in Fig. 1-12. The audio-frequency (a-f) amplifier and the h-f
power amplifiers are quite conventional. However, the high frequency

COMMUNICATION SYSTEMS

Sec. 1-8]

13

used in f-m systems, as already noted,

is usually considerably higher than

This requirement is made necessary by the
broad band of frequencies which is required to transmit all the intelligence
contained in the audio signal. Specifically, to cover the normal audio
spectrum to, say, 7,500 cps, the bandwidth required in the f-m system is
approximately 150 kc. To achieve a bandwidth of 150 kc without excessive loading, the standard broadcast carrier level for f-m transmission
ranges from 88 to 108 Mc.

that for the

a-m system.

amplifier

The elements

Fig. 1-12.

The f-m

may

oscillator

High freq.

Frequency
modulated

Audio

Transducer

power

+side
oscillaior
bands

of

amplifier

wc
+ side

bands

an f-m transmitting system.

take any one of a number of different forms.

"reactance" tube as part of the oscillator

A common form incorporates a

circuit.
With such a device, the effective capacitance or the effective
inductance of the oscillator tank can be changed in a manner dictated
by the audio signal. In this way, the audio signal is converted into
changes in frequency. The f-m oscillator output is usually at a considerably lower frequency than the ultimate station carrier, and the frequency
is

by means of frequency-multiplying
power level required for transthe other systems of producing f-m waves are too

multiplied to the necessary levels

stages, before it is amplified to the high

mission.

Several of

I-f amp.

R-f
amp.

Mixer
+Acoc

and
limiter

<"i

Discrim-

<>>m

inator

Audio

amp.

o>

Local
oscillator

Fig. 1-13. Block diagram of a typical f-m receiver.

complicated for simple discussion, and will be deferred until Chap.

where they will be discussed in detail.

The

14,

an f-m receiver is somewhat similar to that of an a-m

A block diagram is given in Fig.
1-13.
The r-f amplifier, mixer, and beat-frequency oscillator are common
with those in an a-m superheterodyne receiver, except that the over-all
bandwidth is greater, as already discussed. The i-f amplifier is like that
in the a-m superheterodyne, although the last stage is operated at low
basic circuit of

receiver of the superheterodyne type.

14

RADIO ELECTRONICS

potential to produce limiting action.

[Chat.

This serves to eliminate any fluctuno matter how produced. The

other outstanding difference is the circuit used to demodulate the f-m
carrier.
The f-m discriminator serves to convert from frequency moduThe resulting amplitude modulation is
lation to amplitude modulation.
demodulated in a circuit that is peculiar to the f-m system.
ations in the amplitude of the

i-f

carrier,

CHAPTER

CHARACTERISTICS OF ELECTRON TUBES

Before one undertakes a study of many of the
communication systems that have been discussed rather
superficially in the previous chapter, it is desirable to examine the fundamental physical principles which govern the operation of the electron
2-1. Introduction.

details of the

tubes that are used in such systems.

There are two important questions that relate to such devices.

One

and their liberation. The

the electron beam. A brief discussion

relates to the actual source of the electrons

second relates to the control of

of these

matters will be included here.

EMISSION OF ELECTRONS
Source and Control of Electrons. According to modern theory,
matter is electrical in nature. The atom, which is one of the fundamental building blocks of all matter, consists of a central core or nucleus
which is positively charged and which carries nearly all the mass of the
atom. Enough negatively charged electrons surround the nucleus so
2-2.

all

atom is electrically neutral in its normal state. Since all chemsubstances consist of groups of these atoms which are bound to each
other, then all matter, whether it is in the solid, the liquid, or the gas-

that the
ical

eous state,

is

a potential source of electrons.

do, in fact, serve as sources of electrons.

All three states of matter

number

esses serve to effect the release of electrons, those

of different proc-

which are

of

importance

(1) thermionic emission, (2) secondary emission, (3) photoelectric emission, (4) high field emission, and (5) ionization.
These processes will be considered in some detail in what follows.
With the release of the electrons, a means for their control must be

in electron tubes

provided.
tric fields

being

Such control is effected by means of externally controlled elecThese fields perform one or both
fields, or both.

or magnetic

of the following functions: (1) control of the number of electrons that

leave the region near the emitter; (2) control of the paths of the electrons

they leave the emitter. Control method 1 is the more common, and
such a control method is incorporated in almost all electron tubes, except
The cathode-ray tube is a very
those of the field-deflected variety.
after

15

RADIO ELECTRONICS

16

[Chap. 2

important example of a field-deflected tube. However, even in this latter case, a control of type 1 is incorporated to control the electron-tube
current, even though the subsequent motion is controlled by means of
electric or magnetic fields, or both.
2-3. Thermionic Emission.
Consider matter in the metallic state.
Metals are most generally employed in the form of a wire or ribbon filasuch a filament contains electrons and if these are relatively
free to move about in the metal (and this is the case since the application
of a small potential difference between the ends of the wire will result in a
current flow), it might be expected that some electrons might "leak"
out of the metal of their own accord. This does not occur, however.
Consider what happens to an electron as it seeks to escape from a metal.
The escaping, negatively charged electron will induce a positive charge

ment.

If

on the metal. There will then be a force of attraction between the

induced charge and the electron. Unless the escaping electron possesses
sufficient energy to carry it out of the region of influence of this image
force of attraction,

amount

of

it

energy that

be returned to the metal.

will
is

known
minimum amount

attractive force

The minimum

required to release the electron against this

is

as the work function of the metal.

This

of energy may
methods. One of the most important methods is to
heat the metal to a high temperature. In this way, some of the thermal
energy supplied to the metal is transferred from the lattice of the heated
metal crystals into kinetic energy of the electrons.

requisite

number

An

be supplied by any one of a

of different

explicit expression relating the thermionic-emission current density

and the temperature

derived has the form

of

the metal can be derived. 1 *

The

J th = A T 2 e~'">/T

expression so
(2-1)

metals and has the value of 120 X 10 4

amp/(m )(K ) and b a is a constant that is characteristic of the metal.
The quantity 6 is related to the work function E w of the metal by

where

is

a constant for

all

= ll,600iV

(2-2)

been found experimentally that Eq. (2-1) does represent the form
with temperature for most metals, although
the value obtained for A may differ materially from the theoretical value
of 120 X 10 4 amp/(m 2 )(K 2 ).
It follows from Eq. (2-1) that metals that have a low work function
Unforwill provide copious emission at moderately low temperatures.
tunately, however, the low-work-function metals melt in some cases and
It has

of the variation of current

boil in others, at the

emission.
*

temperatures necessary for appreciable thermionic

emitters in present-day use are pure-tungsten,

The important

Superior numbers refer to citations at the end of some chapters.

Sec. 2-3]

CHARACTERISTICS OF ELECTRON TURES

thoriated-tungsten, and oxide-coated cathodes.

The

sion constants of these emitters are contained in

Table

17

thermionic-emis2-1.

TABLE 2-1
THE IMPORTANT THERMIONIC EMITTERS AND THE
THERMIONIC-EMISSION CONSTANTS
Emitter

At, amp/(m')(K')

Tungsten

60

Thiorated-tungsten

Oxide-coated

Tungsten
tively high

is

0.01

X
X
X

Ew

e-v

10*

4.

10 4

2. 63

10 4

52

used extensively for thermionic filaments despite its relafunction.

In fact, this material is particularly impor-

work

tant because

it is

virtually the only material that can be used successfully

as the filament in high-potential tubes.

X-ray tubes,

It is

used in high-potential

and in the large powerradio and communication applications.

in high-potential rectifier tubes,

amplifier tubes that are used in

It has the disadvantage that the cathode emission efficiency, defined as

the ratio of the emission current in milliamperes to the heating power in
is small.
Despite this, it can be operated at a sufficiently high
temperature, between 2600 and 2800K, to provide an adequate emission.
It has been found that the application of a very thin layer of low-work-

watts,

function material on filaments of tungsten will materially reduce the

work function of the resulting surface. A thoriated-tungsten filament is

obtained by adding a small amount of thorium oxide to the tungsten
before it is drawn.
Such filaments, when properly activated, will yield
an efficient emitter at about 1800K. It is found desirable to carbonize
such an emitter, since the rate of evaporation of the thorium layer from
the filament is thus reduced by about a factor of 6.
Thoriated-tungsten
filaments are limited in application to tubes that operate at intermediate
potentials, say 10,000 volts or less.

Higher-potential tubes use pure-

tungsten filaments.

The oxide-coated cathode

cient as tungsten)

is very efficient (about twenty times as effiand provides a high emission current at the relatively

low temperature of 1000K.

It consists of a

metal sleeve of konal (an

and titanium) or some other metal, which is

barium and strontium. These cathodes are

alloy of nickel, cobalt, iron,

coated with the oxides of

number of reasons to use in the lower potential tubes, say

about 1,000 volts or less, although they do operate satisfactorily at higher
potentials under pulsed conditions at relatively low-duty cycle.
They
are used almost exclusively in receiving-type tubes and provide efficient
operation with long life.
limited for a

RADIO ELECTRONICS

18

Curves showing the relative cathode

[Chap. 2

efficiencies of tungsten, thoriated-

It will
tungsten, and oxide-coated cathodes are illustrated in Fig. 2-1.
be seen that tungsten has a considerably lower efficiency than either of

the other two emitters.

The thermionic emitters in their practical form in electron tubes may

be of the directly heated, or filamentary, type or of the indirectly heated
type, and in the case of gas and vapor tubes the cathode may be of the
Typical filaheat-shielded type.
22
mentary cathodes are illustrated in
These filamentary cathFig. 2-2.
1 18
odes may be of the pure-tungsten,
E

14

12

thoriated-tungsten, or oxide-coated

-\y-

type.

<U

10

The

^8
8
2
3

indirectly heated cathode for

vacuum tubes is
2-3.
The heater

use in

Fig.

4
2

60

Fig. 2-1. Cathode efficiency curves of

an oxide-coated, a thoriated-tungsten,

and a pure-tungsten

filament.

wire

is

con-

tained in a ceramic insulator which

ferrr
10
20
30
40
50
Percent rated filament power

illustrated in

enclosed by the metal sleeve on

which the oxide coating is placed. A
cathode assembly of this type has

is

such a high heat capacity that its

temperature does not change with

instantaneous variation in heater current when alternating current is used.

Heat-shielded cathodes, which can be used only in gas-filled electron
tubes for reasons to be discussed in Sec. 2-24, are designed in such a way

This mateas to reduce the radiation of heat energy from the cathode.
Several
different
types of
cathode.
rially increases the efficiency of the
heat-shielded cathodes are illustrated in Fig. 2-4.

HI

Fig. 2-2. Typical directly

heated cathodes.

2-4. Photoelectric Emission.

Fig. 2-3. Typical indirectly heated

cathodes.

The energy

that

is

required to release

an electron from a metal surface may be supplied by illuminating the

There are certain restrictions on the nature of the
surface with light.
surface and the frequency of the impinging light for such electron emisThat is, electron emission is possible only if the fresion to take place.
quency of the impinging light exceeds a certain threshold value that
depends on the work function E w of the surface according to the equation

Fig. 2-4. Different types of heat-shielded cathodes.

Electric Co. and Lowry.)

(General

corresponding threshold wavelength beyond which photoelectric emission

cannot take place is given by
ch

A is
(=3 X 10

where

10

the angstrom unit (10~ 8 cm) and

cm/sec).

to 8000 A, the

than 1.54

12,400

volts.

/n

c is

the velocity of light

For response over the entire

work function

visible region,

of the photosensitive surface

4000

must be

less

RADIO ELECTRON I CS

20

The

[Chap. 2

phototube are the photosensitive cathode

essential elements of a

surface and a collecting electrode, contained in a glass envelope that either

photograph of
is evacuated or contains an inert gas at low pressure.

such a phototube

is

shown

The number

in Fig. 2-5.

of photoelectrons

per square millimeter of area of a photo-

cathode is small, and it is customary to

photocathodes of large area, as

use

shown.
current characteristics of such

The

phototubes for different collecting potentials between the cathode and the
collecting anode, with light intensity as

a parameter, are illustrated.

shows the curves of a

2-6

Figure

vacuum

phototube with light intensity as a

parameter. Note that the current
reaches near saturation values for very
low values of applied potential.
The presence in the glass envelope
of

an

inert gas, such as

neon or argon,

at low pressure materially alters the

set of characvolt-ampere curves.

curves for a gas phototube are

Fig. 2-5. A typical phototube.
given in Fig. 2-7. The presence of the
the current
gas in a phototube increases the sensitivity of the phototube,
potential,
plate
increased
with
increasing
output for a given light intensity
phototube.
vacuum
the
in
constant
sensibly
whereas the output remains
teristic

the output from two phototubes, one of the

type and the other of the gas-filled type, other characteristics

A significant comparison
vacuum

of

20

200 eootca idles

!00

f-io

Plate potential

volts

Fig. 2-6. The volt-ampere characteristics of a type PJ-22

light intensity as a parameter.

vacuum phototube, with

contained in Fig. 2-8. Note that the

is a linear function of the illucell shows deviations from the
gas-filled
for
the
that
whereas
mination,
of the tubes being the

photocurrent for the

same,

is

vacuum phototube

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-7]

linear at the higher illuminations.

the gas-filled

cell is clearly

However, the greater

21

sensitivity of

evident.

2-5. Secondary Emission.

It is possible for a particle, either an
electron or a positive ion, to strike a metallic surface and transfer all or a
part of its kinetic energy in this collision to one or more of the internal
If the energy of the incident particle is sufficiently high, some
the internal electrons may be emitted. Several tubes have been

electrons.
of

20
">

c w

15

$ E
"5
a) P

10

s.a

a:

20

10

Fig. 2-7.

The volt-ampere

light intensity as

40
30
50
60
Plate potential, volts

80

70

characteristics of a type PJ-23 gas-filled phototube, with

a parameter.

designed which incorporate secondary-emission surfaces as part of the

device, and highly sensitive phototubes have such auxiliary elements in
them. Frequently the secondary emission that exists is of a deleterious

This matter

nature.

will

be discussed in explaining certain features of

the characteristics of tetrodes.

2-6.

High Field Emission.

The presence

of a very strong electric field

at the surface of a metal will cause electron emission.

Ordinarily the

20 r

50
75
100
125
150
Illumination, footcondles

175

200

Fig. 2-8. Photocurrent as a function of illumination for a PJ-22

and a PJ-23

vacuum phototube

gas-filled cell.

the average electron tube

is too small to induce such electron

This process has been suggested to account for the electron
emission from a mercury-pool cathode in a mercury rectifier.

field in

emission.

2-7. Ionization.

known

The process in which an atom

The atom that has lost the

as ionization.

loses

an electron

electron

is

is

called a

may occur in several ways.

Consider a free electron, which might have
been released from the envelope or from any of the electrodes within the
-positive ion.

The process

Electron Bombardment.

of ionization

RADIO ELECTRONICS

22

tube by any

[Chap. 2

of the processes discussed above.

Suppose that this free

enough energy from an applied field so that, upon
collision with a neutral atom, it removes an electron.
Following this
action, two electrons and a positive ion exist.
Since there are now two
electrons available, both may collide with gas particles and thus induce
further ionization.
Such a process as this may become cumulative, with
consequent large electron release. This process is very important and
accounts for the successful operation of gas- and vapor-filled rectifier
electron has acquired

tubes.

It is also the basis of the gas amplification in gas-filled phototubes.

If the gas is exposed to light of the proper

frequency, then this radiant energy may be absorbed by the atom, with

Photoelectric Emission.

resulting electron emission.

This process

is

important in initiating

cer-

tain discharges.
Positive-ion

Bombardment.

neutral gas particle

may

The

collision

between a positive ion and a

result in electron release, in

manner as by electron bombardment. This process

and is usually insignificant in normal gas tubes.
Thermal Emission.
electrons

If

may become

the temperature of the gas

is

much

is

high enough, some

dislodged from the gas particles.

gas temperature in electron tubes

normally unimportant.

is

generally low,

the same
very inefficient

and

However, the
this process is

THE HIGH-VACUUM DIODE

2-8. The Potential Distribution between the Electrodes.
Consider a
thermionic source situated in a vacuum. This cathode will emit electrons, most of which have very little energy when they emerge.
Those
electrons which first escape will diffuse throughout the space within the
envelope.
An equilibrium condition will soon be reached when, because
of the mutual repulsion between electrons, the free electrons in the space

prevent any additional electrons from leaving the cathode. The

equilibrium state will be reached when the space charge of the electron
cloud produces a strong enough electric field to prevent any subsequent
will

emission.

The

inclusion of a collecting plate near the thermionic cathode will

allow the collection of electrons from the space charge when this plate is

maintained at a positive potential with respect to the cathode; the higher

the potential, the higher the current. Of course, if the thermionic emission is limited, then the maximum current possible is the temperaturesaturated value.
In addition to such a simple two-element device, which is the diode,
may be interposed between the cathode and plate. If a single grid

grids
is

interposed, the tube

is

a triode.

If

two

grids are present, the tube

is

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-8]

23

Details of the characteristics

such devices will be considered in some detail in the

tetrode; three grids yield a pentode, etc.

and operation

of

following pages.

Consider a simple diode consisting of a plane cathode and a collecting

which is parallel to it. It is supposed that the cathode

plate, or anode,

can be heated to any desired temperature and that the potential between
may be set at any desired value. It is desired to
examine the potential distribution between the tube elements for various
cathode temperatures and fixed anode-cathode applied potential.

the cathode and anode

Suppose that the temperature of the cathode is high enough to allow

some electrons to be emitted. An electron space-charge cloud will be
formed in the envelope. The density of the electrons and the potential
at any point in the interelectrode
space

are

by

related

Poisson's

equation
J>

(2-5)

dx 2

where
is

is

the potential in volts, p

the magnitude of the electronic-

charge density in coulombs per cubic

meter, and o = 10~ 9 /367r is the permittivity of space.

expression

will

study of this

yield

significant

d
Fig. 2-9. The potential distribution between plane-parallel electrodes, for several values of cathode temperature.

information.
It is supposed that the electrons that are emitted from the cathode
have zero initial velocities. Under these conditions, the general character of the results will have the forms illustrated in Fig. 2-9.
At the
temperature 7\, which is too low for any emission, the potential distribution is a linear function of the distance from the cathode to the anode.
This follows from Eq. (2-5), since, for zero-charge density,

dW
dx 2
This

is

or

the equation of a straight

dV
dx

const

line.

At the higher temperature T 2 the charge density p is not zero. Clearly,

the anode-cathode potential, which is externally controlled, will be independent of the temperature, and all curves must pass through the fixed
end points. Suppose that the potential distribution is somewhat as illustrated by the curve marked T 2
All curves must be concave upward,
since Eq. (2-5), which may be interpreted as a measure of the curvature,
is positive.
A positive curvature means that the change in slope dV/dx
between two adjacent points must be positive. Moreover, the curvature
,

RADIO ELECTRONICS

24

greater for larger values of

is

tures.

[Chap. 2

corresponding to the higher temperamaximum current that can be

p,

It is possible to justify that the

drawn from the diode

for a fixed plate potential and any temperature is

obtained under the condition of zero electric field at the surface of the
cathode. Under these optimum conditions,

dV

^=
This condition

is

at x

(2-6)

valid under the assumption of zero initial velocities of

emission of the electrons.

An

2-9. Equations of Space Charge.

between the
applied between the anode
In general, the current density is a measure

current collected and the potential that

and cathode

is

of the rate at

possible.

explicit relation

is

which the electrons pass through unit area per unit time

in the direction of the field.

If v denotes the drift velocity in meters

is the electron density in electrons per cubic meter, and e
per second,
is the electronic charge in coulombs, then the current density in amperes
per square meter is

J = Nev =

pv

(2-7)

Also, neglecting the initial velocity, the velocity of the electron at any
point in the interelectrode space is related to the potential through which
it has fallen by the following expression, which is based on the conservation of energy:

y mv

= eV

By

(2-8)

combining the foregoing expressions, there results

(PV =
2
dx

JV-M
e

7 as

This is a differential equation in

given by

(2-9)

(2e/m)
a function of

x.

The

solution of

it

is

J
2.25

VM

m^

amp/m2

For electrons, and in terms of the boundary conditions

(2 - 10 >

Et at the

anode, there results

J =

2.33

10- 6

amp/m 2

(2-11)

This equation is known as the Langmuir-Childs, or three-halves-power,

It relates the current density, and so the current, with the applied
potential and the geometry of the tube.
It shows that the space-charge
current is independent of the temperature and the work function of the
law.

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-9]

Thus, no matter

cathode.

how many

electrons a cathode

25

may be

able

to supply, the geometry of the tube and the applied potential will deter-

maximum

mine the

current that can be collected

electron supply from the cathode

is

by the anode.

restricted, the current

than the value predicted by Eq. (2-11).

The

If

may

conditions are

be

the
less

somewhat

as represented graphically in Fig. 2-10.

For the case

analysis

tube that possesses cylindrical symmetry, a similar

of a

The

possible.

is

results

of such a calculation lead to the

Space charge
limited

following expression for the current:

1

76

FA
3

amp

= 14.6X10--?-%-

(2-12)

where

is

tube and

the active length of the

2

is a quantity that is
Pig. 2-10. The volt-ampere characterisdetermined from the ratio ra /rk the
tics of a typical diode.
ratio of anode to cathode radius.
For ratios ra /r* of 8 or more, /3 2 may be taken as unity.
Attention is called to the fact that the plate current depends upon the
three-halves power of the plate potential both for the plane parallel and
This is a general
also for a diode possessing cylindrical symmetry.
relationship, and it is possible to demonstrate that an expression of the
form Ib = kEb% applies for any geometry, provided only that the same
ft

2.4

Type 81
Oxide- 1.4

2.0

Type 10
Th.W-1.3

s l,6

12

TypeFP85

W-

"?0.8

/.J

Jo.4
-0.4
0.5

1.0

log

1.5

l0

2.0

2.5

3.0

^(log lo volts)

Fig. 2-11. Experimental results to verify the three-halves-power law for tubes with
oxide-coated, thoriated-tungsten, and pure-tungsten filaments.

restrictions as

imposed

in the

above developments are

true.

The

specific

value of the constant k that exists in this expression cannot be analytically

determined unless the geometry of the system is specified.

The dependence of the current on the potential for any tube may be
determined by plotting the results obtained experimentally on a logarithmic scale. Theoretically one should find, if the expression h = kEJ*

RADIO ELECTRONICS

26
is

[Chap. 2

valid, that

log 10

h=

logio k

+ y2

logic

(2-13)

The logarithmic plots for three commercial tubes are shown in Fig. 2-11.
The type 10 tube is a triode and was converted into a diode by connecting
The other tubes are diodes. It will be observed
grid and plate together.
that the logarithmic plots are straight

lines,

although the slopes

of these

than the theoretical 1.5.

The current and potential ratings of
2-10. Rating of Vacuum Diodes.
u diode, i.e., the maximum current that the tube may carry and the maximum potential difference that may be applied between anode and cathlines are all slightly less

ode, are influenced

1.

emitter.

rent

by a number

of factors.

by the cathode efficiency of the

Thus, for a given input power to the filament, a maximum cur-

limit is set to the tube current

is specified.

is a maximum temperature limit to which the glass envelope

tube may be safely allowed to rise. This is the temperature to
which the tube was raised during the outgassing process. This is about
400C for soft glass and about 600C for pyrex. For higher temperaOwing
tures, the gases adsorbed by the glass walls may be liberated.
to this limitation, glass bulbs are seldom used for vacuum tubes of more

2.

There

of the

than about 1 kw capacity.

3. A very important limitation is set by the temperature to which the
anode may rise. In addition to the fraction of the heat radiated by the
cathode that is intercepted by the anode, the anode is also heated by
the energy carried by the anode current. The instantaneous power
carried by the anode current and supplied to the anode is given by e b ib
where e b is the anode-cathode potential and % is the anode current. The
temperature to which the anode rises will depend upon the area of the
,

anode and the material of its construction.

The most common metals used for anodes are nickel and iron for
receiving tubes and tantalum, molybdenum, and graphite for transmitThe surfaces are often roughened or blackened in order to
ting tubes.
The anodes of many transmitting tubes
increase the thermal emissivity.
may be operated at a cherry-red heat without excessive gas emission.
To allow for forced cooling of the anode, cooling coils may be provided, or
the tube may be immersed in oil. The newer type of transmitting tubes
Two
are frequently provided with radiator fins for forced-air cooling.
different types of transmitting tubes are illustrated in Fig. 2-12.
4. The potential limitation of a high-vacuum diode is also dependent
on the type of its construction. If the filament and anode leads are
brought out side by side through the same glass press, some conduction
may take place between these leads through the glass. This effect is

Sec. 2-10]

CHARACTERISTICS OF ELECTRON TUBES

particularly

marked

if

the glass

is

hot,

and the

27

resulting electrolysis will

The highest potencause the glass to deteriorate and eventually to leak.

upon the spacdepends
tial permissible between adjacent leads in glass
below
kept
1,000 volts.
ing and upon the type of glass but is generally
leads
at one
filament
with
Higher-potential tubes are usually provided
with the anode at the other end.
The glass envelope must be long enough so that flashover on the outIn a diode as a rectifier, no current will
side of the tube will not occur.

end

of the glass envelope,

Fig. 2-12. Photographs of two transmitting tubes.

{RCA Mfy.

Co.)

respect to the cathflow during the time that the anode is negative with
The maximum safe rating of a rectifying diode is known as the
ode.
peak-inverse-potential rating.

Commercial vacuum diodes are made which will rectify current at high
Such units are used with X-ray equippotential, up to 200,000 volts.
and with the highment, with high-potential cable-testing equipment,
equipment for nuclear-physics research. The dimensions and
potential

capacity of the
shape of the glass envelope will depend upon the current
generally
being
tubes
tube and the type of cooling to be used, oil-cooled
smaller than air-cooled types.

RADIO ELECTRONICS

28

[Chap. 2

THE TRIODE
2-11. The Grid.
The introduction of a third element between the
cathode and plate of the diode by DeForest in 1907 was the start of
the extensive developments involving vacuum tubes. This new electrode, called the control grid, consists of a wire mesh, or screen, which
surrounds the cathode and is situated close to it. The potential applied
to the grid in such a tube is usually several volts negative relative to the
cathode, whereas the plate is usually maintained several hundred volts
positive with respect to the cathode.
Clearly, the electric field resulting
from the potential of the grid tends to maintain a large space-charge
cloud, whereas the field of the plate tends to reduce the space charge.
However, owing to its proximity to the cathode, a given potential on the
grid will exercise a greater effect on the space charge than the same potential on the plate.
This would seem to imply that a proportionality should
exist between the relative effectiveness of the grid and plate potentials
on the space charge and that the plate current might be represented
approximately by the equation
(2-14)

where e b is the plate-cathode potential, e c is the grid-cathode potential,

and the factor n is a measure of the relative grid-plate potential effectiveThe factor n is known as the amplification
ness on the tube current.
factor of the grid.

The validity of Eq. (2-14), which is simply a natural extension of the

three-halves-power space-charge equation of the diode, has been verified
experimentally for

No simple,

many triodes.

rigorous theoretical deriva-

of relatively simple
geometry. However, the value of the amplification factor p can be calculated with a fair degree of accuracy from equations that are based on

tion of this equation

is

possible,

even for a triode

electrostatic considerations.

By

maintaining the grid at some negative potential with respect to the

it will repel electrons and will, in part, neutralize the attractive
If the grid potential
field of the anode, thus reducing the anode current.
is made positive, the electron stream will increase because of the combined action of both the grid and the plate potentials. But, with a posicathode,

tive potential on the grid,

some

be attracted to
must be
designed to dissipate the grid power if the grid potential is to be maintained positive; otherwise the grid structure may be seriously damaged.
Generally the grid is maintained negative, although positive-grid triodes

it,

and a current

of the space charge will

in the grid will result.

The

for power-amplifier applications are available.

grid structure

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-12]

The

variations of the plate

and grid currents with variances

29
of grid

potential are illustrated in Fig. 2-13.

In this diagram, the plate potential is maintained constant.

For sufficiently negative grid potential,
cutoff of the plate current occurs.
As the grid potential is made less
negative, the plate current follows a smooth curve, the variation being
expressed analytically by Eq. (2-14).
As the grid potential is made positive, grid

current flows, the

mag-

nitude of this current increasing

rapidly

with

increasing

grid

potential.

For positive grid potentials,

and with the consequent grid current, Eq. (2-14) no longer represents the plate current, although
it

of

Grid volts

does give a good representation

the total space current.

Fig. 2-13. Total space, plate, and grid current in a triode, as a function of grid
potential, with fixed plate potential.

With

increasing grid potentials, the grid

current increases, and the plate current decreases.

2-12. Triode Parameters.

In view of Eq. (2-14), the dependence of

may be represented

the plate current on the plate and the grid potentials

functionally

by the expression
ib

(2-15)

f{eb,e c )

Of course the plate current also depends upon the heater temperature,
but as the heater current is usually maintained at rated value (this is

Fig. 2-14.

The

Fig. 2-15. The transfer

a triode.

plate characteristics

of a triode.

istics of

such as to provide perhaps five to ten times the normal required current),
this term usually does not enter into the functional relationship.
If Eq.
(2-15) is plotted

on a three-dimensional system

representing the function /(4,e&,e c )

is

of axes, a space

obtained.

The

diagram

projections of

these surfaces on the three coordinate planes give three families of characteristic curves.

These curves are given in Figs. 2-14 to 2-16.

RADIO ELECTRONICS

30

The curves

of Fig. 2-14 are

show the variation

known

as the plate characteristics since they

with plate potential for various

grid more negative is

of the plate current

The main

values of grid bias.

[Chap. 2

effect of

making the

to shift the curves to the right, without changing the slopes appreciably.

This is in accord with what would be expected from consideration of

Eq. (2-14).
If the grid potential is made the independent variable, the mutual, or

The effect

transfer, characteristics of Fig. 2-15 result.

potential less positive

is

of

making the plate

to shift the curves to the right, the slopes again

remaining substantially unchanged.

The simultaneous variation of both the plate and the grid potentials
so that the plate current remains constant gives rise to a third group of charThese
acteristics illustrated in Fig. 2-16.

show the

relative effects of the plate

and

grid potentials on the plate current of the

But from the discussion

tube.

2-11

2-16. The constant-current

characteristics of a triode.

Fig.

ratio ot the

m
.

change

fi

is

plate potential to

the change in grid potential for a constant plate current.

cally,

of Sec.

the amplification factor that

Consequently,
relates these two effects.
the amp lifi ca tion factor is defined as the
is

it

Mathemati-

given by the relation

(2-16)

The negative sign takes account of the fact that a decreasing grid potential must accompany an increasing plate potential if the plate current is
to remain unchanged.

Consider the variation in the plate current. This is obtained by

expanding Eq. (2-15) in a Taylor's expansion. But it is here assumed
that the variation is small and that it is adequately represented by the
first two terms of the expansion.
Subject to this limitation, the expression has the form
Aib

(^)

Ae b

+ (p)

Ae<

(2-17)

This expression indicates simply that changes both in the plate potential
Ae b and in the grid potential Ae c will cause changes in the plate current.
The quantity {de b /dib )s expresses the ratio of an increment of plate
potential to the corresponding increment of plate current, for constant
E c This ratio has the units of resistance, is known as the plate resistance
of the tube, and is designated by the symbol r p
Clearly, r p is the inverse
c

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-12]

31

and has been so indicated

slope of the plate characteristics of Fig. 2-14

there.

The quantity

which gives the ratio of an increment of plate

current to the corresponding increment of grid potential for constant
(di b /de c ) Ei

Eb

has units of conductance. It is known as the plategrid transconductance, or mutual conductance, and is designated by the
symbol g m
The mutual conductance gm is the slope of the mutual-, or
plate potential

transfer-, characteristic curves of Fig. 2-15.

To summarize,

the triode coefficients have the forms

plate resistance

mutual conductance

\deJ Ei

(2-18)

amplification factor

easy to show that n

It is

related to r p

is

ft

This

obtained by setting Ai b

is

=
=

and gm by the expression

(2-19)

r Pffm

in Eq.

(2-17)

and then using the

definitions of Eq. (2-18).

The

variations of these parameters for a fixed value of plate potential

6C5 tube

for the

are

shown

in Fig.

It is noticed that the plate

2-17.

resistance varies over rather wide

limits,

being very high at zero plate

current,

22

>20

30

3000

'

18

.micror

and approaches a constant

of

value at the higher plate currents.

20

2000 8

_o
~u

The transconductance varies from a

"1

very small value at zero plate current and tends toward a constant

1000

o
c

1/1

value at the higher plate currents.

The

amplification

factor

remains

reasonably constant over a wide

range of currents, although it falls
off

rapidly

at

the low currents.

The corresponding values

values of

E may
b

20

10

Plate current,

Fig. 2-17.
of a

6C5

The parameters

ma

m, r v,

and g

triode as a function of plate

current.

for other

differ numerically,

but the general variations

will

be

similar.

High-power triodes are used extensively in transmitters. The grid of

is driven positive with respect to the cathode during part
of the cycle, and the current is cut off during part of the cycle.
The
characteristics of importance of such tubes are the plate curves and the
such a tube

RADIO ELECTRONICS
AVERAGE
::

Ef =

ii

PLATE

[Chap. 2

CHARACTERISTICS

volts a.c.

o
o
o

RADIO ELECTRONICS
constant-current curves.

The

are as illustrated in Figs. 2-18

IChap. 2

variations over normal operating limits

and 2-19

for a type

889A

tube.

MULTIELECTRODE TUBES
2-13. Tetrodes.
In the tetrode a fourth electrode is interposed
between the grid and the plate. This new electrode is known as the
screen grid, or grid 2, in order to distinguish it from the "control" grid
of the triode.
Physically, it almost entirely encloses the plate.
Because
of its design and disposition, the screen grid affords very complete elecThis shieldtrostatic shielding between the plate and the control grid.
ing is such that the grid-plate capacitance is reduced by a factor of about
However, the screen mesh does not interfere appreciably
1,000 or more.
with the electron flow. The reduction of the grid-plate capacitance is a
very important improvement over the triode, and this matter will be
considered in some detail in Chap. 5.
Because of the electrostatic shielding of the plate by the screen, the
potential of the plate has almost no effect in producing an electric field
Since the total space current is determined almost
at the cathode.
wholly by the field near the cathode surface, the plate exerts little or no
There is,
effect on the total space charge drawn from the cathode.
therefore, a significant difference between the triode and the tetrode.
In
a triode, the plate performs two distinct functions, that of controlling the
In a tettotal space current, and that of collecting the plate current.
rode, the plate serves only to collect those electrons which have passed
through the screen.
The passive character of the plate makes the tetrode a much better
potential amplifier than the triode.
This follows from the fact that in the
triode with a resistance load an increase in load current is accompanied
by a decreased plate-cathode potential, which results in a decreased space

In the tetrode, the decreased plate-cathode potential still exists,

but owing to the secondary role of the plate the space current is not matecurrent.

rially affected.

The

and the control grid is nearly the same

and the triode, and therefore the grid-plate transnearly the same in both tubes. Also, the plate resistance

disposition of the cathode

in both the tetrode

conductance

is

than that of the triode. This follows from the fact that the plate potential has very little effect on the
plate current.
Thus, with the high plate resistance and with a gm that is
about the same as for the triode, the tetrode amplification factor is very
of the tetrode is considerably higher

high.

2-14. Tetrode Characteristics.

and screen-grid

In the tetrode with fixed control-grid

potentials, the total space current

is

practically constant.

Sec. 2-15]

CHARACTERISTICS OF ELECTRON TUBES

35

Hence, that portion of the space current which is not collected by the
must be collected by the screen; where the plate current is large,
The general character
the screen current must be small, and vice versa.
plate

of the results is illustrated in Fig. 2-20.

Although the plate potential does not affect the total space current to a
very great extent (although a slight effect is noted in the curve at the
lower plate potentials) it does determine the division of the space current
,

between plate and screen. At zero

plate potential, few of the electrons
have sufficient energy to reach the
anode, and the plate current should
be small.

As the plate potential

is

increased, a rapid rise occurs in the

plate current, with a corresponding

reduction
TTT ,

When

the screen

of
,

the plate potential

current.

is

larger

^c 2 ]^

-^g

b
Fig. 2-20. Curves of total space current,
plate current) and gcreen current in a
tetrode,

than the screen potential, the plate

collects almost the entire space current and the screen current approaches zero or a very small value.
An inspection of the curves of Fig. 2-20 shows that the plate current
rises very rapidly with increasing plate potential, but this increase is fol-

lowed by a region of plate-potential variation in which the plate current

This region is one of negative
decreases with increasing plate potential.
plate resistance, since an increasing plate potential is accompanied by a
The kinks, or folds, in the curves are caused
decreasing plate current.
by the emission of electrons from the plate by the process of secondary
This results from the impact of the primary electrons with the
emission.
That is, secondary electrons will be released from the anode, and
plate.
if this is the electrode with the highest positive potential, the electrons
If, howwill be collected by the anode, without any noticeable effect.
ever, secondary electrons are liberated from the anode, and if these electrons are collected by some other electrode, then the anode current will
decrease, whereas the current to the collecting electrode will increase.
It is this latter situation which exists in the tetrode when the plate potential is low and the screen is at a high potential.
When the plate potential is higher than the screen potential, the secondary electrons from the plate are drawn back, without appreciable
If under these potential conditions secondary electrons are libeffect.
erated from the screen, these will be collected by the anode.
The corresponding plate current will be greater than that in the absence of secondary emission from the screen.
2-15. Transfer Characteristics.
Since the plate of a tetrode has no
appreciable influence on the space current, it is expected that the cath-

RADIO ELECTRONICS

36

[Chap. 2

ode, the control grid, and the screen grid should possess characteristics
not unlike those of a triode. This is actually the case, as illustrated in
Fig. 2-21.
These curves show the effect of variations of plate potential
Because of the slight influence of the
on plate current, for fixed E c2
These curves should
plate, the transfer curves are bunched together.
-

be compared with those of the triode in

Fig. 2-15, where the transfer curves are
widely separated.

400
J?c2 'fixed

The

transfer curves

for plate potentials

ential,

and

which

is

In
Fig. 2-21.

The

E =
b

transfer character-

for

istics of a tetrode, for a fixed

screen potential, and with the
plate potential as a parameter.

Tube Parameters.

2-16.

in

practice.

characteristic for

100 volts actually falls below that

50 volts. This anomalous behav-

E =
b

emission effects discussed above.

It

expected, on the basis of the foregoing

is

may

potential of the various electrodes

4 =
e cl is

transfer

ior is directly the result of the secondary-

discussion, that the plate current

where

this is the region of operation

generally avoided

the

fact,

become separated

below the screen pot-

the potential of the

(2-20)

/(e6Ai,e c2 )
or control, grid, e c2

first,

the second, or screen, grid, and

be expressed as a function of the

of the form

by an expression

e b is

is

the potential of

the potential of the plate,

all

with

This functional relationship is just a natural

extension of that which applies for triodes. In fact, an approximate
This form, which is an
explicit form of the dependence is possible.
respect to the cathode.

extension of Eq. (2-14),

may be
ib

written as

7
(
k
I e cl H

ah

and

_L

(2-21)

M2/

Mi

where

e&

m are the control-grid and screen-grid amplification factors,

respectively.

The

variation in the plate current, second- and higher-order terms in

the Taylor expansion being neglected,

Ai-(^)
\de /E
b

*eb
a ,E a

+ (f)
\oe /
cl

Generally, the screen potential

priate value,

and hence Ae c2 =

is

0.

is

given by

Ae cl
b

,E ct

+ (^)
\oe t/E
c

b ,E

(2-22)

maintained constant at some approThe third term in the expansion may

be omitted under these conditions. The partial-differential coefficients

appearing in this expression furnish the basis for the definitions of the

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-17]

tube parameters.

These are

(^iLj

\de c i/ Et,E c2

37

rp

plate resistance

gm

mutual conductance

fi

(2-23)

amplification factor

\de cl /i b ,E M

The two

subscripts associated with each term indicate the parameters

that are maintained constant during the partial differentiation.

It

can

be shown that here too the relation n = r pg m is valid. Nominal values

5
for the various parameters that appear in this relationship are r p = 10
6
to 2 X 10 ohms, g m = 500 to 3,000 M mhos, and M = 100 to 1,200.
2-17. Pentodes.
Although the insertion of the screen grid between the
control grid and the anode in a triode serves to isolate the plate circuit
from the grid circuit, the range of
operation of the tube is limited owing
to the effects of secondary emission.

This limitation results from the fact

if the plate-potential swing is

that,

made too

large,

plate potential

instantaneous
extend into the

the

may

~e

'

The characteristics of a tube

region of rapidly falling plate current,

Fig. 2-22.

with a resulting marked distortion in

when connected

as a tetrode

and

as a

pentode.

the output.

The

kinks, or folds, that appear in the plate-characteristic curves

that limit the range of operation of the tetrode

may

and

be removed by

between the screen grid and

Tubes that are provided with this extra grid are
known as pentodes. The suppressor grid must be maintained at a lower
potential than the instantaneous potential reached by the plate at any

inserting a coarse suppressor-grid structure

the plate of the tetrode.

time in its potential excursions. Usually the suppressor is connected to

the cathode, either externally or internally. Now since both the screen
and the anode are positive with respect to the suppressor grid, secondary
electrons from either electrode will be returned to the emitting electrode.

The main

electron stream will not be materially affected

of the suppressor grid.

The

by the presence

effects of the insertion of the suppressor grid

shown graphically in Fig. 2-22.

The pentode has displaced the tetrode in radio-frequency (r-f) potential amplifiers, because it permits a somewhat higher potential amplificaare

tion at moderate values of plate potential.

Likewise

plate-potential excursion without distortion.

it

permits a greater

Power tetrodes are used

RADIO ELECTRONICS

38

[Chap. 2

extensively in high-power tuned amplifiers for reasons to be discussed in

Chap.

10.

The

transfer curves of a pentode are shown in Fig. 2-23.

It is noted
that the curves are almost independent of the plate potential.
2-18. Remote-cutoff Tubes.
If in a pentode the grid-cathode spacing,
the spacing between grid wires, or the diameter of the grid wires is not

uniform along the entire length

of the control-grid structure, the various

portions of the grid will possess different degrees of electrostatic control
over the plate current. That is, one portion of the grid may cause elec-

tron-flow cutoff, whereas an appreciable current might pass through a

more widely spaced section of the grid. As a result, the plate-current

by the grid is considerably less effective than in a conventional

The general character of the results is illustrated in Fig. 2-24.

control

pentode.

Fig. 2-23. The transfer curves

of a pentode for fixed screen
potential and with the plate
potential as a parameter.

Owing

to

its

Fig. 2-24. The transfer curves of a

6SJ7 sharp-cutoff pentode and a
6SK7 remote-cutoff pentode,

construction, a given grid-potential increment results in a

plate-current change that

mutual conductance

is

is a function of the bias.

This means that the
a function of the bias. For this reason, these

They are also known as remote-cutThey have applications in radio receivers and
frequency-modulation (f-m) transmitters. Some appli-

tubes are called variable-mu tubes.

off

and supercontrol tubes.

may be

used in

cations will be considered in later chapters.

A number of special-purpose tubes conmore grid elements than the pentode are used extensively. These
tubes possess a wide variety of characteristics, depending upon the grids
to which fixed potentials are applied and those to which signals might be
2-19. Hexodes, Heptodes.

taining

These tubes are used extensively as converters in superheteroand find f-m transmitter and other applications. More
will be said about these applications at a later point in the text.
2-20. Beam Power Tubes.
The suppressor grid is introduced into the
pentode in order to extend the range of operation of these tubes beyond
that of the tetrode. These tubes are quite satisfactory over wide limits,
and the range of operation is limited when the instantaneous plate potenapplied.

dyne

receivers

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-20]

39

to the rapidly falling plate-current region at low potentials.

This rapid change in plate current for small changes in plate potential
in the region of low plate voltage results from the overeffectiveness of the
tial falls

suppressor grid at these low plate

potentials.

Beam power tube

shape of the
suppressor grid in some modern pentodes has been so dimensioned that
the effects of secondary emission are
just suppressed or only admitted

Because

of this, the

gr pentode

at the low anode potenFig. 2-25. The plate characteristics of a

This results in an improved
beam power tube and a power pentode.
plate characteristic and is manifested
by a sharper break in the plate characteristic.
The pentode and tetrode beam power tubes were designed with these

slightly
tials.

considerations specifically in mind,

trated in Fig. 2-25.

It will

and a plate

characteristic

is illus-

be noted that the plate current remains inde-

BEAMCONFINING
ELECTRODE

CATHODE

hbbsbhiSsSwShhBBBIIBBSSBHB
Fig. 2-26. Schematic view of the shapes
power tube. (RCA Mfg. Co.)

and arrangements

of the electrodes in a

beam

pendent of plate potential to a lower relative value in the beam power

tube than in the power pentode. The essential features of the beam
power tube are illustrated in the schematic view of Fig. 2-26. One feais
ture of the design of this tube is that each spiral turn of the screen

RADIO ELECTRONICS

40

[Chap. 2

aligned with a spiral turn of the control grid.

This serves to keep
the screen current small and hence leaves the plate current virtually
unchanged. Other features are the flattened cathode, the beam-forming side plates (maintained at zero potential), the shape of the plate, the

curvature of the grids, and the spacing of the various elements.

result of these design characteristics, the electrons flow

As a

between the grid

wires toward the plate in sheets, or beams.

The region between the screen and the plate possesses features which
somewhat analogous to those which exist in the space-charge-limited

are

That

diode.

is,

there

ever, the electrons,

is

a flow of charge between two electrodes.

when they

How-

enter this region do so with an appreciable

For such a case as this, the considerations of Sec. 2-9 would

have to be modified to take account of the initial velocity. If this is
done, it is found that a potential minimum will exist in the region between
the two electrodes. This potential minimum acts as a virtual suppressor
grid, and any secondary electrons that are emitted from either the plate
velocity.

or the screen are returned to the emitting electrode.

The

actual potential distribution in the screen-plate region will depend

on the instantaneous plate potential and the plate current,

screen potential.

The

for a constant

resulting variable suppressor action proves to be

superior to that possible with a mechanical grid structure, as illustrated.

GAS TUBES
2-21. Electrical Discharge in Gases.

There are two important

of discharge in gases that play roles in electron tubes.

the glow discharge, and the second

classes

One of these is
The glow dis-

is the arc discharge.

charge utilizes a cold cathode and is characterized by a fairly high tube
drop and a low current-carrying capacity. The potential drop across the

tube over the operating range

The

is fairly

constant and independent of the

by a low potential drop and

For an arc tube with a thermionic cathode, the
temperature-limited cathode emission may be drawn with a tube drop
approximately equal to the ionization potential of the gas. For a mercurrent.

arc discharge

is

characterized

a high current capacity.

cury-pool cathode, extremely high current densities exist (of the order of

amp/m

), with high total currents possible and a tube drop

approximately equal to the ionization potential of the mercury atom.
Consider a gas tube which consists of a cold cathode and a collecting
anode, between which is connected a source of potential through a curThe volt-ampere
rent-limiting resistance, and an indicating ammeter.
This
characteristic of such a tube has the form illustrated in Fig. 2-27.
curve shows that breakdown occurs at a potential which is somewhat
higher than the maintaining potential but that there is a region where

10 8

Sec. 2-21]

CHARACTERISTICS OF ELECTRON TUBES

41

the tube drop remains substantially constant over an appreciable range

of currents.
Visually, the discharge is characterized by a colored lumi-

nous region, the color being a function of the gas present in the tube.
It is desired to explain the mechanism of operation of these tubes.
Consider, therefore, that a free electron exists within the tube; such an
electron might have been released by ionization due to collision between

a gas molecule and a cosmic ray or by photoelectric emission.

With the
application of the potential between the electrodes, the electron will drift
toward the anode. If the field is large enough, the electron may acquire
it collides with it.
Now two
and also the electron that has
ionization, and a positive ion.
The two

enough energy to ionize a molecule when

electrons will be present, the original one

been liberated by the process of

and the positive ion will move in the applied field, the electrons
moving toward the anode, and the positive ion toward the cathode. If

electrons

the

enough, the resulting cumulative ionization

Once break-

field is large

may

continue

until breakdown occurs.

down

occurs, the potential distribution

within the tube

is

6f

markedly modified,

'

free,

containing as

many positive as neg-

This

ative charges.

is

Maintaining
potential

and most of the region of the discharge

becomes virtually equipotential or force-

Breakdown

the plasma of the

potent/a/

Almost the entire potential

change occurs in the very narrow region
discharge.

near the cathode.

Normal values

for

F^g. 2-27. Volt-ampere characteristic

glow discharge.

or a

range between
potassium surface and helium gas) and 350 volts. The
presence of a low-work-function coating on the cathode will result in a
low cathode fall with any gas. Also, the use of one of the inert gases
cathode-fall

potential

about 59 volts

(a

(helium, neon, argon, etc.) results in a low cathode

fall

with any cathode

The cathode fall adjusts itself to such a value that each positive ion, when it falls through this field, will release an electron from the
cathode by secondary emission. The positive ion combines with this
material.

and thus becomes neutralized.

Another feature of a normal glow discharge is that the current density
at the cathode remains sensibly constant.
For higher currents, a greater
portion of the cathode is covered with glow, the area of the glow on the
cathode increasing directly with the magnitude of the current. Once
the cathode is completely covered with glow, any further current through
the tube depends on an excess of secondary emission from the cathode
over that required to neutralize the positive ions. This is accompanied
by a rising cathode fall. This is the "abnormal" glow and is generally
electron

of small practical importance.

RADIO ELECTRONICS

42

The
tinct.

[Chap. 2

dividing line between an arc and a glow discharge

The

is

rather indis-

arc discharge allows for the passage of large currents at low

potential, the current density at the cathode being high.

Nevertheless
each discharge has associated with it the cathode fall, the plasma, and
the anode fall (which is of minor significance in both types of discharge)

The

discharges differ in respect to the mechanism

by which the

electrons

are supplied from the cathode.

In the glow discharge, as discussed, the

electrons are emitted from the cathode by the process of secondary

emission resulting from positive-ion

bombardment

In

of the cathode.

the arc discharge, the emission of the electrons from the cathode occurs

through the operation

of

a supplementary mechanism other than by

In the thermionic arc, the electrons are

bombardment.

positive-ion

by a cathode that is heated to a high temperature, either by the

by means of an auxiliary heating circuit. The
mechanism for electron release is not fully understood in the arcs that
employ a mercury-pool cathode or an arc between metal surfaces. Howsupplied

discharge or externally

ever, in these discharges the

primary function

of the gas is to supply a

sufficient positive-ion density to neutralize the electron space charge.

Because of this, the normal potential drop across an arc tube will be of
the order of the ionization potential of the gas.

-Cathode
*~

Starting probe

-Anode

c\

T~n

j.

Fig. 2-28. Electrode structube.

ture in a

VR

2-22. Glow Tube.

A glow tube is a coldcathode gas-discharge tube which operates in
the normal glow-discharge region. The poten^ial drop across the tube over the operating
range is fairly constant and independent of the
current.
When the tube is connected in a circuit, a current-limiting resistance must be used

u serious damage to the tube is to be

'

avoided.

One commercial type

of

tube consists of a central anode wire which

coaxial with a cylindrical cathode, as illustrated in Fig. 2-28.

The

is

elec-

trodes are of nickel, the inner surface of the cathode being oxide-coated.

The cathode fall is sometimes lowered by sputtering some misch metal (an
alloy of cerium, lanthanum, and didymium) on the cathode.
The gases
that are commonly used are neon, argon, and helium. The tubes containing neon or helium usually contain a small amount of argon.
The
presence of the argon lowers the starting potential.
available with normal output potentials of 75, 90, 105,
bear the designations OA3/VR-75, OB3/VR-90, etc.

mum

30 ma.

The

These tubes are

and 150 volts and
The normal maxi-

is

attached to the cath-

ode, as illustrated in Fig. 2-28, serves to lower the

breakdown potential

of

current

is

starting probe that

the tube.

Glow lamps

are also available for pilot, marker, and test-lamp service.

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-24]

Such tubes are available in several sizes from

Photographs of a number of these are given in

^5

43

to 3 watts capacity.

Fig. 2-29.

A cold-cathode triode, or grid-glow tube,

2-23. Cold-cathode Triodes.
contains three elements, the cathode, the anode, and a starter, or control,
The spacanode. The control electrode is placed close to the cathode.

Fiq. 2-29. Photographs of several low-capacity glow tubes.

ing of the electrodes

is

(General Electric Co.)

such that a discharge takes place from the cathode

to the control electrode at a lower potential than is required for a

Once the control gap has been
discharge from the cathode to the anode.

broken down, however, it is possible for the discharge to transfer to the

main anode. The cathode-anode potential that is required for this transfer to occur is a function of the transfer
400

current, the current in the control-elec-

trode-cathode circuit.

Such a "trans1300

fer," or "transition," characteristic is

given in Fig. 2-30.

For

zero

transfer

current,

means that the control electrode

required anode-cathode potential

as

increased.

the

transfer

100

<

falls

current

An increased transfer

200

not

is

connected in the circuit, the anode

potential is equal to the breakdown
potential between cathode and anode.
It is observed from the curve that the
rapidly

which

"

25

50 75 100 125 150

Transfer current,

175

200 225

\i.a

Fig. 2-30. Transfer characteristic of

an

RCA OA4G

cold-cathode triode.

is

current indicates the presence of greater

Regardless of the magnitude of the transfer current, however,

the anode-cathode potential can never fall below the maintaining potential
The transfer characteristic approaches this sustaining
for this gap.

ionization.

potential asymptotically.

These tubes are thermionic

an inert gas at low pressure or in which

2-24. Hot-cathode Gas-filled Diodes.

cathode diodes in which there

is

RADIO ELECTRONICS

44

[Chap. 2

mercury vapor is added. In the latter case a few drops of mercury are
added to the tube after evacuation. The pressure in the tube is then a
function of the mercury-vapor condensation temperature. The relationship between the pressure and
the temperature

is

shown

in Fig.

normal operating
conditions, the temperature of the
tube will be 15 to 20C above

2-31.
0.01

Under

that of the surroundings (ambient

temperature).

0.001

As already discussed, the sole

function of the gas in these tubes is

0.0001

20

40

60

80

100

Temperature of condensed mercury,C

Fig. 2-31. Mercury-vapor pressure as a

function of condensation temperature.

to provide ions for the neutralization

of space charge, thus permitting the

current to be obtained at much lower

potentials

than are necessary in

vacuum tubes. If more than saturation current is demanded by the

circuit, then gas amplification, resulting from positive-ion bombardment

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-26]

45

Under these circumstances the cathode fall

The tube drop should not be permitted to exceed the dis-

of the cathode, will occur.

increases.

integration potential of the cathode (about 22 volts for a mercury diode

with either oxide-coated or thoriated-tungsten cathodes); otherwise tht
cathode may be seriously damaged by the positive-ion bombardment.

Two

typical commercial mercury-vapor-filled diodes are illustrated in

Fig. 2-32.

2-25. High-pressure Gas Diodes.

Diodes are available which contain
argon or a mixture of argon and mercury at a pressure of about 5 cm.
The cathodes in such tubes consist of a short, heavy thoriated-tungsten
or oxide-coated filament and are located close to heavy graphite anodes.
These tubes, which are known as tungar or rectigon tubes, are used extensively in chargers for storage batteries.

The presence of the fairly high-pressure gas serves a twofold purpose.

One is to provide the positive ions for reducing the space charge. The
second

is

to prevent the evaporation of the thorium or the coating from

This second factor is extremely important since the filaoperated at higher than normal temperature in order to provide
the large currents from such a simple cathode structure. The high-presthe filament.

ment

is

sure gas in such a tube imposes a limitation on these tubes,

and they are

limited to low-potential operation.

2-26.

The thyratron

The Thyratron.

is

a three-

Anode

electrode tube which comprises the cathode, the

anode, and a massive grid structure between them.

The

grid structure

is

almost complete electrostatic shielding between the

cathode and the anode. In such a tube as this, the
initiation of the arc is controlled by controlling the
potential of the grid.

-Grid

so designed as to provide

The

- -Cathode

grid usually consists of

a cylindrical structure which surrounds both the

anode and the cathode, a baffle or a series of baffles
containing small holes being inserted between the

Fig. 2-33. The elecanode and the cathode. The electrode structure of
trode structure of a
negative-control thysuch a tube is illustrated in Fig. 2-33. The shielding
ratron.
by the grid is so complete that the application of a
small grid potential before conduction is started is adequate to overcome
the field at the cathode resulting from the application of a large anode

potential.

Once the

arc has been initiated, the grid loses complete control over
Grid control is reestablished only when the anode potential is
reduced to a value less than that necessary to maintain the arc. Once
the arc has been extinguished by lowering the plate potential, the grid
once more becomes the controlling factor which determines when conduc-

the arc.

RADIO ELECTRONICS

46

[Chap. 2

tion will again be initiated.

That is, if the grid potential is more positive
than that necessary for the controlling action to prevail, conduction will
take place; if more negative, no conduction will occur. The curve that
relates the grid ignition potential with the potential of the anode for
conduction just to begin is known as the critical grid curve. In fact, a
knowledge of this static curve is all that is required to determine completely the behavior of a thyratron in a circuit.
Typical starting-characteristic curves of mercury-vapor thyratrons are

given in Fig. 2-34. Two distinct types of characteristics are illustrated,

viz., those in which the grid potential must always be positive, and those

1000

600

1c

400

o 200

o
-10

-8

-6
-4 -2
2
4
6
8
D-c grid potential at start of discharge

Fig. 2-34. Critical grid characteristics of a positive-

10

12

and a negative-control thyratron

for different temperatures.

in

which the grid

tials.

trol

The

tubes

is

generally negative, except for very low plate poten-

physical distinction between these positive- and negative-con-

lies

essentially in the

more complete

shielding

by the

grid in

positive control tubes.

In the negative-control tube, where the shielding is far less complete

than in the positive-control type, the effect of the plate potential is clearly

more negative must the grid

For low
potentials must be applied before ioniza-

seen; the higher the plate potential, the

potential be in order to prevent conduction from taking place.

plate potentials, positive grid

and hence conduction, can begin. If the plate potential is reduced

more, even below the potential necessary for ionization, breakdown
can still be obtained by making the grid sufficiently positive. Now, howtion,

still

ever, the function of the tube

place between the cathode

plate.

The thyratron

will

may

and the

be destroyed, since the arc may take

grid, with very little current to the

be converted into a gas diode under these condummy electrode, the cylindrical grid now

ditions, the plate acting as a

It is because of this that a high current-limiting

connected in the grid circuit, as it is unwise to draw a large

serving as the anode.

resistance

is

grid current.

CHARACTERISTICS OF ELECTRON TUBES

Sec. 2-28]

47

In addition to the mercury-vapor- and gas-filled thyratrons of moderate current capacity, small argon-filled low-current-capacity tubes are

The

available.

shielding between the cathode

and the anode

is

not so

complete in these tubes as in the higher-current units. Also, the critical

grid curve is independent of temperature, since the number of gas
molecules in the glass envelope remains
constant
A typical critical grid curve
.

for

an 884

is

given in Fig. 2-35.

Betube occurs, the

current to the grid of a thyratron such
as the FG-27A is a few tenths of a mi2-27. Shield-grid Thyratrons.

fore

breakdown

of the

Although

croampere.

entirely negligible for

this current is

many

applica-

cause trouble in circuits

that require very high grid impedThis is especially true in cirances.
tions, it will

cuits that

employ phototubes.

40 -36 -32 -28 -24 -20 -16

-8

-12

Grid Potential
Fig. 2-35. Critical grid characteristic
of an 884 argon-filled thyratron.

For

a fourth electrode, or shield grid, has been added to the

thyratron.
Such a shield-grid thyratron structure is illustrated in Fig.
2-36.
The massive cylindrical shield-grid structure encloses the cathode,
this reason,

control grid,

and anode.

Owing

reduced to a small fraction of

being of the order of 10 -3 n&.

to the shielding, the grid current

its original

is

value, the preignition current

Shield grid potential

-4
+5
-2 -3

Shield

CONDENSED
He TEMP

grid

40C

Control^

grid

Cafhode
-12

Fig. 2-36. Electrode structure of the

FG-98 shield-grid thyratron.

The

Fig.

-8 -4
4 8 12
Control grid voltage

2-37.

FG-98

16

20

Control characteristics of an

shield-grid thyratron.

such a tube are shown in

be observed that these characteristics are functions of

critical grid starting characteristics of

Fig. 2-37.

It will

the shield-grid potential.

2-28.

which

Ignitron.
The ignitron is a mercury-pool-cathode diode
provided with a third electrode for initiating the discharge

The

is

RADIO ELECTRONICS

48

between the cathode and anode.

The

[Chap. 2

third electrode, or igniter rod,

is

boron carbide, and carborundum) which projects into the mercury-pool cathode.
Such a tube is illustrated in Fig. 2-38.
With an a-c potential applied between the cathode and the anode of
the pool-cathode diode, the arc would be extinguished once each alternate

made

of a suitable refractory material (such as silicon carbide,

Fernico metal alloy^

and pyrex type
glass seal

p^o

Flow-directing-

vanes

Deionization
baffle

Splash - hood -baffle

Auxiliary

anode
Water
connection

Tube support andcathode connection

Fig. 2-38. Section view of a water-cooled metal ignitron.

{General Electric Co.)

The

half cycle, provided that the arc could be initiated regularly.

appli-

cation of a potential to the igniter rod at the appropriate point in the

cycle will permit the regular ignition of the arc.

There

is

a fundamental difference between the control action in a

thyratron and that of the igniter rod in an ignitron. In thyratrons, the

grid prevents the formation of an arc, whereas the igniter initiates the
arc.

In the former case the electrons already exist in the tube, owing

to the presence of an externally heated cathode, but the grid electrostatically prevents the electrons

potential

is

reached.

from flowing to the anode until a critical

is in a nonconducting

In the ignitron, the tube

state until the igniter circuit

is

energized,

when conduction

is

forced.

CHARACTERISTICS OF ELECTRON TURES

Sec. 2-29]

Tube

Current,

49

Temperature. Gas- and

than rms current ratings.
This rating specifies the maximum current that the tube may carry continuously without excessive heating of any of the parts.
The time over
which the average is to be taken is also specified by the manufacturer.
That the average current is important in such a tube follows from the
fact that the instantaneous power to the plate of the tube is given by the
product of the instantaneous anode current and the instantaneous tube
Since the potential is substantially constant and independent
potential.
of the tube current, the average power is the product of the tube drop
and the average tube current. The tubes are also given peak-current
2-29.

Ratings

Voltage,

vapor-filled tubes are given average rather

10,000

9000

8000

&7000
c

^6000
-*S000
o

^4000

^60 3000
40

2000

T5

u 20

s>

1000

100
120
140
60
80
40
Mercury condensation temperature, "C.

20

Fig. 2-39.

Peak inverse potential and tube drop

of

160

an 866 diode as a function

of

temperature.

ratings, these ratings specifying the

maximum

current that the tubes

should be permitted to reach in each conducting cycle.

Such tubes are also given peak-inverse-potential ratings.
largest safe instantaneous negative potential that

may be

This

is

the

applied to the

tube without the possibility of conduction in the inverse direction arising

because of breakdown of the gas in the tube. This potential is also
referred to as the flash-back potential.

potential with temperature for

The maximum peak forward

only for thyratrons.

The

variation of the inverse peak

an 866 diode
potential

It specifies

is

is

shown

in Fig. 2-39.

a quantity that

is

significant

the largest positive potential that

may

be applied to the anode before the grid loses its arc-initiating ability.
That is, for potentials higher than this, a glow discharge may occur

between anode and grid, which will immediately initiate the cathodeanode arc.
The condensed-mercury temperature limits are specified for the safe
and efficient operation of mercury-vapor tubes. The range usually

RADIO ELECTRONICS

50

[Chap. 2

extends from about 30 to 80C. The upper temperature limit is determined by the allowable peak inverse potential. The lower limit is set
by the allowable tube drop, which increases with decreasing temperature and which

may

cause serious cathode disintegration, as well as a

decreased efficiency.
2-30. Deionization and Ionization Times.
The ionization time of a
tube specifies the time required for conduction to be established once the
potentials have been applied.
It seldom exceeds 10 ^sec and is approximately 0.01 Aisec for the 884 thyratron.
The deionization time is a measure of the minimum time that is
required after removal of the anode potential before the grid of a thyratron again regains control.
It represents the time that is required for the
positive ions to diffuse away from the grid and recombine with electrons
to form neutral molecules.
The deionization time depends on many
factors, such as gas pressure, electrode spacing, and exposed areas.
For
commercial tubes that are operated under rated conditions, it varies
between 100 and 1,000 usee. This is considerably longer than the ionization time and may offer a serious limitation to the use of such tubes in

many

applications.

REFERENCE
1.

For more
Millman,

details, see:
J.,

pany, Inc.,

and

New

S. Seely,

"Electronics," 2d ed., McGraw-Hill

Book Com-

York, 1951.

PROBLEMS
A

tungsten filament, 0.0085 in. diameter, 3 l }{ 6 in. long, is operated at

What is the temperature-limited current? If the temperature is
increased by 50K, by what percentage does the emission current increase?
2-2. The filament of an FP-400 tungsten-filament tube is 1.25 in. long and
0.005 in. in diameter.
If the total emission current is 30 ma, at what temperature
is the filament operating?
2-3. A simple inverted-V oxide-coated cathode is made of tungsten ribbon
0.125 by 0.020 in. and is 1.4 in. long. It is maintained at a temperature of
1100K. What is the thermionic-emission current?
2-4. An oxide-coated emitter is operating at 1100K.
Calculate the relative
thermionic-emission current if & has the value 12,000; the value 11,000.
2-5. At what temperature will a thoriated-tungsten filament give as much
current as a tungsten filament of the same dimensions which is maintained at
2-1.

2650K.

2650K?
2-6. At what temperature

will an oxide-coated cathode give the same emission

as a thoriated-tungsten filament of the same physical dimensions which is maintained at 1750K?

Monochromatic light of wavelength 5893

Cesium, with a work function 1.8 volts.

2-7.
a.

falls

on the following surfaces:

CHARACTERISTICS OF ELECTRON TUBES

51

Platinum, with a work function 5.3 volts.

emission possible in both cases? Explain.
2-8. A PJ-22 vacuum photocell is to be used to sound an alarm when the light
What
at a given region of a room falls below 40 ft-c or increases above 120 ft-c.
A collecting potential of 45 volts is used.
are the corresponding photocurrents?
2-9. Plot it, vs. e b of the 6H6 diode (see Appendix B) on log paper.
From this
plot, determine the quantities k and n in the expression ib = ke b
2-10. The anode current in a type 5U4G diode with 54 volts applied between
What is the required potential for a current of
the plate and cathode is 200 ma.
100 ma? The tube operates under space-charge conditions.
2-11. Suppose that the FP-400 tube is operating under rated filament power
The operating temperature is 2700K, and the anode
input (see Prob. 2-2).
diameter is 0.50 in.
a. Calculate the saturation current.
b. At what potential will the current become temperature-saturated?
2-12. Plot it vs. e b
ne c on log paper of the 6J5 triode (see Appendix B).
n
From this curve, find the quantities k and n in the expression ib = k(eb
Mc)
2-13. The 6J5 triode is operated with E b = 135 volts.
Determine and plot
curves of n, g m and r p as a function of e c
2-14. The rating of a certain triode is given by the expression
b.

Is photoelectric

With

E = -20
c

2-15.

The

volts,

plate

4 =

130

E =

350

volts, find

and grid

The current

E = 8
c

volts

2-17.

+
h,

0.125ei,)

rp, g m

li8

fx.

eb

fie c )

fie c

power triode are given in

on log paper, and find the quantities

n
.

which n = 20 and which is operating with

Estimate the current when E b = 200
8.7 ma.

in a 6J5 triode for

E = 250
E = 6 volts.

volts,

and

10-\e c

characteristics of a type 851

Appendix B. Plot i s = ib + i c vs.

k and n in the expression i s = k{e b
2-16.

volts

is

6J5 triode for which n

20

is

operating with

Eb =

250

grid potential is required to reduce the current to zero?

2-18. Evaluate the value of n, g m and r of the 6SJ7 pentode for
,

Ecc2 =

volts.

What

E = 3 volts,
c

100 volts, Eb = 150 volts.

2-19. Evaluate the values of n, g m r p of the 6SK7 supercontrol pentode for
Ecc2 = 100 volts, Eb = 250 volts, with Ec = -1 volts; with Ee = -10 volts.
2-20. Plot a curve of g m vs. E cl of a 6SK7 with E cc2 = 100 volts, E b = 250 volts.
2-21. Use the plate characteristics of the 6SJ7 and the 6SK7 pentodes to construct mutual characteristics on the same sheet, with Eb = 200 volts for each
Determine the maximum and minimum values of g m for each tube in the
tube.
range of your sketch.
2-22. Plot a curve of g m vs. E a of a 6L7 with E ccl = 6 volts, E cc2 = 150 volts.
2-23. Refer to Sec. 2-15 for a discussion of secondary emission caused by elecWhat happens to the secondary electrons that are produced by
tron impact.
the impact of the primary current on the anode in a diode? In a triode?
2-24. The mercury-condensation temperature in a General Electric type FG57A thyratron is 40C. If the volume of the tube is 300 cm 3 calculate the mass
of the mercury vapor in the tube.
2-25. a. An OA4G cold-cathode triode is used in the circuit shown, with
Ebbl = 80 volts, Ebb2 = 120 volts. Determine the largest value of R for which
the current will transfer to the main anode, for no input pulse.
,

RADIO ELECTRONICS

52
b.

rated

Determine the value

maximum of 25 ma.

[Chap. 2

of the load resistance to limit the load current to the

2-26. A VR-105 regulator tube is incorporated in the circuit shown to maintain

a constant output potential. The supply potential remains constant at 250 volts,
but the load fluctuates between 40 and 60 ma. Find the value of R so that the
load potential is steady at 105 volts. Assume that the normal operating range
of the tube is 5 to 40 ma.

2-27. Suppose that the plate supply potential to a circuit containing a thyraIf the rms value of the potential is 220 volts, draw a sketch
is sinusoidal.
showing the supply potential and the corresponding critical grid curve for (a) a

tron

negative-control tube;

(6)

a positive-control tube.

CHAPTER

RECTIFIERS AND FILTERS

RECTIFIERS
3-1. Volt -ampere Relation ; Plate Resistance.

basic characteristics of
2-10.

diode,

Some

of the

the vacuum diode were discussed

important

in Sees. 2-8 to

As noted, space charge limits the flow of current in a vacuum

Ordiat least to the point where temperature saturation occurs.

vacuum diode

is operated under space-charge concurve

over the range of operation follows
ditions, and the volt-ampere
an approximate three-halves-power law. That is, over the range of operation when the plate is positive with respect to the cathode, the threehalves-power law specifies the approximate relation between the plate
Of course, when the plate is negative
current and the plate potential.

narily,

however, the

with respect to the cathode, electrons are repelled by the plate and no
Clearly, the relation between the potential across
plate current exists.
it is nonlinear, whence the vacuum diode
fundamentally a nonlinear element.
The volt-ampere curve is a static curve and gives the relationship
between the instantaneous potential across the device and the instantaneThe slope at any point of the curve is an important
ous current through it
quantity, although the reciprocal of the slope, which is known as the

the tube and the current through

is

plate resistance,
is

is

generally preferred.

By definition, the plate resistance

given by the relationship

(3-D

Note that the plate resistance

for

every point on the curve.

is

not a constant and

The

plate resistance

ward direction when the plate potential

cathode;

it

is

is,

in fact, different

is infinite

in the back-

negative with respect to the

possesses the varying value r p in the forward direction.

meaning only for alternating or varying currents.

In

practice r v has

The gas diode has been discussed in Sees. 2-24 and

gas is there shown to cause the volt-ampere

of the

gas diode to be materially different from that of the

2-25.

The influence

characteristic of the

vacuum

diode.

The

gas diode also possesses an effective infinite resistance in the backward

53

RADIO ELECTRONICS

54
direction.

In the forward direction there

is

[Chap. 3

a substantially constant

potential drop across the tube, independent of the tube current.

Both the vacuum diode and the gas diode, by virtue of their nonlinear
volt-ampere characteristics, possess the ability to convert an a-c current
into a current which contains a d-c component in addition to a-c comIn fact, any electrical device which has a high resistance to
ponents.
current in one direction and a low resistance to current in the opposite

An

direction will effect rectification.

ideal rectifier

would be one with

zero resistance in the forward direction and an infinite resistance in the

backward

direction.

No known

A number

device meets the conditions of the ideal

do possess nonlinear characterand many have been adapted as rectifying elements. The important ones in widespread use are the high-vacuum thermionic diodes, gasfilled and vapor-filled thermionic diodes, pool-cathode mercury arcs,
certain crystals which have been incorporated into diodes, and a number
rectifying element.

of devices

istics,

of metallic

semiconducting contact

rectifiers.

The discussion in this chapter will be limited to circuits which employ

the vacuum diode and the gas diode as rectifiers. Attention is called
to the fact that the resulting operation in a circuit of these elements

owing to the different volt-ampere

3-2. Single-phase Half-wave Vacuum Rectifier.

slightly different,

for half-wave rectification

is

shown

in Fig. 3-1.

is

characteristics.

It is

The

basic circuit

assumed that the

+ eb

Fig. 3-1.

simple half-wave

rectifier circuit.

load is a pure resistance.

Also, it is supposed that the power transformer
with
negligible
ideal,
resistance
and leakage reactance.
is

An

application of Kirchhoff's potential law to the load circuit yields

e

where

e is

eb

ib Ri

(3-2)

the instantaneous value of the applied potential,

e b is

the

instantaneous potential across the diode when the instantaneous current is

This one equation is not sufficient for
i h and Ri is the load resistance.
the determination of the two unknown quantities i b and e b that appear
Here, as for triodes and multielectrode tubes, a second
contained in the static plate characteristic of the tube. Con-

in the expression.

relation

is

sequently a solution
characteristic.

is

effected

by drawing the load

line

on the plate

RECTIFIERS AND FILTERS

Sec. 3-2]

55

one significant difference between the solution of the diode as

With the rectifier,
vacuum tube as an
amplifier converts direct current from the plate supply into alternating

There

is

a rectifier and that for the other tubes as amplifiers.

an a-c source supplies the power to the circuit. A
current.

The dynamic
what

The procedure
current

is

potential
is

is

obtained in a some-

is

for an amplifier.
For an applied potential e, the

illustrated in Fig. 3-2.

the intersection of the load line with the static characteristic,

That

say point A.

and

characteristic of the rectifier circuit

manner from the corresponding curve

different

e results in

drawn

is,

for the particular circuit, the application of the

a current

vertically

This

i.

above

e in

is

one point on the dynamic curve

The slope of the load

the diagram.

does not vary, although the intersection with the e b axis varies with e.
Thus, when the applied potential has the value e', the corresponding curThe resulting curve so generated is the dynamic characteristic.
rent is tA'.
If the static characteristic of the tube were linear, the dynamic characNote from the construction, however, that
teristic would also be linear.
line

there

is

considerably less curvature in the dynamic curve than there

the static characteristic.

dynamic curve

To

find the

like that

is

of the current in the

that illustrated in Fig. 3-3.

in

output circuit, the proThis procedure is very

used to find the waveshape in a general amplifier circuit;

in fact, the situation here

8-6),

is

be assumed in what follows that the

is linear.

waveshape

cedure followed

much

It will

is

quite like that of a class

amplifier (see Sec.

except that cutoff of the tube exists at zero input.

If it is

assumed that the relation

eb

is

valid during conduction,

is

linear,

and

then from Eq. (3-2)

e

eb

i b Ri

(3-3)

i br v

this supposes that the static characteristic

it

follows that
i b (r p

Ri)

= Em

sin

oit

(3-4)

RADIO ELECTRONICS

56

[Chap. 3

or
lb

=
Ri

it

sin
rv

I m sin

when

u>t

when

o<

<

<

o>t

<

2tt

(3-5)

where

ir

<

Ri

rp

The d-c power supplied to the load is defined as the product of the readammeter in the load circuit and a d-c voltmeter across the

ing of a d-c

cut

Fig. 3-3.

The method

of obtaining the output-current

waveform from the dynamic

characteristic.

load.

Thus
Pic

Clearly, the reading of the d-c

Id-c

~Zir

P'

Ed.Id. e

ammeter

da =

ib

Jo

(3-6)

represented by

is

['

1
pr2ir

I m sin a da

Jo

(3-7)

ir

and so

p -

rs

lY

ElJii
(3-8)

The power supplied to the circuit from the a-c source, and this is the
power that would be read by a wattmeter with its current coil in the line
and with the potential coil across the source, is given by the integral
Pi

This becomes, by Eqs.

Pi

which

may be

and

(3-3)

eiida

h jo

(3-9)

(3-4),

2r

I(r,

Ri)

da

(3-10)

written in the form

Pi

IL.(r P

Ri)

(3-11)

RECTIFIERS AND FILTERS

Sec. 3-3]

57

where the rms current has the value

The

- >/i

42

efficiency of rectification is defined

* - 7? X 100% =

7- sin2

by the

TLff^)

(3 - i2)

relation

X 100%

which becomes
100

"=(fe)V + r,/B
By combining

this with Eqs. (12-6)

_ (IJttY

and

(12-11), there results

100

(3-13)
t

40.6

This indicates that the theoretical maximum efficiency of the single-phase

half-wave rectifier is 40.6 per cent. But it may be shown that maximum

power output occurs when

R =
t

rv

with a corresponding theoretical

plate-circuit efficiency of 20.3 per cent.

of such a rectifier circuit that warrant special
Refer to Fig. 3-1, which shows the complete wiring diagram
On the inverse cycle, i.e., on that part of the cycle during
of the rectifier.
which the tube is not conducting, the maximum potential across the

There are several features

attention.

tube is equal to the transformer maximum value. That is, the

peak inverse potential across the tube is equal to the transformer maxirectifier

mum

value.

Note

also from the diagram that with the negative terminal of the output connected to ground the full transformer potential exists between the
primary and the secondary windings of the filament heating transformer.
This requires that the transformer insulation must be adequate to withstand this potential without rupture. Evidently if the positive terminal
is

grounded, then the transformer need not have a high insulation

strength.
3-3. Ripple Factor.

Although

it is

the object of a

rectifier to

convert

a-c into d-c current, the simple circuit considered does not achieve this.

Nor, in

fact,

do any

of the

more complicated

rectifier circuits, either

single-phase or polyphase, accomplish this exactly.

What

is

achieved

is

a unidirectional current, periodically fluctuating components still remaining in the output.

Filters are ordinarily used in rectifier systems in order

and these

will receive

of the fluctuating

components

to help decrease these fluctuating components,

detailed consideration below.

measure

RADIO ELECTRONICS

58
is

given by the ripple factor

T

rms value
avg

'

which

r,

defined as

is

components

of a-c

[Chap. 3

or d-c value of

of

wave

wave

and which may be written as

where

An

I'raxa

/'

E'

4=5-'

'

9=2!

and E'm denote the rms values

(3-15)

of the a-c

analytical expression for the ripple factor

is

components

noted that the instantaneous a-c component of the current

i'

But by

only.

readily possible.
is

It

is

given by

Id-o

definition

This expression is readily interpreted. The first term of the integrand

when evaluated yields the square of the rms value of the total wave I?ms
The second term yields
.

The rms

By

Air

Jo

ripple current then

^rms

2"

2ih.< da

= 2IL

becomes
2/|. c

\/^?m

"f"

I d-c

I,d-o

V^-^rmj

combining these results with Eq. (3-15)

r

= VjL,

IL =

J(^\

(3-16)

This expression is independent of the current waveshape and applies in

general, since the development was not confined to a particular waveshape. In the case of the half-wave single-phase rectifier the ratio
^rms

I in/ 2

7T

and hence
r

a/1-57

1.21

(3-17)

This shows that the rms value of the ripple potential exceeds the d-c
This merely tends to show that a single-phase
potential of the output.
half-wave rectifier without filter is a relatively poor device for converting
a-c into d-c potential.

3-4. Single-phase Full-wave Rectifier.

The circuit of the singlephase full-wave rectifier, given in Fig. 3-4, bears some resemblance

RECTIFIERS AND FILTERS

Sec. 3-4]

to a push-pull circuit.

59

Actually the circuit comprises two half-wave

which are so connected that conduction takes place through

during one half of the total power cycle and through the other
tube
one
tube during the second half of the power cycle. The output current
circuits

A-c Input

Fig. 3-4. Schematic wiring diagram of a single-phase full-wave rectifier.

through the load has the form illustrated in Fig. 3-5, where the portions
of the wave marked 1 flow through tube Tl and the portions of the wave
marked 2 flow through tube T2.
The d-c and rms values of the load current are found from Eqs. (3-7)
and (3-12) to be

Id-c
I rms

x
(3-18)

Im

V2

where Im is the peak value of the curThe d-c output power is

rent wave.
then

= ILRi
Fig. 3-5.

The transformer

and output load current

phase full-wave

in

is

a single-

higher

Ri)

(3-19)

potential

rectifier.

delivered to the load

EIRi
(r,
p

By comparing

this

Eq.

seen that the power

by a

However, the power depends on the

(3-8)

it

is

expression with

factor of 4 in the full-wave case.

circuit

parameters in the same

way

as for the half-wave circuit.

The input power from the

form as Eq.

a-c source

Pi

The

is

readily found to have the

same

(3-11), viz.,

IL.(r p

(3-20)

Ri)

efficiency of rectification is

81.2
Vr
1

r p /R,

(3-21)

RADIO ELECTRONICS

60

maximum

This expression shows a theoretical

half-wave

The

that

is

twice that of the

rectifier.

ripple factor

is

From Eq.

when

readily found

Im .
Id-c

it is

_ IJy/2

noted that

1.11

2/m/x

(3-16),

= Vl-11 2 -

Thus the

[Chap. 3

0.482

(3-22)

dropped from 1.21 in the half-wave rectifier to

What has been accomplished in the full-wave
rectifier, therefore, is that the rectification process has become more
efficient, with a higher percentage of the power supplied to the circuit
being converted into the desired d-c power, and with a consequent smaller
ripple factor has

0.482 in the present case.

fraction remaining in a-c form, which, while producing heating of the

load, does not contribute to the desired d-c power.

study of Fig. 3-4 indicates that when one tube is conducting, say
T2 is in the nonconducting state. Except for the tube drop
i b r p in Tl, the peak inverse potential across T2 is 2E m or twice the transformer maximum potential measured to the mid-point, or the full transformer potential. The potential stress between windings of the filament
transformer is seen to be the full d-c potential, if the negative is grounded,
Tl, then tube

and

sensibly

is

zero,

if

the positive

is

grounded.
with Gas Diodes. Gas
be used in the half-wave and

3-5. Circuits

diodes

may

full-wave circuits discussed above.

Owing

to their different plate characteristics, the

results are
Fig. 3-6. The applied potential
current waveshape in a
half-wave rectifier circuit using
a gas diode.

ance

is

different.

For these

down
consequently conduct for

somewhat

tubes a sensibly constant potential appears

across the tube when the tube is conductbut conduction does not begi n untu
,
,
the applied potential exceeds the break-

less

potential of the tube.

than 180 deg in each

cycle.

The tube will

The situation

illustrated in Fig. 3-6.

The equation of the potential across the load during conduction

obtained by applying Kirchhoff 's law to the plate circuit,
et

ib Ri

= Em

and the corresponding expression

ib

where Ed

is

= E

sin

Eq

for the current

* in

"

~ g

is

(3-23)

is

the constant tube drop during conduction.

(3 .24)

RECTIFIERS AND FILTERS

Sec. 3-5]

The

d-c plate current

is

instantaneous current and

61

found by taking the average value of the

is

Em

sma-E

2r J ai

da

is the angle at which the tube fires and a 2 is the angle at which
conduction ceases. Ordinarily the applied plate potential is much larger

where a x

than Ea, and the angles a x and jt a 2 are very nearly zero. Consequently the limits of the integral of Eq. (3-25) may be changed to
and
x without appreciable error in the result. When this is done and the
integral is evaluated, it is found that

^m _ ^

wRi

The load

potential

E^ may

-hi -It)
1

2Ri

<

3 -26 >

be written as

^ = ^(i-it;)

(3-27)

7T

This equation does not contain the load current. This means, of course
that Ed-c is independent of the load current, with consequent perfect
;

regulation.

To

is necessary to calculate
given by

calculate the efficiency of rectification, it

the input power to the plate circuit.

r

D =

1
2lr

f
Jo

eib

This

['

1
2tt

is

Em

v Sm

sin

**

and

where the limits are again taken as

r.

efficiency of rectification is

may be

<3 -28)

then

ir

which

- E

This expression reduces to

('-;)
The

Ri

Jo

Em

reduced to the form

Vr

40.6 (l

1.87

(3-30)

Note that this value is independent of the load current or load resistance.
To the same approximation, namely, Em 3> Eo, the ripple factor is given

RADIO ELECTRONICS

62

[Chap. 3

by
r

1.21

(3-31)

than the value with the vacuum diode. This

is less than 180 deg.
The corresponding properties of the full-wave circuit with gas tubes
will follow a completely parallel development and yield results that bear
the same relation to the vacuum-tube case that the foregoing results do

which

is

slightly higher

increased ripple results because the tube conduction

to the corresponding half-wave vacuum-rectifier case.

3-6. Miscellaneous

Single-phase Rectifier Circuits.

variety

of

Among these are

other rectifier circuits exist which find widespread use.

bridge rectifier circuits, potential-doubling circuits, and potential-multiplying circuits.

The

finds extensive use

rectifier

and

bridge circuit

both as a power

also as the rectifying

system in rectifier-type a-c meters.

The

rectifiers for

Fig.

3-7.

Single-phase

full-wave

rectifier circuit.

bridge

power use

utilize

both the
vacuum and gas varieties, whereas
those for instrument use are usually of the copper oxide or crystal
thermionic

diodes

of

types.

To examine

the operation of the bridge circuit, refer to Fig. 3-7. It is

observed that two tubes conduct simultaneously during one half of the
cycle

and the other two tubes conduct during the second

The conduction paths and

half of the cycle.

directions are such that the resulting current

through the load is substantially that shown in Fig. 3-5.

The primary features of the bridge circuit are the following: The currents drawn in both the primary and secondary of the plate-supply transformer are sinusoidal. This permits a smaller transformer to be used for
a given output power than is necessary for the same power with the
single-phase full- wave circuit of the two-tube type.* Also, the transformer need not have a center tap. Since each tube has only transformer
potential across it on the inverse cycle, the bridge circuit is suitable for
high-potential applications.
However, the transformers supplying the
heaters of the tubes must be properly insulated for the high potential.
A rectifier meter is essentially a bridge-rectifier system which utilizes
copper oxide elements. The potential to be measured is applied through
a multiplier resistance to two corners of the bridge, a d-c milliammeter
being used as an indicating instrument across the other two corners.
*

For a discussion of transformer

'Electronics," 2d ed., chap. 14,

utilization factor, see J. Millman and S. Seely.

McGraw-Hill Book Company, Inc., New York, 1951,

RECTIFIERS AND FILTERS

Sec. 3-6]

63

But as the d-c milliammeter reads average values of current, the scale
of the meter is calibrated to give rms values of sinusoidal waves by applying a sinusoidal potential to the input terminals.
The indication on such
an instrument is not correct for input signals that contain appreciable
harmonics.

A common

potential-doubling circuit

is

shown

in Fig. 3-8.

The

out-

approximately equal to twice the transformer

maximum potential. It operates by alternately charging each of the two
capacitors to the transformer peak
potential E m current being continually drained from the capacitors
through the load. This circuit is
characterized by poor regulation unput 1 from such a circuit

is

less

very large capacitors are used.

inverse potential is twice

The peak

the transformer peak potential.

If

ordinary rectifiers are used, two sepa-

Fig. 3-8.

A full-wave potential-doubling

circuit.

rate filament sources are required.

If

is built, and these are used extensively

such as 25Z5 are available. These tubes

a relatively low potential system

in a-c /d-c radio sets, special tubes

are provided with separate indirectly heated cathodes.

The cathodes

in

these tubes are well insulated from the heaters, which are connected in
series internally.

The

regulation of the potential doubler can be improved, particularly

by employing a bridge doubler rectifier, 2 which is

3-9.
The feature of this rectifier circuit is that at

at the higher loads,

illustrated in Fig.

Fig. 3-9.

A bridge

doubler circuit.

output potential is approximately twice the transformer

However, the potential will never fall below the output
the bridge circuit at any load, nor will the ripple factor exceed that of

light loads the

peak potential.
of

the bridge circuit,

viz., r

0.482.

Most other

features of this circuit

are like those in the normal bridge circuit, such as the peak inverse
potential to which each tube is subjected, and the heater-cathode insulation problems.

An

3
alternative potential-doubling circuit

is

shown

in Fig. 3-10.

The

RADIO ELECTRONICS

64

[Chap. 3

output potential from this circuit, like that from Fig. 3-8, is approximately equal to twice the transformer maximum potential. It operates
by charging capacitor Ci during one half cycle through tube Tl to the
transformer peak potential Em and during the next half cycle charges C 2
through tube T2 to the potential determined by that across C\ and the
The peak
transformer in series, the peak being approximately 2E m
inverse potential across each tube is twice the transformer peak potential.
.

Fig. 3-10.

A half-wave potential-doubling

circuit,

This circuit

two

may be extended

shown
multiplication, odd

capacitors, as

n-fold

Fig. 3-11. A half-wave potential-quadrupling circuit.

to a quadrupler

in Fig. 3-11.

It

by adding two tubes and

may

be extended to provide

or even.

FILTERS
3-7.

The Harmonic Components

in Rectifier Circuits.

It is usually

the requirement of a power supply to provide a relatively ripple-free

source of d-c potential from an a-c line. However, as seen above, a
rectifier actually provides an output which contains a-c components in
addition to the d-c term that is desired, a measure of the a-c components
being given by the ripple factor. It is customary to include a filter
between the rectifier and the output to attenuate these ripple components.

Often an electronic regulator

is

also included,

to be substantially constant with load or

The

if

the potential output

analysis of the action of such rectifier niters

fact that the rectifier as a driving source

is

solution of circuits with nonlinear elements.

to

if

make

is

the ripple must be small.

is

complicated by the

nonlinear, thus requiring the

It is possible in

most cases

reasonable assumptions in order to effect an approximate

engineering solution.

In consequence, the results obtained are only

approximate.

An

analytic representation of the output of the single-phase half-wave

rectifier is

obtained in terms of a Fourier series expansion,

This series

RECTIFIERS AND FILTERS

Sec. 3-7]

65

representation has the form

CO

00

+ ^

ka

b k cos

+ ^

t=i

where a = wt and where the

by the integrals

coefficients that

bk

6o

a* sin

&a

(3-32)

=i

appear in the

series are

given

2*

&io

PT

cos &a da

sin fca

(3-33)

Jo
"2x

a*,

ir

It

da

Jo

should be recalled that the constant term bo that appears in this Fourier
average or d-c value of the current.

series is the

The
which

the current in a half-wave rectifier circuit,

obtained by performing the indicated integrations using Eqs.

explicit expression for

is

(3-5) over the

two

specified intervals, yields

1

V
I

+ -sm^--2
.

IT

cos

(fc

*== 2,4,6,

where Im

= Em /(r

+R

t)

and

Em

The lowest angular frequency that

is

is

fco;<

,
(3-34)

]
1} J

the peak transformer potential.

present in this expression

is

that of

term of frequency oi,

the expression are even-harmonic terms.

Also, except for this single

the primary source.

other terms that appear in

The corresponding Fourier

full-wave rectifier which

is

all

output of the
be derived from

series representation of the

may

illustrated in Fig. 3-5

Eq. (3-34). Thus, by recalling that the full-wave circuit comprises two
half-wave circuits which are so arranged that one circuit is operating
during the interval when the other is not operating, then clearly the currents are functionally related
total load current,

which

is i

by the expression

-MM

ii

*= 2,4,6

i^,

12(a)

z\(a

+ t).

The

then attains the form

cos
(k

koit

1)(*

(3-35)

1)J

r p ) and where E m is the maximum value of the

where Im = E m /(R
transformer potential measured to the center tap.
t

A comparison of Eqs. (3-34) and (3-35) indicates that the fundamental

angular-frequency term has been eliminated in the full-wave circuit, the
lowest harmonic term in the output being 2w, a second-harmonic term.
This will be found to be a distinct advantage in

filtering.

RADIO ELECTRONICS

66

[Chap. 3

The Fourier series representation of the half-wave and full-wave circuits

using gas diodes can be obtained in the same way as above, although the
form will be more complex. This is so because conduction begins at some
small angle

<po

and ceases at the angle

it

<po,

when

it is

assumed that

the breakdown and the extinction potentials are equal. But since these
angles are usually small under normal operating conditions, it will be

assumed that Eqs.

vacuum

(3-34)

of a controlled rectifier

and

(3-35)

The Fourier

or gas diodes.

is

are applicable for circuits with

series representation of the

also possible, although the result

is

output

quite com-

However, such controlled rectifiers are ordinarily used in services

which the ripple is not of major concern, and, as a result, no detailed
Some results will be given below covering
analysis will be undertaken.
plex.

in

these rectifiers, however.

The operation

3-8. Inductor Filters.

on the inherent property

of

of an inductor filter depends

an inductance to oppose any change of current

Fig. 3-12. Half-wave rectifier circuit with inductor

that

may

tend to take place through

it.

That

is,

filter.

the inductor stores

magnetic field when the current is above its average value

and releases energy when the current falls below this value. Consequently any sudden changes in current that might otherwise take place
in the circuit will be smoothed out by the action of the inductor.
In particular, suppose that an inductor is connected in series with the
load in a single-phase half-wave circuit, as illustrated in Fig. 3-12. For
simplicity, suppose that the tube and choke resistances are negligible.
Then the controlling differential equation for the current in the circuit
during the time of current conduction is
energy in

its

dib

dt

Rii b

=Em

sin

cat

(3-36)

A solution of this differential equation may be

effected.
This solution is
complicated by the fact that current continues over only a portion of the
The general character of the solution is shown graphically in Fig.
cycle.
3-13, in which is shown the effect of changing the inductance on the wave-

of the current.
Since a simple inductance choke is seldom used with
a half-wave circuit, further details of the analysis will not be given.
Suppose that an inductor filter is applied to the output of a full-wave

form

RECTIFIERS AND FILTERS

Sec. 3-8]

67

and a sketch of the output-current waveshape are

Since no cutout occurs in the current, the analysis
assumes a different form from that for the half-wave case. Now, instead
of considering the circuit differential equation, as in Eq. (3-36), and

The

rectifier.

circuit

given in Fig. 3-14.

'

<?<7
90

180

u)t(degrees)

The

Fig. 3-13.

half-wave

270

360

changing inductance on the waveform of the current in a

with inductor filter. The load Ri is assumed constant.

effect of

rectifier

adjusting the initial conditions to

an approximate solution

is

fulfill

effected.

the required physical conditions,

It is

the potential that is applied to the filter

over, it is noted that the amplitudes

supposed that the equation of

given by Eq. (3-35). More-

is

terms beyond the first, and

second-harmonic frequency,
are small compared with that of the
In particular, the fourthfirst term.
harmonic frequency term is only 20
of the a-c

this is of

per cent of the second-harmonic term.

Furthermore, the impedance of the

inductor

increases

and better

filtering

with

frequency,

action exists for

Consethe higher-harmonic terms.

quently it is assumed that all higherorder terms

it

may

Fig. 3-14. Full-wave rectifier circuit

with inductor filter, and the waveshape of the load current.

be neglected.

In accordance with' this discussion,

is supposed that the input potential to the

filter

and load has the

approximate form
e

The corresponding load current

ii

4Em

= 2Em

= 2Em
ttRi

is,

in

4E m

COS

(3 _ 37)

2o)t

accordance with a-c circuit theory,

cos (2wt

3t y/R\

\p)

L2

(3-38)

RADIO ELECTRONICS

68

where
tan

The

2wL

+ =

ripple factor, defined in Eq. (3-15),

+ 4o, L
2

(4#m /3x)(l/V#?

may be

the ratio

becomes
2R,

y/B\

+ 4a> L
2

expressed in the form

r

If

(3-39)

2EJwRi
which

[Chap. 3

uL/R

is large, this

(3-40)

reduces to
1

Ri

3 y/2

oiL

(3-41)

This expression shows that the filtering improves with decreased load
At no load,
resistance or, correspondingly, with increased load current.
R = * and the filtering is poorest, with r = 2/3 y/2 = 0.47. This is
also the result which applies when no choke is included in the circuit.
[Compare this result with Eq. (3-22), which gives 0.482. The difference
arises from the terms in the Fourier series that have been neglected.]
The expression also shows that large inductances are accompanied by
,

decreased ripple.

The

d-c output potential

is

E^ = I^R =
t

given by

0.637tf m

0.90fl rm .

(3-42)

where E^, is the transformer secondary potential measured to the center

Note that under the assumptions made, viz., negligible powertap.
transformer leakage reactance, transformer resistance, tube resistance,
and inductor resistance, the output potential does not change with load,
with consequent perfect regulation. Because the neglected effects are
not negligible, the output potential actually decreases with increased
current.

Filtering is frequently effected by shunting

During the time that the rectifier output is
capacitor.
with
a
the load
charging
is
to the rectifier output potential and
the
capacitor
increasing,
During the time that the rectifier
the
capacitor.
stored
in
is
energy
capacitor,
the capacitor delivers energy
that
of
the
below
falls
potential
potential
at a high level for a longer
the
maintaining
thus
the
load,
to
3-9. Capacitor Filter.

period than without the capacitor.

decreased.

The

ripple

is

therefore considerably

Clearly, the diode acts as a switch, permitting charge to

RECTIFIERS AND FILTERS

Sec. 3-9]

when the

flow into the capacitor

69

exceeds the capacitor

rectifier potential

and then acts to disconnect the power source when the potential
falls below that of the capacitor.
To examine the operation in some detail, refer to Fig. 3-15, which shows
a diagram of the circuit. The tube current during the conducting portion
potential,

of

the cycle

is

U =

ic

(3-43)

i,

I,

1
I

where

]+

-^
i

+ e6
.

ei_

Ri

ec
(3-44)

Ri

and where
dq c

where qc
of the

is

Fig. 3-15.

n dec

(3-45)

The

the capacitor charge.

single-phase half-wave ca-

pacitor-filtered rectifier.

controlling differential equation

charging current through the tube

r

is

then

dec
(3-46)

But the potential

ec during the time that the

the transformer potential,

tube

the tube drop

if

is

is

conducting

neglected.

is

simply

Hence the

^Cut-out
{ point

Fig. 3-16.

The tube current and the load

potential in a single-phase half-wave

capacitor-filtered rectifier.

capacitor potential during this portion of the cycle

ec

The corresponding tube current

= Em

This

may

sinusoidal

and

is

cot

is

Em

ib

sin

is

-5- sin

ait

+ wCEm cos

cot

be written in the equivalent form

JS
E,m yjco

\p

tan -1 coCRi

C2

+ -5s sin

lb

(at

+ i)

(3-47)

where

sketch of the current

wave

is

illustrated in Fig. 3-16.

(3-48)

RADIO ELECTRONICS

70

Equation
improve the

(3-47)

[Chap. 3

shows that the use of large capacitances, in order to

accompanied by large tube currents. Therefore,

filtering, is

a large capacitance is used for a given load in order to maintain the

output potential more nearly constant, a very peaked current exists. In
fact, for a certain required average current demand by the load, the tubecurrent pulse becomes more and more peaked as the capacitance is made
if

This imposes serious duty conditions on the tube, since the average current through the tube may be well within the tube rating and yet
the large peak current may injure the cathode. Vacuum diodes would
not be appreciably damaged by the high peak-current demands, since
temperature-saturated currents may be drawn without seriously injuring
larger.

the cathode. In the case of gas tubes, however, any attempt to draw
higher than temperature-saturated current will usually be accompanied
by severe positive-ion bombardment of the cathode, with consequent

cathode disintegration. It is for this reason that large-capacitance input

be used with rectifiers that employ gas diodes.
When the tube stops conducting, i h = and the controlling differential
equation during the nonconducting portion of the cycle is, from Eq.
niters should not

(3-46),

C
The

%=

if +

solution of this differential equation

ec

(3 " 49)

is

Ae-' /R,c

(3-50)

This shows that the capacitor discharges exponentially through the load.
To determine the value of the constant A that appears in this expression, use is made of the fact that at the time t = t\, the cutout time,
ec

Combining

and Eq.

= Em

this result with Eq. (3-50) gives

(3-50)

A = Em

sin

whe h/RlC

(3-51)

= Em

sin

uhe-w*

(3-52)

becomes
ec

The quantity h that appears in this

is zero.
From Eq.

expression

sin
it

ti

from Eq.

known, since at

= h

ip)

follows that

cofi

If

(co<i

is

(3-47) this requires

the tube current

from which

sin wti

(3-53)

is

7T

- ^ =

tan" 1 wCRi

(3-53)

substituted in Eq. (3-52), there results

ec

= Em

sin & ,< ie -c<+*-*>/ers,

(3.54)

RECTIFIERS AND FILTERS

Sec. 3-10]

71

To find the cutin " point, it is noted that ec equals the impressed transformer potential e at this point. This requires
'

'

Em

sin

o>t 2

= Em

w tie-^^ /uCRl

sin

or

sin ut 2

sin

u>

tie-^

^~

170
ir)/'-,CR '

(3-55)

The

evaluation of the cutin time

not

be

solved

explicitly,

for

The

evaluation.

can-

this

150

is

140

Graphical

^130

effectively in this

jo 120

transcendental equation.

methods can be used

<2

160

are given in

results

110

Included on this graph are a

plot of Eq. (3-53) for the cutout angle and
a plot of Eq. (3-55) for the cutin angle.

-5100

The foregoing analysis gives a complete

specification of the operation of the capa-

j> 60

Fig. 3-17.

citor filter, the current

= Em

o)< 2

and by Eq.
ec

= Em

30

20

f2 ( full

wet ve)

10

<

<*>t

<

(3-56a)
10

coii

(3-54)

sin coher

for uti

ifA*/f vav eU

. 40

sin ut

for

90

through the tube

being given by Eqs. (3-47) and (3-48),

the potential across the load resistor being
given by
ec

<

20 30 40 SO 60 70 80 90
fan' 1 u>CR l

Fig. 3-17. Plot of cutin angle o>h

"'-"

wt

<

^^
2w

and cutout angle

(3-566)

coti

vs.

circuit

parameters for the capacitor

filter.

With this information it is possible to evaluate the d-c output potential,

the ripple factor, the peak tube current, etc.
These quantities may then
be plotted as functions of the parameters R h C, E m
Such an analysis is
.

quite involved, but

it

has been carried out, 4 and the results are given in

graphical form.
3-10. Approximate

Capacitor Filters.

make

Analysis

of

It is expedient to

several reasonable approxima-

tions in order to obtain an approxi-

mate analysis

of

the behavior of the

Such an approximate analysis possesses the advantcapacitor

Fig. 3 18. Oscillogram of the load potential in

a single-phase full-wave

filter.

capacitor-filtered rectifier.

age that the important factors of the

operation are simply related to the circuit parameters.
Moreover, the

most engineering applications. The

evident by an inspection of Fig.
an oscillogram of the load potential in a

results are sufficiently accurate for

character of the approximation

3-18,

which shows the trace of

is

made

RADIO ELECTRONICS

72

[Chap. 3

single-phase full-wave capacitor filtered rectifier. The potential curve

may be approximated by two straight-line segments, as shown in Fig.
3-19.

If

the total capacitor discharge potential is denoted as E r then,

from the diagram, the average value
,

of the potential is
Ed-o

Also, the

= Em

rms value

of the triangle

may

be shown to be

ripple potential

The approximate load-potenwaveform corresponding to the

EL,,.

Fig. 3-19.
tial

=
2

curves o Fig. 3 18.

Also,

(3-57)

if it is

(3-58)

V3

assumed that the capaciwhich is

tor discharge continues for the full half cycle at a constant rate

equal to the average load current h-c, the

That is, approximately
cycle is E r

fall in

potential during this half

Er =
The

ripple factor

is

(3-59)

2/C

then given by

E'rm.

Ei-c

But

since Ed-c

IiRi,

iV3fCR

(3-60)
t

This expression shows that the ripple factor varies inversely with the
load resistance and the filter capacitance. At no load, Ri = and the
As Ri decreases, corresponding to increasing current, the
ripple is zero.
Also, for given Ri, the ripple is smaller for large
ripple becomes larger.
Actually, Eq. (3-60) is more nearly correct for small
capacitances.
,

values of ripple than for the larger values, the value of ripple being
The results are
generally larger than that obtained experimentally.

adequate for most purposes.

The regulation curve is obtained by combining Eqs. (3-57) and (3-59)
This yields
Ei-v

En

Ij-<!

4/C

(3-61)

This expression represents a linear fall in potential with d-c output current.
Also, it shows that the simple capacitor filter will possess poor
regulation unless the capacitance

is

large.

RECTIFIERS AND FILTERS

Sec. 3-11]

73

Now refer to the circuit of Fig. 3-15 to ascertain the peak inverse potential

be twice the transformer peak potential.

peak inverse potential is also twice the transpotential, as measured from the mid-point to either
It is seen to

across the tube.

For the full-wave

former

case, the

maximum

Thus the presence

end, or the full transformer potential.

of the capacitor

peak inverse potential in the half-wave circuit from Em to

2E m but does not affect the peak inverse potential in the full-wave circuit.
increases the

An

3-11. L-section Filter.

L-section

filter

consists of a series induc-

tor and a shunt capacitor, as shown in Fig. 3-20.

that the inductor offers a high impedance to
the harmonic terms, and the capacitor

This

filter is

so arranged

>^

shunts the load, so as to by-pass the har-

monic currents. The resulting ripple is

markedly reduced over that of the relatively
,

simple

The

n1j

&

filters of Sees.

ripple factor

oo
3-8

is

BoFIG

3-9.
andjon

readily approximated

applied to the input terminals of the

filter

3-20.

An

by taking for the potential

the first two terms in the

Fourier series representation of the output potential of the

4E m

= %Em

.L-section filter,

rectifier, viz.,

cos n2ut

/o

/<-%

(3-62)

6w

But since the filter elements are chosen to provide a high series impedance
and a very low shunting impedance, certain plausible approximations
may be made. Thus, since the choke impedance is high compared with
the effective parallel impedance of the capacitor and load resistor, the
net impedance between terminals AB is approximately Xl and the a-c
current through the circuit

is

Likewise, since the a-c impedance of the capacitor

itor

may

be assumed that
and none through the

Ri, it

(the ripple potential)

is

all

ripple factor

is

small compared with

resistor.

The

a-c potential across the load

the potential across the capacitor and

^. =
The

is

the a-c current passes through the capac-

then given by

^ -cg
d

is

(3-64)

RADIO ELECTRONICS

74
which

may

be written, at 60 cps, with

[Chap. 3

in henrys

and C

in microfarads

0.830
(3-66)

LC

should be noted that the effect of combining the decreasing ripple of the
inductor filter and the increasing ripple of the simple capacitor filter for
increasing loads is a constant ripple circuit, independent of load.
It

y(d)

Fig. 3-21.
filter,

when

The tube-current waveform

(a)

The above

L =

0, (6)

L < Lc

(c)

in the full-wave rectifier

L = L

analysis assumes that

c,

(d)

L > L

e,

with an L-section

for constant Id-c-

no current cutout

exists at

any time

the analysis would follow along the lines of Sec.

But since with
3-9 and Eq. (3-62) for the potential would not apply.
no inductance in the filter cutout will occur, whereas with sufficient inducof the cycle.

If it did,

tance there will be no cutout, it would be expected that there would be

some minimum inductance for a given current below which cutout would
occur, although for larger values than this critical value the conduction

would continue
cally.

for the full cycle.

The situation is best illustrated

graphi-

Figure 3-21 shows the expected tube current for various amounts

of series inductance L.

RECTIFIERS AND FILTERS

Sec. 3-11]
If

75

to pass current throughout the entire cycle, the peak

must not exceed the d-c component. But the d-c value
Hence for
Also, the peak a-c current is (2E d . /3)(l/X L ).

the rectifier

is

current delivered
is

E&JRi.

current flow during the full cycle

it is

necessary that

^ 2E d

Ed-c

Xl

Ri
or

XL
from which the value for the

>^
inductance

critical

Lc

(3-67)

2R

which has the value

L =
c

is

found to be

o
ou

(3-68)

power frequency, where Ri is in ohms and L c is in henrys.

However, owing to the approximations that have been made in this

for a 60-cps

analysis,

it is

advisable for conservative design to use a larger value of

than that given in Eq. (3-68). A good practical figure

denominator as 1,000 instead of 1,130.
The effect of the cutout is illustrated

is

to choose the

in Fig. 3-22,

curve

of

which shows a regulation

the system,

and varying

Ri

Clearly,

for

inductance.

series

when the

series

Edc
m

constant

inductance

FlG 3 _ 22 ; The regulation curve ofH

with L-section filter as a func-

tance type and the output potential is

With increasing
approximately E m

rectifier

tion of series inductance, for constant

out P ut curren t.

L = Lc

falls,

is

zero, the filter is of the simple capaci-

inductance, the potential

Q^

until

that corresponding to the simple L filter

For values of L greater than L c there is no
with no cutout, or 0.637 m
change in potential, except for the effects of the resistances of the various
at

the output potential

is

elements of the circuit.

It is

not possible to satisfy the conditions of Eq. (3-68) for all values
no load this would require an infinite inductance. How-

of load, since at

ever,

when good

potential regulation

is

desired,

it is

customary to use a

bleeder resistance across the load so as to maintain the conditions of

Eq. (3-68) even

more

if

power loss.
method than using a high bleeder

this respresents a

efficient

current, with its

attendant power dissipation, is to make use of the fact that the inductance of an iron-core reactor depends, among other things, on the amount

RADIO ELECTRONICS

76
of d-c current in

[Chap. 3

Chokes for which the inductance is high

and which decrease markedly with increased
"swinging" chokes. The swinging choke is pro-

the winding.

at low values of d-c current

d-c currents are called

vided with a closed iron core, whereas the core of the inductor which is to
possess a more nearly constant inductance is provided with a narrow air
gap.
A typical curve for such a
reactor

20

is

illustrated in Fig. 3-23.

The advantage

of

such a choke

is

that for high Ri, and therefore low

d-c current, the inductance is high.

-C

810

As a result, the conditional equation

(3-68)
20

40

60

80

100

120 (mo)

Fig. 3-23. The inductance of a swinging

choke as a function of the d-c current

through

it.

of

is

satisfied over

swinging choke
factor

a wider range

Clearly, however,

Ri.

is

is

when a

used, the ripple

no longer independent

of

the load.

The above analysis for the critical inductance of the L-type filter applies
which conduction continues for 180 deg in
Consequently the results so obtained are not applicable
when an L-section filter is used with a controlled rectifier. The analysis
for a full-wave controlled rectifier is considerably more complicated than
that above, owing to the fact that
the amplitude of the harmonics in the
for the full-wave rectifier for

each cycle.

Fourier series representation of the

output depends on the delay angles,

of such amplitude
that they cannot be neglected in the
The results of such an
analysis.

and these are

analysis are given graphically 1 in Fig.

The curves give a measure of

both the critical inductance and the
output potential.

3-24.

3-12. Multiple L-section Filters.

10

If it is desired to limit

the ripple to a

is less than that possible

with a single L-section filter using
commercially available elements, two
or more L-section filters may be con-

value that

20 30 40 50 60 70 80 TO
Delay angle, dteq.

Fig. 3-24. Critical inductance and d-c

output potential as a function of the
delay angle in a full-wave controlled
rectifier.

nected in cascade, as shown in Fig. 3-25. An approximate solution is

It is assumed, therefore,
possible by following the methods of Sec. 3-11.
that the choke impedances are much larger than the reactances of
the capacitors. Also, it is assumed that the reactance of the last capacitor is small

compared with the resistance

of the load.

Under these

RECTIFIERS AND FILTERS

Sec. 3-12]

77

A 3 and B 3 is Xa- The impedand the impedance between Ai and Bi is

assumptions, the impedance between

ance between

XL

A 2 and B

is

XC

i,

approximately.

i,

The

a-c current I x

is

approximately

The

a-c potential across Ci

The

a-c current 7 2

is

is

approximately

approximately
x

12

The

a-c potential across the load

_ E AlB
~ ~Y
is

approximately

^Cl _
TV
lA C2 TP
A CI 1TV

The

ripple factor

V^2
5

Ali

C'd-o

<j

Xci Xc2

^?
An Ai2

given by the expression

is

^f^
Ah
2

(3-69)

Aii

comparison of this expression with Eq. (3-64) indicates the generalmade in obtaining an expression for the ripple
factor of a cascaded filter of n secization that should be

The

tions.

expression would have

a,o

^
UWiT

the form

^Cl

-^C2

X/,!

Xij

the sections are

^'

Xx, n

all similar,

It follows

from

factor r

given by

is

A"

12

-Xcn

&

<>,

Fig. 3-25.

(3-70)
If

t^WtiF -

two-unit L-section

filter.

then Eq. (3-70) becomes

Wf
\X
LJ

this that the required

LC =

1.76

LC

(16rV*LC)*

(3
^

71)

product for a specified ripple

(^r^y

"
(

3 " 72 )

Note also that, to the approximation that the impedance between A 2

and B2 is simply Xci, the critical inductance is given by Eq. (3-68), as
for the single-section unit,

RADIO ELECTRONICS

78

[Chap. 3

3-13. n-section Filter.

The use of a II-section filter provides an output potential that approaches the peak value of the a-c potential of the
Such a filter is illussource, the ripple components being very small.
trated in Fig. 3-26. Although such filters do provide a higher d-c output
potential than is possible with an L-section filter, the tube currents are
peaked and the regulation is generally poor, these results being common

with the simple capacitor

filter.

study of the oscilloscope patterns at various points of such a filter

shows that the action can be understood by considering the inductor and
the second capacitor as an L-section filter
L
-^ffOTT
that acts on the triangular output potential
r
wave irom t Qe first capacitor. The output
\r'
-^c2
1
potential is then approximately that from the
1

input capacitor, the ripple contained in this

0U tp U t being reduced by the L-section filter.
the ripple factor of the II-section filter is given approximately by
.

n-section

ig. 3- 6.

That

is,

'

ter.

rr

where rc
becomes

is

all

(3-73)

r c rh

given by Eq. (3-60) and Tl

r,

with

is

given by Eq. (3-65).

0.855

(3-74)

reactances calculated at the second-harmonic frequency.

60-cps power source, this

This

For a

is

with the capacitances in microfarads, the inductance in henrys, and the

This result is only approximate, since it assumes in
resistance in ohms.
effect that the ripple output from the capacitor filter is sinusoidal rather

than triangular.

somewhat more accurate evaluation

Arguimbau, 6

is

possible.

of the ripple factor,

The technique employed

is

due to

similar to that used

power of a class C amplifier. For the filter

connected to a rectifier at the power frequency a, the important ripple
term is of second-harmonic frequency. Consequently, it is required to
find the peak value of the second-harmonic current I'2m to the input
capacitor of the II filter.
This is given by the Fourier component
to evaluate the grid driving

IL =

P*

- /
t Jo

cos 2ut d(ut)

(3-76)

Now assume that the current pulse is significant only near the peak value
of the

integral

cosine curve.
is

Therefore, the cos 2wt factor appearing in the

replaced by unity, and approximately

RECTIFIERS AND FILTERS

Sec. 3-13]

f 2*

y/2

= -

7'
2

i b d(o>t)

T JO

= 27d

79

(3-77)

.c

Hence, the upper limit of the rms second-harmonic potential

E't

But the potential E\

logic as before, the

is

/'
2

X C1

V2 /d- X cl
L

section, so that,

by the same

Hence, the ripple factor

E'^Xa/Xi,.

is

(3-78)

applied to the

output ripple

is

is

rT

V2

ff

C2

V2

(3-79)

where all reactances are calculated at the second-harmonic frequency.

At 60-cps primary frequency, this reduces to

TOE

* '

(3 - 80)

here, as in the previous analysis, the effects of higher harmonics

than the second have been neglected. This result is probably more
accurate than that given in Eq. (3-75) owing to the more reasonable

Note that

approximation in the analysis.

If the inductor of the II-section filter is replaced by a resistor, a practice
that is often used with low-current-drain power supplies, the ripple factor
Thus
is given by Eq. (3-79) with Xl replaced by R.

VZ

U _~

2.5

RRi

or

10*

C C 2 RR
1

REFERENCES
1.

For an analysis

of the operation, see Waidelich,

D.

L., Proc.

IRE,

29,

554

(1941).
2.

Corbyn, D.

3.

Cockroft,

4.

5.

Waidelich, D. L., Proc. IRE, 30, 534 (1942).

Mitchell, R. G., Wireless Eng., 22, 474 (1945).
Waidelich, D. L., Trans. AIEE, 60, 1161 (1941).
Schade, 0. H., Proc. IRE, 31, 341 (1943).
Overbeck, W. P., Proc. IRE, 27, 655 (1939).

6.

Arguimbau,

418 (1952).
Walton, Proc. Roy. Soc. (London), 136, 619

B., Electronic Eng., 24,

J. D.,

and E. T.

S.

(1932).

L.,

"Vacuum Tube

Circuits,"

John Wiley

&

Sons, Inc.,

New

York, 1948.

PROBLEMS
3-1. A type 5U4G is connected in a half-wave circuit to supply power to a
1,500-ohm load from a 350-volt rms source of potential.

RADIO ELECTRONICS

80

[Chap. 3

a. On a plate characteristic of the tube, plot the load line, and from this find
the dynamic curve.
b. Obtain a plot of the output-current waveshape for a sinusoidal applied

potential.
c. Estimate the value of the plate resistance r p from the static characteristic
Use the average of
at four different values of current (50, 100, 150, 200 ma).
these as the r p of the tube.
d. Plot on the curve in part b the value obtained from Eq. (3-5), and compare.
3-2. The two sections of a 6X5 diode are connected in parallel and supply

power to a 5,000-ohm load from a 325-volt rms source of potential. The effective
plate resistance of the parallel combination of diodes is approximately 125 ohms.
Calculate the following:
a.
b.
c.

d.
e.

/.
g.

The d-c load current.

The a-c current (rms).
The reading of a d-c voltmeter placed across the diode
The total input power to the plate circuit.
The efficiency of rectification.
The regulation from no load to the given load.
The ripple factor.

terminals.

6X5 tube supplies power to a 5,000-ohm load from a

325-0-325 transformer. Repeat Prob. 3-2 under these conditions.
3-4. Show that the input power to a rectifier using gas diodes may be expressed
in the form
3-3. Suppose that a

Pi

= Im.RL

-Eo/d-o

3-5.
gas diode for which the breakdown and maintaining potential is taken
to be 10 volts supplies power in a half-wave rectifier circuit to a 1,000-ohm load
from a 325-volt rms source. Calculate the following:
a. The d-c current through the load.

The a-c (rms) current through the load.

The reading of a d-c voltmeter placed across the diode.
d. The reading of an rms a-c voltmeter across the diode.
e. The power input to the plate circuit.
/. The efficiency of rectification.
g. The ripple factor.
3-6. The peak inverse plate potential rating of a 2X2/879 half-wave highvacuum rectifier is 12,500 volts. Calculate the maximum d-c potential possible to
a load, without exceeding the peak inverse potential, when such tubes are used in
b.

c.

a.
b.
c.

d.
e.

A half-wave circuit.
A full-wave circuit.
A full-wave bridge circuit.
A full-wave potential-doubling circuit.
A half-wave potential-doubling circuit.

Specify in each case the insulation strength of each filament transformer

positive terminal is grounded.
3-7. Analyze the operation of the potential-quadrupling circuit of Fig. 3-11.
/.

when the

Calculate:
o.
b.
c.

The maximum possible potential across each capacitor.

The peak inverse potential of each tube.
The required insulation strength of each filament transformer.
is planned to use a type 83 gas diode in a single-phase full-wave rectifier
with capacitor filter. The transformer potential is 350 volts rms to

3-8. It
circuit

RECTIFIERS AND FILTERS

81

The load consists of a 16-jif capacitor in parallel with a 2,500-ohm

The tube drop and the transformer resistance and leakage reactance

center tap.
resistor.

may

be neglected.
Calculate the cutout angle.
b. Determine the cutin point.
Should the type 83 tube be used? Comc. Calculate the peak tube current.
pare the peak current per plate with that given (1 amp) in the tube manual.
3-9. Given two 20-henry chokes and two 16-pf capacitors.
Calculate the
output potential and ripple factor under each of the following conditions:
o. The two chokes are connected in series with the load.
6. The two capacitors are connected across the load.
c. A single L-section filter, consisting of the two chokes in series and the two
a.

capacitors in parallel,

is

used.

filter, consisting of two sections, each of one choke and

one capacitor, is used.
The load is 2,000 ohms, and a 375-0-375 transformer is used in a full-wave
circuit.
Assume a 25-volt drop occurs across the tube.
3-10. A power supply has the form shown in the diagram.

d.

double L-section

Determine the approximate secondary potential of the power transformer.

cps; 400 cps?
3-11. In the power supply shown in the figure:
a. What is the output d-c potential?
b. What is the ripple potential in the output?

a.
6.

C.

What would be the ripple potential if the power frequency is 60

What is the minimum load current below which current cutout in the filter
What is the corresponding load potential?
Make allowance for the tube drop, but assume a perfect transformer.

occurs?
Note:
3-12.

shown

typical circuit for the high-potential supply for a cathode-ray tube

Estimate the output ripple potential.

in the diagram.

is

RADIO ELECTRONICS

[Chap. 3

The circuit shown in the accompanying diagram is to supply two

different

82
3-13.

potentials.

If the

transformer

is

375-0-375,

what

are the output potentials?

CHAPTER

VACUUM TRIODES AS CIRCUIT ELEMENTS

The analysis of the behavior of a vacuum-tube
be accomplished by two different methods, both of which are
to be examined in some detail.
In one method, use is made of the static
characteristics of the tube.
The second method achieves two forms. In
one, the tube is replaced by an equivalent potential source and a series
resistance.
The source potential depends on the amplitude of the input
signal, the internal resistance depending on the tube that is used.
In
the other form, the tube is replaced by an equivalent current source and
a shunting conductance. The magnitude of the source current depends
on the amplitude of the input signal, the shunting conductance depending on the tube that is used.
Although the second methods assume that the tube characteristics are
linear, the ultimate analyses allow a very clear insight into the operation
of the circuit.
Because of this, the equivalent-circuit methods of analysis are usually considerably more important than the method involving
the tube characteristics. Moreover, it is possible to estimate the inaccuracies in the method, when large signal operation is involved.
Actually, the form of analysis is dictated in large measure by the bias of the
It must
tube, the signal amplitudes, and the characteristics of the load.
be noted that the equivalent-circuit techniques provide no means for
establishing d-c bias and current levels, and direct recourse to the static
4-1. Introduction.

circuit

may

characteristics

is

necessary

for

this purpose.

An

introduction of the methods

be made in terms of the operation of the triode, but these will

later be extended to the operation
will

of other

4-2.

types of tubes.

Symbols and Terminology.

The

The simple triode amplifier is illus-

basi

circuit

of a triode

Before proit is necessary to discuss the meaning of the

symbols and the general terminology of vacuum-tube circuits.
trated in Fig. 4-1.

ceeding with the analysis,

83

RADIO ELECTRONICS

84

[Chap. 4

The input circuit of the amplifier usually refers to all the elements of
the circuit that exists between the grid and the cathode terminals of the
Similarly, the output, or plate, circuit usually refers to the eletube.
ments that are connected between the cathode and the plate terminals
In the circuit illustrated, the input circuit comprises the
of the tube.
input potential source d, the grid resistor R g and the bias battery E cc
The plate circuit consists of the load resistor R and the plate-supply
.

battery Ebb- In many applications, the input signal e x is a sinusoidally

varying potential, although the waveshape may be nonsinusoidal, and is
frequently very carefully chosen for a particular application.
A variety of potentials, both d-c and varying, are involved simultaneously in a vacuum-tube circuit, making it necessary that a precise method
In what follows, lowercase letof labeling such quantities be established.
ters will be used to designate instantaneous values and capital letters

denote either d-c or rms values of sinusoids. The subscripts c and g

and the subscripts b and p will refer to the
Examples of the notation follow
plate circuit.
will

will refer to the grid circuit,

E =

d-c grid, or C bias

= d-c plate supply, or B supply
= rms value of a-c input excitation

cc

En,

Ei

potential

if

this excitation is

sinusoidal

E =
d =

rms value of a-c output potential for a sinusoidal output

instantaneous input signal potential; measured with respect to
the input terminals

ec

eg

instantaneous potential that appears between grid and cathode

of tube
instantaneous-signal component that appears between grid and

d =

cathode of tube
instantaneous signal potential that appears across output

ment

ele-

of circuit

iv

instantaneous potential between plate and cathode of tube

instantaneous-signal component of plate current; positive in
direction from cathode to plate through load

ib

instantaneous total plate current; positive in direction from

cathode to plate through load

eb

I =
E =
b
b

average or d-c current in plate circuit

average or d-c potential from plate to cathode

Figure 4-1 illustrates the reference positive-potential polarities and the

reference direction of current.
tial

These reference conditions are an essen-

part of the diagram.

As a

specific illustration of

the notation, suppose that the input signal

VACUUM TRIODES

Sec. 4-3]

AS CIRCUIT

potential to the amplifier of Fig. 4-1

ex

Then the instantaneous grid-cathode

= Ecc

ee

Eg sin wt

potential

+ V2 E

Circuits will be discussed in which

85

is

\/2

ELEMENTS

is

sin ut

no such simple relation between grid

driving signal and grid-cathode potential exists, owing to an involved

among the tube elements.

Refer to Fig. 4-1, and suppose that the grid
Owing to the d-c sources Ecc and Em, it will be supa current in the plate circuit. This is true only if

interconnection of circuit elements

4-3. Graphical Analysis.

input signal e x = 0.
posed that there is
the plate supply Ebb and the grid
supply Ecc are properly chosen. The
value of this current
In fact,
graphically.

may
it is

be found
essential

that a graphical solution be used.

This follows from the fact that the

plate circuit of Fig. 4-1 yields the
relation

Eb
eb

However,
sufficient

this

Ebb

ibRl

one equation

is

ec

2?&&,

Ecc.

is

given by the plate characterisand the plate

solution of Eq. (2-1)

characteristics will yield the desired current.

drawing Eq.

(4-1)

in Fig. 4-2.

The

two unknown quanti-

eb

The simultaneous

the triode.

since there are

and tj.
between e& and %

the expression,

second relation

tics of

not

to determine the current

corresponding to the potential

ties in

Ebb

Fig. 4-2. The operating point Q is

located at the intersection of the load
line and the plate characteristic for

(4-1)

on the plate

This

characteristics, in the

line that passes

4 =

is

accomplished by

manner

illustrated

through the points

Ebb

Ebb
(4-2)

Ri

known

It is obviously independent of the tube chardepends only upon elements external to the tube. The
intersection of this line with the curve for e c = E cc is called the operating,
The grid-bias supply Ecc is usually such as to
or quiescent, point Q.
maintain the grid negative relative to the cathode. The Q current in
is

as the load line.

acteristics, for it

the external circuit

is

Eb

is

I b and the corresponding plate-cathode potential

,

RADIO ELECTRONICS

86

Fig. 4-3.

The output current and

potential

waveforms

Suppose that the grid-cathode potential

ec

= E cc

+ V2 Eg

[Chap. 4

for a given input grid signal.

is

sin

co<

The maximum and minimum values of e c will be E cc

E cc \/2 Eg respectively. The plate current and the
,

will

The

>

\/ 2

and

plate potential

vary about the values

of

I b and

graphical construction of Fig.

shows the

details of the variations.

b.

4-3

The

and ib for any given value of e c

are obtained from the intersection of the
load line and the ib -e b curve for the specified
values of

ec

The

eb

points

a', b', c', etc.,

of the out-

put current and the points a", b"

of the output-potential

c", etc.,

wave correspond,

respectively, to the points A, B, C,

the input-grid-signal waveform.
It

is

instructive to

show the

etc., of

several

waveshapes in their proper phase relation.

This is done in Fig. 4-4. It should be
noted in particular that the variations
about the quiescent values have been
labeled.
Fig. 4-4.

The

grid input

The

quantities so labeled are

wave-

shape and the corresponding output current and potential waveshapes.

ep
ip

=
=

eb
ib

- E
Ib

(4-3)

These quantities give a measure of the amplification property of the

it is a direct measure of the a-c output variations for a

amplifier, as

given a-c input variation.

VACUUM TRIODES

Sec. 4-4]

The curves
the current

If

of Fig. 4-4 indicate the following

87

very significant results:

sinusoidal, then i p and e p are 180 deg out of phase

Also, the grid driving potential e g and the plate cur-

are in phase with each other.

ip

positive signal

ELEMENTS

i p is

with each other.

rent

AS CIRCUIT

is

This simply states that, when a

applied to the grid,

More-

the tube current increases.

with an increased current in

the plate circuit, the potential of the
over,

plate falls.

curve of the intersection of the

load line with the static-characteristic

curves, which

current

specified

is

a measure of the

as a function of e c for the

hb

and load

Ri, is

known

as the

impor-

"dynamic" characteristic of the tube

circuit and yields directly the outtant.

It

is

put current for a given input signal.

The construction

is

Fig. 4-5.

The dynamic curve and

in determining the

its

use

output waveshape for

a given input signal.

directly related

and is given in Fig. 4-5

The correspondon both curves are similarly marked.
4-4. Potential-source Equivalent Representation of a Triode.
In most
electron-tube problems, one is interested in the "a-c response" of the
tube, rather than in the total instantaneous variation of the potentials
and current. That is, the values of i p and e p for a given e g are ordinarily
desired.
It is possible to deduce this information directly from the static
characteristics of the tube, as discussed in Sec. 4-3.
Of course, if the
potential variations are small, the accuracy of the results will be poor,
as small changes cannot be read with any degree of accuracy from the
curves.
Moreover, the process may become quite tedious to perform,
particularly for a reactive load, since the load curve is no longer a straight
to the construction of Fig. 4-3

ing points

line.

For small variations in the input potential, the tube parameters n,

and g m will remain substantially constant over the operating range.
Under such conditions, it will be shown that the graphical solution may
be replaced by an analytic one. Actually, the equivalent analytical solution depends on the constancy of the tube parameters, rather than on the
magnitude of the signals involved. The analytic method may be used
even under large signal operation provided that the tube parameters
rp

remain substantially constant.

Reference is made to Eqs. (2-17) and (2-18), which specify the variation in current about the quiescent point in terms of the variation resultThis expression,
ing from the changes in the plate and grid potentials.

RADIO ELECTRONICS

88

which

[Chap. 4

is

only approximate. It specifies only the first two terms of the Taylor
expansion of the function ib ib (eb ,ec). ^ n * ne general case, the result is

is

more informative form

is

by

possible,

relating the higher-order terms

in the expansion explicitly as variations in the plate resistance r p or in the

mutual conductance gm
Consider the third term in the expansion. By
.

combining this with Eq.

(1-18), there results

When the tube parameters are sensibly constant over the operating
range of Aeb and Aec Eq. (4-4) is an adequate representation of the variation.
This may be written, by Eqs. (2-17) and (2-18), as
,

Aib

+ gm Ae

Aeb

(4-7)

But as the changes about the quiescent values

Aib
Aec
Aeb

=
=
=

ip

iP
Ecc = e g
= ep

(4-8)

~\~

(4~9)

ib ~~

ec
e&

are, respectively,

lb

then Eq. (4-7) becomes

rp

@p

gm&Q

or
ep

i*e g

ip r p

(4-10)

This expression shows that the potential e p comprises two components;

one is an equivalent generated emf, or electromotance, which is /x times
as large as the grid-cathode potential e g and the second is a potential
difference across the tube resistance rp resulting from the current ip
through it.
Equation (4-10) may be used as the basis for drawing an equivalent
network for the tube. This is done in Fig. 4-6. Observe that the plate
circuit of the tube is replaced by a fictitious potential source with an
electromotance fieg and an internal resistance rp
Two points are empha,

VACUUM TRIODES

Sec. 4-5]

AS CIRCUIT

ELEMENTS

89

and reference current

direc-

tion are essential parts of the equivalent-network representation.

Sec-

sized.

First, the reference positive polarities

ond, no d-c quantities appear on the diagram, since the equivalent-circuit

representation applies only for changes about the

Fig. 4-6.

The vacuum

triode

4-5. Current-source

and

its a-c

point.

potential-source equivalent representation.

Equivalent Representation of a Triode.

The

current-source equivalent representation of a triode replaces the tube by

a constant-current source shunted by a conductance, instead of the poten-

source with a series resistance. The form of the result

obtained by rearranging Eq. (4-9) into the form

tial

gmeg

ip

-^

is

easily

(4-11)

This expression shows that the current i p comprises two components; one
is a generated current which is g m as large as the grid-cathode potential
e g and the second is a current through the shunting tube resistance r v
because of the potential e p across it.
,

Fig. 4-7.

The vacuum

triode

and

its

current-source equivalent representation.

Equation (4-11) may be used as the basis for drawing the equivalent
network of the tube. This is done in Fig. 4-7. Observe that the plate
circuit of the tube is replaced by a current source with generated current
Note also that the reference positive
g m e g and a shunting resistance r p
polarities and the reference current direction are essential parts of the
.

equivalent-network representation, as before.

The reader will observe a striking parallel between the discussion in
this section and that in the previous section, except that one section confines itself to a potential source and a series internal resistance, whereas
the present section confines itself to a current source and a shunt con-

RADIO ELECTRONICS

90

[Chap. 4

ductance (or resistance). This is actually part of a larger pattern which

*
exists in general network analysis and which is given the name of duality.
If the varying quantities are sinusoidally varying ones, and this will
ordinarily be assumed unless otherwise explicitly stated, the analysis proceeds most easily in terms of the phasors (sinors and complex-number
representation for impedances) of elementary a-c circuit theory. The circuit notation and certain elements of general network analysis which will

be found useful in this text are discussed in Appendix A. The reader is

urged to refer to this appendix before proceeding. For sinusoidally varying signals, therefore, the tube potentials are expressed in terms of the
symbols E, Ej,, and I p where these boldface symbols are employed to
denote sinor quantities, a symbolism that is adopted in this text. For
sinusoidally varying quantities, the vacuum triode is given in the accompanying two equivalent forms (see Fig. 4-8).
,

(a)

(b)

Fig. 4-8. (a) The potential-source and (6) the current-source equivalent circuits of
the triode for sinusoidally varying quantities.

4-6. Linear Analysis of Electron-tube Circuits.

As discussed

in Sec.

determine the a-c response of a vacuum-tube circuit,

when the parameters of the tube remain substantially constant over the
range of operation, by replacing the tube by either its potential-source
equivalent or its current-source equivalent, and then employing the techniques of general network analysis in the complete analysis of the circuit.
The technique of drawing the equivalent network of any tube circuit
is a straightforward process, although care must be exercised in carrying
out the details. To avoid error, the following simple rules will be found
4-4, it is possible to

helpful
1.

2.

Draw
Mark

the actual diagram neatly.

the points G, P, and

K on this diagram.

as the start of the equivalent circuit.

Locate these points

Maintain the same relative posi-

tion as in the original circuit.

3.

Between points

and

K include either the

potential-source repre-

sentation of Fig. 4-8a or the current-source representation of Fig. 4-8&,

depending on the preferred form.
* Some amplification of the principles of duality is given in Appendix A.
For
an extensive discussion see W. LePage and S. Seely, "General Network Analysis,"
McGraw-Hill Book Company, Inc., New York, 1952.

Sec. 4-6]

VACUUM TRIODES

AS CIRCUIT

ELEMENTS

91

Transfer

all circuit elements from the actual circuit to the equivawithout altering the relative positions of these elements.
5. Replace each d-c source by its internal resistance, if any.
Several examples will be given to illustrate the foregoing techniques.

4.

lent circuit,

Example 1. Calculate the output potential E 2 of the simple amplifier circuit

given in Fig. 4-9. Note that the technique of drawing the equivalent circuit is in
accord with the rules given above.

+
E 2

Fig. 4-9. A simple amplifier circuit and

equivalent of the tube has been used.

its

a-c equivalent.

direct application of the Kirchhoff potential law,

algebraic

sum

and

of the potential rises

The

potential-source

which requires that the

completing a closed loop must be

falls in

zero, yields directly

ME
Note

I(r p

+R =

(4-12)

t)

also that

Ex

(4-13)

It therefore follows that

I

and the output potential

E
^

is

IRi

= -

The

ratio of the

(4-14)

R,

given by

=
rv

+R

1+ r

output to input potentials

Therefore

/Ei

Ei

(4-15)

r /R,

is

the amplification, or gain,

of the amplifier.

eT

(4-16)
r p /R,

It is of interest to plot this expression,

which has the form given in Fig. 4-10.

It should be observed from this dia3 4
5
6
7
8
9 10
I
gram that gains which approach p are
quite feasible with moderate Ri/r v raFig. 4-10. The gain of the amplifier of
tios.
For values of the ratio K//i
Fig. 4-9 as a function of load resistance.
which are nearly unity, it is required
that Ri be large.
In this case, however, for the tube to be operated at the proper
d-c quiescent levels, the source i? must be large, and the heating of the tube or
load resistor may become unduly high.

RADIO ELECTRONICS

92

Example

2.

Calculate the output potential

E2

[Chap. 4

of the amplifier circuit given in

Fig. 4-11.

R3

Rl G

Fig. 4-11. A simple amplifier circuit and

equivalent has been used.

The

a-c equivalent.

its

current-source

The Kirchhoff current law is applied to the two junctions. This law requires
that the algebraic sum of the currents at any junction must be zero.
The equations are

E
R

Ej

Junction

Eo

Ea

R2

Ri

Ei

=
(4-17)

Junction P:

E2

E2

rv

Ri

Collecting terms gives two equations,

Ei

E2

Ri

Rz

'

(ffl

i?2

^3^
(4-18)

Combine the equations

Ei
Ri

The output

+E

to get,

(gi

J_
Ra

potential

R2

+ r + b)
v
=

(4-19)

I/-R3
l/Rz

is

-(g m
l_

gm
R~z

E,,

+
Qm

E2 =

The

by eliminating

R)
3

{ri

potential gain of this amplifier

1/B,)(1/B0
Ei

Ri

+R

)(r,

gm

R)R +
s

+ r)
rP

(4-20)

is

-(g m - \/R 3 ){\ /R

+
{fti

R2

(4-21)

+ R )(r +
s

rv

+ r)

It should be noted that if the resistors Ri and R 3 were absent, the circuit will
reduce to that of Example 1. Thus, by setting Ri = and R 3 = o in Eq. (4-21),
the result reduces to that given in Eq. (4-16). Because of the presence of Ri
and Rz, the gain of this amplifier stage is lower than that given in Example 1.

Sec. 4-7]

VACUUM TRIODES

AS CIRCUIT

ELEMENTS

93

discussion of the effects of introducing these resistors, which have introduced

is known as negative feedback, will be deferred until Chap 7.

what

Measurement

As several additional illusof Triode Coefficients.

methods of analysis just discussed, the circuits for obtainIt should be
ing the values of pt, r p and g m of a triode will be analyzed.
recalled that the triode coefficients, first discussed in Sec. 2-13, were shown
4-7.

trations of the

to be related to the slope of the static-characteristic curves, according to

However, the accuracy with which these quantities

Figs. 2-14 to 2-16.
can be measured in this way is not high. Not only do the methods now
to be discussed yield results which are made under dynamic conditions,
but the results are usually more accurate than those deduced from the
static characteristics.

^earphones

Ecc ,r~

Fig. 4-12. The Miller bridge and its equivalent circuit for determining the amplification factor of a triode under operating conditions.

amplification factor n is readily determined by means of the cirThe operations involved in balancing the bridge
cuit given in Fig. 4-12.

The

consist simply in varying

Ri and

heard in the earphones.

When

0.

Then the

R 2 until

no

signal

from the

oscillator is

this condition prevails, the plate current

potential E

LRi.

-!xE g

1R 2 =

By

applying Kirchhoff's law to

the plate circuit,

or

+nE = IR =
2

It follows

from

nIR!

this that

= R2

(4-22)

Rl

This measurement may be effected for any desired d-c current in the tube
simply by adjusting the grid bias E cc
The transconductance g m is measured by means of a bridge circuit that
The addition of a resistor R 3
is a slight modification of Fig. 4-12.
.

between the plate and cathode makes this measurement possible. The
schematic and equivalent circuits of this bridge network are given in
The measurement is accomplished by adjusting the resistors
Fig. 4-13.
until no signal is heard in the earphones.

RADIO ELECTRONICS

94

By

[Chap. 4

applying Kirchhoff's law to the several meshes, there results

+ l p rp
But the potential Ep

nE g =

is

Then
l P (Rz

Also,

it

rp )

iiLRi

(4-23)

follows that

IR2

IpRz

or

IR 2 =

The

ratio of Eq. (4-23) to Eq. (4-24)

Rs

(4-24)

is

Ri

rp

Rs
from which
(4-25)

Although this bridge may be used to evaluate r p the result would be

dependent on the measurement of n. If, however, R 1 is chosen in such
,

K MydC-
1

'

vV/^

Fig. 4-13. The Miller bridge for determining the transconductance of a triode under
operating conditions.

way

that

uRi/Rz

>S> 1,

then approximately
rP

R3R1
R2

or

R2
RzRi

9m

(4-26)

The plate resistance r p of the tube can be measured directly by incorporating the plate circuit of the tube as the fourth arm of a Wheatstone
bridge, as

shown

in Fig. 4-14.

When

the bridge

R R
2

Ri

is

balanced,

(4-27)

VACUUM TRIODES

Sec. 4-8]

The above

AS CIRCUIT

ELEMENTS

95

do not yield perfect balance owing to the capacitive

it is sometimes necessary to provide a means for
balancing these effects. Basically, howeffects of

circuits

the tube, and

ever, the circuits are those given.

4-8.

Tube.

Harmonic Generation in a
The equivalent-linear-circuit

an
adequate solution of an amplifier circuit
analysis of Sec. 4-4 usually permits

when the

limitations of the

not exceeded or

if

method are

relatively slight dif-

ferences are considered of

no import-

There are occasions when it is

desirable to examine critically the effects
ance.

of

Fig. 4-14. A Wheatstone bridge for

determining the plate resistance of a
triode under operating conditions.

the assumptions.

of linear operation, which is implied in Eq. (4-4) and

which assumed that higher-order terms in the Taylor expansion of the
This assumption,
current ib(e b ,e^) are negligible, is not always valid.
which allowed the graphical solution of Fig. 4-5 to be replaced by the
analytical one of Fig. 4-6, requires that the dynamic characteristic of
Actually, the
the amplifier circuit be linear over the range of operation.
dynamic characteristic is not linear in general but contains a slight
This nonlinear characteristic arises because the (u,eb) static
curvature.
characteristics (see Fig. 4-3) are not equidistant lines for constant
The effect of this nonlinear
e c intervals over the range of operation.
dynamic characteristic is a nonsinusoidal output waveshape when the
input wave is sinusoidal. Such an effect is known as nonlinear, or ampli-

The assumption

tude, distortion.

obtain a measure of the degree of nonlinearity that

from the existence of the nonlinear dynamic curve. To do this,
it is observed that the dynamic curve with respect to the Q point may be
expressed by a power series of the form
It is possible to

results

ip

aie

a 2 el

a s e%

(4-28)

terms in this series vanish except the first, then the linear
concept result. It will be found
assumptions
conditions, may be adequately
under
normal
when
operated
that triodes,
When a
terms
in the expansion.
first
two
retaining
the
by
expressed
instantaneous
gridsignal
that
the
large
with
such
a
operated
triode is
with
such
triode
is
operated
positive,
or
if
the
becomes
potential
cathode
Clearly,

if

all

of the equivalent-circuit

a bias that the very curved portions of the plate characteristics must
be employed, more than two terms must be retained in the expansion.

found that the parabolic approximation is not adequate

the
dynamic curve of a tetrode or a pentode under normal
to represent
Likewise,

it is

RADIO ELECTRONICS

96
operating conditions.
ture or

if

the dynamic curve contains an extreme curvaover an extreme range, it is sometimes found

If

the operation

is

preferable to devise special

methods

cial

methods

For example, such spea tuned class C amplifier in

of analysis.

be used in the analysis

will

[Chap. 4

of

Chap. 10.
Suppose that the dynamic curve may be represented as in Eq. (4-28),
and consider that the input wave is a simple cosine function of time, of
the form
eg

By combining

= E gm

cos wt

(4-29)

with Eq. (4-28) and expanding the higher-

this expression

order powers of the cosine that appear in the resulting series, the result

may

be shown to have the form

ib

h+

Bo

+B

cos wt

+B

cos

2oit

+B

assumed that the excitation potential

If it is

is

cos

3o><

(4-30)

a sine function of the time

instead of the cosine form chosen, the resulting Fourier series representing the output current will be found to contain odd sine components and

even cosine components.

The construction for obtaining the plate-current values to

five-point schedule for determining the Fourier coefficients.
Fig. 4-15.

A
B\,

number

2,

of different

One

etc.

of

reference to Fig. 4-15.

terms,

Bi,

methods

the more

in the

exist for obtaining the coefficients

common methods

is

best discussed

jBo,

by

be assumed for convenience that only five

In order
exist in the resulting Fourier series.

It will

B Bh B
2,

be used

4,

to evaluate these five coefficients, the values of the current at five differ-

ent values of

eg

are required.

The values chosen

are 7 mo %,

and Imin and correspond, respectively, to the following values

maximum

positive value; one-half the

one-half the

maximum

It is evident

at the angles

from the

lb,

of e g

I-h,
the

maximum positive value; zero;

maximum negative value.

negative value; the

figure that the currents are those chosen as

shown

VACUUM TRIODES

Sec. 4^8]

=
_
~

wt

AS CIRCUIT

4 = la

3
IT

ib

ih

I-.y,

2x
cot

By combining these

with Eq. (4-30),

results

(4-31)

<j)t

five

The simultaneous

unknowns are obtained.

97

1 max

lb
IT

ELEMENTS

equations containing five

solution of these equations

yields

= ^ (Imzx + 27)4 + 27_j^ + Zmin) /(,

= M (7m ax + Ift I-Vz Imin)
B = K (7m - 2Ib + 7 mia
Bz = M (7m ax 27^ + 27_1.<; 7min )
Ba, = K2 (7m .x - 47 H + 67i, - 47_^ + 7min
Bo
Bl
2

The percentage
7) 2

|-

(4-32)

harmonic distortion

of

X 100%

7) 3

is

defined as

X 100%

7>4

iii

>i

I#1

X 100%

(4-33)

where 7),, (s = 2, 3, 4,
.) represents the per cent distortion of the
sth harmonic and the total distortion is defined as
.

D =

Vm+ Dl + >!+

(4-34)

For the case where a three-point schedule is sufficient, and, as already

indicated, this would apply for a triode under normal operating conditions, the analysis yields the expressions

Bl

B =
2

}/2

(I max

Bo =

Imin)

27 6

7min )

(4-35)

PROBLEMS
4-1. A 6C5 triode is used in the circuit of Fig. 4-1, the plate characteristics of
which are given in Appendix B.
a. With Ebb = 300 volts, Ecc = 8 volts, Ri = 20 kilohms, draw the load line,
and locate the operating point. Plot the dynamic characteristic of the circuit.
6. If ei = 6 sin 10,000t, determine the output current graphically, and plot the

curve as a function of

oit.

From

these curves, determine and plot the instantaneous plate potential

for the same interval as in part b.
Check the phase relation between a-c components of grid potential, plate current, and plate potential.
4-2. The characteristics of a given triode may be represented by the expression
c.

ib

It is to

8.8

10- 3 (e

be operated at a plate potential

E = -9
c

volts.

16e c ) 16

Eb =

ma

250 volts and a grid-bias potential

98

RADIO ELECTRONICS

[Chap. 4

Calculate the plate resistance of the tube.

If this tube is used in the circuit of Fig. 4-1 with a load resistance
Ri = 10 kilohms, determine the plate supply potential necessary for the tube
to be operating under the specified conditions.
c. Suppose that the grid driving source applies a potential e-i = 8 sin wt to the
Determine the a-c potential across the load resistor.
grid.
4-3. Draw the potential-source equivalent circuit of the electron-tube circuits
a.

b.

in the

accompanying diagrams.

VACUUM TRIODES

AS CIRCUIT

ELEMENTS

99

4-4. A type 6A3 triode is used in an amplifier circuit to supply power to a

3,000-ohm resistor. In this circuit E cc = 45 volts, Eu, = 350 volts. A 45-volt
peak a-c signal is applied to the grid.
o. Plot the dynamic curve of the tube.
b. Assume that only the fundamental and a second harmonic exist in the output.
Determine the magnitudes of each.
c. Plot a curve showing the output current for the sinusoidal input.
On the
same sheet, plot the corresponding results from the calculations in part 6.
4-5. It is possible to obtain a five-point schedule for determining the coefficients Bo, Bi, Bi, Bi, B t by almost any sensible choice of angle.
Determine the
five-point schedule for determining the coefficients B in terms of 1^, /0.707, h,
^0.707, -^min'

CHAPTER

BASIC AMPLIFIER PRINCIPLES

Tubes and Amplifiers. The classification of an
somewhat involved, owing to the fact that a complete

5-1. Classification of

amplifier

is

usually

classification

must include information about the tubes that are

used, the

conditions of the bias, the character of the circuit elements connected to

the tubes, the function of the

circuit,

and the range

tain of these factors will be discussed here, but

many

Cerbe deferred 'for

of operation.
will

later discussion.

Apart from the wide variety of vacuum tubes of the diode, triode,
beam, hexode, heptode, and multiunit types and the
varied power capacities of each type, it is possible to classify the tubes
according to their principal applications. Tubes may be classified
tetrode, pentode,

roughly into

five groups, viz., potential-amplifier tubes, power-amplifier

tubes, current-amplifier tubes, general-purpose tubes,

and special-purpose

tubes.
1.

Potential-amplifier tubes have a relatively high amplification factor

and are used where the primary consideration is one of high potential
gain.
Such tubes usually operate into high impedance loads, either
tuned or untuned.
2.

Power-amplifier tubes are those which have relatively low values of

amplification factor and fairly low values of plate resistance.

They

are

capable of controlling appreciable currents at reasonably high plate

potentials.
3.

Current-amplifier tubes are those which are designed to give a large

change

of plate current for a small grid potential; i.e., they possess a high
transconductance. These tubes may be required to carry fairly large
plate currents.
Such tubes find application as both potential and power

depending on the tube capacity.

General-purpose amplifier tubes are those whose characteristics are
intermediate between the potential- and the power-amplifier tubes.
They must have a reasonably high amplification factor and yet must be
able to supply some power.
5. Special-purpose tubes include a wide variety of types.
The hexode,
heptode, and multiunit tubes are of this type.
amplifiers,
4.

100

BASIC AMPLIFIER PRINCIPLES

Sec. 5-1]

101

Amplifiers are classified according to their frequency range, the method

For example,
of tube operation, and the method of interstage coupling.

may

be classed as direct-coupled amplifiers, audio-frequency (a-f)

amplifiers, video amplifiers, or tuned r-f amplifiers if some indication of
Also, the position of the quiescent
the frequency of operation is desired.
characteristic
that is being used will
tube
point and the extent of the
will
specify whether the
operation.
This
determine the method of tube
or class C.
These
B,
class
AB,
class
operated
in
class
A,
tube is being
they

definitions are illustrated graphically in Fig. 5-1.

1.

A class A

amplifier

is

an amplifier in which the grid bias and the a-c

grid potentials are such that plate current flows in the tube at

all

times.

<5

Fig. 5-1. Amplifier classification in terms of the position of the quiescent point of
the tube.
2.

class

AB

amplifier

is

one in which the grid bias and the a-c grid

potentials are such that plate current flows in the tube for appreciably

more than
3.

half but less

class

than the entire electrical cycle.

is one in which the grid bias

amplifier

is

approximately

equal to the cutoff value of the tube, so that the plate current is approximately zero when no exciting grid potential is applied, and such that
plate current flows for approximately one-half of each cycle
grid potential
4.

class

is

when an

a-c

applied.

amplifier

is

one in which the grid bias

is

appreciably

greater than the cutoff value, so that the plate current in each tube
zero

when no

a-c grid potential

is

applied,

flows for appreciably less than one-half of

potential

is

is

and such that plate current

each cycle when an a-c grid

applied.

To indicate that

grid current does not flow during

cycle, the subscript

class identification.

any part

of the input

frequently added to the letter or letters of the

The subscript 2 is added to denote that grid current

is

RADIO ELECTRONICS

102

[Chap. 5

does flow during some part of the cycle.

For example, the designation
class ABi indicates that the amplifier operates under class AB conditions
and that no grid current flows during any part of the input cycle.
Potential amplifiers, whether tuned or untuned, generally operate in

Low-power audio amplifiers

with special connections, under class
class A.

r-f

power amplifiers are operated

conditions.

AB

either

Oscillators usually operate

detailed discussion

When

may

a tube

is

be operated under
or class

class

conditions.

A and

Tuned

under class B or under class

under class C conditions.

deferred until the appropriate points in the text.

used essentially as a switch, no classification is ordinarily
is

specified.

The

5-2. Distortion in Amplifiers.

to the grid of an ideal class

application of a sinusoidal signal

amplifier will be accompanied

by a

sinus-

Frequently the output waveform is not an exact

replica of the input-signal waveform because of distortion that results
either within the tube or from the influence of the associated circuit.
The distortions that may exist either separately or simultaneously are
nonlinear distortion, frequency distortion, and delay distortion. These
oidal output wave.

are defined as follows:

1. Nonlinear distortion is that form of distortion which occurs when
the ratio of potential to current is a function of the magnitude of either.

Frequency distortion is that form of distortion in which the change

magnitudes of the different frequency components of a
wave, provided that the change is not caused by nonlinear distortion.
3. Delay distortion is that form of distortion which occurs when the
phase angle of the transfer impedance with respect to two chosen pairs of
2.

is

in the relative

terminals

is

not linear with frequency within a desired range, the time of

transmission, or delay, varying with frequency in that range.

In accordance with definition

1,

nonlinear distortion results

when new

frequencies appear in the output which are not present in the input signal.

These new frequencies arise from the existence of a nonlinear dynamic

curve and were discussed in Sec. 4-8.

Frequency distortion arises when the components of different frequency are amplified by different amounts. This distortion is usually
a function of the character of the circuits associated with the amplifier.
If the gain vs. frequency characteristic of the amplifier is not a horizontal
straight line over the range of frequencies under consideration, the circuit
is

said to exhibit frequency distortion over this range.

Delay distortion, also called phase-shift distortion, results from the fact
that the phase shift of waves of different frequency in the amplifier is
different.

Such distortion

is

type, since delay distortion

not of importance in amplifiers of the a-f

not perceptible to the ear. It is very

is

objectionable in systems that depend on waveshape for their operation,

BASIC AMPLIFIER PRINCIPLES

Sec. 5-3]

example, in television or facsimile systems.

as, for

If

103

the phase shift

is

proportional to the frequency, a time delay will occur although no distortion

is

amplifier

periodic

d = Eml
If

To see this, suppose that the input

may be expressed analytically by

introduced.
is

the gain

signal to the

and

sin (ut

+ Em2 sin

X)

(2wt

2)

(5-1)

K is constant in magnitude but possesses a phase shift that is

proportional to the frequency, the output will be of the form

KE ml

e2

sin (at

0!

+ KEm2 sin

(2cot

d2

2$)

This output potential has the same waveshape as the input signal, but a
time delay between these two waves exists. By writing

cot'

U)t

\j/

then
e2

This
to a

is

KEml sin

(at'

+ KEm2

simply the expression given by Eq.

new time

scale

t'.

Delay

vacuum

(3-1),

<9 2 )

except that

it is

(5-2)

referred

distortion, like frequency distortion, arises

from the frequency characteristics

It is

sin (2a?

of

the circuit associated with the

tube.

not possible to achieve such a linear phase characteristic with

it may be approximated with special phase-equal-

simple networks, but

izing networks.

Sensitivity.
In many problems where two
be compared, it is found very convenient to compare
The
the relative powers on a logarithmic rather than on a direct scale.
A decibel, which is abbreunit of this logarithmic scale is called the bel.
viated db, is
bel.
By definition, the logarithm to the base 10 of the
bels; that is,
ratio of two powers is

5-3.

power

The Decibel; Power

levels are to

Number

of bels

p
logio p1

or

Number

of

db = 10

(5-3)

logio

should be emphasized that the bel or the decibel denotes a power ratio.
Consequently the specification of a certain power in decibels is meaningIn comless unless a reference level is implied or is explicitly specified.
munication applications, it is usual practice to specify 6 mw as the zero
It

However, any power may be designated as the zero

any particular problem.
Suppose that these considerations are applied to a power amplifier,
with Pi the output power and Pi the input power. This assumes that
reference level.

reference level in

the input circuit to the amplifier absorbs power.

If

the grid circuit does

RADIO ELECTRONICS

104

[Chap. 5

not absorb an appreciable power, then the term decibel gain of the amplimeans nothing. Under such conditions, it is customary to speak of
power sensitivity, which is defined as the ratio of the power output to the
square of the input signal potential. Thus
fier

Power
where
If

is

sensitivity

mhos

the power output in watts and

Z?i is

(5-4)

the input signal rms volts.

the input and output impedances are equal resistances, then

P = E\/R

and P x = E\/R, where E 2 and Ei are the output and input

Under this condition, Eq. (5-3) reduces to

potentials.

Number

db = 20 log 10

of

(5-5)

In general, the input and output resistances are not equal. Despite
this, this expression is adopted as a convenient definition of the decibel
potential gain of an amplifier.
It is essential, however, when the gain of
an amplifier is discussed, that it be clearly stated whether one is referring
to potential gain or power gain, as these two figures will be different, in
general.

Many of the considerations

by

of the foregoing sections are best illustrated

several examples.

Example

1.

Calculate the gain of the grounded-grid amplifier circuit of Fig.

5-2.

Fig. 5-2. Schematic and equivalent circuits of a grounded-grid amplifier.

Solution.

The equivalent circuit

and is that shown in

rules of Sec. 4-6

of the amplifier is

Fig. 5-26.

The

drawn according

to the

application of the Kirchhoff

potential law to the equivalent circuit yields

E,

ixE

\(r v

+R +R =
g

{)

(5-6)

Also from the diagram

= E, +

IR,

(5-7)

BASIC AMPLIFIER PRINCIPLES

Sec. 5-3]

Combine the two equations to

The

plate current

is

find

- M (-B,+Ifi,)

E,

I(r,

R,

Ri)

then given by
E,( M
r

The corresponding output

potential

Eo

IRi

(jjl

1)

l)R s

0*

+R

(5-8)
t

is

=
(ji

The

105

l)fl(Ei

\)R,

rp

(5-9)

R,

which

gain, or potential amplification of this amplifier,

output to the input potential,

E2

is

rP

+R
+

(5-10)
t

given as the ratio Ei/I and

Z,

the ratio of the

Ri

R +
The input impedance

is

is

Ei

rP

+ Ri
+ +

is

Ra

(5-11)

This means, of course, that heavy loading of

which, for small R g is quite small.
the driving source may exist if R g is small.
It is of interest to compare the results of this example with Example 1 of Sec.
4-6.
Observe that it is possible to apply the signal either in the grid circuit or
in the cathode circuit and still achieve operation of the tube, although the input
,

impedance

Example
in

an

is

different in the

2.

two

cases.

which n = 20, r v = 7,700 ohms is employed

which consists of an inductor for which R L = 1,000
Calculate the gain and phase shift of the amplifier at
= 10,000 rad/sec. Draw the complete sinor diagram

A type 6J5 triode for

amplifier, the load of

ohms and L =

henry.

2,000 rad/sec and o>

The input signal is 6 volts rms.
of the system.
Solution.
The schematic and equivalent circuits are
ing diagrams. At w = 2,000 rad/sec,
a)

120

=
7,700

+ j0
=
+ i2,000)

(1,000

13.1

shown

j'3.01

in the

ma

accompany-

RADIO ELECTRONICS

106

The output

potential

is

E =
=
2

The gain

is

-(1,000 +j2,000)(13.1 -i3.01)

-(19.1 + j23.2) = 30.1 /-129.5

given by

K=

E2
eT

10~ 3

-^W = = 5mt^l
30.1/ -129.5

"

potential sinor diagram has the form

The

[Chap. 5

shown

in the sketch.

,120

At

co

10,000 rad/sec,
I

120

=
7,700

The output

potential

E2 =
=
=
The gain

is

+ jO
+ jl0,000)

5.94

j'6.83

ma

(1,000

is

-(1,000 + il0,000)(5.94
-(74.2 + j52.6)
90.8 / -144.7

j6.83)

10~ 3

given by

K
The

90.8/ -144.7

- 15.1^144^!

potential sinor diagram has the form of the

accompanying diagram.

The results are tabulated for convenience. An examination of the results indicates the presence of frequency distortion, since the gain at co = 2,000 rad/ sec is
Also, phase-shift distortion exists in
different from that at co = 10,000 rad/sec.
this amplifier.

CO

2,000
10,000

Gain and phase

Potential db gain

5.01/-129.5
15.1/-144.7

23.6 db

14

db

BASIC AMPLIFIER PRINCIPLES

Sec. 5-4]

5-4. Interelectrode

Capacitances in a Triode.

It

107

was assumed

in

the foregoing discussions that, with a negative bias on the grid, the grid
This is generally true if one
driving-source current was negligible.

examines only the current intercepted by the grid because of its location
Actually though, owing to the
within the region of the electron stream.
physical proximity of the elements of the tube, interelectrode capacitances
between pairs of elements exist. These capacitances are important in the
behavior of the circuit, as charging currents do exist.
Owing to the capacitance that exists between the plate and the grid,
it is not true that the grid circuit is completely independent of the
Since the capacitance between plate and grid is small, the
plate circuit.
approximation that the plate circuit is independent of the grid circuit
However, at the higher frequencies,
is valid at the lower frequencies.

may

interelectrode capacitances

seriously affect the operation.

Fig. 5-3. Schematic and equivalent circuits of an amplifier, including the interelectrode capacitances.

A more complete schematic diagram and its equivalent circuit are given
in Fig. 5-3.

In this circuit,

C gp

denotes the capacitance between the grid

and the plate, C g k is the grid-cathode capacitance, and Cp k is the capacitance between the plate and the cathode. The solution for the gain of
this circuit is readily obtained with the aid of the Millman theorem (see
Appendix A, Sec. A-56). The point 0' corresponds to the plate termiFour branches must
is the cathode terminal K.
nal P, and the point
be considered between these points the load impedance with zero poten:

tial;

C 2 with
potential E in

the capacitor

zero potential; the potential rise /aE 9 in series

which exists
The capacitor
with C 3
The result is
across the input Ei does not appear in the equation.
with

rp

the

17
E2

series

- mEY + EiY
- Y7 + Y, + Y + Y
2

where

Yp =
Y2 =
Y =
Yj =
E2 =
3

\/r v

is

juCi
j^Cs

is

l/Zi

is

is

admittance
admittance
admittance
admittance

(5 " 12)
3

corresponding to
corresponding to
corresponding to

rv

C
C

2
3

corresponding to Zi

potential difference between

load impedance

and K,

or potential across

RADIO ELECTRONICS

108

Note that E =
x

The

E.

potential gain

_
and may be written

in the

is

[Chap. 5

given by

_ E2

output potential
input potential

Ei

form

- y,

+ y:;y; +

<->

y.

In this expression use has been made of the fact that gm = n/r p
In this analysis a number of factors have been neglected. It has been
assumed that no conduction or leakage currents exist between tube terminals.
Such leakage current will depend upon many variable factors,
such as the spacing between electrodes, the materials of the base, the
conditions of the surface of the glass and the tube base, and perhaps the
Ordinarily the error is small
surface leakage between connecting wires.
If this assumption is
in neglecting the effects of this surface leakage.
not true, the effect can be taken into acount by writing for each interelectrode admittance g s + jcoC s instead of jiC s where g s takes account of
.

the leakage current and also dielectric losses. Interwiring and stray
capacitances must be taken into account. This may be done by con-

them to be in parallel with C h C%, and C3. Additional considThese are discussed in
erations are necessary at the high frequencies.
sidering

Sec. 5-8.

made

by neglecting the interThese interelectrode capacitances are usually 10 /u^if or less, which corresponds to
admittances of less than 2 ^mhos at 20,000 cps. This is to be compared
with the mutual conductance of the tube of, say, 1,500 /initios at the
normal operating point. Likewise Y 2 + Y 3 is usually negligible compared with Yj, + Yi. Under these conditions, the expression for the gain

The

error

in the calculation of the gain

electrode capacitances

is

very small over the

a-f

spectrum.

[Eq. (5-13)] reduces to Eq. (4-16).

5-5. Input

Admittance of a Triode.

Owing

to the presence of the

no longer isolated from the

plate circuit.
In fact, with a positive signal on the grid and with the
consequent negative potential on the plate, an appreciable change of
interelectrode capacitances, the grid circuit

potential appears across the capacitance

ciable current flow.

Cg k

is

C gp with
,

a consequent appre-

Also, the potential change across the capacitance

accompanied by a current flow. Clearly, therefore, the input-sigmust supply these currents. To calculate this current, it is
noted from the diagram that
is

nal source

Ii

I3

= E BPY3 =

E1Y1

and
(Ei

E2)Y3

BASIC AMPLIFIER PRINCIPLES

Sec. 5-5]

But from the

fact that

= KEi

E2

then the total input current

h = h

is

I,

Yi

Yi

;Y,

The input admittance, given by the

If Y, is to

109

(1

ratio Y<
(1

K)Y,]E,

K)Y

I,-/Ei, is

(5-14)

cannot,
be zero, evidently both Yi and Y 3 must be zero, since
be 1/0 deg. Thus, for the system to possess a negligible input

in general,

admittance over a wide range of frequencies, the grid-cathode and the

must be negligible.
Consider a triode with a pure resistance load. At the lower frequencies, the gain is given by the simple expression [Eq. (4-16)]
grid-plate capacitances

g _ uRi
Ri + r p
In this case, Eq. (5-14) becomes

(5-15)

Thus the input admittance

cathode of magnitude

c =
<

that from a capacitor between grid and

is

Cl

wfd

(5 " 16)

Attention is called to the very large contribution to the input capacitance by the grid-plate capacitance C 3 owing to the fact that its magnitude is multiplied by the amplifier gain. As a result, the total input
,

capacitance
tances.

is

very

much

The presence

higher than any of the interelectrode capaci-

of this

input capacitance will be found to affect

make operation impossible,

the operation of the amplifier, and often will

Methods of compensation have

been devised to overcome this effect, and these will be examined later.
For the general case when the gain of the amplifier K is a complex
quantity, the input admittance will consist of two terms, a resistive and
a reactive term. For the case of an inductive load, the gain K may be
written in the form (see Sec. 5-3, Example 2)
especially at the higher frequencies.

K=
and Eq.

(5-14)

-(*!

+jk

(5-17)

2)

becomes
Y,

-aCJc+Md +

(1

fcOC]

(5-18)

RADIO ELECTRONICS

110

[Chap. 5

This expression indicates that the equivalent input circuit comprises a

resistance (which is negative in this particular case, although it will be
positive for a capacitive load) in parallel with a capacitance C,-, as shown
The equivalent elements have the form
in Fig. 5-4.

Ri

(5-19)

and the capacitor

d
Fig. 5-4.

The

equivalent input
circuit of a triode.

= Ci+(.l+

Ai)C,

(5-20)

As indicated
for the

in the above development, it is possible

term k 2 to be negative (with an inductive load).

Under these circumstances the input

resistance Ri will

be negative. Physically, this means that power is being fed back from
the output circuit into the grid circuit through the coupling provided by
the grid-plate capacitance. If this feedback reaches an extreme stage,
These feedback effects in an amplifier will
the amplifier will oscillate.
be examined in some detail in Chap. 7.
5-6. Input Admittance of a Tetrode.
The equivalent circuit of the
tetrode is essentially that of the triode, even though a screen grid exists
A schematic diagram of a simple amplifier circuit
in the tetrode.
employing a tetrode is given in Fig. 5-5. In drawing the equivalent

Fig. 5-5. Schematic

and equivalent

circuits of a tetrode in

an amplifier

circuit.

given in Sec. 4-6 have been appropriately extended and

This requires the introduction of a point S, the screen terminal, in addition to the points K, G, and P.
Notice that the screen potential is maintained at a fixed d-c potential
circuit, the rules

employed.

with respect to cathode and is at zero potential in so far as a-c variations

about the Q point are concerned. As indicated in the figure, this effectively places a short circuit across C* and puts C g k and C gs in parallel.
This parallel combination is denoted C\. The capacitance C ps now
appears from plate to cathode and is effectively in parallel with C p kThis parallel combination is denoted C 2
Also, from the discussion in
.

Sec. 2-14, the shielding action of the screen

is

such that the capacitance

BASIC AMPLIFIER PRINCIPLES

Sec. 5-7]

Cps between

111

and plate is very small. If this capacitance is assumed

and it is less than 0.001 nyi in the average potential tetrode, then Fig. 5-5 may be redrawn in the form shown in Fig. 5-6.
In
this figure, the capacitances have
G
the values
grid

to be negligible,

Cz

Cps

~\~

cpk

The input admittance

is

(5-21)

of the tube

then
Fig. 5-6. The ideal equivalent circuit of a
tetrode amplifier.

(5-22)

The mere substitution of a tetrode for a triode may not result in a very
marked improvement in the amplifier response. This follows from the
fact that the stray and wiring capacitances external to the tube may allow
significant grid-plate coupling.

It is necessary that care be exercised in

order that plate and grid circuits be shielded or widely separated from

each other in order to utilize the inherent possibilities of the tube.

5-7. Input Admittance of a Pentode.
The discussion in Sec. 2-14
showed that, even though the tetrode had a significantly smaller gridplate capacitance than the triode, the presence of the screen grid was

accompanied by the

secondary emission from the plate when

fell below the screen potential.
As
overcome by the insertion of a suppressor
grid between the screen grid and the

effects of

the instantaneous plate potential

discussed, the effect of this

is

plate.

When used in a circuit as a potential

amplifier, the

pentode

is

connected in

the circuit exactly like the tetrode with

the addition that the suppressor grid is

connected to the cathode.

Fig. 5-7.

The equivalent

circuit of

By

draw-

ing the complete equivalent circuit of

pentode amplifier.

the pentode amplifier, by appropriately

extending the rules of Sec. 4-6, and by including all tube capacitances, it is
easy to show that the equivalent circuit reduces to that shown in Fig. 5-7.

In this diagram

Cl
C2

where

Cp

is

Cg S

Cgk

~f~

Cpk

+ C ps

-\-

Cpz

(5-23)

the plate-grid No. 3 capacitance.

much smaller than the plate

convenient to use the current-source
equivalent-circuit representation of the tube, as shown.
For the range
The

plate load

impedance
and

resistance of the tube,

Zi

is

it is

frequently

RADIO ELECTRONICS

112
of frequencies over

are negligible,

through

which the input and output capacitances Ci and Ci

r p 2> Z i; the total generator current passes

and with

The current-source equivalent

Fig. 5-8.

is

potential

is

given by the simple form

K = gmZ
If

pentode amplifier,

= QmElZl

E2

and the gain

circuit of the

Under these circumstances the output

Zi.

[Chap. 5

(5-24)

the assumed conditions are not valid, the gain becomes

K = gm Z

(5-25)

where Z is the combined parallel impedance in the

output circuit.
5-8. High-frequency (H-F) Considerations.
In addition to the effects of the interelectrode
capacitances in affecting the performance of an
amplifier, several other factors are of importance,
Fig. 5-9. Circuit for ex-

amining the effect of

cathode lead inductance

These
and include the effects
lead inductances and also the effects due to

particularly at the higher frequencies.

were mentioned
of

in a pentode.

in Sec. 5-7

transit time.

To examine

the effect of the cathode lead inductance, Fig. 5-9

is

ana-

For convenience, it will be assumed that the grid is negative

throughout the cycle and that transit-time effects are negligible. Then

lyzed.

for I 2

Ii
I2

gm

Eg

and
E

Ii

jwCi
Also

El

Combine equations

Ji
= J-fT
+ jgmEguL

to get

Ei

II

jut 1

II

jojC

BASIC AMPLIFIER PRINCIPLES

Sec. 5-8]

The input admittance

is

jwdjl
1

If

co

^L

1,

113

+ jgmoiL

jwg mL)

+ "W>

(5-26)

then
Y,

= ^gmLC,

+ ja,C

(5-27)

Observe, therefore, that the cathode lead inductance introduces an input

conductance of amount u 2 g mLCi.

A second component of input conductance arises because of the transit
time of the electrons between cathode and plate. An exact calculation
is difficult, but a qualitative explanation is possible which indicates the
To understand grid loading, consider an
grid-loading effects involved.
electron that has left the cathode and is approaching the grid in its
flight to the anode.
Suppose that the grid potential is negative relative
to the cathode so that

no electrons are collected by the

grid.

As the

electron approaches the grid, a changing image-charge density will be

induced on the grid (see Sec. 2-4 for a discussion of image charges).
This changing image charge represents an instantaneous grid current, the
The
direction of flow of charge being such as to charge the bias battery.
power for this charging process is supplied by the moving electron, and
as a result the electron

is

decelerated.

Once the electron has passed the grid, the process is reversed, and the
moving electron receives energy from the grid, and it is accelerated
thereby.
The amount of energy lost by the electron as it approaches the
grid is just equal to that which it gains as it moves away, and the net
energy change is zero. As a result, the net grid loading is zero.
If

the transit time of the electron in the cathode-anode space is of the

of the applied grid potential, the grid loading

order of the period

becomes important, for now the electron can no longer be considered to

It is possible for the energy that
be in a field which is constant in time.
supplied to the grid by the moving electron to exceed the amount of
energy that is returned by the grid in its interelectrode flight, with a
This energy is supplied by
resultant net energy loss in the grid circuit.
the grid driving source, and it represents a load on this source.
From a circuits point of view, the foregoing may be described in terms
is

an induced current in the grid. At the lower frequencies, the induced

is 90 deg out of phase with the grid potential, with a consequent zero net power loss. At the higher frequencies, an inphase component exists. This inphase component reduces the input resistance, and
this may produce an appreciable loading of the input circuit.
The foregoing concepts may be employed to indicate in a qualitative
way the effect of the various factors on the input resistance. If T
of

grid current

RADIO ELECTRONICS

114

[Chap. 5

denotes the transit time, / denotes the frequency of the applied grid
and gm is the mutual conductance of the tube, it is expected
that the grid current I g is proportional to T and /, since I g is small if
potential,

is small.
Also, I g should be proportional to g m since g m
determines the a-c component of plate current for a specified E g and the
total grid current is proportional to this a-c component of the plate cur-

either of these

rent.

If

a denotes the

transit angle,

the inphase component of I g

deviations from 90 deg.

is

But a

the inphase component of I g

is

gt

where k

which

is

now

I g sin a, which

than 90 deg, then

simply I g a for small

less

is

also proportional to

is

T and

Thus

/.

proportional to g m T 2P, or

kg mpT*

(5-28)

a constant depending on the geometry of the tube and elec-

is

trode potentials.

This relationship agrees with the complete analysis of

Ferris. 1
It will be seen from Eqs. (5-27) and (5-28) that g { and the conductance
component of the cathode inductance depend on the frequency in the
same way. Consequently, these components cannot be separated readily
in measurement of input resistance or conductance.
Tubes for use at the high frequencies are made in a manner to reduce
transit time, interelectrode capacitances, and lead inductances.
This is
done by means of very close electrode spacing, and generally small phys-

dimensions of electrodes. Among such tubes are the so-called

"acorn," "doorknob," "pencil," and "disk-seal," or "lighthouse," tubes,
with upper limits in frequency of approximately 2,000, 1,700, 3,000, and
ical

3,500 Mc, respectively. These names are indicative of the external

envelope shape, the first three possessing essentially cylindrical electrode
structures, the last being essentially of a planar construction.

two have the leads brought out

of the

The

first

envelopes at widely spaced points,

The latter two bring the leads out in the

At the higher frequencies these tubes are incorporated in

in order to reduce capacitances.

form

of disks.

coaxial line resonators, lead inductances being unimportant as these form

part of the resonant cavities.

5-9. Potential Sources for Amplifiers.
tial

sources are required in an amplifier.

A number

filament, or A, supply; the plate, or B, supply

supply

cc

the screen supply

E cci

of different poten-

These are the following: the

E bb

the grid-bias, or C,

These potentials are supplied

in

different ways.

The Filament, or A, Supply. The most common method of heating the

cathodes of indirectly heated tubes is from a low-potential winding on a
transformer which operates from the a-c supply lines. Storage batteries
may be used if d-c heating is necessary, but this is ordinarily not neces-

BASIC AMPLIFIER PRINCIPLES

Sec. 5-9]

sary except in special applications.

115

Special low-drain tubes are avail-

and are fed from dry batteries.

The Plate, or B, Supply Ebb
Most equipments involving the use of
electron tubes are operated from the a-c supply mains, and the d-c plate
supply is then secured by means of a rectifier and filter unit (see Chap.
3 for details).
For applications with severe requirements on regulation or low ripple, the power supply must be electronically regulated.
able for use in portable radio sets
.

For low-drain requirements, dry batteries may be used.

The Grid, or C, Supply E cc
The grid circuit of most amplifiers ordinarily requires very little current, and hence low-power
dry batteries may be used. In most cases, however,
self-bias is used (although this is restricted to class A
and class AB amplifiers). Self-bias is achieved by
including a resistor R k in the cathode of the amplifier
tube and shunting this resistor with a capacitor C k the

, ,
* IG 5-10. Obtain7
c
ii
n
reactance oi which is small compared with R k over the
ng se lf-bias by
operating frequency range.
The quiescent current Ib means of a cathode
.

flows through this resistor,

The

provides the grid bias.

The capacitor Ck

and the potential difference

correct self -biasing resistor

serves to by-pass

any

a-c

resistor

Rk = E

components

cc

/Ib

in the plate

current, so that no a-c component appears across the resistor R k

If such
an a-c component, or varying bias, does exist, then clearly there is a
reaction between the plate circuit and the input circuit.
Such a "feed.

back"
is

effect will receive detailed consideration in

to be avoided, large-capacitance capacitors

larly

if

the frequency

is

Chap. 7. If this effect

be required, particu-

may

High-capacitance low-potential electrolytic

and are quite small

low.

capacitors are available for this specific service

physically.

The Screen Supply

E ciC

the plate-supply source.

The
In

screen supply

many

is

ordinarily obtained from

is lower
than the plate supply, and it is usual practice to connect the screen to the
plate supply through a resistor.
The resistor is chosen of such a size
that the potential drop across it due to the screen current will set the
screen at the desired potential.
A capacitor is then connected from the
screen to the cathode so as to maintain this potential constant and independent of B-supply variations or variations in the screen current.

cases the screen potential

customary to use a common B supply for all tubes of a given

Because of this, the possibility for interactions among
the stages through this common plate supply does exist and might be
troublesome unless the effective output impedance of the power-supply
It is

amplifier circuit.

is very small.
It is necessary in some applications to include RC
combinations known as decoupling filters so as to avoid this interaction.

unit

A typical

resistance-capacitance coupled-amplifier circuit which

is

pro-

RADIO ELECTRONICS

116
vided with

self-bias,

decoupling

filters,

and screen dropping

[Chap. 5
resistors is

illustrated in Fig. 5-11.

Fig. 5-11. Resistance-capacitance coupled amplifier, with self-bias, decoupling

and screen dropping

filters,

resistors.

REFERENCE
1.

Ferris,

W.

R., Proc.

IRE,

24,

82 (1936).

PROBLEMS
5-1.

Two

waves, one of amplitude 10 volts and frequency 1,000 cps, the second
and frequency 3,000 cps, are applied to the input of a certain
The two waves are so phased that they both pass through zero in the

of amplitude 5 volts

network.

positive direction together.

Sketch the resulting input potential.

Suppose that the fundamental component suffers a phase delay of 10 deg
on the fundamental scale and that the third-harmonic component suffers a 50-deg
delay on the third-harmonic scale, although neither amplitude is effected. Sketch
the output wave.
5-2. a. The output potential of a given amplifier is 18 volts, when the input
a.
b.

What is the decibel potential gain of the

potential is 0.2 volt at 5,000 cps.
amplifier?
b. The output potential is 7 volts when the input potential is 0.2 volts at
By how many decibels is the response of the amplifier at 18 kc
18,000 cps.
below that at 5 kc?
5-3. Prepare a table giving the power sensitivity of the following tubes (assume
that the output power and the grid excitation are those specified in the tube
manual): 6A3, 6F6, 6V6, 6L6, 6AG7.
5-4. An a-c excitation potential of 5 volts rms at a frequency of 2,000 cps is
The load is a pure
applied to a 6J5 tube for which n = 20, r p = 7,700 ohms.
Calculate the following:
resistance of 15,000 ohms.
a.
b.
c.

d.

The a-c current in the plate circuit.

The a-c output potential.
The gain of the amplifier.
The a-c power in the load resistor.

5-5.

Repeat Prob. 5-4

if

the load

is

an inductive reactance of 15,000 ohms.

BASIC AMPLIFIER PRINCIPLES

Find expressions

5-6.

for the indicated quantity in the

Find I

5-7.
fied

Find

type 6SF5 high-mu triode

100

rp

is

circuits.

E2

factors are

C =

ohms

66,000

accompanying

operated as a simple amplifier under speci-

The important

conditions at 30,000 cps.

117

Cpk =

2.4

Cgk =

ju/xf

4.0

3.6 ^/if

a. Calculate the input capacitance and the input resistance of the tube alone
when the load is a resistor Ri = 100 kilohms.
b. Repeat when the load impedance is of the form 60,000 + j60,000 ohms.

5-8.

type 6J5 triode

M = 20

is

operated as a simple amplifier under specified condi-

The important

tions at 22,000 cps.

rp

7,700

factors are

C =

ohms

Cph =

3.4

M /zf

Cgk =

3.4

3.6 mil

Calculate the input capacitance and the input resistance of the tube when
is a resistor Ri = 20 kilohms.
b. Repeat when the load is an impedance of the form 10,000
yi0,000 ohms.
5-9. A type 6J5 tube is operated in the circuit of the accompanying diagram.
Calculate the output potential.
(See Prob. 5-8 for the important factors of the
a.

the load

tube.)

5-10.

Show

that Fig. 5-7 does represent the complete equivalent circuit of the

pentode.
5-11.

type 6SJ7 pentode

is

operated as a simple amplifier under specified

conditions.
a.

When

connected as a pentode, with Ri

25 kilohms, the important factors

are

gm

1,575 MHihos

rp

0.7

megohms
Coutput

Cgp =
7.0 nixi

Calculate the input capacitance of the amplifier.

0.005 nnl

Ciapat =

6.0

jiyuf

RADIO ELECTRONICS

118

b.

When

19

this
rp

tube

is

8,000

[Chap. 5

reconnected as a triode, the factors become

ohms

C =

2.8 utii

Calculate the input capacitance with Ri

C gk =

3.4 nrf

Cpk =

11

jujuf

25 kilohms, and compare with the

results of part a.

5-12. A 6AC7 pentode is to be used as a class A amplifier with Et

Determine the value of the self-biasing cathode resistor to set E cc
the screen dropping resistor to set E c2 at 150 volts if Ebb = 350.

= 250 volts.
= 2 volts;

CHAPTER

UNTUNED POTENTIAL AMPLIFIERS

6-1. Basic Considerations.

It is frequently

higher gain in an amplifier than

is

necessary to achieve a

possible with a single amplifier stage.

In such cases, the amplifier stages are cascaded to achieve this higher
gain, the output potential

from one stage serving as the input potential

to the next stage.

A number of factors influence the number and the characteristics of

the individual stages which must be used to meet certain previously specified requirements.
Among the factors which must be taken into account
in amplifier design are the total over-all gain required, the

shape of the

frequency-response characteristic, and the over-all bandwidth.

Certain

which impose limits to the sensitivity which may be achieved,

among these being the inherent noise generated in such devices. The
requirements for stability of operation impose severe practical restrictions on the techniques of construction.
Because of the several factors
that play a part in amplifier design, gains in excess of about 10 6 or 120 db
potential gain, are extremely difficult to achieve.
Depending on the
bandwidth considerations, amplifiers seldom exceed six to nine stages in
factors exist

cascade for stable operation.

of

Extreme caution

is

required in the design

such multistage amplifiers.

To

calculate the over-all gain

and frequency response of a multistage

must be drawn. The

amplifier, the equivalent circuit of the amplifier

The resultant equivathen analyzed as a conventional problem in a-c circuit

rules for accomplishing this are given in Sec. 4-6.

lent

network

is

analysis.

variety of coupling networks between the cascaded stages are posand a few have become very common, either because of their sim-

sible,

plicity or

because of some especially desirable characteristic.

ber of the more

common

A num-

types will be considered in this chapter in some

detail.

6-2. Resistance -Capacitance

(RC)

Coupled Amplifier.

The

resist-

ance-capacitance (RC) coupled amplifier, illustrated in Fig. 6-1, is one

of the more common and more important amplifier circuits.
This amplifier circuit is

used when a sensibly constant amplification over a wide

is desired.
By the use of tubes with high amplifica-

range of frequencies

119

RADIO ELECTRONICS

120
tion factors,

it is

[Chap. 6

possible to achieve a gain of 50 or

more per

stage.

It

be found that high-gain triodes possess certain inherent disadvanIf pentodes

tages, and it is frequently desirable to use pentodes instead.
are used, the screen potential must remain constant; otherwise the following analysis will no longer be valid.
The capacitors Ci, C 2 and C 3 in this schematic diagram are known as
coupling, or blocking, capacitors and serve to prevent any d-c potentials
that are present in one stage from appearing in another stage. That is,
capacitor Ci serves to prevent any d-c potential in the input from appearing across the grid resistor R gi and thus changing the d-c operating
Capacitor C 2 serves a similar function in coupling
level of the amplifier.
stage 1 to stage 2. The value of the coupling capacitors is determined
They ordinarily range from about
primarily by the 1-f amplification.
The resistor R g which is
0.001 fif to 0.1 /xi for conventional a-f stages.
will

Output

'-ICC

Fig. 6-1.

known

Schematic diagram

of a two-stage

as the grid resistor, furnishes a path

RC

coupled amplifier.

by which the

grid-bias supply

path through which any

electrons that may be collected by the grid from the electron stream
within the tube may be returned to the cathode. If such a leak path
were not provided, the grid would acquire a negative potential with the
is

applied to the grid.

It also serves as a leak

collection of the electrons, thus influencing the operation of the tube.

negative-bias supply potential

is

ordinarily used,

and the

grid current

This permits the use of relatively large resistances

Large values of R g are desirfor R g say from 50 kilohms to 2 megohms.
able in achieving a wide frequency response. The load resistor Ri is
determined principally by the gain and the frequency bandwidth that
is

usually very small.

,

are desired, as will be

shown below.

equivalent circuit of the amplifier of Fig. 6-1 is shown in Fig. 6-2.

This circuit is valid for triodes, tetrodes, or pentodes provided that the
In this cirscreen potential of the latter two is maintained constant.

The

cuit Ei denotes the a-c input potential applied to the grid of the first

This potential appears across the parallel combination consistR g i in parallel with the input impedance to the amplifier.
The interelectrode capacitances are not shown on the diagram, but
their effect is contained in the effective input capacitance C g to each
stage.
That is, the input impedance of the stage is considered to comstage.

ing of the resistor

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-2]

121

a resistance (assumed positive) in parallel with the input capaciIt is also supposed that the impedance of the driving source is
low, so that the loading by the total input impedance of the first stage
does not affect the input potential. The output circuit of the first stage
prise

tance.

C 2 output tube
and wiring capacitances, and the total input impedance of the second
This is denoted as R g and Cg for the total resistive and capacistage.
The output of the amplifier is the potential across the
tive components.
output impedance, which is denoted by the symbol Z. This impedance
cannot be specified more completely until the nature of the output circuit
is known.
The coupling between the grid and the plate of the tubes through the
interelectrode capacitances can be neglected over a wide frequency range
consists of the load resistance, the coupling capacitance

<QJ/r
Fig. 6-2.

The equivalent

circuit of the

RC

amplifier of Fig. 6-1.

with pentodes and over the a-f range with triodes. Consequently each
stage may be considered as independent of the following stage, but the
output of one stage is the input to the next stage. As a result, it follows
that since

g _E _
2

Ei

output potential of 1st stage

input potential to 1st stage

and

_
K =E
E
3

output potential of 2d stage

input potential to 2d stage
output potential of 2d stage
output potential of 1st stage

then the resultant over-all gain

g _E _
3

E!
It follows

is

output potential of 2d stage

input potential to 1st stage

from these expressions that

(6-1)

By

taking twenty times the logarithm of the magnitude of this

expression

20 logio

20 log 10 #1

20 log 10

(6-2)

RADIO ELECTRONICS

122
It follows

from

stage amplifier

[Chap. 6

this that the total decibel potential gain of the multiis

the

sum

of the decibel potential gains of the separate

independent of the type of interstage coupling.

6-3. Analysis of RC Coupled Amplifier.
A typical stage of the RC
coupled amplifier is considered in detail. This stage might represent any
of a group of similar stages of an amplifier chain, except perhaps the output stage. Representative subscripts have been omitted. The equivalent circuit is given in its two forms in Figs. 6-3.
The typical stage will be analyzed by two methods in order to show the
features of the methods.
One method will employ the Millman theorem,
The second method will employ a straightforas applied to Fig. 6-3a.
This fact

stages.

is

ward junction solution

of Fig. 6-36.

HP

(6)

Fig. 6-3. A typical stage of an

(6) current-source equivalents.

RC

A direct application of the

K yields

coupled amplifier,

(a)

Potential-source and

Millman theorem between the points G 2 and

the expression

E2

= v

1
,

(6-3)

where Y c = joiC, Y Rf = l/R g Yc, = juC g

An application
rem between the points P and K yields the expression
.

of this theo-

- M E 0lYp + E Y C
Yp + Y + Y c
2

(6-4)

where Yp = l/r p and Y,

and solving for the gain

K
This

is

stage.

l/Ri.

K =

By combining Eq. (6-3) with Eq.

= Ei, there results
since E

E2./E;,!,

ff

(6-4)

ytiYpYc

=
(Y c

+ Y R + Y Cg )(Y p + Y,) + Y c (Y Ra + Y c
,

(6-5)
.)

the complete expression for the potential gain of such an amplifier

If

the constants of the circuit are known, the gain and phasefunction of frequency may be calculated.

shift characteristics as a

Now

refer to Fig. 6-36,

analysis.

The

and apply the standard techniques

of junction

controlling equations, obtained from considerations of the

Kirchhoff current law, are

(Y p + Y + Y c )E pl - Y C E = -g m E gl
-Y cE pl + (Yc + Y Co + Y )E =
2

ft

(6-6)

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-3]

By determinantal methods,

it

follows that

+ Yc

Y,

-Y c
Y c + Y Ca + Y R

Yp + Y + Y c
-Yc
t

The expansion
it

-g

E2

as

123

of these

(6-7)

determinants by Cramer's rule yields Eq.

(6-5),

must.

It will

be found convenient to analyze the response of the amplifier for

an interpretation of

limiting regions of frequency instead of attempting

i-m-

ME,

Fig. 6-4.

The mid-frequency equivalent

Eq. (6-5) directly.

In

many

fact, in

circuits of the

cases

it is

RC

amplifier.

more convenient

to ana-

lyze the appropriate equivalent circuit, rather than attempting the analysis

from Eq.

(6-5).

The intermediate

Intermediate Frequencies.

quencies, are those for which

Yc

is

and

large

frequencies, or mid-fre-

Yc

is

small.

Subject to

these conditions, the equivalent circuits of Fig. 6-3 reduce to those

shown

in Fig. 6-4.

For the range

of frequencies over

the expression for the gain

which

this equivalent circuit is valid,

becomes

K =K =
1,

Y,

+Y

(6-8)
flf

independent of the frequency, since no

Each parameter in the equation
is a conductance, and because of the negative sign the relative phase
angle between the input and output potentials is constant and equal to
This expression for the gain

is

reactive elements appear in the circuit.

180 deg.

L-F Region.
Yc,

may

be

At the low frequencies the

made

zero.

The

effect of

Cg

is

negligible,

effect of the coupling capacitor

and

C becomes

RADIO ELECTRONICS

124

Fig. 6-5.

The

1-f

equivalent circuit of the

[Chap. 6

RC amplifier.

very important. The equivalent circuit under these conditions has the
form shown in Fig. 6-5. The general expression for the gain [Eq. (6-5)]
reduces to

Ki =

ixY pYc
Y C (Y P +

Y,

+ Y Rg + Y Rg {Y p + Y
)

found convenient to examine the 1-f gain

quency gain. The ratio Kx/Ko becomes

relative to the mid-fre-

It is

Y (Y p

+ Y C (Y +
P
e

This

may

(6-10)

Y,

(6-9)
;)

Y,)

+ Y Bf

be written in the simple form, for any frequency

/,

(6-11)

if.//

where
f
11

+ Y,)
+ + Yn,)

Y B ,(Y P

2xC(Y p

(6-12)

Y,

Equations (6-11) and (6-12) are meaningful only if the load is a pure
resistance, since then f i is a real number and the magnitude of the relative
gain becomes

Ki =

K
where

fx

Vi +
/l

C-f

Af\22

0V/)

(6-13)

2xC

[R

r p fi,/(r p

Ri)]

This shows that the parameter f represents the frequency at which the
This
gain falls to l/\/2, or 70.7 per cent of its mid-frequency value.
frequency is usually referred to as the l-f cutoff frequency of the amplifier.
The relative phase angle di is given by
x

tan

0i

(6-14)

This approaches 90 deg as the frequency approaches zero.

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-3]
It

should be noted that the

1-f

125

cutoff value [Eq. (6-12)] depends,

among

other terms, on the size of the coupling capacitor C.

Since the value of
C appears in the denominator of the expression for/i, then, for a decreased
1-f

cutoff, larger values of

ultimately

fall

C must be

Of course, the gain must

chosen.

to zero at zero frequency.

There are several practical limitations to the

size of the coupling capacibe used. The capacitor must be of high quality so that
any leakage current will be small. Otherwise a conduction path from
the plate of one stage to the grid of the next stage may exist.
But good-

tance that

may

quality capacitors in sizes greater than

relatively expensive.

Also,

if

are physically large

0. 1 fii

the coupling capacitance

is

and are

large, a phe-

nomenon known as blocking may result if grid current flows. This arises
when the time constant CR g is much larger than the period of the highest
frequency to be passed by the amplifier. Thus if an appreciable charge

Fig. 6-6.

The

h-f equivalent circuit of the

RC

amplifier.

flows into the capacitor with the application of the input signal

and

if

cannot leak off quickly enough, a charge will build up. This may
bias the tube highly negatively, perhaps even beyond cutoff.
The
amplifier then becomes inoperative until the capacitor discharges.
This
condition is sometimes desirable in special electron-tube circuits.
Howthis

it is a condition that must be avoided in an amplifier that

reproduce the input signal in an amplified form.

ever,

is

to

grid resistance R g must be made high to keep the gain high, since
one stage represents a loading across the plate resistance Ri of the
previous stage.
The upper limit to this value is set by the grid current.

The

Rg

of

Ordinarily the grid current

negative.

ohms

is

But

if

is

when the grid bias is

too high, and several meg-

small, particularly

the grid resistance

is

made

the usual limit, the potential across this resistance will act as a

spurious bias on the tube.

While special low-grid-current tubes are availand would not ordinarily

able, these are designed for special operations

be used in conventional

H-F

circuits.

At the high frequencies, the admittance of C is very

large, and the admittance of C g becomes important.
The equivalent circuit corresponding to these conditions becomes that shown in Fig. 6-6.
The general expression for the gain reduces to
Region.

K = K = Y +Y,+mYYp
B + Yc
p

(6-15)

RADIO ELECTRONICS

126

The

gain ratio

K /K
2

becomes

K2 =

1
j

_|_

Yc
Yp + Y + Y B

case.

It

may

(6-16)

This expression

[Chap. 6

be written in a form similar to Eq. (6-11) for the

1-f

becomes

K
K

(6-17)

1+jf/U

where

Y + Y R<
_
= Y p +2^
;

f2

(6 " 18)

In this expression Cg denotes the total capacitance from grid to cathode

and comprises the input capacitance of the following stage, the output
wiring, and the output tube capacitance.
In Eqs. (6-17) and (6-18), as in Eqs. (6-11) and (6-12), the expressions
Then f 2 is a real
are meaningful only if the load is a pure resistance.

number, and the magnitude

of

the relative gain becomes

K2 =
Ko
It follows

gain

falls

from

r
\1 +

(6-19)

(f/h)

which the h-f

mid-frequency value. This

this that / 2 represents that frequency at

to l/y/2, or 70.7 per cent, of

its

frequency is usually referred to as the h-f cutoff of the amplifier.

relative phase angle 2 is given by
tan

= -

The

(6-20)

J2

This angle approaches 90 deg as the frequency becomes very large

compared with /2
Note from Eq. (6-18) that the h-f cutoff value depends on the value of
C g among other factors. Since the value of C g appears in the denominator of the expression, then clearly a high h-f cutoff requires a small value
Moreover, since the input capacitance of a pentode is appreciably
of C g
.

than that of a triode, the pentode possesses inherently better possiNote above
bilities for a broad frequency response than does the triode.
that the h-f cutoff is improved by the use of large Y p> Y h and Y Rl which
implies the use of small values of resistance Ri, R g and a tube with a small
less

plate resistance.

Curves for RC Amplifiers. 1 The foregoing analysis shows that the gain of an RC coupled amplifier is substantially constant over a range of frequencies and falls off at both the
6-4. Universal Amplification

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-5]

high and the low frequencies.

127

typical frequency-response curve has

the form sketched in Fig. 6-7.

Since the relative gain and the relative phase-shift characteristics

depend only upon the two parameters /i and/2 it is possible to construct
curves which are applicable to any such amplifier. Such universal curves
,

are given in Fig. 6-8.

270

1.0

180

&0.5
ii
a-

90

Fig. 6-7.

typical frequency response

and phase characteristic

of

an

RC

coupled

amplifier.

The frequency-response

any

characteristics of

RC

(6-12)

and

Then the values

(6-18).

coupled amplifier
first step in the

The

can easily be obtained with the aid of these curves.

analysis is to calculate the values of the parameters

/i

and /2 from Eqs.

and the relative

of the relative gain

phase angle are obtained from the curves for a number of values of the
These are plotted as a function of /. It must be
ratio fi/f and /// 2
.

remembered

in using Fig. 6-8 that the ordinate

abscissa is/i//.

Also, the ordinate

is

K2/K0

or

is

K /K
1

when the

or di

when the

abscissa is/// 2

1.0

<^ 90

80

Phase-

0.8

2 70
0.6

tso
.50.4
0)

a.

%
1

30
20

(X.

0.2

10

J.
0.1

0.2

0.3

0.5

1.0

Frequency

ratio

5
10
fj/f or f/fz

Fig. 6-8. Universal gain and phase-shift curves for an

RC

20 30

50

100

coupled amplifier.

Suppose that it is desired to extend

coupled amplifier. According to the universal
By
gain characteristic, this requires that the quantity / 2 be increased.
Eq. (6-18) this increase in /2 may be accomplished by increasing any of
6-5.

Gain -Bandwidth Product.

the h-f response of an

Y,,, Y;, or Yb or by decreasing

varying these parameters.

the terms
effect of

RC

Cg

It is desired to

examine the

RADIO ELECTRONICS

128

[Chap. 6

Consider the factor Yp

An increase in Yp implies that the plate resistance rp is reduced. This would seem to favor the use of triodes with low
values of r p
However, tubes of this type are power triodes, which are
low-/t tubes.
Consequently, in addition to the low gain inherent in such
tubes, and the corresponding high grid driving signal that would be
required for reasonable output potential, the use of a triode is inadvisable because of the relatively large total input capacitance which such a
stage would possess [see Eq. (5-16)], so that the influence of the increase
in C g would more than overcome the gain possible by increasing Y p
An increase in Yi, which implies a reduction in the load resistance Ri,
will also be accompanied by an increased value of /2
Thus while there
is an increase in the bandwidth of the amplifier, the gain is thereby
reduced.
Suppose that the tube that is used is a pentode, and this is
generally the case for a broad-band amplifier.
For the pentode, since
r p is large (and of the order of 1 megohm) and R g may also be made large,
the h-f cutoff value is given with good approximation by
.

Yi

Moreover, for the pentode, the gain

approximation by

of the stage is given

gm Ri

with good

(6-22)

If it is assumed that the 1-f cutoff is small, so that fi denotes the

bandwidth of the amplifier, then the gain-bandwidth product is

KB
2

total

(6-23)

Observe from this expression that the gain-bandwidth product of the RC

coupled amplifier is a constant that depends only on the tube. This
means that, by changing a circuit parameter to increase the gain, the
bandwidth of the system is reduced; one is obtained at the expense of
the other for a given tube and a given circuit configuration.
Since the gain of the stage is proportional to g m of the tube, and since
the bandwidth, for a given gain, is proportional to 1/C g in a given circuit configuration, the limit to the bandwidth is dictated fundamentally
by the interelectrode capacitances of the tube. Thus even if the wiring
and socket capacitances were reduced to zero, an impossible practical situation, the sum of the output capacitance of the one tube and the input
capacitance to the following stage would provide the ultimate limitation.
That is, the ultimate limit is imposed by an effective capacitance Ct,
which is the output capacitance of one tube and the input capacitance to

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-6]

129

the next tube, or

From

is

M
M=

this expression, a quantity

known

is

denned as
(6-25)

as the figure of merit of the tube.

For service requiring a large gain-bandwidth product, the tube should

possess a large transconductance in proportion to the input plus output
The 6AK5 and the 6AC7 are both highly satiselectrode capacitances.
factory in this respect, the 6AK5 being slightly superior to the 6AC7.
When allowance is made for socket and wiring capacitances, an average
6AK5 has a gain-bandwidth product of approximately 55 Mc (for the
tube capacitances alone this figure is approximately 117 Mc) and an
average
6-6.

6AC7

has a corresponding value of 50 Mc.

When

Cascaded Stages.

identical stages are connected in cas-

by Eq. (6-1). However, this

accompanied by a narrower bandwidth. It is desired to
obtain expressions which show the effect of cascading identical amplifiers.
This is done piecewise for the important frequency regions.
Consider that n identical stages are connected in cascade. In the midfrequency range the resultant gain is constant and is given by
cade, a higher gain

higher gain

is

provided, as required

is

Km
For the

1-f

= (K

(6-26)

region, the relative gain for the

n stages

is

given by

1
[i

The

resulting

tive gain

is

1-f

cutoff value

+ (M)

]" /2

(6-27)

defined as that value for which the rela-

is

reduced by 1/ y/ 2.

This requires that

(/)

2ii

(6 " 28)

from which
1

so that the ratio of single to

^\ =

J
[

2 1/n

n cascade stages

=
J In

V2

1/B

is

(6-29)

RADIO ELECTRONICS

130

The
the

1-f

relative h-f gain for the w-stage amplifier

region and

is

[Chap. 6

obtained exactly as for

is

given by
1

[i

The corresponding

relative h-f cutoff -value ratio

Ip.

(6-30)

+ (f/m n/i
then

is

V2 ^r="l
T

(6-31)

Table 6-1 gives the values of the cutoff frequency reduction function

TABLE

6-1

BANDWIDTH REDUCTIO N FACT OR

V2

3
4
5
6

\/2 1/n

It is seen, for

1.

ingly, the

1-f

cutoff value

of course, that the total

is

1/n

example, that the h-f cutoff value of two

reduced by a factor 0.643. Correspond-

is

by

increased

bandwidth

ber of cascaded stages increases.

1-f

"

1.0
0.643
0.510
0.435
0.387
0.350

identical stages in cascade

l/

of

this

same

factor.

This means,

the amplifier decreases as the

To

num-

achieve specified over-all h-f and

cutoff values, the single-stage cutoff values

must be correspondingly

high and low, respectively.

6-7. Direct-coupled Amplifier. 2

It is possible to build a type of

cascaded amplifier without reactive elements and, in principle at least,
The potential gain of such an amplisecure a very broad-band amplifier.
fier does not depend on the frequency, at least to a first approximation.
However, the effect of tube and wiring capacitances imposes the same
limitations on the h-f cutoff of the amplifier as in the RC coupled amplifier.
It might appear that such amplifiers would find very widespread
However, such amplifiers
use because of these desirable characteristics.
do possess certain disadvantages, and their use is limited, though they

find extensive

employment as

d-c amplifiers

and as

amplifiers for very

slowly varying inputs.

A battery-coupled cascade-amplifier circuit of basic design, together

with the equivalent plate circuit for small changes in voltage and current, is shown in Fig. 6-9.
The gain of such an amplifier stage is readily

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-7]

131

found to be

K =
rv

tii

be observed that the circuits are quite like the

It will

amplifier except that the coupling

Because

(6-32)

of the fact that

(blocking)

the grid of one stage

plate circuit of the previous stage,

is

RC

coupled

capacitors are absent.

directly connected to the

necessary to include d-c sources

it is

at the various critical points in the circuit in order that the quiescent conditions be those of class

operation.

The battery-coupled amplifier has the outstanding feature that it will

amplify a steady component in the input voltage, but it suffers from three
main disadvantages. The first is the cost of the relatively high-potential
grid-bias batteries.
These are required when a common plate and a

common

filament supply are used.

In an alternative arrangement,

indirectly heated cathodes having different potentials are used,

obviating the necessity for large grid-bias potentials.

thus

However, sepa-

rate plate supplies are required in this case.

The second disadvantage

instability

associated

coupling.

The

the tubes in
slightly

of the direct-coupled amplifier

with direct

characteristics

the

with

circuit

time;

the

is

the inherent

of

change
battery

potentials, or the a-c line-operated

rectified

power

supplies,

likewise

change with time. Since such

changes are amplified, the d-c amFlG 6 " 9 Basio battery-coupled amplifier
plifier is not feasible unless precaucircuits
tions can be taken which tend to
overcome this instability. For this reason, balanced circuits and circuits
with degenerative feedback are used, since they tend to minimize this
-

difficulty.

The

third disadvantage arises

bias batteries

from the capacitance between the

and the cathodes.

grid-

This, plus the interelectrode capaci-

and stray inductance, influences the

and materially affects the rapidity with which

tances, stray wiring capacitance,

transient-response time

the amplifier output responds to rapid changes of input potential.

consequence, even though the amplifier

must be taken to ensure a broad

is

In

direct-coupled, precautions

h-f response in order to provide a short

response time.
It is possible to build a direct-coupled amplifier that uses a positive

plate supply, a negative bias supply,

This overcomes the

is

first

disadvantage.

illustrated in Fig. 6-10.

and resistance coupling networks.

The circuit of such an amplifier

The equivalent

circuit of a typical stage of

RADIO ELECTRONICS

132

this amplifier is given in Fig. 6-11.

The gain

of

[Chap. 6

such an amplifier

is

readily found to be

Rn
2

K
rP

RlljRcl
Rl\

(6-33)

Rcl

Rgl)

Rg2

For an appreciable potential gain, the parallel combination of Rn and

Rci + Rg2 should be large compared with r p and R gi should be large
compared with R c i. This will necessitate the use of a large bias supply.
,

Fig. 6-10.

A resistance-coupled

amplifier.

Direct-coupled amplifiers are used extensively as the amplifier in a cirwhich has a very high internal resistance

cuit the grid exciting source of

or which is capable of supplying only a very small current.

In this case,
the grid current must be very small. In particular, the grid current is
significant

when the grid-cathode

resistance of the tube, though high,

might not be large in comparison

with the resistance of the circuit
that supplies the grid signal voltage.

AM

in

Special electrometer tubes

which the grid current

order of 10~ 15

amp

for such applications.

Fig. 6-11. The equivalent circuit of a
typical stage of the resistance-coupled
amplifier of Fig. 6-10.

the

The

grid

current of the typical negativegrid tube is of the order of 10~ 8

amp
applied to the tube electrodes.

is of

are available

With the

with normal rated potential

electrode potentials at very low

be reduced as low as 10-12 amp.

6-8. The Cathode-coupled Amplifier.
A two-tube circuit which is
used extensively as a direct-coupled amplifier, owing to certain self-balancing features, and which is often used as an a-c amplifier, is illustrated
in Fig. 6-12.
This circuit overcomes the first disadvantage of the previous section and permits the use of a common battery supply for all
values, the grid current

may

stages.

To

analyze the operation of the

circuit,

the Kirchhoff potential law

is

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-8]

The tubes

applied to the loop circuits shown.

133

assumed to be iden-

are

Hence, there follows

tical.

= Ei Ea = Ei (Ii I^Rk
E2 = Ei = (Ii l2)Rk
Iir p juEsi + Ejt =
I.(r, + Ri) + mE, - E* =
Ei

(6-34)

Write the equations in the form

HrP +

-IiGu

The

l)R k

(ji

i)R t ]

h[rp

(m

+
+ l)Rt +
1.0*

Ri]

solution of these equations yields, for current I 2

I2
[r,

The output

0i

l)^4]tr P

potential

(m

+
+

rp

0*

fEi

l)fl*E,

l)t

Ri]

[(m

1)*]

(6-35)

is

n(n

E = hR,

=
=

l)fl*][r p

+
(m

VRkRiE!

Km

l)RkV
(6-36)

Now

write this as

E2

uRiEi

=
2r p

+
(ji

parameters are so chosen that

mately

If the

+
rp

(6-37)

Rl)

l)Rk

Ri

Ri

<3C (ji

1)^4, then approxi-

which is a form quite like that for the ordinary single-tube amplifier,
except for the appearance of the factor 2r p in the denominator instead of

Fig. 6-12.

cathode-coupled amplifier and

its

equivalent circuit.

simply rp
Note also that the output potential has the same phase as the
input potential. A typical circuit showing cascade cathode-coupled
.

amplifiers

is

given in Fig. 6-13.

RADIO ELECTRONICS

134
It

may

results

[Chap. 6

be shown that the h-f cutoff value for cascaded stages, which

from the

effects of the interelectrode,

wiring,

and distributed

considerably higher than in a single-tube amplifier.

However, such amplifier stages are not used for broad-band or video
capacitances,

is

pentode proves to be superior, both as regards gain and

bandwidth possibilities. Moreover, pentodes are seldom used in this
circuit from bandwidth considerations alone.
Such cathode-coupled
amplifiers, since a

amplifiers are used for very

1-f

100

or d-c amplifier service.

too

,200

(100)

200

2Mn

MMr

-200

Fig. 6-13.

*-20O
multistage d-c amplifier employing cathode-coupled amplifiers.

6-9. Inductance -Capacitance (LC)

The

Coupled Amplifier.

circuit of

the inductance-capacitance coupled amplifier differs from that of the

resistance-capacitance coupled amplifier only in the use of an inductor

The schematic diagram of the

The use of an inductor instead of a

plate load instead of a plate resistor.

amplifier

is

given in Fig. 6-14.

resistor in the plate circuit makes possible the use of a smaller platesupply potential for a particular tube operating condition, since the d-c

2 Output

Input

Fig. 6-14. Schematic diagram of an inductance-capacitance coupled amplifier.

resistance of the inductor

inductor

The
6-15.

is

is

small and the d-c potential drop across this

also small.

equivalent circuit of one stage of this amplifier

Observe that

is

given in Fig.

from the corresponding equivalent

the plate-circuit impedance.
Consequently

this circuit differs

circuit of Fig. 6-3 only in

the analysis leading to Eq. (6-5)

the plate-circuit admittance

is

is

valid in the present case provided that

interpreted to be

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-10]

Yi

j*>C L

135

+ R L + juL

(6-39)

Here C L is the distributed winding capacitance, which has been assumed

shunt the inductor, and R L is the resistance of the inductor.

to

The frequency-response characteristic of this amplifier may be examsame way as that for the RC system. The analysis shows that

ined in the

the response

is

RC amplifier, except that

narrower, although the mid-frequency gain is
gain drops

generally similar to that for the

the sensibly flat region

is

somewhat higher. The

off more rapidly than for the RC
amplifier at the low frequencies owing

in addition

to the high reactance of C.

The gain

to the

drops

RC

low reactance
off

of

more rapidly than

-x-Cz.

for the

amplifier at the high frequencies

because of the shunting effects of both

Cl and C g

The mid-frequency

Fig. 6-15. The equivalent circuit of a

typical stage of an LC coupled amplifier.

gain

higher than for the RC amplifier

provided that the impedance of the inductor at these frequencies
than the resistance of the plate load of the RC amplifier.
is

is

higher

The LC coupled

amplifier is seldom used as a potential amplifier owing

narrow frequency band and the cost of the inductor. This type
of coupling, which is also referred to as shunt- or parallel-feed coupling, is
frequently used in power amplifiers.
6-10. Transformer -coupled Amplifiers.
The circuit of a transformerto the

coupled amplifier

is

given in Fig. 6-16.

The transformer

as a coupling

Input

Fig. 6-16. Schematic diagram of a transformer-coupled amplifier.

device

possesses

several

desirable

features.

Owing

to

the

step-up

character of the transformer, a total amplification per stage greater

than the n of the tube can be achieved. Also, the d-c isolation provided

by the transformer automatically removes the requirement

for a blocking
High-quality interstage transformers generally have potential
ratios of 1:3 or less.
Higher transformation ratios usually are accom-

capacitor.

panied by distributed winding capacitances and by interwinding capacitances that are excessive.
The effects of these capacitances will be
considered below.

RADIO ELECTRONICS

136

The equivalent

[Chap. 6

circuit of a typical transformer-coupled stage

is

given

in Fig. 6-17.

An

approximate expression for the gain per

assumed that the transformer is ideal.
An ideal transformer is one for which unity coupling exists between
primary and secondary windings and in which the losses and stray
capacitances are negligible.
If it is assumed that the secondary of
Mid-frequency Response.

stage

may

readily be found

Fig. 6-17.

The

if it is

equivalent circuit of a typical transformer-coupled stage.

the transformer, which feeds the grid of the following stage,

then the plate-circuit impedance is infinite.

tially open-circuited,

essen-

is

Conse-

quently the full effective potential ixE g of the stage appears across the
transformer primary. The corresponding output potential at the second-

ary terminals
former.

is

ntxE g ,

The gain

where n

is

of the stage is

the transformation ratio

then simply

K=K =

of the trans-

nn

(6-40)

a constant, independent of the frequency.

L-F Region. The foregoing considerations are only approximate, owing
to the character of the assumptions that were made.
To examine the
response at the lower frequencies, the effective distributed winding

K
Fig. 6-18.

The

1-f

equivalent circuit of a transformer-coupled amplifier.

capacitances and leakage inductances may be neglected.

The corresponding equivalent circuit has the form shown in Fig. 6-18. The resistance of the primary winding is denoted as R h and its inductance is L
Similarly, i? 2 and L 2 represent the secondary resistance and inductance,
x

respectively.
It is

evident from this diagram that the primary current

is

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-10]

The potential

difference across the

--

e- -

primary inductance

137

is

::\)/ mL,

J(r

But the potential that appears across the secondary winding E bk

The secondary

is

may be in

phase or 180 deg

out of phase with E t and depends upon the relative winding direction
of the primary and secondary windings.
The gain of the transformercoupled stage is then
times as large as

ojfc

The

ratio of

1-f

to mid-frequency gain

=
which

may

voltage

j(r P
is

Ri)/<aLi

then

- j{rP + K

(6_43)
1

)/coL 1

be written in the form

where
/,

(6-45)

Observe that the gain ratio has the same general form as for the RC
coupled amplifier. Now the gain drops off at the lower frequencies
because the reactance of the primary winding decreases and no longer
will
rp

be large compared with the total resistance of the primary circuit

Ri.

Equation

(6-45) contains

an explanation why high-gain tubes, which

inherently possess high plate resistances, are not employed in transformer-

such high-gain tubes were used, then the 1-f

If an attempt were made to improve the
response by designing the transformer to have a very high primary

coupled amplifiers.

If

response would be very poor.

1-f

inductance, the associated distributed capacitances would

The

method

become

exces-

high primary inductance

through the use of core material of very high permeability has led to the
development of high-permeability alloys such as permalloy and hiperm.
sive.

H-F

alternative

of obtaining

At the high frequencies the shunting effects of the

may be neglected. However, the effects
the leakage inductances, and also the interwinding and distributed
Region.

primary and secondary windings

of

capacitances, are important.

The

circuit of Fig. 6-17

appears in Fig.

with the transformer replaced by its T equivalent, and referred to

the primary. At the high frequencies the circuit reduces to that of
6- 19a

RADIO ELECTRONICS

138
Fig. 6-196, in

which the winding, interwinding,

[Chap. 6

interelectrode,

and stray

wiring capacitances are lumped into an equivalent capacitance

across

the primary of the transformer.

The total effective shunting capacitance

ponents by the approximate expression

C =

[(1

ra)

Cm

2
ra

is

C2

related to the several

n*Cg

com-

(6-46)

The sign depends on the relative disposition and connection of the

primary and secondary coils. When properly connected, the minus

To

sign usually applies.

examine Fig. 6-19a.

justify this expression,

fa)

(b)

The complete and approximate

Fig. 6-19.

If

h-f equivalent circuit of a transformer-

coupled stage.

the potential across the input terminals

terminals

is

E2

is

Ei,

Ec* = Ei + E 2 = E,(l
where n

and that across the output

Cm

the difference of potential across

= N3/N1.

The energy
pp-

is

n)

stored in this capacitor per cycle

_ Cm E%

(1

is

+ nYE\

The equivalent capacitor across the primary terminals which will store
The other capacitances in
the same energy per cycle will be (1 + n) 2 Cm
C B ). The total
the secondary are reflected into the primary as n 2 (C 2
.

shunting capacitance is that given in Eq. (6-46).

An analysis of the approximate equivalent circuit yields for the gain
of the amplifier the expression
jnixXc

K, =

R + j(X L - X B
L = L\ + L' /n

where R = rv +
+ fl 2 /n 2
The primary leakage inductance
;

ance

is

L'2

The gain

ratio

of

XL

= wL;

Xc

1/aC.

and the secondary leakage inductmay be written in the form

K
The magnitude

(6-47)

L'1;

-R+j{XL -Xc)

the gain ratio

K2 =
Ko

is

(6-48)

is

Xc

Vr +
2

(x L

- xcy

(6-49)

UNTUNED POTENTIAL AMPLIFIERS

Sec. 6-10]

139

At the lower end of the region of frequencies where this analysis is

X L is small and Xc is large, so that the gain ratio approaches unity,
At the higher frequencies, Xc is small, X L is large, and the
as it should.
valid,

Notice, however, that the secondary circuit may pass

through a maximum, owing to a resonance condition that exists. The
maximum is found to occur when
gain falls to zero.

Xc

+R

= 2X1

2XL

R will be much smaller than L and the maximum occurs when

Xl, the condition for series resonance, But as the frequency at
which this resonance occurs is
Usually

Xc

wo

Vlc

then the corresponding value of the gain

V^O/mai

is

(l>oCR

(6-50)

typical frequency-response curve, the dependence on the primary

inductance L\, and the total leakage inductance

are illustrated in

Fig. 6-20.

Max

Fig.

6-20.

typical frequency-response

characteristic

of

a transformer-coupled

amplifier.

The peaking

of the transformer-coupled-amplifier

frequency-response

may

be suppressed to a considerable extent by several methods.

The simplest way consists in shunting the secondary of the transformer
curve

with an appropriately chosen resistance.

resonant peak, although

than

rifi.

it

Another method

This reduces the height of the

also causes the mid-frequency gain to be less

for

improving the response characteristic is

In modern transformers the shunt

in the design of the transformer.

capacitance

is

decreased by decreasing the contributing factors.

The use

grounded shield between windings reduces the interwinding capacitance.

The design of the windings reduces the distributed capacitances.
Further, the leakage inductance is reduced, so that the peaking will occur
of a

RADIO ELECTRONICS

140

[Chap. 6

beyond the normal operating range of the amplifier. This

reduction has been accomplished both by proper design of the windings
and by use of suitable high-permeability core materials. Also, the use
of a high-resistance secondary winding tends to suppress the resonant
at frequencies

peak.

is to be avoided owing to the

High-quality transformers are availdb over a range of frequencies from approxiThe peaking may also be reduced materially

high-resistance primary winding

adverse effect on the

able that are

fiat

1-f

response.

within

mately 20 to 10,000 cps.

through the use of inverse feedback in the amplifier, as will be shown
in Chap. 7.
Transformers are seldom used as interstage coupling devices merely to
obtain higher gain. This follows because the use of a pentode in an RC
coupled amplifier will ordinarily provide a higher gain than is possible
with a triode with a step-up transformer. Furthermore, the RC coupled
stage requires less space and

ment.

The

makes use

of relatively inexpensive equip-

principal uses of transformers are as the coupling stage

between the driver and a push-pull amplifier, so as to provide the required

potentials that are 180 deg apart in phase (although this application
is largely being supplanted by tube circuits), and also as the output transformer in a power amplifier. Here the principal function is one of impedance matching and permits the matching of low impedance loads to
These applications will be discussed in
high-internal-resistance tubes.
some detail in Chap. 8.

two

REFERENCES
1.

Terman, F.

E., Electronics, 10,

34 (June, 1937).

PROBLEMS
6-1. The important constants of one of a chain of
employing pentodes (see Fig. 6-3) are

Ri

75 kilohms

Cgk =

rp

11

ju/xf

10 6

ohms

CP k =

gm

coupled amplifier stages

1,600 /zmhos

Rg =

y-iii

RC

C =

0.01

yuf

500 kilohms

Calculate the mid-frequency gain and the upper and lower cutoff frequencies.
is the amplifier-stage phase 180 15 deg?
6-2. The frequency response of a three-stage cascaded RC amplifier employing
pentodes is to be constant within 0.5 db up to 18 kc. Calculate the h-f cutoff of
a.
b.

Between what frequencies

each stage.
6-3. Consider a chain of similar triode circuits in cascade.
limiting gain-bandwidth product is given by

R2B =

Qm_
2ttC,

What

can be said about the use of triodes in a cascade chain?

Show

that the

UNTUNED POTENTIAL AMPLIFIERS

Compute the figure of merit
6AK5, 6AC7 pentodes.

6-4.

6F4

triodes;

141

6AK5, 6AC7, 6J6,

any tube handbook for characteristics.)

of the following tubes:

(See

The

6-5. Refer to Fig, 6-10 showing a resistance-coupled amplifier.

constants are

Rn =

250 kilohms

= 0.5 ma, Eib =

-8 volts?

If Tbi

be

300

R e2 =
volts,

500 kilohms

R cl =

what must be the value

of

circuit

500 kilohms

cc if

of

T2

is

to

6-6. Calculate the gain of the series balanced d-c amplifier shown in the
diagram.

6-7.

The LC coupled

amplifier of Fig. 6-14 uses a triode.

The important

fac-

tors are
rp

Rg = 10 6 ohms
G
g m = 2,000 nmhos
40 henrys
Distributed capacitance = 200 nni

10 kilohms

L =

0.01

/if

Determine the upper and lower cutoff frequencies and the maximum gain.
6-8. A transformer-coupled amplifier is to be constant within 3 db over the
frequency range from 100 to 8,400 cps.
a. Specify the required values of primary inductance, leakage inductance
The tube
(reduced to unity turns ratio), and frequency of secondary resonance.
Neglect the winding resistance in the calculations.
is a 6J5 with r v = 7,700 ohms.
b. If the turns ratio is 3 and the total input and wiring capacitance of the next
tube is 25 nixi, what is the permissible equivalent capacitance across the secondary
of the transformer?

6-9. Obtain an expression for the output potentials of the cathode-coupled twotube circuit shown in the figure. Compare the results with Eq. (6-38) when Ri of

Tl

is

zero.

6-10. Calculate the input admittance and the output-terminal impedance of

the cathode-coupled amplifier when connected as shown in Fig. 6-12. Neglect
tube and wiring capacitances.

RADIO ELECTRONICS

142
6-11.

The

diagram.

circuit of an inverted amplifier

Calculate the following:

a.

The

6.

Input impedance.
Ratio of output to input power

c.

is

illustrated in the

gain.

if

Z e and

Zi are resistors

[Chap. 6

accompanying

CHAPTER

SPECIAL AMPLIFIER CONSIDERATIONS

NOISE
7-1. Noise in Amplifiers.
The term noise is used to describe anyspurious signals which ultimately appear in the output of an amplifier.

Noise may be produced by causes which may be external to, or which

might be internal to, the system.
External noise, which will be only superficially examined, includes
atmospheric interference with reception and occasional large amplitude
pulses.
Atmospheric interference arises principally from lightning discharges in the lower atmosphere.

Owing

magnitude of the fields

very large. However, such
interference is most troublesome at the longer wavelengths, although
considerable energy exists even in the short-wave band.
Since the noise
is present at all frequencies, the wider the bandwidth of the receiver, the
greater will be the energy picked up from the atmospheric disturbance.
If the interference arises from short pulses, it proves most desirable
to have a relatively wide band system, in order to avoid having the system
extend the pulse. It is possible to provide suppression circuits in a
involved, the effective area of disturbance

receiver to

to the
is

overcome to a considerable extent noise from

External noise

may

also be of the

sparking of any sort exists,

as, for

man-made

this cause.

variety and arises

when

example, the sparking at the brushes

equipment may be the source

Diathermy machines, which are generally
simple power oscillators, often with inadequately filtered power supplies,
produce considerable amounts of noise over a relatively broad band.
The ignition systems of automobiles and aircraft engines are' strong
sources of noise.
The noise from such sources is most severe in the 30- to
100-Mc frequency band. While screening of the ignition leads will reduce
this type noise, this is generally not done, except in mobile systems which
of a motor.

Also, poorly shielded electronic

of considerable external noise.

are directly affected.

In addition to the external sources of noise, a number of internal

It might happen that the amplifier is oscillating
because some part of the output is being fed back into the input. This
may be caused by a common impedance between stages, for example,
sources of noise exist.

143

RADIO ELECTRONICS

144

[Chap. 7

common plate supply.

Appropriately placed decoupling filters

prevent such interaction. Coupling may occur through the
Methods of neutralization have
interelectrode capacitances of the tubes.
been developed for avoiding this. If the coupling occurs through stray

through a

will usually

wiring capacitances, electrical shielding is usually resorted to in order to

prevent such coupling.
Some noise may appear in amplifiers the cathode heaters of which are
If the heat capacity of the cathode-heater assemfed from the a-c lines.
bly is too low, there may be some variation in cathode current due to the
This is usually low in indifluctuations in temperature of the cathode.
rectly heated cathode tubes, although it may be appreciable in tubes

heated type which are fed from an a-c source.

Noise, related to the power-line frequency, may also appear in an
Such
amplifier if the rectifier power supply is not adequately filtered.
radio
receiver.
home
ripple hum is often detectable in the average
Microphonics, which causes output potentials due to the vibration
of the electrodes produced by acoustic or mechanical jarring of the tube,
may be very serious in high-gain systems. The microphonic property
of a tube depends on the character of its construction, certain tubes being
considerably more microphonic than others. Special low-microphonic
tubes are available. The judicious use of shock mounts, loaded supports,
of the directly

or special vibration-isolation

methods

will usually

overcome

this problem.

7-2. Internal Noise.

Even with the elimination
sources of noise, a limitation exists to the useful amplification possible
of

The noise generated in

the minimum amplitude that can be

the foregoing

with an amplifier.

the receiver

limit to

amplified

signal/noise ratio

is

if

itself

sets a

a satisfactory

to be obtained in the output.

There are two fundamental sources of internal noise. The first results
from the fact that the electrons in the circuit elements of the amplifier are
This activity produces fluctuations
in the electron distribution within the elements, and this appears as a
The second cause arises within the tube
thermal-agitation potential.
itself and is produced by the random motion of the electrons in the
in a state of continuous agitation.

cathode-anode space.
Thermal-agitation Noise.

The

free or conduction electrons in a con-

random motion, the amplitude

of the motion being

of this random
Because
dependent upon the temperature of the circuit.
most-probable
distribustatistical
motion, slight deviations exist in the
within
are
produced
fluctuating
potentials
tion, with the result that small

ductor are

all in

the conductor. The small-fluctuation, or thermal-agitation, potentials in

the input circuit of an amplifier will give rise to an appreciable output if a
high-gain system

is

used.

Since these thermal-agitation motions are random,

it

is

reasonable

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-2]

145

to expect that thermal-agitation potentials of all frequencies are produced.

Consequently, the total rms potential expected in a circuit which

is

upon the bandwidth of the circuit, among

An analysis of this problem by Johnson and Nyquist 2
other factors.
shows that the square of the rms potential produced in a circuit component having constant resistance R over the frequency band used is
given by
frequency-sensitive will depend

= 4kTRB

(7-1)

where k is the Boltzmann constant (= 1.38 X 10~ 23 jouIe/K), T is the

temperature of the resistor in degrees Kelvin, R is the resistance in ohms,
and B is the pass band of the channel over which the noise is evaluated,
measured to the 3-db, or half-power, points. This expression is a special
case of the following

more general

where Re
noise

is

expression,

= 4kT

Re

(Z) df

(Z) denotes the real part of the

being produced and

is

(7-2)

impedance element

in

which

the effective resistance component of the

circuit.
If

the measuring circuit

is

an amplifier having a gain characteristic

the input circuit to the amplifier of gain

characteristic K(J), then the total rms noise is given by the expression
K(f), or

if

the noise source

is

4fcT

"

fQ

K*(f)

Re

(Z) df

(7-3)

Of course, noise is generated in all resistance elements of the circuit.

However, it is usually only the potential produced across the input terminals which is of importance, since it is this potential which is subject to
If this Johnson noise is
the full amplification of the amplifier stages.
then
noise in subsequent
the
input
signal,
sufficiently below the level of
amplified
output signal.
with
the
stages will be negligible compared
potentials
that are produced
of
the
To get some idea of the magnitude

by
k

choose the following values:

10" 23 joule/K
T = 300K R = 10 5 ohms

this source of noise,

then

1.38

it is

found that

EimB =
Shot Noise.

Among

B =

100 kc

4.0 fiv

the various sources of noise in the tube

itself,

most important. The shot effect results from the fact that
the current from the cathode to the anode consists of the flow of electrons,
and since the electron emission from the cathode surface may not be
completely uniform for a given cathode temperature, instantaneous
If
fluctuations may occur, even if the time average current is constant.
shot noise

is

RADIO ELECTRONICS

146

[Chap. 7

the tube is operating under space-charge-limited conditions, the irregularities in the cathode emission, and so in the arrival of the electrons at the
anode, are considerably reduced.
The magnitude of the shot noise in a temperature-limited diode can
be calculated, and as a result, temperature-limited diodes have been used
as sources of noise.
An analysis of this problem yields the following
expression for the noise produced in such a temperature-saturated diode, 3
7"2

= Y\2eI

B)

(7-4)

where Ib = d-c current, e = electronic charge, B = bandwidth, and

T = constant depending on the randomness of the electron flow ( = 1 for
a completely

and

less

random distribution as exists in a temperature-limited

1 if some degree of space-charge limitation occurs).

diode,

than

A quantitative treatment of shot noise in multielectrode tubes working

under normal conditions is difficult. It is possible to deduce certain
significant results from physical reasoning.
The total shot noise will be
proportional to the anode current.
On the other hand, the signal in
the output will depend on the transconductance g m
It appears, therefore, that for a large signal/noise ratio the tube should have large g m
and yet draw a small anode current. It is likewise found that the partition of the cathode current between the screen and the anode also
produces noise. Consequently pentodes with small screen current are
best from this point of view (the beam tube is probably best in this
regard.)
In general, triodes operate with a lower noise than a com.

parable-type pentode.

convenient to specify the shot and partition noises produced

tube in terms of a fictitious resistance R ei which, when placed in the
grid line, will produce in the plate circuit the same noise by thermal agiIf R denotes the parallel combitation as is produced by the shot noise.
nation of r v and Ri of the given amplifier, the equivalent noise potential
squared in the output is
If is often

in a

PR =
2

But from Eq.

T 2 (2eIb B)R 2

(7-1) for a noise resistance

ing potential squared in the output

K E>
2

If

R eq in the input,

the correspond-

is

= (4kTR elB)glR*

these expressions are equated, the equivalent grid noise resistance

is

found to be
eq

But

since I b varies directly with g m

_TH
2kTg*

it

would be expected that

R eq

should

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-2]

vary inversely with gm

It

has been shown, in

R eq =

2 5

fact,

147

that for triodes

ohms

(7-5)

Qm

As noted above, tetrodes and pentodes have more shot noise than do
owing to the random interception of electrons by the screen grid.
The expression for the equivalent noise resistance R eq has been shown to
triodes,

be 4

+
r^-r
h + I a (~
\g m

(7-6)

gl )

h is the plate current in amperes, 7 c2 is the screen current in

amperes, and g m is the transconductance in mhos. These formulas enable
a direct comparison to be made between the input circuit noise and the
shot and partition noise.
Typical values follow:
where

Type

Req

tube

Triode
Tetrode and pentode
Mixer and converter

ohms

200-1,000
5 000-20 000
200 000-300 000
,

Another important source of noise in the grid circuit

from the motion of charges in the grid circuit due to the motion
of electrons between cathode and anode.
The following formula has
been developed for the mean square induced grid noise, 6
Other Noise.

arises

7|

where I g

is

= AkTG g fiB

the noise current in amperes and

(7-7)

Gg

is

the electronic portion of

the input conductance in mhos.

This implies that, for noise purposes,

the induced grid noise is represented by a current generator I g shunted
The value of is approximately 5.
by a conductance G g
As indicated in the table above, a tube used as a mixer or converter is
This greater noise arises
noisier than the same tube used as an amplifier.
from the effects of the local oscillator injection. As is discussed in Sec.
As a
13-13, the oscillator provides several volts to the converter grid.
Since the
result, the g m of the tube varies over a very wide range.
equivalent noise resistance R eq and the average input conductance G g
depend upon the average gm of the tube during the injection cycle, then
.

because of the wide excursions in g m the resulting noise is high.

Another important source of noise in vacuum tubes is the flicker

effect.

This effect arises from the changes of emission from various portions of
It has been found that the flicker effect produces
components that are most important at the lower frequencies, and

the cathode surface.

noise

the resultant noise varies roughly inversely with the frequency.

It

may

RADIO ELECTRONICS

148

be large in oxide-coated cathode tubes, where

it

[Chap. 7

overshadows the regular

shot noise.

Another source of noise arises from the random neutralization of space

charge by the comparatively few molecules normally present in a vacuum
tube.
In a tube with a good vacuum this effect produces about as much
Secondnoise as thermal-agitation noise in the plate circuit of the tube.
ary emission from the plate and grids in a vacuum tube also introduces

some

noise.

customary to specify the quality of a receiver

This quantity measures the noise generation
within the receiver circuits. The noise figure F is a measure of the ratio
of the actual available noise power output of a receiver to that of an ideal
receiver which is free of noise sources, the only source of noise being the
thermal-agitation noise that has been applied to the input terminals of
The term "available" denotes that the
the receiver by the antenna.
power source is matched to the load for optimum power transfer. By
definition, F is the ratio of the available signal to noise power at the input
(Si/Ni) to the available signal/noise power at the output (S /N ); thus
7-3. Noise Figure.

in

terms

It is

of its noise figure.

F =

Ww.

(7 " 8)

But the available power gain K p of the receiver is the ratio S /Si. MoreThus
over, by definition, the available noise power input Ni is kTB.
Eq.

may

(7-8)

be written as

F =

N
kJTb

(7_9)

This expression shows that the noise figure is the ratio of the available
output noise power of the actual receiver to the available output noise
power of an otherwise ideal receiver. The noise figure is often expressed
in decibels, as Fab

Many

10 logio F.

factors contribute to the over-all noise figure of a receiver;

the input circuit, the mixer (in a superheterodyne receiver), the beat-

frequency

oscillator,

and

others.

The mixing problem becomes more

the higher frequencies, with the result that receivers for very
h-f use (in the microwave bands) usually have higher noise figures at
the higher frequencies. For example, a representative noise figure of a
radar receiver having a gain of 120 db and a bandwidth of 2 Mc and
difficult at

operating at 3,000

Mc

is

15 db.

Of

this figure several decibels of noise

arise in the mixer.

Networks in Cascade. Suppose that a network

cascaded with a network of noise figure Ft. An
expression is desired for the noise figure F u of the combination. As
already discussed in Sec. 7-2, the noise in the input circuit to the first
7-4. Noise Figure of

of noise figure

Fi

is

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-4]

network

is

usually the most significant source of noise, and

pated, therefore, that the noise figure

nearly equal to Fi, particularly

network

is large.

To examine

if

F i2

of the

the power gain

combination

Kp

149
antici-

it is

will

be very

of the first, or input,

the matter in some detail, refer to Fig. 7-1.

The input to the first

In Fig. 7-1 two networks are shown in cascade.
is shown as an equivalent noise resistance and would be the

network

equivalent output-terminal noise resistance of the antenna or other noise

This resistance supplies an available noise power kTB to the

The available power at the input terminals of the second
network is evidently JcTBF\K p i, and the available output power at the
source.
first

network.

output of the second network is kTBFiK pl p2

In addition to the noise produced in the input circuit to network 1
there is an available noise power produced in the input circuit to network
2.
This results in a component available power at the output of net-

work 2

of

kTBF 2 K p2
Network

However, a certain part

of this latter available

Network 2

lcTBFzKpJ -

Fig. 7-1. Determination of the noise figure of

two networks

in cascade.

power has already been included in the term involving Fx. To avoid
counting the available power kTBK p2 twice, it must be subtracted from
the total available power at the output of network 2.
Therefore the
noise output due to the second network is kTBF 2 K p2 kTBK p2
The
over-all noise power available is
.

kTBF 12 K pl Kp2 = kTBFiK pl Kp2

from which

it

+ kTBF K p2 - kTBKp2
2

follows that

Fu =

Fi

F 2

+ K

(7-10)

As anticipated, the over-all noise figure F 12 will be very nearly equal to

However, if K pl is small, then the
Fi if the gain of network 1 is large.
over-all noise figure will depend on F 2
In the development of Eq. (7-10) it was assumed that the bandwidths
.

of both networks were equal.

Careful consideration of this matter will
show that this expression is also valid if the bandwidth of the first network is greater than and contains the bandwidth B of the second network.
The reasoning that led to Eq. (7-10) can be extended to yield expressions for the case of three or more networks in cascade.
Specifically,
for three networks in cascade, the over-all noise figure is

Ft

+ K
p

1
i

F3 -

Kp\K p2

(7-11)

RADIO ELECTRONICS

150

[Chap. 7

FEEDBACK IN AMPLIFIERS*
combined with the input
feedback

effect of the

back
nal

is

is

When

Feedback. 6

7-5. Principles of

is

signal,

a part of the output signal

feedback

is

said to exist.

If

is

the net

to increase the effective input signal, the feed-

called positive, direct, or regenerative.

If

the resultant input

reduced by the feedback potential, the feedback

is

sig-

called negative,

inverse, or degenerative.

The

principle of feedback

is illustrated in the schematic diagram of

For simplicity, series injection is shown at the input, but other
forms of network coupling may be employed. In the diagram shown, a
potential Ej is applied to the input terminals of the amplifier.
Suppose

Fig. 7-2.

that the resultant potential at the output terminals

Amplifier
nominal gain

Now suppose

2.

C2 (ouipuf)

is

Feedback network
Feedback fraction

f}

Fig. 7-2.

The

principles of feedback in amplifiers.

that a fraction (5 of this output is fed back in series with the input signal
in such a way that the resultant signal that appears between the grid and

cathode terminals has the form

E,

But the nominal gain

K=

Ei

?E 2

of the amplifier is given

(7-12)

by

E2

output potential
potential between grid and cathode

Then

E 2 = KE

(7-13)

Observe that the nominal gain requires the injection of a potential E a

between the grid-cathode terminals, with an evaluation of the output
potential E 2 with the
network acting as part of the total output load
(5

of the circuit.

Equation (7-13)

is

combined with Eq.

E = KEi
2

from which

it

K(3E 2

follows that

KEj
2

1
*

-f-

(7-12) to yield

(7-14)

K0

Part of the contents of this section was originally prepared with Dr.

tor the second edition of "Electronics,"

McGraw-Hill Book Company,

York, 1951, although the material was not included in this

text.

J.

Millman

Inc.,

New

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-5]

The

resultant gain of the amplifier with feedback

f _
Therefore

it

output potential
input signal potential

is

_ E

151

defined as
2

Ei

follows that

K' =
r^K?

This equation expresses the resultant gain of the amplifier with feedback

terms of the nominal gain of the amplifier without feedback K, and

the feedback fraction [3.
It is noted that often in a practical situation
the feedback path is so involved that it is not possible to isolate the [5
network. The discussion to follow still applies in principle, but a direct
application of the mathematical expressions may not be possible.
It can be seen that if |1 K(?| is greater than unity, then K. f is less
than K. The feedback is then said to be negative, or degenerative. The
application of negative feedback to an amplifier results in a number of
It tends to
characteristics that are highly desirable in the amplifier.
flatten the frequency-response characteristic and to extend the range of
uniform response. It materially reduces nonlinear and phase distortion.
It improves the stability of the amplifier, making the gain less dependent

K/

in

on the operating potentials or on variations of the tube characteristics.

it tends to make the gain less dependent on the load, so that load

Also,

variations do not seriously influence the operating characteristics of the

The use

amplifier.

of feedback

networks

of special design will provide

selective attenuation, thus permitting a frequency response of desired

characteristic.

Conversely,

The feedback

if

is

jl

detailed discussion of these features will be given later.

K(J| is less

now termed

than unity, then K/

positive, or regenerative.

is

greater than K.

The

application

feedback has effects opposite to those with negative feedback.

Thus positive feedback tends to sharpen the frequency-response curve
and to decrease the range of uniform response. This permits an increased
Positive feedback in any amplifier is critical of
gain and selectivity.
adjustment. Too much regenerative feedback in any system may result
Ordinarily, negative feedback is more common than posiin oscillation.
tive feedback in amplifiers, although oscillators of the feedback variety
depend for their operation on the presence of positive feedback.
Observe that, for the case when Kg = 1
jO, the gain becomes infiof positive

nite.

In this case the amplifier becomes an oscillator, and the output

independent of any external signal potential.

is called to the fact that the action of a feedback path
depends upon the frequency of operation. That is, the feedback may
remain regenerative or degenerative throughout the range of operation
of the circuit, although the magnitude and phase angle of the feedback
signal may vary with the frequency.
It is also possible for the feedback
potential

is

Attention

RADIO ELECTRONICS

152

[Chap. 7

to be positive over a certain range of frequencies and negative over

another range of frequencies.

Example. The circuit of a simple triode amplifier with an impedance in the

cathode lead is illustrated in Fig. 7-3. This circuit is to be analyzed by two
methods. One method is a direct application of the feedback equation [Eq.
The second method is a direct application of electron-tube circuit
(7-5)].
principles.

Fig. 7-3.

simple amplifier with cathode degeneration.

Refer to the equivalent circuit of the amplifier which is given in

Observe that a part of the output is fed back into the input circuit
through the impedance Z k
It follows from the figure that
Solution.

Fig. 7-36.

But

E,

Ex

E2

IZ t

(7-16)

since
(7-17)

Ri
then

E
If this expression is
it is

compared with Eq.

Ei

=-

E2

(7-18)

ni

which defines the feedback

(7-12),

fraction,

seen that

(7-19)

Ri

Note

also that

rP

Zk

(7-20)

Ri

and the nominal gain then becomes

K=
The

resultant gain

K/

is,

=
1

by Eq.

rv

Zfc

(7-21)

+ Ri

(7-15),

-iiRil{r v

- Kg

-liRi

Zk

+R

r,

Zk

-uRi

{)

-ixRi

rP

Ri

Gi

l)Z t

+R

(7-22)
t

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-6]

155

These results follow of course from direct considerations of the equivalent

circuit.

The

plate circuit yields the expression

+ I(rp + Z +

UES

Ri)

(7-23)

But

E5 =
The solution

E!

+ IZ

(7-24)

of these equations gives

+ IZ*) + I(r, + Z + R
k

t)

from which

~
and the resultant gain

r,

0*

1)Z*

+ flj

'" 25)

is

K/ =

E =

fiRi

ET

rp

0*

(7 " 26)

l)Z7+fii

which is the same as above.

Often Z k consists of the parallel combination of R k and Ck the value of Rk
being so chosen that hR k is just equal to E c the quiescent d-c basis of the tube.
The capacitor C k is so chosen that its reactance is very small over the operating
As a result, Z k is very small and may be omitted in the
range of the amplifier.
In this case, the usual simple amplifier formula is obtained,
above expression.
since the feedback factor (3 is zero, and no feedback exists.
,

7-6.

The presence

Feedback Amplifier Characteristics.

feedback in an amplifier results in a

number

of negative

of desirable characteristics.

These are discussed below.

Suppose that the feedback is negative

1. Stability of Amplification.
and that the feedback factor K(S is made large compared with unity.
The resultant gain equation (7-15) becomes

K
This

>*-h =

-\

means that when the magnitude Kfi ^>

with negative feedback

network only.

is

1,

(7 " 27)

the actual amplification

a function of the characteristics of the feedback

In particular, if (3 is independent of frequency, then the

be independent of the frequency. This permits a sub-

over-all gain will

frequency and phase distortion of the amplifier.

of feedback network, it is possible to achieve
a wide variety of frequency characteristics.
= K/K(J
K, so that the over-all
Note that if Kp
1, then
f
gain of the amplifier with inverse feedback is less than the nominal gain
without feedback. This is the price that must be paid to secure the
advantages of negative feedback. This is not a serious price to pay,
stantial reduction of the

In fact,

by the proper choice

since the loss in gain

Clearly,

if

K/3

is

can be overcome by the use of additional tubes.

greater than unity, then Eq. (7-27) shows that the

RADIO ELECTRONICS

154
over-all gain will not

[Chap. 7

change with tube replacements or with variations

Even if Eq.
(5 is independent of the tube.

in battery potentials, since

(7-27)

not completely valid, a substantial improvement results in genThis follows from the fact that a change in the nominal

is

eral stability.

gain

dK

for

whatever reason results

in a

change dK/ in the resultant gain

by an amount

dKf _

Kf
where

|1

K(J|

This equation
sion,

dKf/Kf

is

dK

1
|1

Kff|

(7-28)

the magnitude of the quantity

represents

the logarithmic derivative of Eq. (7-15).

gives the fractional change in

fractional change in K.

Kf

and

for example, the quantity

If,

K(5.

In this expres-

dK/K
|1

gives the

K($|

5 in

a particular feedback amplifier, then the variation in any parameter that

might cause a 5 per cent change in the nominal gain will result in a change
of only 1 per cent in the resultant gain of the amplifier.

Reduction of Frequency and Phase Distortion. It follows from Eqs.

and (7-27) that the over-all gain of the amplifier is almost independent of frequency, provided that (3 is frequency-independent. In such
2.

(7-15)

cases the frequency and phase distortion of an amplifier are materially

reduced below the nonfeedback value.
One effect was omitted in the
3. Reduction of Nonlinear Distortion.
above considerations. It was implicitly assumed that the dynamic curve
was linear and that the output potential was of the same waveshape as
the input. If an appreciable nonlinear distortion exists, then the output
contains harmonic components in addition to the signal of fundamental
frequency.
Suppose, for simplicity, that only a second-harmonic component B 2 is generated within the tube when a large signal potential is
impressed on the input. Because of the feedback, the second-harmonic
component B'2 that appears in the output is different from that generated
within the tube. To find the relationship that exists between B\ and B 2
the procedure parallels that for the gain considerations.
Thus, for a second harmonic B 2 in the output, a fraction B'2 is supplied to the input.
As a result, the output actually must contain two components of secondharmonic frequency, the component B 2 that is generated within the tube
and the component K(32?'2 that arises from the signal that is fed back to
the input. This requires that
,

K$B'2

+B =
2

B\

or

B'2

Note that

since both

and

($

j-A^

(7-29)

are functions of the frequency, in general.

Sec. 7-7]

SPECIAL AMPLIFIER CONSIDERATIONS

the appropriate values that appear in this equation

155

must be evaluated at

the second-harmonic frequency.

It should be pointed out that this
harmonic distortion generated within
grid swing of the fundamental signal
additional distortion that might arise

derivation has assumed that the

the tube depends only upon the

The small amount of

because a fraction of the secondharmonic component is returned to the input has been neglected. Ordinarily this procedure will lead to little error, although a more exact calcupotential.

account is readily possible. 7

Another feature of Eq. (7-29) should be noted. According to this
expression, if |1 K(3| = 10, then the second-harmonic distortion with
feedback is only one-tenth its value without feedback. This is the situation when the total output-potential swing is the same in each case otherwise the harmonic generation within the tube could not be directly compared.
This requires that the signal, when feedback is applied, must be
As a practical consider|1 K($| times that in the absence of feedback.
ation, since appreciable nonlinear distortion is generated only when the
signal potential is large, then the full benefit of the feedback amplifier in
reducing nonlinear distortion is obtained by applying negative feedback
lation taking these successive effects into

to the large-signal stages.

4.

Reduction of Noise.

Considerations such as those leading to Eq.

feedback amplifier will

(7-29) for the resultant nonlinear distortion in a

show that the resultant noise generated in the input to an amplifier chain
reduced by the factor 1 (3K, when feedback is employed. This would
seem to represent a real reduction in noise. However, if the requirement
is for a specified output signal, the resultant gain with feedback will have
to be adjusted, by adjustment of the circuit parameters or by the addition of amplifier stages, to give the same over-all gain as the amplifier
is

Consequently, the noise will be amplified as well as

Moreover, since the noise is independent of the signal, additional amplifier stages to compensate for the loss of gain due to feedback
will introduce additional noise.
In such cases, the over-all noise of the
amplifier with feedback might be higher than one without feedback.
If
the required gain is achieved by the readjustment of the circuit parameters, a reduction in noise will result in the negative feedback amplifier.
Noise generated at some intermediate point of the amplifier chain, say
at a point whose gain relative to the output point is Ki, will be reduced
by the factor Ki/(1 - K(J).
5. Modification of Input and Output (Effective Internal) Impedances.
These topics will be the subject of detailed consideration in several of
without feedback.
the signal.

the following sections.

7-7.

Feedback

Circuits.

the amplifier into the input

The

may

potential fed back from the output of

be proportional either to the potential

RADIO ELECTRONICS

156

[Chap. 7

across the load or to the current through the load.

In the first case, the
feedback is called potential feedback; in the second case, it is called current feedback.
In either case, the feedback may be positive or negative,
depending upon the connection. Often the feedback loops are so
involved and interconnected that it is not possible to specify directly
whether the feedback is of the potential or the current types or whether
a combination of both exists.
It is possible to state rules

which help to specify more uniquely the

existence of potential or current feedback.

7-4,

which

Consider the circuits of Fig.

two amplifiers employing current feedback. The

these diagrams is identical with that of the illustrative example

first of

illustrate

E,

Fig. 7-4. Circuits employing current feedback.

cathode impedance

is

now shown

in the example, the feedback ratio

is

(5

of the foregoing section, except that the

as a resistance

Note

Rk

As

K/ =
Therefore the output potential

2a

h.

(7-31)

Rk

which

is proportional to the load impedance.

given by

which

is

- -

Also, the output current

<

seen to be independent of the load impedance.

tions are characteristic of current feedback.

rent feedback the ratio of the feedback potential

pendent of the load impedance.

The

(7-30)

Rk

is

E ~

is

= R k /Zi.

that, for large feedback ratios, the resultant gain approaches

These condi-

Hence with negative

to the

cur-

load current is inde-

condition that the output current should be independent of the

is fulfilled when the internal impedance of the generator

high compared with the impedance of the external load. Conse-

load impedance
is

quently, negative current feedback has the property of increasing the

impedance of the network. In fact, from the complete expres-

internal

sion for the current, from Eq. (7-25), namely,

Sec. 7-7]

SPECIAL AMPLIFIER CONSIDERATIONS

-/tEi

rp
it is

possible to

draw

(n

which

Fig. 7-5,

follows from this that the

l)R k
is

+Z

the equivalent of Fig. 1-Za

source

E (/ =

nal impedance

juEi

-o-

with an inter-

rp

(n

En

l)Rk-

Since the internal impedance without

R k the effect
feedback is simply r p

of the

feedback

is

The
impedance by the term nR k
and without feedback is given by

nal

A
this

Fig.

The equivalent

7-5.

circuit

of

Fig. 7-3a.

to increase the inter.

hi

It

circuit in-

cluding feedback comprises a potential

157

rp

ratio of internal

(n

l)R k
=
Rk

which employs potential feedback

circuit, the resistance combination 13 R

circuit

\e2

Zl

impedances with

flRk

+R

rp

(7-33)
k

given in Fig. 7-6.

is

(1

p)R =

In

which

\ti-m

i/3R

Fig. 7-6. Circuit employing potential feedback.

shunts the output is made large compared with the load impedance Z;.
The capacitor C has a reactance that is negligible compared with R at
Its sole purpose is to block the d-c
the frequencies to be employed.
potential from the plate circuit from appearing in the grid circuit.
The feedback ratio is shown as (S in the diagram. Also for large feedback ratios, the resultant gain approaches

The output

potential

is,

therefore,

E2

Ei

which is seen to be independent of the load impedance, since [$ is independent of Zi. Observe therefore that potential feedback is directly proportional to the output potential, and the ratio (3 of the feedback potential

RADIO ELECTRONICS

158

[Chap. 7

A generator whose outoutput potential is independent of the load.

put potential is substantially independent of the load impedance must
Consequently negative potential
possess a very low internal impedance.
feedback has the property of decreasing the internal impedance of the
to the

amplifier.

obtain an expression for the resultant gain of the amplifier, the

To

feedback method will be employed. By neglecting the shunting effect

of the feedback resistance network on the load impedance, it follows that
the nominal gain of the amplifier is given by

rp

Then the

resultant gain with feedback

This expression

may be

2,i

is

transformed to the form

(7-36)

where

But

l+rt

this is exactly the

output that

Consequently, the circuit behaves

is

obtained from the circuit of Fig.

like a potential source

E tf =

with an internal impedance Z v

fjL \

t+jup

Ej

C^)

r/(l

^)+

7-7.

The equivalent

circuit

of

Ei

impedance

without feedback

is

The effect of potensimply Z = rp

tial feedback is to reduce the internal
impedance

Fig.

internal

effective

of the amplifier
.

Z *[^I >E2

_^

The
1

7-7.

in the ratio
7...

TT79

(7 " 37)

From the form of Eq. (7-36), the circuit gain appears to be that
obtained from a tube whose amplification factor is y' and whose plate
Note that the

is reduced
This indicates that
a tube possessing a high plate resistance can be effectively converted into
a low-plate-resistance tube and thereby permit an impedance match to a
low impedance load. This is accomplished, of course, at the expense of
effectively converting the tube into a triode, with low n and low r p
The combination of current and potential feedback in an amplifier is

resistance

in the

is r'
v

same

effective amplification factor

ratio as the plate resistance of the tube.

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-7]

The

frequently called compound, or bridge, feedback

amplifier

given

is

The feedback

Fig.

in

fraction

circuit of

such an

7-8.

H(-

found to

is

159

be
5

+
,

ffi

Rk

As in the analysis
assumed that the
bination
Zi

9t

Zi

is

(7-38)

of Fig. 7-6, it is

resistance

much

and that the reactance

capacitor

is

com-

A circuit employing bridge,

compound, feedback.
Fig. 7-8.

greater than
of the

The

negligible over the frequency range of operation,

result-

ant gain of the amplifier has the form

nZi

K, =
This

may

{n

l)R k

(1

n$i)Zi

be written in the form

y
K? =
r' '

" =

where

6i

+
rp +

(7-39)

The corresponding equivalent

(,

circuit

l)R k

m3i

shown

exactly this expression for the gain and

Fig. 7-8.

The

effect of

in the effective n

and

the feedback

r v of

is

is

seen to reflect

The

the tube.

in Fig. 7-9 gives rise to

therefore the equivalent of

as a change

itself

effective potential

and internal

impedance are given by the expression

(J

Z tf =
Fig.

7-'

+
+

(m

n&i

l)R k

(7-40)

equivalent circuit

Owing
Z tf

Fig. 7-8.

to the form of the expression

this quantity may be made

value without feedback.
Feedback can be effected over several stages and need not be limited
A two-stage .RC-coupled amplifier which
to a stage-by-stage practice.
combines current feedback in the first stage through resistor R t and

for

greater than, equal to, or less than

potential feedback between stages

its

is

illustrated in Fig. 7-10.

careful

consideration of the polarity of the potentials which are fed back will

show that both types

of

feedback are negative.

RADIO ELECTRONICS

160

[Chap. 7

It is not always evident what type of feedback is being employed in a

given amplifier. The following tests will serve to clarify the situation:
1. If the ratio of feedback potential E/ to output potential E 2 is nearly
independent of the load impedance, then potential feedback is employed.

= E//E 2
This ratio is the feedback fraction
2. If the ratio of feedback potential E/ to load current I is nearly independent of the load impedance, then current feedback is employed. The
ratio Z/ = E//I is the feedback impedance.
3. If the feedback potential E/ is the sum of two terms of the form
(5

E,

= ?E

Z/I

and Z f are nearly independent

both
compound feedback is employed.

where

(5

of the load impedance, then

Fig. 7-10. A two-stage RC coupled amplifier with current feedback in the

potential feedback between stages.

7-8. Effective Internal

Impedance with Feedback.

The

first

stage and

discussion in

the foregoing section has shown that the effective internal impedance of
the equivalent plate circuit of an electron-tube circuit with feedback

depends on the type of feedback that is employed. As shown, current

feedback increases the effective internal impedance, and potential feedback decreases the effective internal impedance. These results will be
generalized.

The

following notation,

some

of

which has already appeared,

will

apply

in the following development:

($

is

feedback ratio

is

potential gain without feedback, with load connected

K/

is

potential gain with feedback, with load connected

K(

is

potential gain without feedback, with load open-circuited

is

E/

is

effective internal potential source without feedback (this is the

Helmholtz-Thevenin potential source obtained on open circuit)
effective internal potential source with feedback
input-terminal impedance without feedback
corresponding input-terminal impedance with feedback

Zi

is

Zif

is

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-8]

Z< is effective internal

impedance without feedback

161

(this is the

nin impedance of the equivalent network and

is

TheVe-

the impedance

looking back into the output terminals of the amplifier, with the
load open-circuited)

Z/
Z
Z /

is

corresponding effective internal impedance with feedback

is

output-terminal impedance without feedback

is

output-terminal impedance with feedback

is

output potential

is

input potential to amplifier

feedback potential

Ei
E/

is

A for a general discussion of the Helmholtz-Thevenin

Refer to Fig. 7-11, which shows a general feedback network

(Refer to Appendix

theorem.)

Amplifier
nominai gain

Feedback
network
3
Fig. 7-11.

The

ft

general potential or parallel feedback circuit.

which is provided with potential feedback. An expression for the interimpedance Z v of this feedback network will be derived in terms of the
internal impedance Z without feedback.
Consider first the amplifier with the feedback potential removed. This
is accomplished by removing lead A from the feedback network and connecting it to the cathode K. The
QThevenin potential-source equivanal

lent of this circuit is given in Fig.

En
Eg
diagram is the
effective internal impedance without feedback, and K is the gain
Fig. 7-12. The equivalent circuit of Fig.
without feedback on open circuit
7-11 with feedback removed.
(with Z; omitted from the diagram).
To deduce the equivalent circuit of Fig. 7-11 with feedback present,
note that the effect of feedback appears in the form of the potential E.
With feedback, E 9 = E x + (JE 2 Clearly,
Without feedback, E a = E x
from this discussion, the equivalent circuit of Fig. 7-11 with feedback
It should be noted that this figure,
present is that shown in Fig. 7-13.
even though it is the equivalent circuit of Fig. 7-11 when feedback is
present, is not a Thevenin potential-source equivalent representation,
because the apparent Z and K(Ei + (JE 2 ) are functions of the load.
7-12.

in

this

RADIO ELECTRONICS

162

[Chap. 7

Note from the diagram that

E =
2

K,(Ei

0E 2 )

K,Ei

IZ

Therefore

E 2 (l -

QKt)

IZ

0K,

or
1

The Thevenin

&K

(7-41)

equivalent network for the circuit with feedback

ing to this expression, that

shown

is,

accord-

in Fig. 7-14.

-oKt(E,+pE2 X

a
Fig. 7-13.

The equivalent

circuit of Fig. 7-11

with feedback present.

As a check of these results, it is noted that

E 1 /(1 (3K ) represents
the open-circuit potential with feedback, K/Ei. This agrees with the
result obtained in Eq. (7-15) that K, = K/(l - (JK).
The internal

impedance with potential feedback

is

z,

Z tf =

(7-42)

0K,

1 K is greater than unity, the impedthan that without feedback. Note, moreover,
that the effective internal impedance is reduced by the same factor as
the gain, when feedback is applied.

Since, for negative feedback,

ance with feedback

The Thevenin equivalent network

Fig. 7-14.

It

is

is less

of Fig. 7-11.

interesting to apply these results to the potential-feedback cir-

is open-circuited, then the magnitude of the

= n. Also, the
simply the m of the tube and
internal impedance without feedback is r p whence Z = r p
The results
so obtained agree with those in Fig. 7-7.

cuit of Fig. 7-6.

If

gain of the circuit

the load

is

It is now desired to examine the results of current or series feedback

on the effective internal impedance. Refer to Fig. 7-15, which shows a
general feedback circuit with current feedback. Although the impedance Z f is shown isolated from the remainder of the circuit, it is a part

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-8]

feedback circuit and

of the

is

not part of the external load

shown

equivalent circuit will be of the form

Amplifier
nominal gain

The general current

Fig. 7-15.

gram Z

is

163

The

;.

In this dia-

in Fig. 7-16.

if

-o-

or series feedback circuit.

the total internal impedance looking back from the load and
Z s From Fig. 7-16 it follows that

includes the effect of

K (Ei +
(

IZf )

I(Z (

Z,)

from which
K,E,
Z/K,

But this is the current that exists in the circuit

Thevenin equivalent with current feedback.

of Fig. 7-17,

which

is

the

-D-

K < E=Kt rE/IZ/j(

Fig. 7-16.

The

The equivalent

with feedback present.

impedance

effective internal

with current feedback

circuit of Fig. 7-15

is

of the equivalent TheVenin generator

thus seen to be

Z = Z

K,Zf

(7-43)

Note that the open-circuit potential with feedback

open-circuit

the

potential without

feedback, in view of the significance

of

This result

is

the observation that

consistent with
Z;

if

is

re

K Ei,

which equals

removed

from Fig. 7-17, an open circuit reand I, and so the feedback, is

sults,

FlG 7-!7- The Thevenin equivalent of

-

Fig.
s 7-15.

zero.

These

results are applied to Fig. 7-4.

open-circuit gain

back

is

is

Z =
t

rp

is

+R

K
k.

p..

The

If

the output

is removed, the
impedance without feed-

internal

Hence

Z/ =
t

which agrees with the previous

rp

+ R + nR

result.

RADIO ELECTRONICS

164

Example 1. Analyze the amplifier

when the output is taken across Ri.

of Fig. 7-18

[Chap. 7

by the use

(b)

(a)

Fig. 7-18.

The

circuit of a specific example.

The equivalent circuit is given in Fig. 7-186.

Solution.
the grid-cathode potential E is given by the expression

E =
But

since

E*

of feedback methods,

Ei

It

is first

noted that

E*

IR*, then

E*

Rk
I

which

is

independent of the load impedance.

This indicates, according to the

criterion given in Sec. 7-7, that current feedback exists.

To analyze the circuit completely, it is

To evaluate these requires an evaluation

K/ and Z^.
Note from the

desired to calculate both

of

K,

(5,

(,

and

Z,.

equivalent circuit that

E2

The nominal gain

is

Ri

found by noting that

^+(E

-E,)(l +

i-)=-|

But
R2

E*
R,

Then

^. + E,(l + |)(l + l)

+ ||

or

Therefore

_1_

Ri

rp

R +
t

(Ri

+
rv

fl)(fli

R\

(7-44)
rP)

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-8]

The equivalent gain on open

is

+ R J R2 + r "
...p
r P + iti

r Rl
=- -~
+
r p + Ki

7 = K
T>
Z
2

_L

It follows

simply

circuit is

Also the equivalent impedance

fll

R +

(Ri

(Ri

rp

Ri
if
Ri

r P Ri

+R

)(R

,"7 Ad\
( -46)

from these expressions that the gain with feedback

165

rp)

rp)

is

QmRl

+R

)(Ri

TpRi

This reduces to

-fiRiRi

K =
'

Also, the effective internal

Zj/

r(g.

may

l)Bii

+ RiRi

(7

is

fo)

r p gi

fiRi

Ri

rp

Qi

Ri

l)BiB,

(7-18)

effects of the

The gain

ratio,

feedback on the gain and on

given by the ratio of Eq.

+ R + Ri) + RiRi + RiR

+ R + Ri) + RiRi + 0* + l)RiR

r(Bi
r p (i

Ri)R 2
rp

examine the

interest to

K/

which

(m

the effective internal impedance.

(7-47) to Eq. (7-44), is found to be

fii)

impedance

(r>

Z" =
some

fi 2

= Z K/Z/
_

It is of

r,(fl!

be written in the form

K.

"

fiRiR 2
1

r p (Ri

+R + Ri) + RiRi + R1R2

2

This expression shows that the resultant gain with feedback is less than thai
without feedback, as expected.
In a somewhat similar way, the effective internal impedances may be compared,
to examine the effects of the feedback.
By Eqs. (7-48) and (7-46), the ratio is
readily found to be

Zf _
Z
t

The

effect of the

feedback

is

nRJR 2

r p {R x

+R +
t)

RA

to increase the effective internal impedance, which

characteristic of negative feedback.

is

RADIO ELECTRONICS

166

[Chap. 7

Example 2. Analyze the circuit of Example 1 when the output is taken across
the cathode resistor i? 2
In the present case, since E* is the output potential, then since
Solution.
.

Ee =
it

Ei

Efc

follows that

= -i

But since (J is independent of the load, then potential feedback now exists.
The nominal gain is obtained from a study of the equivalent circuit of
7-186.

It is

Fig.

observed that

^+

-E,)(l +

(E 2

i-) =

Also

Ri

E2

Ek

=r-

Xl2

Then

**-*(1+ s)te + s)-s"

or

+ R )(R +

(R 2

r P)

fpRi

Therefore

K=

= =
E

_L_ rpi?!

(fl,

^2
which

+R

)(R l

+ r,)

*"pjRl

is

ixRiRz
r P Ri

The equivalent gain on open

(Ri

Tp

Also the equivalent impedance

It follows

)(Ri

(7-49)
rp)

circuit is

K =

+R

(7-50)

til

(r p

is

'pKj

RQRi

r P fli

from these expressions that under feedback conditions

liRrRt/lrpR!

K/
1

.RiRi/[r v Rx

(Rt

+ B )(fli + r p
+ R )(Ri + r)\
2

(Ri

)]

which reduces to
1

r p (R,

R,

R{)

<ji

ljfl.fl,

"

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-9]

The corresponding

effective internal

impedance
(r p

0K
1

(r p

it is

rP

Ri

(7-53)

{lx

draw certain conclusions from a comparison of the results

and 2, the same conclusions are possible from the simpliin Fig. 7-19, in which Ri = i? 2 and Ri is set to infinity, or

possible to

obtained in Examples
fied circuit illustrated

an open

+ r p R,
+R

+ gQfl; + rpR!
+ + l)Rt

rP

While

is

Ri)Ri
rP

Z =

167

This circuit

circuit.

known

is

as a

single-tube "paraphase" amplifier and provides two equal output potentials of opposite

polarity from a single excitation source.

For
the case when the output potential is E 2 the
,

significant
(7-47)

and

deduced

expressions

from

Eqs.

(7-48) are the following:

V=
Zu =

-uRi
rv
rv

+
+

(ft

+
+

When the output potential

Eqs. (7-52) and (7-53),

(7-54)

2)fl;

l)B
is

Fig. 7-19.

phase"

Et

single-tube

para-

the appropriate expressions become, from

Ri

K, =
rp

+
+

Ri

Z tt =

amplifier.

(ji

2)Ri
(7-55)

It will be observed that the gain of the amplifier with respect to

each output

the same.
However, it is also noted that the effective internal
impedances looking back from these terminals are quite different, one being much
higher than the other.
pair of terminals

is

7-9. Effect of Feedback on the Output-terminal Impedance.

The
output-terminal impedance of a circuit is the impedance looking back
into the output terminals of the network when the load impedance is in
place, but with the input potential reduced to zero.
Clearly, the out-

put impedance of an amplifier

is the parallel combination of the effecimpedance and the load impedance. Since the equivalent
internal impedance Z (/ depends on the type of feedback that is incorporated in the amplifier, then the output impedance will also depend on

tive internal

the type of feedback.

The

situation

is

illustrated

schematically in

Fig. 7-20.

The output impedance Z

/,

which

is

given as the ratio of the current I

RADIO ELECTRONICS

168

when a

into the output terminals

potential

is

[Chap. 7

impressed,

is

clearly

Zt/Z;
'0/

Ztf

E0/
where, for the case
Eq. (7-42),
Fig. 7-20.

The output-terminal

impedance

of

the

by

Z,
1

feedback amplifier.

and

(7-56)
Zi

of potential feedback,

Z =

general

by Eq.

for the case of current feedback,

Zi/

(7-43)

= Z K Z/
(

It is desired to obtain an expression for Z

ance without feedback, Z where

in

terms of the output imped-

= JnZn
Z

(7-57)
Z,

a. Potential Feedback.
Suppose that the input source to the general
feedback amplifier is reduced to zero and that a potential source is applied
This
to the output terminals. The situation is illustrated in Fig. 7-21.

Feedback
network
Fig. 7-21. The modifications of the general potential feedback amplifier for calculating
the output-terminal impedance.

output-impedance deterIn view of Fig. 7-14, which gives the equivalent circuit of
the general amplifier with potential feedback, then Z has the form of

diagram

is

Fig. 7-2 appropriately modified for

minations.

Eq. (7-57).

The current

Io

from the applied source

and the

effective

is

similar in

seen to be

K(?E

output impedance with feedback

z" which

Eo

is

form to Eq.

- r^TE5

(7-15)

is

<"">

This shows that the output imped-

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-10]

ance

is

169

reduced by the same factor as the potential gain with the applica-

tion of potential feedback.

The calculation for the output impedance of an

which employs current feedback follows a similar pattern. In
this case, as before, the input signal is reduced to zero, and a potential
The current-feedback circuit
source is applied to the output terminals.
for the output-terminal impedance calculation then becomes that shown
b.

Current Feedback.

amplifier

in Fig. 7-22.

Fig. 7-22. The modifications to the general current feedback circuit for outputterminal impedance determination.

In this circuit Z denotes the output impedance of the circuit without

K is the gain without feedback,
feedback and includes the effect of Z f
but with Z; in position. The potential E is the drop across Z f and
.

is

I/Z f

It follows

from the diagram, by taking account

of the current

through

the load impedance, that

Eo

But

this

IqZq

KI/Z,

KZ/

^Io

becomes

Eo

IoZo

2^

This gives

Io(Z

from which

it

-K

KZ/) = Eo [1

follows that the effective output

Zo/

KZ/
Zo
=
1

KZ//Z,

Z,

**" 1

impedance with feedback

- KZ//Z
- KZ//Z;

is

(7-59)

Feedback on the Input-terminal Impedance. It is of

some importance to examine how the input impedance of an amplifier is
It will be found that the effective
affected by the presence of feedback.
potential
and current feedback.
input impedance increases for both
It follows directly from Fig. 7-11 that the
a. Potential Feedback.
input-terminal impedance with feedback is simply
7-10. Effect of

Ex

RADIO ELECTRONICS

170

may

This

[Chap. 7

be written as
Zl/

*!

But the input impedance without feedback

(1

KB)

is

E
Z -
Then

Zv =

Zx(l

Kg)

(7-60)

Therefore, owing to the feedback, the input impedance with feedback

is

greater than the input impedance without feedback, and in the same

degree as the gain and distortion decrease.

As a specific example, suppose that Zi is the impedance due to a capacitance between the grid-cathode terminals, and this may be the actual
tube capacitance modified by the Miller effect. Since the impedance
increases with feedback, this
is

decreased.

means that the

Cu =
b.

Current Feedback.

back, but

effective input capacitance

Clearly, therefore,

now with

By

(7-6D

j-^f

proceeding as in

(a) for

the potential feed-

reference to Fig. 7-15 for the general current-feedback

circuit,

Ei

_ E

IZf

4(e--H
But

=
Then

it

E = KE,
2

follows that

Zv =

Zi(l

Kg)

(7-62)

Note that the input impedance with current feedback is greater than the
input impedance without feedback in the same degree as for the case of
potential feedback.
7-11.

Feedback and Stability. A great deal of information about the

an amplifier can be obtained from an analysis of the factor

stability of
1

Kg

that appears in the general gain expression [Eq. (7-14)].

This

is

best analyzed through the use of the polar plot of the expression Kg.

This is
Observe that
the total open-loop gain, including the amplifier and the feedback

Attention

is first

best examined

Kg

is

by

called to the significance of the quantity Kg.

reference to the diagram of Fig. 7-23.

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-11]

171

network, but with the feedback connection open. In network parlance,

open-loop transfer function o. the ampl fier and the feedback
network. In essence, therefore, consideration of the open-loop performance of the amplifier and feedback network is to be used to provide significant information regarding the performance of the amplifier under
this is the

closed-loop operation.

Kg

a function of the frequency,

is

and, in general, points in the complex

plane are obtained for the

values of

Kg

corresponding to

Amplifier

gain K.
1

all

values of / from to
The locus
of all these points usually forms a
.

closed curve for amplifiers.

Feedback
network

+^

Fm

7 . 23

Illustration of ihe significance

As a particular example, suppose of the factor Kg.

that the locus of Kg in the complex
plane is drawn for the amplifier illustrated in Fig. 7-6.

To do this, the
complete expression for the nominal gain, including the effect of the feedback circuit, must be written, rather than the simple form given in Eq.
Also, the value of g must include the effects of the blocking
(7-34).
Certain of the features of the response of this amplifier are
capacitor C.
known. At the mid-frequencies, the gain is substantially constant and
has a phase of 180 deg. For the low and high frequencies, the gain falls
to zero, and the phase approaches + 90 deg, respectively.
At the 1-f and
h-f cutoff values the

phase

is

+ 135

deg, respectively.

It

may

be shown

v6

Fig. 7-24.

The

locus in the complex plane of K(5 for the circuit of Fig. 7-6.

that the general locus of

The

result is

shown

Suppose that a phasor

This

is

Kg

of this amplifier for all frequencies is a circle.

in Fig. 7-24.

the quantity

is

drawn from the polar

Kg, as shown.

For

locus to the point (1J0).

this particular case, its

magnitude is greater than unity for all frequencies, and it has its maximum magnitude at the middle range of frequencies. Moreover, since
the resultant gain varies inversely with 1 Kg, then the effect of the
feedback

is

to cause a general flattening of the frequency-response

characteristic.

RADIO ELECTRONICS

172

The

and negative feedback

criterion for positive

First note that the quantity

plex plane.

circle of unit radius

with

Clearly,

if

|1

is

evident on the com-

Kgj

represents a

center at the point (l,jO), as illustrated.

its

feedback
gain.

[Chap. 7

for a given amplifier

|1

K(}|

>

1,

then the

negative, with an over-all reduction of

is

Likewise,

if

|1

K(3|

<

1,

there

is

an over-all

and the feedback is positive. These

considerations show that if K(? extends outside of the
unit circle for any frequency, then the feedback is
increase in gain

Fig. 7-25.
of

|1

K3|

A
=

plot
1

negative at that frequency.

If K(J lies

within the unit

in

then the feedback is positive. If K(5 passes

the complex plane.
through the point (1 JO) then 1 K(5 = 0, and, as will
later be shown, the amplifier is unstable and oscillates.
A more general
analysis by Nyquist 6,8 shows that the amplifier will oscillate if the curve
K(3 encloses the point (1 JO) and is stable if the curve does not enclose
circle,

this point.

That is, if the magnitude

than unity when its phase
zero, no oscillations are

of K(3 is less

angle

is

possible.

As a specific example for discussion,

suppose that the plot of a given amplifier

is

that illustrated in Fig. 7-26.

The feedback

is

negative for this am-

the frequency range from

Positive feedback exists in the

plifier in

to/i.

Fig.

7-26.

The

polar

plot

of

an

amplifier.

Note, however, that since the locus of

frequency range from /i to
K(5 does not enclose the point (lj'O), then, according to the Nyquist
criterion, oscillations will not occur.
7-12. The Cathode Follower. 9 The cathode follower is illustrated in
This feedFig. 7-27 a, the equivalent circuit being given in Fig. 7-276.
.

Fig. 7-27. Schematic and equivalent circuits of the cathode follower.

back

circuit is singled out for detailed consideration

because of

its

exten-

These applications stem from the

fact that the cathode follower possesses a high input impedance and a
sive use in a variety of applications.

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-12]

low output impedance and

may

173

therefore be used as a coupling device

between a high impedance source and a low impedance load.

The cathode follower is similar to the single-tube paraphase amplifier
The equivalent circuit is that
of Fig. 7-19, but with a zero plate load.

shown

The

in Fig. 7-276.

grid circuit

isolated, since interelectrode

is

capacitances are neglected, and the input impedance

tive internal
if

ix

1,

impedance

then Z (/

is

r p /n

Z v = rp /(n +
= l/g m But
.

from 1,000 to 10,000 /imhos, then Z v

is

1),

which

is

since g m for

is

g.

The

effec-

very low. In fact,

most tubes varies

of the order of 1,000

ohms

to

Fig. 7-28. Schematic and equivalent circuits of the cathode follower, including the
interelectrode capacitances.

double cathode-follower circuit has been devised which

has a greatly reduced effective internal impedance. 10
The gain is obtained from an analysis of Fig. 7-276 and is given by
100 ohms.

/iZfc

=
rp

(m

fl

1)Z*

r p /Z k

(7-63)

approaches the limiting value m/ (m + 1) as the ratio r p /Z k

For tubes with large value of n, and
rp
approaches zero, or as Z k
Clearly

the gain approaches unity. For values of Z k and r p found

In fact, it is
is of the order of 0.9 or higher.
in most normal cases,
because of this unity-gain feature that the circuit derives its name, since
the output potential is almost equal to the input potential, whence the

with Z k

p,

cathode and grid rise and fall together in potential by almost equal
amounts (or the cathode follows the grid)
The interelectrode and wiring capacitances have been neglected in these
discussions, as their effects are usually negligible for frequencies below
about 1 Mc/sec. For purposes of our study, these will be taken into
The schematic and equivalent circuits are now given in
account.
Fig. 7-28.

The

expression for the gain of the amplifier

the circuit.

It is

E Y Clk

Ye,*
it

deduced by analyzing

noted that

E2

But

is

+ Yp +

mE,Y p
Yc ph +c,t + Y Zl

follows that

Ei

E2

(7-64)

RADIO ELECTRONICS

174

and Eq.

[Chap. 7

becomes

(7-64)

jcoC^Et

v =

+ Cph

jo,(C gk

Solving for the gain

K/ which

+ mY p (Ei E
+ Cft + l/r p +
2)

(jur v C gk

v
jur p Z k (Cg k

For those values

of

C pk

Z k which

= E

given by K/

is

(7
'- 0>
{

l/Z k

/Ei, there results

n)Z h

Cf k)

rp

(ju

(7-66)

l)Z k

are normally used, the effect of the inter-

electrode and wiring capacitances on the potential amplification

below about 1 Mc, as already noted.

seen by writing Eq. (7-66) in the form

ligible for frequencies

so

is

K,

(gm

+ juCgk)Z

That

is

neg-

this

is

(7-67)

+(~^ +iC )z
r

But the effect of the capacitances will

where C T = Cgk + Cpk + Cfk
become important only for those frequencies for which o>Cr becomes comIf CV is taken as 30 ppf and gm = 1,000
parable with (n + l)/r p = gm
==
gr
=
jiimhos, then /
5 Mc.
m /2xCr
To find an expression for the input capacitance of the cathode follower,
It is seen that the current flowing through the source
refer to Fig. 7-286.
comprises two components. One of these is the current through the
capacitance C gv and is
.

11

The second

is

= jwC^E,

the current through the capacitance

But as E = Ei

- E

and

Kf = E

and the

is

(7-69)

/Ei, then

Ii

+ I = MC^E, +
2

effective input capacitance

many

This

(7-70)

is

d
Since in

= juC gk (l - K^Ei

I2

total current

C gk

= jwCe *E e

12

The

(7-68)

circuits

K/

= Cgp
is

(1

K,)C,*EJ

is

(1

K,)C gk

approximately

0.9,

(7-71)

then C; has the approxi-

mate value
Ci

= C gp +

0.1C o4

(7-72)

comparison of this expression with the corresponding form given by

Eq. (5-16) for the conventional amplifier stage shows a roughly similar

Sec. 7-13]

SPECIAL AMPLIFIER CONSIDERATIONS

175

dependence on the tube capacitances, although the numerical value for

is considerably smaller than that for the conven-

the cathode follower

tional amplifier stage.

The

impedance

effective internal

can be determined by finding the

current I as a consequence of the

Ztf

application of

an

a-c potential

to

the output terminals of Fig. 7-28.

The

grid exciting potential

The equivalent

zero.

drawn

as Fig. 7-29,

impedance
admittance

of

is

G,P,0

made

Fig. 7-29. The equivalent output circuit of the cathode follower.

circuit is that

the internal

if

the grid driving source

of the

tube alone
Io

But under the conditions

is

The

low.

is

effective internal

found from

EoYt

mE

-|

(7-73)

specified

E = Eo
Then
Ytf

+ Yp + gm

= YT

(7-74)

where Yr = jwCr- It is of interest to compare this result with that

which applies without capacitances being considered, viz.,

Y tf =

^=
1

Qm

~T"

Yp

The effect of the interelectrode capacitances is the addition of the term

Here, as for the gain, Y T does not become comYr to the terms Y p + gm
.

parable with the other terms except at the

higher video frequencies.
7-13. Graphical Analysis of the Cathode

Suppose that the cathode ima resistance R k and this is the

usual situation. A graphical solution of
Follower.

pedance

7-30.

the operation of the circuit is possible on

the plate characteristics of the tube. This

The

cathode
lower with a cathode resistor
Fig.

is

necessitates drawing the dynamic charfrom the known plate characteristics. Refer to
7-30 for notation.
The controlling equations of the grid and plate

acteristic of the circuit

Fig.

circuits are
6n

Gc n

~\~

tbnRk

(7-75)

&bn

+ IbnRk

(7-76)

and
Ebb

RADIO ELECTRONICS

176

[Chap. 7

Equation (7-76) is the equation of the load line for the plate supply Ebh
and the load resistance Rk. The procedure for constructing the dynamic
characteristic follows
1.

On

(7-76).

the plate characteristics draw the load line specified by Eq.

This

is

illustrated in Fig. 7-31.

Fig. 7-31. Graphical construction for finding the

follower with cathode resistor Rk.

dynamic

characteristic of a cathode

2. Note the plate current at each point of intersection of the load line
with the plate characteristics. For example, the current at the intersection of the load line with E c2 is labeled 4 2
3. Now relabel the plate characteristics with the appropriate symbol e,
according to Eq. (7-75). Thus
-

+iM

E -+ei{ = Ecl
E e (= Eel. +
el

C2

>

%lRk)

etc.
4.

The dynamic

where % n

is

characteristic

is

a plot of the (4,e) characteristic,

the current corresponding to the input

culating the value of e n for each value of

E cn

and

e.

its

This requires calcorresponding i in

.

Often the complete dynamic curve is not required, but only the current I b for a specified value of e, say E.
By Eq. (7-75) this is

_ E - E
b

For several values

of i b is calculated,

Fig. 7-32.

The

of e c (and the available

and noted on the plate

intersection of the load line

values are used), the value

characteristics, as

and the

shown

in

connecting the
calculated points is the appropriate current
for the specified E.
It should be specifically noted that the value of the input signal E will
be quite large before E c the actual grid-cathode potential, becomes positive with the consequent grid current.
That is, since the cathode potential follows the grid potential rather closely (for a gain almost equal to
line

unity), the input signal

may

swing considerably positive before the onset

SPECIAL AMPLIFIER CONSIDERATIONS

Sec. 7-13]

The

of grid current.

able positive swing.

larger the values of

When

Rk

cutoff occurs,

177

the larger will be the allowno potential difference appears

,

Consequently, the applied signal required to reach cutoff

across Rk.

independent of

Rk

is

Fig. 7-32. Graphical construction for obtaining the value of

potential Ei.

for a specified input

Example. Consider a 6J5 tube with E bb = 300 volts and R k = 10,000 ohms.
Find the maximum positive and negative input swings for positive grid-cathode

and

potential

Solution.

cutoff, respectively.

From

Eu and

specified

the plate characteristics of the 6J5 (see Appendix B-9) and the

Rk, the following data are found:

For
For

E =
E =
c

0:

It

-18:

h=

15.7

ma

This shows that the cathode follower may swing from +157 volts to
without drawing grid current or driving the tube beyond cutoff.

18

volts

Clearly, the operation of the cathode-follower circuit of Fig. 7-30 is

unsymmetrical. For small potential excursions, this causes no difficulty.

Two ways

Fig. 7-33.

Also,

if

ever,

if

lish

of achieving

more symmetrical operation

of a

cathode follower.

only positive signals are to be used, no difficulty exists. Howit is necessary to estab-

large negative signals are to be applied,

the grid at a large positive potential with respect to the bottom end

Ec will still be
be achieved in several ways, as illustrated in
For symmetrical operation, the bias will be established to set
Fig. 7-33.
the d-c level across R k at about half of the peak-peak potential swing.
of

Rk

(ordinarily ground), although the actual tube bias

negative.

This bias

may

RADIO ELECTRONICS

178

[Chap. 7

REFERENCES
Johnson, J. B., Phys. Rev., 32, 97 (1928).
Nyquist, H., Phys. Rev., 32, 110 (1928).
Moullin, E. B., "Spontaneous Fluctuations of Voltage," Oxford University

1.

2.
3.

London, 1938.
H. T., Proc. IRE, 32, 419 (1944).
North, D. O., and W. R. Ferris, Proc. IRE, 29, 49 (1941).
Black, H. S., Elec. Eng., 53, 114 (1934).
Peterson, E., J. G. Kreer, and L. A. Ware, Bell System Tech.
Press,

4. Friis,
5.
6.

J., 13,

680

(1934).
7. Feldkeller, R., Teleg. fernsp. Tech., 25, 217 (1936).
8. Nyquist, H., Bell System Tech. J., 11, 126 (1932).

9.

Bode, H. W., "Network Analysis and Feedback Amplifier Design," D. Van

Nostrand Company, Inc., New York, 1945.
Goldman, S., "Transformation Calculus and Electric Transients," Prentice-Hall, Inc., New York, 1949.
Reich, H. J., "Theory and Application of Electron Tubes," 2d ed., sec.
6-11, McGraw-Hill Book Company, Inc., New York, 1944.
Reich, H. J., Proc. IRE, 35, 573 (1947).
Kraus, H. L., Electronics, 20, 116 (January, 1947).
Schlesinger,

10.
11.

K,

Hammock, C,

Electronics, 21, 103 (February, 1948).

MIT Radiation Lab.

Rept. 469 (1943).

For an extensive treatment of noise, see:
Goldman, S., "Frequency Analysis, Modulation, and Noise," McGrawHill Book Company, Inc., New York, 1948.

Moullin, E. B. (see

ref. 3).

PROBLEMS
7-1. a. Evaluate the

rms thermal noise potential across the

parallel

network

shown.
b.

Comment on

the results.

R
c
Determine the rms shot noise current in a temperature-limited diode
bandwidth for an average tube current of 10 ma.
this current flows through a temperature-limited diode, how much noise

7-2. o.

in a 10-kc
6.

If

developed in a 10-kilohm series resistance in the diode circuit?

7-3. Consider the circuit shown, for which r v = 7,700 ohms, g m = 2,600 ^mhos,
Ri = 25 kilohms, and C is chosen to yield a 12-kc amplifier bandwidth.

is

SPECIAL AMPLIFIER CONSIDERATIONS

179

Direct-current sources have been omitted for convenience. Calculate the

when R = 0; when R = 10 kilohms.
7-4. Compare the value of R eq of a 6AC7 when triode-connected with the value
when pentode-connected. The operating potentials are those specified for normal
operation.
7-5. Consider a resistor having a resistance R at room temperature.
When
this resistor is connected in the circuit that includes a temperature-limited diode,
passes through R.
Under these circumstances the effective
the d-c current
Calculate the d-c potential across R.
fluctuation noise is doubled.
7-6. Consider a simple grounded-grid amplifier stage to have fluctuation noise
to be produced in the grid circuit, the cathode resistor, and both the plate resistance and the plate load resistance. Neglect transit-time effects.
a. Find an expression for the mean-square noise in the output.
b. Denote the ratio of the mean-square noise potential of (a) to that of an
ideal tube (one with zero grid and plate noise) by the quantity N.
For a specified Ri find the approximate value of R k for which
is a minimum.
Caution:
Recall that the component potentials are random.
c. Deduce the value of N^ in under the conditions of (6).
7-7. A radar receiver has an over-all bandwidth of 1.2 Mc, a noise figure of
Find the rms noise potential across a 500-ohm out11 db, and a gain of 110 db.
noise potential in the output

put resistance.
7-8. A type 1N21 silicon crystal diode is used in the converter of a microwave
receiver.
If the converter has a loss of 5.5 db and a noise figure of 9 db and is
used with an i-f amplifier that has a noise figure of 3.5 db, calculate the over-all
Assume that the converter band contains the i-f band, which is
noise figure.
the usual case in radar receivers.
7-9. An amplifier has a gain 3,000/0.

gain

is

When

negative feedback

is

applied, the

Determine the feedback network.

reduced to 2,000/0.

7-10. An amplifier without feedback gives an output of 46 volts with 8 per cent
second-harmonic distortion when the input is 0.16 volt.
a. If 1 per cent of the output is fed back into the input in a degenerative circuit,
what is the output potential?
b. If an output of 46 volts with 1 per cent second-harmonic distortion is permissible, what is the input potential?
7-11. Given the amplifier stage with cathode degeneration shown in the accompanying diagram,

=
=
Rg =

Eu,
gm

a.
b.

250 volts
1,200 /imhos
1

R,

=
=

100 kilohms
70

megohms

What should be the value of Rk to give an over-all gain of 8?

What is the value of Ecc and the largest value of e g to yield an output without

distortion?

RADIO ELECTRONICS

180

[Chap. 7

7-12. Plot the gain as a function of frequency of the simple amplifier shown in
the accompanying figure. Also plot on the same sheet the gain of the stage when
fixed bias is used.

7-13.

The

E =
b

first

The second stage

The other factors

Rn =

C =

stage of the circuit of Fig. 7-10 uses a 6SJ7 pentode with

250 volts

0.04 M f

is

ccl

= -3

a 6C5, with

volts

Ei,

250

Ecc2 =
volts,

100 volts

cc \

= 8

h = 3 ma
volts, h = 8

ma.

are

100 kilohms
Ckl = 10

R =
R gX = R g2 = 250 kilohms
CM = 2.5 /if
Cd = 0.1 juf
= 150 kilohms
Rkl R k2 Ebb2

juf

ffi

25 kilohms
= 200 ohms

fij

Specify the values of

b. Draw the complete mid-frequency equivalent circuit.
Calculate and plot a
c. The total shunting capacitance across R g2 is 80 nnf.
gain-frequency-response curve over the range from 20 to 50,000 cps.
d. Repeat (c) if Ri = 0, R 2 = 150 kilohms.
7^14. Given the two-stage circuit which is provided with negative-potential
feedback. The tubes have r P = 10" ohms, g m = 1,200 /imhos.
a. Calculate the output impedance.
6. Calculate the impedance between points AA.
a.

7-15. Given a simple pentode amplifier stage as illustrated, the screen by-pass
Derive an expression for the gain of the amplifier
capacitor being omitted.
Assume that
is independent of Eb and that y.. g of the screen grid is
stage.

the same relative to plate and to screen currents.

SPECIAL AMPLIFIER CONSIDERATIONS

7-16. Calculate the gain of the inverse feedback pair.*
are identical

and that R S>

181

Assume that the tubes

Ri.

7-17. Obtain an expression for the output potentials of the cathode-coupled

two-tube circuit shown in the accompanying figure.

An RC coupled amplifier has a mid-frequency gain K = 27. Potenfeedback is applied, the amount of feedback being such as to reduce the lower
half-power frequency to one-tenth its no-feedback value.
o. Find the feedback fraction /3.
b. Find the resulting mid-frequency gain under feedback conditions.
7-19. Given a three-stage RC coupled amplifier, each stage of which has an
Plot
1-f cutoff of 20 cps, and h-f cutoff of 84 kc, and a mid-frequency gain of 220.
7-18.

tial

the locus of the complex potential gain.

7-20. The locus of the complex potential amplification of a certain amplifier is
illustrated.
If 1 per cent negative feedback is applied, determine the value of
the gain K at the following frequencies: 100 cps, 10 kc, 40 kc. Assume that p is
independent of frequency and that the potential fed back is in phase with the

output potential.
f*S700^

'Kj/OOO
f=!00.

7-21. Three identical

RC

coupled amplifier stages are connected in cascade,

inverse feedback being provided between the output and input, as indicated in
the diagram at the top of page 182. The per stage constants are (see Fig. 7-4
for a typical stage)

Ri

Tube: r v

=
=
=

20 kilohms
20 fi/ii
1

megohms

G. R. Mezger, Electronics, 17, 126

C =
=
gm =

(April, 1944).

0.01
1

juf

megohms

1,000 /imhos

RADIO ELECTRONICS

182
a.

Determine the

maximum

value of

for

[Chap. 7

which the amplifier

will

operate

without oscillations.
b.

Calculate the mid-band gain under these conditions.

3 stage R C
coupled amplifier

E,

-I

7-22. Consider the amplifier stage illustrated in the accompanying figure.

Calculate the following (neglect the effects of interelectrode capacitances)
a. Input impedance Z v
Compare this with the value R a
.

b.

Effective internal impedance Z,f

c.

Output-terminal impedance Z

/.

gm

1500/j.mhos

Tp =66 kn

1100 ka

6SF5,

Ev

'JMn

E2

I5kn

7-23. The circuit of Prob. 7-22 is rearranged by connecting the grid resistor
to the cathode rather than to ground, as shown.
Repeat Prob. 7-22 for the

rearranged circuit.

gm -l500jjjnhos

rp =66kn

7-2i. a. Refer to the cascode circuit illustrated.

following: gain K, output-terminal impedance Z /.
b.

and

Compare the
its

results

cathode resistor

under

(a)

Find expressions for the

with the corresponding results when tube Tl

are replaced

by a

resistor

t.

SPECIAL AMPLIFIER CONSIDERATIONS

7-25. Calculate the effective input

the
rp

of the circuit

in

Each tube has the value g m = 2,000 /tmhos,

Neglect all tube and wiring capacitances.

accompanying
10 kilohms.

impedance at 10 kc

183

figure.

7-26. Derive an expression for the effective input impedance of the cathodecoupled amplifier shown. Under what conditions is this impedance negative?

7-27.

What must be the

6J5 tube

is

value of

R k in Fig.

7-27

if

Z 0/ =

300 ohms at 1,000 cps?

used.

R g tied from
accompanying figure. Derive expressions for
the effective input impedance Zy and the output-terminal impedance Z /.
Neglect tube capacitances. Show that Zi/ can be written in the form
Find an expression for K.
Zi/ = R g /(l K).
7-28. Given the cathode-follower circuit with the grid resistor

grid to cathode, as

7-29.

shown

in the

Compare the values

of gain, effective input impedance, effective internal

impedance of the two cathode-follower stages

illustrated.

RADIO ELECTRONICS

184

[Chap. 7

7-30. Calculate and plot as a function of frequency on semilog paper the gain,
output impedance, and input impedance of the cathode-follower amplifier shown

at the following values of w: 250; 2,500

Choose a 6J5 tube

rp

7,700

for

ohms

ft

250,000; 2.5

10 6

10" rad/sec.

which

20

C =

3.4

wi

C sk =

3.4 nnf

plc

3.6 nrf

Also choose

Rk =

10 kilohms

200 kilohms

C =

0.01

juf

shown in the accompanying diagram.

The positive signal that will drive ec to zero.
The negative signal to drive the tube to cutoff.
The mid-frequency gain.
The input admittance when Cgp = 3.4 fi/xi, CBk = 4 /ijuf.

7-31. Consider the circuit

a.
b.
c.

d.

40 mil

Determine:

2S0

7-32. Repeat Prob. 7-31

when the tube

is

changed to a 6AC7 pentode.

The

linear region, with g m = 9,000 jumhos, r v = 1 megohm.

7-33. Plot the dynamic characteristic of a 6J5 tube in a simple cathode-

tube operates in

its

Ebb = 250 volts, for the following values of cathode resist5,000 ohms; 25,000 ohms; 100,000 ohms.
7-34. A video amplifier is coupled to a cathode follower, as shown in the figure
at the top of page 185. The frequency-response curve of this amplifier is also
sketched.
Choose the transconductance g m = 9,000 /xmhos.
a. Calculate the approximate value of the coupling capacitor between the two
follower circuit, with

ance

Rk

stages.
6.

Calculate the approximate value of the total shunt capacitance.

SPECIAL AMPLIFIER CONSIDERATIONS

100

1.0

0.1

1000

Frequency, cpsxIO 3

The essentials of a double cathode follower 10

Show that the expression for the gain is

7-35.
a.

K,

M
b.

Show

+ nr p/Ri
+ 1 + (m +
/i

=
2

that the output admittance

Ri

2)r p/R,

is

+ 1)
+ r,/R,

m(m
r

are illustrated.

186

RADIO ELECTRONICS

7-36. Consider the double cathode follower shown in the diagram.

the gain and effective output-terminal impedance of this amplifier.

[Chap. 7
Calculate

CHAPTER

UNTUNED POWER AMPLIFIERS

The

potential amplifiers discussed in Chap. 6 are designed to increase

a potential signal from a low level to one which is adequate for operatSuch amplifiers are generally operated in
ing some low-power circuit.
class

since the amplification

is

to be accomplished without distortion.

power amplifier serves to supply an appreciable amount of power to

some power-absorbing circuit, although in general it must be accomPower amplifiers
plished under very low grid-driving-power demands.
may be operated as class A, B, or C or at any point between these limits,
the choice of operating conditions being determined by the ultimate
If the amplifier is to reproduce the audio
spectrum without distortion, then the amplifier must be operated in class
A if a single tube is used. If two tubes are used in a push-pull circuit,
then the amplifier may also be operated in class AB or class B. If the
amplifier is to reproduce the input waveshape over a very narrow range of
Only
frequencies, tuned class B or tuned class C amplifiers may be used.
a-f amplifiers will be considered in this chapter.
The basic schematic diagram
8-1. Class A Triode Power Amplifiers.
of a typical series-fed power amplifier and its equivalent circuit is given

purpose of the amplifier.

Fia. 8-1.

The schematic and equivalent

in Fig. 8-1.

It is

circuits of a simple series-fed

observed that this circuit

is

power

amplifier.

identical with that of Fig.

6-1 for the simple potential amplifier.

If it is

assumed that the dynamic curve

of operation,

then the plate current

is

=
Ri
187

is

linear over the entire range

given by

(8-1)
rp

RADIO ELECTRONICS

188
since E

Ei,

and the power supplied to the load

P = PR =
t

[Chap. 8
is

Ri
4rp (1

(8-2)

Ri/rp yr p

A sketch showing the variation of the output power as a function of

is

mum

Ri/r p

can be seen that the power curve reaches a maxiat the point at which R /r p = 1, although this maximum is quite

given in Fig. 8-2.

It

broad.

P is a slowly varying function

of Ri in the

neighborhood
the power

cent of

its

maximum* value

of Ri/r p

ranging from 0.5 to 2.0.

is

Ri/rp
.

maxi-

at least 88 per
for values

This

condition shows that the power loss

is

than 2.25 db for

between 0.5rp and 2rp

less

Fig. 8-2. Variation of power output

as a function of resistance ratio Ri/rp

of the

mum, and

Since the
fer occurs

all

values of

maximum power

when the load

trans-

resistance

equals the internal plate resistance of the tube, it is necessary to use tubes
with low values of r P in order to obtain reasonable amounts of power with
nominal values of plate supply potential. Since the gm of a tube cannot
be designed over very wide limits, then tubes with low rp also possess low
values of

ix.

As a

result, large grid excitation potentials are required for

appreciable amounts of power output.

Note from Eq.

(8-2) for

a given

value of Eg that tubes which possess large values of

possess high output-power output.

In fact, the power sensitivity which

was defined by Eq. (3-4) becomes,
Fig. 8-3. The output current and potential
under the conditions of maximum
waveforms in a triode power amplifier.
power transfer, simply ngm /^.
To determine the power output directly from the static plate characteristic of the tube, it is necessary only to draw the appropriate load line on
these characteristics and read the significant information from the diagram. Thus, by referring to Fig. 8-3, it follows that
I

=
(8-3)

and

E =

V2

2\7i

UNTUNED POWER AMPLIFIERS

Sec. 8-2]

The power output

is

Em Im
which

may

189

ItRi

be written in the form

(8-4)

not negligible, the harmonic components must be evalu-

If distortion is

The

ated according to Eqs. (4-32).

P =
This

may

{B\

B\

total

Bl

power output
''

is

then

Ri
2

be written as

which becomes, by Eq.

+ (!:)*+(!)' +

Ri

(4-33),

P =

(1

D\

P =

(1

> 2 )Pi

D\

)Pi

or

where
tortion

is

is

the total distortion.

(8-5)

Notice, however, that

high, say 10 per cent, then

distortion represents a

power

of

P =

only

That

if

the total dis-

a 10 per cent
per cent of the fundamental.

l.OlPi.

is,

Thus, with little error, the output power is approximately that of the
fundamental-frequency component only.
8-2. Output Circuits.
It is not always feasible, nor is it generally
desirable, to connect the load directly in the plate lead, as shown in Fig.
8-1.

Among

the reasons for this are:

that the quiescent current through the

load resistor represents a considerable

waste of power, as it does not contribute to the a-c component of power and
that the quiescent current may cause a
;

serious polarization of the output.

example,

it

is

For

inadvisable to pass a

large d-c current through the voice coil

Fig.

8-4.

power

parallel

or

shunt-fed

amplifier.

For these reasons, the transformer-coupled load is

of a loud-speaker.
used extensively, although the parallel-feed system may be used.
The circuit of the parallel-feed system is illustrated in Fig. 8-4. It is
clear that this is just the impedance-capacitance coupled system discussed in Chap.

5.

In this system the plate supply

is

connected to the

RADIO ELECTRONICS

190

[Chap. 8

plate of the tube through a high inductance L, the load resistor being

connected across the output through a blocking capacitor C. The inducL must be so chosen that wL ;$> R h and C must be so chosen that
1/ojC <SC Ri over the operating range of frequencies.

tor

The potential and current

relations for the parallel-feed system are illus-

trated in Fig. 8-5.

Dynamic load line
Slafic

tube

load line

lb

is

determined by Rl, the

re-

sistance of the inductor, although

Ian

Rt

bb

Fig. 8-5.
of

shunt-fed

or

resistance
is

working

into
is

Ri.

Since the static resistance of the

The dynamic and

a

"dynamic"

the

which the tube

Eb\,
E Emax
lines

Observe that

the quiescent current through the

static load
transformer-

coupled amplifier.

choke or the transformer winding

is

usually small, the static load line

is

almost vertical.

Suppose that the load resistance into which the tube works is small; for
example, the resistance of the voice coil of a dynamic loud-speaker usually
ranges from about 5 to 15 ohms. If such a low resistance load were used
in either the series- or the shunt-feed circuits, only a

very small power

output would be possible, most of the power being lost within the tube
resistance.
In this case, and in fact in any case in which the load resistance does not properly match the tube resistance, the use of a transformer
as an impedance-matching device will permit optimum power transfer.
Such a system is illustrated in Fig. 8-6.
The impedance-transforming property of an ideal transformer follows
from the simple transformer relations

(8-6)

and

r.

l2

Fig. 8-6.

simple

transformer-coupled
where E 2 and E 2 are the primary load in a power amplifier.
and secondary potentials, respectively, and Ij and I 2 are the primary and secondary currents,
The ratio
of the above yields

which

may

Ei

(NjY
^"E
Ej

r,

\nJ

'

be written as
Ri

since E'2 /F2

which are

i,

and E 2 /I 2 represent the

resistances.

n2

Ri

effective input

(8-7)

and output impedances,

When the turns ratio Ni/N 2 is greater than unity,

UNTUNED POWER AMPLIFIERS

Sec. 8-3]

191

is called a step-down transformer; with the ratio N\/N 2

than unity, it is a step-up transformer.
Equation (8-7) is true for an ideal transformer. In general, however,
the coupling is not perfect, the primary and secondary resistances are not
By taking these facnegligible, and the core losses cannot be neglected.
tors into account, the input impedance is given by

the transformer

less

n2

'

(8-8)

n1

n 2/

where Ri and R 2 are the primary and secondary winding resistances, L\

and L 2 are the primary and secondary leakage inductances, and Zi is the
load impedance.

The same distinction between the static and dynamic load lines must
made for transformer-coupled loads as for the shunt-feed circuit, and,
But for the reasons discussed
as noted, Fig. 8-5 applies for both circuits.
be

frequency response of the transformer is not flat for all

effects are less severe than for a transformer
interstage coupling, since over the audio range the transformer capacitances, tube capacitances, and stray capacitances appear across a relatively low plate-resistance tube or across the low-output load resistance.
in Sec. 6-10 the

However, the

frequencies.

8-3.

Maximum

Undistorted Power.

The

foregoing analyses, which

are based on the linear equivalent plate circuit, are not completely valid

owing to the curvature of the dynamic characteristic, particularly

in the region of small plate currents.

order

In

maximum

possible

to

obtain

the

power output

without making the instantaneous

Eo

Emm

^>wax

>**6

plate current too small during the

most negative part of the applied

and without driving the
grid positive at the positive peak

Fig. 8-7. Graphical construction for deter-

mining the operation conditions and maximum undistorted power output from an

signal,

amplifier.

of the applied signal, it is neces-

sary to maintain a careful balance

plate supply potential,

To find

and plate

among

the grid bias, load impedance,

resistance.

the expression for the output power under these conditions, and

also to determine the appropriate conditions in order to achieve the

present results, refer to the graphical construction of Fig. 8-7.

Since the

from the curvature

eliminated by setting 7 min at an

distortion that results at small plate currents arises

of the static characteristics, this region is

appropriate value.
grid

potential

This specifies the crosshatched area

of

the diagram.

allowed to swing from any point Q, corresponding to the

(which would be i?w if an ideal shunt-fed or transformer-

Thus the

is

RADIO ELECTRONICS

192

[Chap. 8

coupled load were used) between zero grid bias and that bias which corresponds to 7min. It will be assumed that the distortion is negligible in
this region.

The

characteristics are essentially linear in the operating

region, with a slope corresponding to r p

To find the value of load resistance for
.

mum,

refer to Fig. 8-7.

Eb

But

which the power

be a maxi-

will

noted that

It is

= Ed

+ Em

2Im r p

since

Em =
then

it

Rllm

follows that

Eb

= Ed +
= ED +

2Im rp
I m (Ri

+ Im R,
+ 2rp
)

Solving this expression for Im there results

,

The power

to the load

is

_ Eb Ed

Ri

(8-9)

2rp

then given by the expression

P = URi _

- EdY

{Eb

16r p

A sketch showing the variation

of the
is

(1

m/2r p
+ Ri/2r P Y

(8-10)

output power as a function of Ri/rp

given in Fig. 8-8. This curve, like

that of Fig. 8-2, reaches a

maximum,

but in this case at the point at which

Ri = 2r p although the variation is not
,

rapid in the region of the

The power remains

maximum

of its

maximum.

at least 88 per cent

value for load

resist-

ances Ri ranging from r p to 4r p

To find the appropriate bias for
these conditions, combine Eq. (8-9)
.

Fig. 8-8. Variation of output

as a function of load Ri.

power

with the fact that the current changes

from Ib to Im when the signal potential

j

pE
Ri

The

result,

by equating Eq.

equal to

is

c.

Thus

(8-11)

+ rp

(8-9) to (8-11), yields

ED )
in

which Ri has been set equal to 2rp

from the curves.

directly

The value

(8-12)

of

ED

is

obtained

UNTUNED POWER AMPLIFIERS

Sec. 8-4]

193

The maximum undistorted power output becomes, from Eqs.

and

(8-10)

(8-12),

P max =
J-

Further, since

Egm

9r,

becomes

this

P max =

(8-13)

*-

9r

and the power sensitivity at optimum power output

P_

Power
which

is

slightly less

sensitivity

is

mhos

El

than that for the conditions

(8-14)

of Sec. 8-1.

The results showing the power output and second-harmonic distortion as

a function of Ri of a type 6A3 triode are illustrated in Fig. 8-9. Optimum
output is obtained at about Ri = 2,500
ohms, which is approximately three
times the plate resistance rp of the tube.
Although the second-harmonic distortion is not negligible at this point, a 5
per cent distortion

The above

is

usually tolerable.

analysis

is

based on the

use of a plate-supply source of so-called

If it
nominal value (about 300 volts)
.

4
5
2
3
Load resistance, kilohms
I

is

assumed that a plate source

potential

then with

available,

is

any

of

in-

Ebb

the ultimate

limitation will be imposed

by the allowUnder these

creases in the value of

able

dissipation.

plate

circumstances,

it is

found that the

of Ri that greatly exceeds 2r p

Fig. 8-'
Power output and secondharmonic distortion of a type 6A3
triode

as

a function

of

load

re-

sistance.

circuit should

be operated with a value

1
.

8-4. Plate-circuit Efficiency.

The foregoing

discussion gives the meth-

ods for calculating the output power of a power amplifier. The a-c power
so obtained is converted from the d-c plate supply by the vacuum tube.
The ratio of these quantities is called the plate-circuit efficiency of the
amplifier.

Thus
Vp

power output to the load

d-c power input to the plate circuit

Suppose that

Then by the

a-c

Pv

X 100%

(8-15)

denotes the average power dissipated by the plate.

principle of the conservation of energy

Etth =

im + EJ

+ Pp

RADIO ELECTRONICS

194

where Ri

is

Pp =
But

By

the static load resistance.

Evblh

IIR\

Eh

[Chap. 8

solving for

Pp

there obtains

EP I P

as

IbR\

then P has the form

PP =

Eblb

EIp

(8-16)

This equation expresses the amount of power that must be dissipated by

the plate and represents the kinetic energy of the electrons that

is

con-

verted into heat at the plate.

Notice in particular that the heating of the

anode is reduced by the amount of the a-c power that the tube supplies
Hence, a tube is cooler when delivering power to a load
to the load.

\Eb
Fig. 8-10.

than when there

is

The

\Em

plate characteristics of an ideal triode.

no such a-c power

transfer.

factor in the operation of high-power

r-f

This

is

a very important

transmitting tubes, since such

tubes are ordinarily operated close to the rated allowed plate dissipation.
If for any reason the output circuit becomes slightly detuned, with a

consequent decrease in output power, the plate power

dangerously high.

The plate-circuit
From Eq. (8-15),

efficiency

Vp

This

may

may

may become

be written in several different forms.

E pi p X 100%

(8-17)

Ebb lb

also be written as

Po

+ PP

X 100%

(8-18)

Clearly, a large value of rj p means a small value of P p for a given output.

This means that a smaller tube with a smaller plate-supply source may
be used.
It is possible to obtain an approximate expression for the theoretical
value of r] p for the series-fed and the shunt-fed circuits.
Consider that
an ideal tube is used in an amplifier circuit. The plate characteristics of
such a tube would have the form illustrated in Fig. 8-10. Suppose that
the grid does not swing beyond E c =
and may swing to give zero cur-

UNTUNED POWER AMPLIFIERS

Sec. 8-4]

rent.

Then, by the proper choice of

195

c,

= Im

lb

and

Ep l p
For the

E m Im

series-fed circuit, the point

Em
marked

E^

= Ebb and
,

im

from the

diagram,

Emai

Hence Eq.

2Imrp

-\-

becomes

(8-19)

Vp

= 2Em

Ebb

Em

rrv

2{E m

_
~

I m rv )

25
1

I m r p /E m

/o

from which, since

Em =

Im Rl

then

* = r+rb^
The

theoretical

fier is

maximum

plate-circuit efficiency for the series-fed ampli-

For the conditions

25 per cent.

(8 - 20)

Ri Tp, V P = 12.5 per cent.

operation without distortion,

of

maximum power

output,

when

Actually, owing to the limited range of

r)

seldom exceeds 10 per cent in practice.

is an inefficient device for

Evidently, the linear vacuum-tube amplifier

converting d-c into a-c power.

In the shunt-fed system a means has been devised for eliminating the
power loss in the load. This results in an improved plate efficiencjr.

d-c
If

the static resistance

Ebb

is

assumed

= Eb = Em

negligible,

then

2Im rp

and

^=50^

50

lm

This reduces to the form

50

*
The

theoretical

maximum

transformer-fed amplifier
static

power

is

r+2^

(8 " 21)

plate-circuit efficiency of the shunt-fed or

50 per cent.

loss in the load reflects itself as

Thus the elimination

an improved plate

of the

efficiency.

For the conditions of maximum output power, when Ri = 2rp t) p = 25 per

However, since the static resistance is not negligible and since the
current 7 min cannot be taken as zero if distortion is to be avoided, the
actual plate-circuit efficiency will be less than the 25 per cent figure.
,

cent.

RADIO ELECTRONICS

196
8-5.

[Chap. 8

Power Pentodes and Beam Power Tubes.

istics of

The

plate character-

the power pentode are markedly different from those of a triode,

and the graphical analyses given above are not valid for pentodes. Since
beam power tubes are similar to those of the
power pentode, the discussion to follow applies for both the power
pentode and the beam power tube.
the characteristics of the

100 r

Rt =2.5 k
-Ec =0

80

5
60
-10

-K

;40

-20
20

-25
-30
--35
100

200

300

400

500

plnte volts

e/j,

Fig. 8-11. Plate characteristics of a

6F6 power pentode.

Power pentodes differ from triodes principally in the character of the

dynamic curve with increasing load resistances. In the triode, the disThis
tortion decreases as the magnitude of the load resistance increases.
follows from the fact that the dynamic curve becomes increasingly linear
as the load resistance becomes
80
In the power pentode, the
higher.
dynamic characteristic is critically
60 E
dependent on the load resistance,
40 3

Z0-S

-40

-30

-20

-10

..bias

Fig. 8-12. Dynamic characteristics of a

for three values of plate load resistance.

6F6

with excessive curvature at both

the high and the low values of load
Moreover, the critical
resistance.
load resistance to be used cannot
be related analytically with the
This
plate resistance of the tube.
resistance is always less than the
plate

resistance

of

the tube.

It

might appear therefore that the

output-power capacity of the tube would be too low to make the tube feasible.
Actually, owing to the large ng m product, even with the appropriate Ri the power output is usually higher than with the tube connected

and this with a smaller grid driving-potential amplitude.

To examine the matter in somewhat greater detail, refer to Fig. 8-11,

as a triode,

UNTUNED POWER AMPLIFIERS

Sec. 8-6]

197

which gives the plate characteristics of a GF6 power pentode. It will be

supposed that the load is transformer-coupled to the tube and that the
is maintained at 300
and 10,000 ohms. The
corresponding dynamic curves are

plate potential

Three load

volts

lines are

shown,

2,500, 7,000,

The

given in Fig. 8-12.

example

will

help

following

clarify

the

situation

Example. Calculate the output

power, the plate-circuit efficiency, and
the second, third, and fourth harmonics
2 20
for the 6F6 that supplies power to a
loudspeaker, the effective resistance of
which is changed to have values of 2.5
kilohms, 7 kilohras, and 10 kilohms. A
300-volt source is available, and the
5j<
jo*
tube is biased at 20 volts.
load resistance
Solution.
The important data from
Fig. 8-13. Operating characteristics of a
Figs. 8-11 and 8-12 are included in the
6F6 pentode as a function of load retabulation.
The general character of the sistance.
results is given graphically in Fig. 8-13.
Notice that optimum power transfer occurs somewhat above 10 kilohms, which is
very small compared with the tube resistance of 78,000 ohms.
Ri

2.5
kilohms

h
Imin

Ri

\\

D
D,

10

kilohms

kilohms

64
49
25
6

53
46
25
6

86
52
25

hi

"\

2.5
kilohms

7
kilohms

20.5

11.5

-2.3

-11.5
-1.4

+1

Ri

B
Bo
Bi

B,
B>

10
kilohms

3.2

-16

!-,

Pa-c
Id-c

Pd-c
Vp

8-6. Push-Pull Amplifiers.

The

2.5
kilohms

7
kilohms

9
44
9

10

kilohms

35
4

31

-4

-5

-M

2.5
kilohms

10

kilohms

kilohms

35.5
4.4

31
4.8
26
7.8

45.0
2.5
34
10.2
24.5

29

8.7
50.6

61.6

use of two tubes in parallel provides

the possibility of obtaining twice the output power of a single tube with
the same distortion. A push-pull amplifier circuit is a much more
desirable connection for

two

tubes.

In this

circuit,

the two tubes are

RADIO ELECTRONICS

198

The

arranged as shown in Fig. 8-14.

[Chap. 8

excitation potentials to the grids of

the two tubes must be of equal magnitude, but of opposite phase. In the
circuit shown, a center-tapped transformer is used to provide these two

equal potentials that differ by 180 deg.

number

vacuum-tube

of

are possible for achieving

circuits

these results, and several of these will

be examined below.
To examine certain of the features
of such an amplifier, suppose that the
input grid potential to one tube is of
the form
Fig. 8-14.

The

The

basic push-pull circuit.

= E gm

eg i

cos

plate current of this tube will be represented in general

cot

by the expres-

sion (see Sec. 4-8)

%i

= B +

The corresponding

i?i

cos

cot

= Egm

and the output plate current

Bo

+B

cos

cos 2wt

+B

signal to the second tube

e g2

%i

+B

(cot

7t)

cos

cot

Egm

cos

3cot

(8-22)

is

cos

(cot

t)

is

+B

cos

2(cot

+B

ir)

cos

Z(cot

w)

which has the form

ib2

Bo

Bi cos wt

+B

cos 2wt

cos

Scot

(8-23)

But, from Fig. 8-14, the currents are in opposite directions through the
output-transformer windings. The total output is then proportional to
the difference between the plate currents in the two tubes.
i

k(ibl

ib2 )

= 2k(B

cos

cot

+B

cos 3w<

This

is

(8-24)

This expression shows that the push-pull circuit balances out all even
harmonics in the output and leaves the third-harmonic term as the
principal source of distortion.

Another feature of importance in the push-pull system is evident

from the circuit of Fig. 8-14. It is observed that the steady components
of the plate currents flow in opposite directions in the windings, thus

opposing each other magnetically in the transformer core.

This elimi-

nates any tendency toward core saturation and the resulting distortion
that might arise from the magnetization of the transformer core.

The

effects of ripple potentials that

supply due to inadequate

filtering will

may be contained in the power

be balanced out in the push-pull

UNTUNED POWER AMPLIFIERS

Sec. 8-7]

This

circuit.

is

199

so because the currents that are produced

potential are in opposition in the output transformer

by this ripple
and hence will not

appear in the load. Of course, the effects of the ripple potential that
appear on the grids of the tubes will not be balanced out and will be
noticeable with the signal.
Another feature of this amplifier is that under self-biased conditions
there is no need for a by-pass capacitor across the cathode resistor.
This
follows from the fact that the potential which appears across the self-bias
resistor

(4i

Rk

is

(4i

ibi)Rk

+ ib2)Rk-

But

2R k (Ib

+B +

this

is,

Z? 2

from Eqs.

cos 2wt

(8-22)

Z? 4

cos

and

4cod

(8-23),

But for tubes operating in class A the harmonic amplitudes are very small
and are therefore not significant.

One
is

of the particularly significant features of

that the output power possible with the

the push-pull amplifier

two tubes

for a given total

higher than twice that of the single tube.

This results
from the fact that with the automatic cancellation of even harmonics in
the output the tubes may be driven harder until the third harmonic terms
distortion

become
Note

is

significant.

also that, with the increased grid drive, the rectification

ponent 2B

Rk

As a

becomes

significant

and adds to the bias 2lbRk,

if

com-

self-bias

with consequent reducThis means that the output power of a push-pull

amplifier under otherwise similar conditions will be higher with fixed bias
than with self-bias.
is

used.

result, the effective bias increases,

tion of output power.

8-7. Equivalent Circuit of a Class A Push-Pull Amplifier.

Suppose
that both tubes of the push-pull amplifier are identical and that n and rp

Fig. 8-15. Equivalent circuit of the class

A push-pull amplifier of Fig. 8-14.

Fig. 8-16. The simplified equivalent circuit of the class

push-pull amplifier.

are constant over the range of operation.

The equivalent circuit of the
system then has the form given in Fig. 8-15. Observe that the connection
between the cathode terminals and the mid-point of the output transformer does not carry a fundamental-frequency component of current,
owing to the cancellation that occurs. This connection may be omitted
from the diagram without influencing the operation. The resulting cir-

RADIO ELECTRONICS

200
cuit then has the

form given

[Chap. 8

In this diagram

in Fig. 8-16.

(8-25)

The

resultant current

is

given by
2fiEg

2r p

which

may

/*E g

rP

total

power delivered to the load

P = PR'
This expression
is

(8-26)

R\

be written in the form

t

The

may

(8-27)

R\/2
is

then

VK

(
\r v

R\l2)

(8-28)

be interpreted to mean that the total output power

twice the power of each tube considered to be working into the equiva-

lent load resistance R'i/2.

more

significant expression results

by writing Eq.

(8-28) in the

form

(8-29)

This may be interpreted to show that the class A push-pull amplifier may
be represented by a single composite generator which has emf ^E 9 with
an internal resistance rp /2, and which works into a load resistance equal
,

to

R'i/4:.

It

is

possible, in fact, to derive a set of static characteristics of

the composite tube from the tube plate characteristics and to obtain

from this.
Composite Static -characteristic Curves.

significant operating information

8-8.

characteristics of the push-pull amplifier

Fig. 8-17.

The

may

The composite static

be obtained from the plate

potentials in the push-pull amplifier.

by a graphical construction. 2 It is
assumed that the output transformer is ideal, whence the potentials across
each half of the transformer are equal. The situation is illustrated in
characteristics of the individual tube

UNTUNED POWER AMPLIFIERS

Sec. 8-8]

The load

Fig. 8-17.

potential

is

201

given by

irr

(8-30)

e,

primary ampere-turns are equal

from which

Also, in the ideal transformer, the total

to the secondary ampere-turns,

Nxihx

Thus the load

potential
e

By combining
eP

which

may

Niib2

Ni
2

(8-31)

is

iRi

(8-32)

(4i

Eqs. (8-30) and (8-32) there results

- iM

(ii

'*

(8-33)

Ri

be written in the form

eP

where R[

is

(8-34)

(ibi

the plate-plate

resist-

ance.

The

following relationships are

evident from an inspection of the

diagram

of Fig. 8-17:
e&i
Vb2
@cl
6c2

It follows

that

= Eb e p
Eb + e p
= Ecc -f- 6g

cc

(8-35)

Bg

from these equations

when the plate-cathode po-

tential e b \ of tube Tl decreases

from the quiescent-point value Eb
by an amount e p then the corresponding potential eb2 of tube T2

^tr lb2)

beyond Eb by a like Fig. 8-18. To construct the composite

amount. Also, when the grid- static characteristics of a push-pull amplifier from the plate characteristic.
cathode potential e c \ increases
beyond E cc by the signal potential e g the corresponding value of e c2
decreases below Ecc by eg
These conditions are shown in Fig. 8-18 for
three different values of e p
In these diagrams point A corresponds to
conditions e M = E b e p e ci = E + e g the tube current being ibl
increases

RADIO ELECTRONICS

202

B is that for 652 = Eb eP) ec2 = Ecc eg and the tube current is
Point C has the ordinate in ibi for the chosen value of e p
The composite static characteristics are the family curves of in-in vs. e&,

Point

%2

[Chap. 8

with grid signal potential as a parameter. Clearly, point

on the composite static curve for the signal potential e g

is one point
Other points

Composite
curve

static

Fig. 8-19.

on

The composite

static

this characteristic are

curve derived from the constructions of Fig. 8-18.

found by maintaining

eg

constant and by vary-

The construction for two other values e'p and

ing e p
Fig. 8-186 and 8-18c.
These locate two other points
.

composite static characteristic.

ep

are given in

and C" on the

The complete composite

static is given

in Fig. 8-19.

Several significant features are evident'from Fig. 8-19.

static

characteristic extends

The composite

above and below the zero-current

axis.

Also, the composite static curve

much more

linear

characteristics

of

is

than the plate

the

individual

tube.

An

alternative

method

for

ob-

taining the composite static characteristics

was described by Thomp-

According to this method,

the plate characteristics of the tube
are plotted in the usual way.
The
curves are also plotted in an inverted
manner, with the potential scale
son. 3

Fig. 8-20. The Thompson method of

obtaining the composite static characteristics of a push-pull amplifier.

shifted so that the potentials

both

of

sets of curves are aligned with

each other. This construction is shown in Fig. 8-20. The inverted

curves represent the plate family of tube T2. The two methods are
essentially equivalent.

The foregoing

discussion of the graphical construction is general and

any type of tube under class A, AB, or B conditions of operaAlthough the illustrations are for triodes, the curves for other

applies to
tion.

UNTUNED POWER AMPLIFIERS

Sec. 8-9]

tube types are obtained in the same way.

203

In fact, owing to their shape,

the composite static characteristics for pentode-type tubes are

more

easily

obtained than the triode curves. The simplification results because the
current i b2 remains substantially constant for large variations in e b
The
composite static characteristics for the 6F6 pentode are shown in Fig.
.

8-21.

e3 -20

.Ec -0

eg-IO

eg-0
Fig. 8-21. Composite static characteristics of a

6F6 pentode push-pull

Composite Dynamic Characteristic.

8-9.

amplifier.

The composite dynamic

obtained from the composite

static characteristics in precisely the same way that the dynamic curve
is obtained from the plate characteristics of a single-tube amplifier.
This
characteristic of a push-pull amplifier

requires drawing the effective

is

comLoad line of lube

posite load line on the composite

and noting the points

of intersection of the two, which
are points on the composite dynamic
static curves

This construction also per-

curve.

mits obtaining the load line into

which each tube is working and

hence also the dynamic characteristic

of the individual tubes.

construction
Clearly,

the

is

shown

The

in Fig. 8-22.

intersection

of

Fig. 8-22.

the

circuit,

push-pull load line with the composite

plate

characteristics

points on the composite

The load

composite
one tube of

line of the

and the load

line of

the push-pull amplifier.

gives

dynamic

That is, points A, B,

dynamic curve. These points are
replotted in Fig. 8-23 to give the composite dynamic curve.
To find the dynamic characteristic of each tube, the procedure is
C,

characteristic.

are points on the composite

essentially the reverse of that of Sec. 8-8.

point

in Fig. 8-22.

This point

is

In particular, consider the

a representation of

%i-ii2 corre-

RADIO ELECTRONICS

204

[Chap. 8

But the point in lies

sponding to the appropriately chosen value of ep
vertically above this point, by an amount 4 2 and must lie on the plate
This defines the point A\. The other points
characteristic of the tube.
are obtained in a similar way, and the results are shown in Fig. 8-22.
These points are plotted also in Fig. 8-23. The points A 2 B 2 C 2 D 2
for tube T2 are obtained by symmetry from the corresponding points of
tube 21.
It is interesting to note that the composite dynamic curve is practically
a straight line, although the individual tube dynamic curves are markedly
curved.
Thus, for a sinusoidal input, the total output current is closely
sinusoidal, although the current in each tube is not sinusoidal, in general.
.

Fig. 8-23. The single-tube dynamic characteristics and the composite dynamic of the
push-pull amplifier. The waveshapes in each tube and in the output are also shown.
If

only the output current

individual tube dynamics.

is

If

desired, there

is

no requirement

the plate-circuit efficiency

is

for the

required, the

individual tube dynamics are required in order to calculate the value of

Ib and the d-c component Bo that results from the partial rectification

due to the curvature

amplifier

is

2Ebb (Ib

dynamic.

of the

The

total d-c

power input to the

Bo)

Power Output and Distortion in Push-Pull Amplifiers. Owing

to the fact that the dynamic curve must be an odd function, by virtue of
the manner of its construction, then for the composite circuit
8-10.

h=
=
-^min
In = I-tf

^msi

Under these circumstances the five-point schedule

Bo = B 2 = 5 4 =
B =
+ la)
- 2I)
B3 =
l

(^8-36)

of

Eq. (4-30) reduces to

(8-37)

UNTUNED POWER AMPLIFIERS

Sec. 8-11]

The fundamental power output

*
By

is

205

given by the expression

mi

b
-

-p

neglecting the harmonic components of power, the total

power

is

given by

Pi

(8-39)

Em

where

and Im denote the peak values of the a-c output voltage and
These values are obtained directly from the curves
=
since
8-22,
Im
I A and E m = Eb Ea, whence

current, respectively.
of Fig.

P =

(Eb

- E^ 1*

(8-40)

output power in the push-pull class A system

of the fact that the load resistance should
equal the internal resistance of the equivalent or composite generator.
This follows from Eq. (8-29) for the class A amplifier and requires that
This requires that the slope of the effective load line must
R'i/4: = r p /2.
be equal to the reciprocal of the composite static characteristic, which
has a value of rP /2,
Suppose that the tubes are operated in push-pull class B. Now, since
the tubes are biased to cutoff, then either one or the other of the two tubes
will be supplying current to the circuit and each contributes power for
Consequently, the equivalent generator will be
one-half of each cycle.
one with an internal resistance equal to rp of the tube. The maximum
It is
power under these conditions will be obtained for R'i/4 rp
reasonable to expect that the internal resistance of the equivalent source
of a push-pull class AB amplifier will lie between the value for the class
A circuit r p /2 and that for a class B circuit rp In all cases, however,
recourse should be had to the composite static characteristics, and then
R'i/i should be set equal to the reciprocal of the slope of these lines.
The situation for pentodes is different from that discussed above for
triodes and follows roughly the reasoning of Sec. 8-5.
The optimum load
The optiis that which yields the maximum power with low distortion.
mum load line is drawn through the point Q' so that it intersects the peak
composite grid-voltage curve in the neighborhood of the knee of the
This is illustrated in Fig. 8-21 for the 6F6 tube.
curve.
8-11. Driver Stages for Push-Pull Amplifiers.
The driver may be
considered to comprise the circuit that supplies the two potentials of equal
magnitude but in phase opposition to the grids of the push-pull power

To

find the

maximum

utilizing triodes, use is

made

amplifier.

variety of suitable circuits exist, the most direct of which

RADIO ELECTRONICS

206

[Chap. 8

illustrated in Fig. 8-24.

This circuit consists of a simple amplifier
with a transformer in the output, the secondary of which is center-tapped.
The characteristics of this stage are determined by the grid driving-power

is

requirements of the power amplifier. If the push-pull circuit requires

substantially no driving power, then there are no serious requirements
imposed on the driver stage. If the driver stage is called upon to supply
power, and this would seldom exceed 15 per cent of the output of the
push-pull stage, although

it is

ordi-

narily less than this amount, the

o
o ,

\~'
J=-

f 1

1*
J

Fig. 8-24. A simple driver circuit for

a push-pull amplifier.

driver stage must have a relatively

low internal resistance if no distortion is to be introduced into the

am-

grid circuit of the push-pull

,. n
i
T
L
i_
,

P llfier

It is advisable

m such cases

that a step-down transformer be

used to couple the driver stage to the push-pull input in order to reduce
the effective resistance in the grid circuit.
If the power requirements are low, then any one of a wide variety of
"paraphase" circuits may be used. A paraphase circuit is one which
provides two equal output potentials which are 180 deg apart in phase
from a single signal source.

A single-tube amplifier in which the

divided equally between the plate and cathode circuits

Single-tube Paraphase Amplifier.

load resistor
is

is

the simplest form of paraphase amplifier.

The

circuit,

redrawn

for

convenience in Fig. 8-25, is discussed

The reat some length in Sec. 7-8.

and R k have the same value,

whence the amplitude of the potential
sistors Ri

developed across each is the same,

since the same current flows through
each.

The

polarity

is

cause the cathode output

E20

Ei

opposite beis

taken from

Fig.

8-25.

single-tube

amplifier.
the more positive end of R k and the
plate output is taken from the less positive end of Ri.
this circuit in Sec. 7-8 shows that the gain of the stage is
and is given by the expression

K' "
U

* JTTl

paraphase

The

analysis of

less

than unity

<8- 41)

Two-tube Paraphase Amplifiers. In the two-tube paraphase amplifier,

one tube is used as a conventional amplifier, and a second tube is used as
a phase-inverter amplifier. Figure 8-26 illustrates such a circuit. The
resistors

Rx and

comprise a potential divider across the output

of a

UNTUNED POWER AMPLIFIERS

Sec. 8-11]

207

conventional amplifier, the ratio of the resistances being chosen so that

Tl is equal to the output of T2. Also, the operating

the output of

conditions of the tubes are carefully chosen to allow the curvature of

T2

the characteristic of

to compensate for the curvature of Tl.

Thus

the output potentials relative to ground are both slightly distorted to

provide a comparatively undistorted po-

between the output teris difficult to apply

practice because the adjustments neces-

>

tential difference

This method

minals.
in

sary to reduce distortion to a

minimum

are critical.

second form of two-tube paraphase

employs the differential potential

between the outputs of two tubes as the
amplifier

input signal to the phase-inverter section.

This

circuit,

which

also referred to as the

is

floating paraphase amplifier,

in

two versions

is

illustrated

26

"

tw0 " tube paraphase

in Fig. 8-27.

In the circuit of Fig. 8-27a tube Tl

tude of the applied waveform.
will

FlG

is

an amplifier to increase the ampli-

The cathode resistors

Rk,

if

not by-passed,

provide some degeneration, which will help to reduce distortion.

The

coupled through Ci to Ri and R it both of which have

The potential which appears across R 3 is applied
the same value as R 2
The output of T2 is passed through C 2 and is applied
to the grid of T2.
across ^2 and R 3
Thus half the output of both Tl and T2 appears
output from Tl

is

across R3.

Since these potentials are of opposite polarity, the resultant

potential across

fi 3 is

the difference between these two.

The output

of

Tl is larger than the output of T2, and

be kept as small as possible, pentodes are used, so as to take advantage
in order that this difference should

of their

high amplification.

RADIO ELECTRONICS

208

[Chap. 8

is that the difference between

taken care of in the choice of the resistors Ri and
Ri so that the output potentials have the same amplitude. To do this
requires that the following condition be satisfied:

The

feature of the circuit of Fig. 8-276

the output potentials

is

KK+
where

K is the

R2

(8-42)
1

gain of the stage.

02

*7<

>R*
Fig. 8-28.

cathode-coupled paraphase amplifier.

The cathode-coupled paraphase

which
was discussed

amplifier,

the cathode-coupled amplifier which

is

closely related to

in Sec. 6-8,

is illus-

used extensively to provide

potentials
for
the
plates
of
a cathode-ray tube, and
push-pull deflection
amplifier.
also as the driver of a push-pull
An analysis of this amplifier circuit is readily effected in terms of the
This circuit

trated graphically in Fig. 8-28.

Thevenin equivalent

is

viewed from the cathode-ground terminals.

circuit as
I7-

_JWWV-

-V\AMr

^P+Rl +
/+/

JU.+)

IR*

Fig. 8-29.
terminals.

The

The equivalent

circuit of Fig. 8-28 as

resulting equivalent circuit

from

From

given in Fig. 8-29.

is

drop across

this figure that the potential

E4 =

-I-

viewed from the cathode-ground

+
M +

rp

is

It follows directly

given by

+ Ri
+1
r p + Ri
Rk +
M + 1
IiR,

Ri

Rk
rp

this
Ii

rP

+ Ri
+ 1)

Rkitx

(8-43)

UNTUNED POWER AMPLIFIERS

Sec. 8-11]

209

Notice from this expression that the current Ii is always greater than I 2
For example, for a typical triode with, say, rp = 10,000 ohms and n = 19,
and with R = R k = 10,000 ohms, Ii/I 2 = 1.10. In general, if the
magnitude of Ii is not to exceed the magnitude of I 2 by more than, say,
10 per cent (this requires than Ii/I 2 ^ 1.1), then
.

Tp

Rk ^

10

f'

should be observed that the more nearly Ii and I 2 are to be the same,
must R k be. In fact, the two currents are equal only if

It

the larger

Rk

<,

As a result, this circuit can never

by making R k large compared with

an impractical condition.

give exactly equal outputs, although

(r p

Ri)/(n

To determine

1)

almost equal outputs are possible.

it is noted from the diagram that

the output potentials,

E+h
M+l'

rp

+ R _ ^ rp + R =
l

M+l

m+1

from which
T M
If it is

assumed that

Ii

=
T

T
i-2

^E

+
1

rp

tT"

Ri

approximately, as discussed above, then

- -T -

^E

whence

E " = ~ Eo2 =

^fTM

(8 " 44)

REFERENCES
1.

2.

3.

Nottingham, W.
Millman, J., and

B., Proc.

IRE,

29, 620 (1941).

"Electronics," 1st ed., chap.

Book Company, Inc., New York, 1941.
Thompson, B. J., Proc. IRE, 21, 591 (1933).
S. Seely,

XX, McGraw-Hill

PROBLEMS
8-1.

A 6F6 tube is operated as a triode and supplies power to a 4,000-ohm load.

E = 25 volts and with a peak signal of 25 volts, calculate the

Etb = 300,
following:

With

cc

Output power.
b. Plate-circuit efficiency.
Per cent second-harmonic distortion, d. Plate dissipation.
8-2. Repeat Prob. 8-1 when the load is transformer-coupled to the tube.
8-3. It is supposed that the plate dissipation at the operating point is kept
constant. Prove that for class A operation the plate load is made larger with
a.
c.

increasing values of

and the plate

efficiency increases.

RADIO ELECTRONICS

210
8-4.
volts.

A 6V6
The

is

operated with

grid signal

is

Ea = Ec2 =

[Chap. 8

250 volts and with

peak value
5,000 ohms:

sinusoidal, with a

E =
cc

-12.5

Calculate

of 12.5 volts.

the following for a shunt-fed load of

a. Output power.
b. Total distortion.
d. Plate-circuit efficiency.
c. Plate dissipation.
8-5. Repeat Prob. 8-4 if the load is 2,000 ohms; 8,000 ohms.
8-6. Two 6F6 tubes are connected as triodes and are operated in push-pull
class A from a 350- volt plate source, with a grid bias of 30 volts.
A 30-volt

peak signal

is

used.

Draw the composite static characteristics.

From this, determine the plate-plate resistance

a.

for maximum output power.

Calculate the power output, third-harmonic distortion, and plate-circuit

b.
c.

efficiency.

Two 6F6

tubes are connected as pentodes and are operated

from a 350-volt plate source, with a grid bias of 25 volts.
potentials are maintained at 250 volts.
8-7.

in push-pull

AB

class

The

screen

Draw the composite static characteristics.

Plot the composite dynamic and the dynamic characteristic of each tube.
c. Calculate the output power, third-harmonic distortion, and plate-circuit
efficiency for a plate-plate resistance of 10,000 ohms.
The peak grid signal is
a.
b.

40 volts.
8-8.

Two 2A3

triodes are operated in push-pull with Ebb

300,

E = 60
cc

volts.

Draw
From

the composite static characteristic.

determine the plate-plate resistance for maximum power output.
c. Calculate the power output under these conditions.
d. Repeat c for 2^ and 2 times the optimum value.
e. Construct the paths of operation for the individual tubes.
8-9. A 6N7 zero-bias tube is connected as a class B push-pull audio amplifier
and is to furnish 10 watts into a dynamic loud-speaker, the voice coil of which
has a resistance of 8 ohms. A 35 1 step-down transformer is used. The plate
supply is 325 volts. Determine the following:
b. Grid driving potential.
a. D-c plate current.
8-10. The typical operating characteristics of a 6L6 beam power tube when
used in push-pull class A are shown below. Values shown are for two-tube unless
otherwise specified.
a.
b.

this,

Plate supply

Screen supply

Cathode

270 volts
270 volts

Zero-signal plate current

125
134

Maximum-signal plate current

145

Zero-signal plate current

11

resistor

ohms

ma
ma
ma
ma

Maximum-signal screen current

17

Plate resistance

23,500 ohms
5,700 /amhos

Transconductance
Effective load resistance (plate to plate)

Maximum-signal power output

5,000 ohms
18 5 watts
.

The 6L6 tubes are to supply the 18.5 watts to the grids of a pair of 806 triodes
which are operating in class B push-pull. The required peak grid driving potential is 660 volts.

UNTUNED POWER AMPLIFIERS

211

Calculate the turns ratio of the output transformer.

is the peak a-c plate potential on each tube?
c.
is the peak a-c grid-potential swing on each tube?
d. Does grid current flow during any part of the input cycle?
e. Calculate the plate-circuit efficiency.
8-11. The diagram gives the basic circuit of what has been called a singleended push-pull amplifier.*
o. Prove qualitatively that push-pull operation is achieved with this amplifier.
b. Choose E e i =
What is the output potential 2J 2 ? From this find an
g i.
expression for the power output, and show that it is the same as Eq. (8-29).
a.
b.

What
What

8-12. Verify the conditions (8-42) imposed

for balanced

output potentials.

What

on the

floating paraphase amplifier

conditions are imposed on

i2 3 ?

8-13. It is suggested that the paraphase principle be combined with a push-pull

amplifier to yield push-pull operation without a separate driving source.
The

push-pull amplifier feeds a dynamic speaker.

Discuss the suggested operation
from the point of view of class of operation possible; of distortion.
*

A. Peterson, and D. B. Sinclair, Proc. IRE, 40, 7 (1952).

CHAPTER

TUNED POTENTIAL AMPLIFIERS

Tuned

potential amplifiers are used in those cases in which

to amplify a relatively narrow

band

of frequencies centered

it is

desired

about some

designated mean or carrier frequency. Potentials whose frequencies lie

outside of this range are undesirable and are to be rejected. The use of
tuned networks accomplishes this, as it is possible to adjust the tuned
network so that the impedance falls steeply to low values outside of the

CcJ
Fig. 9-1. The three basic tuned amplifier circuits: (a) single-tuned, direct coupling;
(6) single-tuned, transformer coupling; (c) double-tuned.

desired frequency band, with the consequent reduction in amplifier gain

The resulting nonlinear distortion that is
to negligibly low values.

produced in these amplifiers is very small, both because the stage is

operated under class A conditions and because the tuned plate-circuit
impedance may be very low for any harmonic frequencies that might be
generated within the tube.

There are three basic amplifier circuits, and these are illustrated in
9-1.
Pentodes are ordinarily used in such amplifiers, and the
In two of these types, a single
circuits are drawn showing such tubes.
resonant circuit is used, which may be included directly in the plate
circuit (direct-coupled) or which may be inductively coupled to the plate
In the third type, a double-tuned bandcircuit (transformer-coupled).
Fig.

212

TUNED POTENTIAL AMPLIFIERS

Sec. 9-1]

pass arrangement

An

tuned.
will

is

used, both the primary

and secondary

213
circuits being

analysis of the operation of each of these amplifier circuits

be given.

9-1. Single -tuned Direct -coupled Amplifier.

The

equivalent circuit of

Included
a typical single-tuned direct-coupled stage is given in Fig. 9-2.
in this diagram are the output tube capacitances [see Eq. (5-23)] and the
The coupling
input and wiring capacitances to the following stage.
capacitance

may

be neglected, as this

presumed to be

large.

The equivalent

Fig. 9-2.

is

circuit of a single-tuned direct-coupled class

amplifier

In accordance with the discussion of Sec. 5-7, the gain of the amplifier

can be written directly as

K = gm Z
where Z is the
form given by

where Z

is

This impedance has the complex

impedance.

total load

(9-1)

the impedance of the antiresonant circuit and comprises the

inductance L and the

sum

of the various capacitances

C = C2
t

+ Cg + C

,.

where C 2 is defined as in Eq. (3-23).

The impedance Z has the form
(

Z =

Rl
which

may

be written

in the

7T

(9-3)

+j

form

(9-4)

Ri

['+'5e('-sto)

RADIO ELECTRONICS

214

By

[Chap. 9

writing
1

03

(9-5)

coo

_1

/:

the impedance function becomes

RlQ-

Zt

Hi)
/

+jQ

z,

At resonance

Since

co

co

0.

Zt

RlQ-

and

for the circuit

used

too

CO

co/

\O>0

+jQ8(2

(9-6)
5)

Then
(9-7)

usually high, with

is

Q =

10,

then with

good approximation

Z =
(

R =

tf L

(9-8)

This result shows that the shunt impedance Ro of the antiresonant circuit
Q > 10 is essentially resistive at the resonant frequency.
By combining Eq. (9-8) with Eq. (9-2), the gain at resonance becomes

for circuits with

-g m aioLQ

rv

may

This expression

T R T R L Q>

Qe

be written in the form

the effective

Qe

This

is

the

ances

res

= g m oioLQ

(9-10)

of the amplifier, is

Q
1

is

i?

K
where

(9-9)

-f ojoLQ/r p

uoLQ/R g

(9-11)

the equivalent Q of the resonance curve of the tuned amplifier and

of the actual resonant circuit as modified by the shunting resist-

Rg

and

rv

TUNED POTENTIAL AMPLIFIERS

Sec. 9-1]

215

To find the gain of the amplifier when the input frequency and the
resonant frequency of the tuned circuit are slightly different from each
1.0

+0.9-

~X_08'

0.7-

'

'q=/o
C

-50-|

espon:

i_

Q=IO

V.O

Q-oo

=0.4;

yoi

/V
*

0-5

//

a>

oc

Q=ro

S "5
c s

25

Q=oo

xJ?

!0.l -50

Q=oa,

0.1-

o5T

-75

Q=I0
3.0

2.5

2.0

1.5

05

0.5

1.0

1.5

1.0

2.0

2.5

3.0

K/K

Fig. 9-3. Plots showing the amplitude

re , and phase of the output of a single-tuned
(From F. E. Terman, "Radio Engineering," 3d ed., McGrawdirect-coupled amplifier.
Hill Book Company, Inc., New York, 1947.)

other,

it is

supposed that

o>

co

whence

8 is

From Eq.

small.

(9-6), it

follows that

RlQ

The corresponding value

2j5Q

of the gain, given

K = The gain

ratio

K/K

res is

Re

res

from which the amplitude ratio

K
K

re!

plot of these results

is

(9-12)

2j5Q

by Eq.

(9-9), is

(9-13)

+ 2jSQ
RL Q

then

K _
K ~

versal resonance" curve.

J_

J_
rp

1
1

(9-14)

j2SQ e

is
1

Vl +

(9-15)

(25Q e ) 2

given in Fig. 9-3.

Note from Eq.

25Q.

This

is

essentially the "uni-

(9-15) that

when

RADIO ELECTRONICS

216

[Chap. 9

then

K
K

re .

V2

But since the bandwidth of the circuit is the frequency width between
the 3-db power points, then for a symmetrical gain characteristic

B =

2(/3db

2(/3db

/o) =

/o)/o
/o

or

253db/o

= h

(9-16)

In order that the potential gain at resonance be

large, the resonant

impedance of the tuned circuit, and the grid resistor R g must be large
It might appear from Eq. (9-10) that higher gains
compared with rp
Note, however, from
are accomplished by choosing large values of L/ C.
Eqs. (9-5) and (9-8) that an increase of Z at resonance by increasing the
L/C ratio is accompanied by a decreased Q e with a corresponding increase
This is an undesirable
of bandwidth or decreased frequency selectivity.
condition if a narrow bandwidth is desired, but it is an important conNote also that if the circuit
sideration in wide-band tuned amplifiers.
Q of the tuned circuit is increased at fixed values of wo and L/C ratio,
then the effective Q e of the amplifier is increased, with a corresponding
increase of frequency selectivity or decreased bandwidth.
The general be9-2. Single-tuned Transformer -coupled Amplifier.
,

havior of the single-tuned transformer-coupled amplifier is quite similar

to that of the direct-coupled circuit.

To examine
circuit in
9-4,

the operation of the

some

detail, refer to Fig.

which gives the equivalent

mate expression
Fig. 9-4. The equivalent circuit of the
single-tuned transformer-coupled poten-

cir-

An approxi-

cuit of this amplifier.

for the potential

gain of the amplifier

is

readily ob-

noted that r v is
tial amplifier.
usually large compared with Ri and
coLi, then the potential induced in the secondary of the transformer is given
tained.

If it is

by
Etna

since in this circuit E

tial

across the capacitor

E2

= jwM(gJ& B ) = juM(gm Ei)

Ei.

C2

is

The output

potential,

which

is

the poten-

then
jo>M(suEi)

R + j(wL 2 - l/ud) + w
2

(9-17)

At
M /r juC
2

(9-18)

TUNED POTENTIAL AMPLIFIERS

Sec. 9-2]

where w 2

M /r
2

p is

The

ary circuit.

the reflected impedance of the primary into the secondexpression for the gain then becomes

+ i(o>L -

r p [R 2

Ei

res

may

uW

Af
R +
/-

jj^

rp

This result

l/oCs)]

for the potential gain at resonance

The corresponding expression

and

217

to

also be expressed in

Q 2 = woL 2 /R 2 =

of

is

(9-20)

5-^75
2
2

terms

'

l/cooC 2 R 2

is

X^res

which

may

M */r

be written as

K
where the

MQ

a
w%

effective value

res

Qe

is

Qe

=
,

uoMQ e

(9-22)

p2
<alM*/r p R

(9-23)

gm

+
,

A comparison of this expression with Eq. (9-10) shows that transformer coupling modifies the amplification by the ratio M/L. This
provides a means for controlling the gain of the stage and still retaining
It might appear that there are no
the high Q required for selectivity.
limits on the gain and that it continued to increase with increasing values
in the denominator
of M.
This is not so, owing to the appearance of

Eq. (9-20). An optimum value of gain exists, and this occurs when
has the value required to make dK re ,/dM = 0. This yields, for the
optimum value of M,
of

dKres

dM

n/C 2

=
rp

R +
2

_ (nM/C )2ulM =
(r p R +
M
2

o>

co

or

opt

(9-24)

Equation (9-21) becomes

^res,opt

Qn

To

find the

y/r v R 2

\r

(9-25)

bandwidth of the amplifier, consider the general expression

by Eq. (9-19). By writing, as before,

for the gain given

RADIO ELECTRONICS

218

CO

[Chap. 9

Vlc
CO

(9-26)

coo

Qx

COoL 2

CO0C2R2

R,2

and noting that

in the

neighborhood

C0L2

\C,

of resonance

L2

/ CO

\ ^2

C0C2

72,

O)o\

\COo

CO

25# 2 Q 2

(9-27)

Eq. (9-19) becomes

fiM/d

K=
rp

The

(l

(9 " 28)

+j ^2 S ^y>M>

gain ratio [Eqs. (9-28) to (9-20)] then becomes

K =
K ~
res

1
1

+ j28Q

(9-29)
e

which has the same form as for the direct-coupled connection [Eq. (9-14)].
The bandwidth of this amplifier is, following the same reasoning as that
which led to Eq. (9-16),
fi

(9-30)

is hot of much importance, owing to practical

of
This follows from Eq. (9-24), which shows that for penwould be
todes, with the corresponding large values of r p the value of
the distributed capacitances
large.
In fact, to achieve these values of
of the windings may become excessive, and the self-resonant frequency
may be so low as to make the coils useless. Owing to this, the mutual
inductance is usually chosen far below the optimum value in the pentode

The optimum value

limitations.

amplifier.

9-3.

The Double-tuned

Amplifier.

Both the single-tuned

direct-

coupled amplifier and the double-tuned amplifier are extensively used

For the i-f ampliin radar, television, and communication receivers.

both a-m and f-m types, the double-tuned amplifier is commonly

This is so because such an amplifier can provide substantially
constant amplification over a band of frequencies and the gain falls more
sharply outside of this band than does the single-tuned stage.
To examine the operation of the circuit, refer to the equivalent circuit
fiers of

used.

TUNED POTENTIAL AMPLIFIERS

Sec. 9-3]
of

219

This circuit can bs further simplified

of the circuit to the left
The equivalent generator has the potential

the amplifier given in Fig. 9-5.

by applying Thevenin's theorem to the portion

of the points aa.

E =

*f{3f$

(9-31)

= C'v + C'J. But since

where
> 1/wCi for the pentode, then
with good approximation

E,

rp

mEi

E =
The

Fig. 9-5.

gmE
(9-32)

-Q-

internal

impedance

to the

is,

tQ

same approximation

as above,

M_

Z =

jwCt

Then the equivalent

Fig. 9-6.

The equivalent

series

form

of

circuit of Fig.

9-6.

standard methods of network analysis.

Pnli
P12I1

then the current in the secondary

I2

where

P12I2

(9-34)
P22I2

Epl2

= -

(9-35)
p?2

p,2

= Ri+j("Li- Jr)
= jwM

>22

= R2 + j

11

The gain

+
+

This circuit is analyzed by the

Accordingly, if one writes

is

P11P22

of the amplifier

(9-36)

becomes

K = EEi
2

it

(9-33)

reduces to the form of Fig.

9-5

Fig. 9-5.

But

circuit of a double-

generator will have the value

of the equivalent

rp

which

The equivalent

tuned amplifier.

(gJj<*Ci)(l/juCt)juM

must be noted that both

Pll>22

circuits are

(9-37)

Q12

tuned to the same resonant

RADIO ELECTRONICS

220

[Chap. 9

Thus

frequency.

co

Vuci

Vl c
2

Also define*

Qi

Ri

Q2 = K =
2

oeoCiRi

-T^r
CO0O2K2

(9-38)

and write

Then, by Eq.

(9-27),

Similarly

p 22

Ri+j (uLi = R (l + j28Q

and

P12

= juM

Qu =

The

+ j2<2,)

Rr(l

(9-39)

2)

expression for the gain [Eq. (9-37)] then becomes for frequencies near

resonance

-jg m (M/u C 1 C 2 )

K=

+ j28Qi)(l + j2Q) +

K=
1

rr

a,%M i

w'M

VRlRi
/tt-rVR1R2
/R R + 2j5(Q + Q 1

2)

or finally

The gain
There

-jagm QiQ 2

VRiR

+ j28(Q +

Q t) -

a2

at resonance

is

obtained by setting

4S 2 Q 1 Q 2

p 40 '

in this expression.

results

-jag
mQiQ 2
J^-VIV^V
1

The gain

fli.fi

ratio at frequencies slightly different

K
K-

(9 _ 41)

from resonance

is

given by

45 2 Q 1 Q 2 \

+Q
+a

2S(Q l
1

(9-42)
t)

The exact shape of the response curve of the double-tuned system

depends upon the parameter a, or, correspondingly, on A;, the coefficient
of coupling between the primary and secondary coils.
The resonant gain
* If the amplifier is

be included in

Qt.

one of a chain, the loading

effect of

the following stage should

TUNED POTENTIAL AMPLIFIERS

Sec. 9-3]
is

maximum when

Moreover,

if

may

as

1,

221

=
=

be verified by examining dK/da

Q values are the same and a

the primary and secondary

the response curve has the

maximum

0.
1,

possible single-peak flatness in the

This is the condition for critical coupling k c

vicinity of resonance.
A
value of a greater than unity (overcoupling) results in double peaks,
.

whereas a value of a less than unity (undercoupling) causes the response

If the circuit is considerably undercoupled,
to be rounded on the top.
then the gain may be less than that at resonance. The situation discussed

An

is

illustrated graphically in Fig. 9-7.

analytic expression for the positions of the peaks in the output of

the overcoupled circuit

These are given,

readily possible.

is

the values of frequency at which Eq. (9-40)

is

maximum.

of course,

by

To find these

|JT|

Units of

9m QiQz VRjR2

r0.5-

a*2/

7^-0.3

-2.5

-2.0

-1.5

<

0.2-

0.1-

-0.5

-1.0

0.5

1.0

2.0

1.5

25

<SQ

Fig. 9-7. The response characteristics of a double-tuned amplifier for various values of
coupling.

values,

it is

noted that the gain

is

maximum without regard to the phase.

Thus the square of the absolute value of Eq. (9-40)

The results are
respect to 5 and maximized.
|K|

(g m QxQ*

VWiY

-jy
(1

and the derivative d\K\ 2/d8 =

1

a2

a2

45 2 Q!Q 2 ) 2

is

differentiated with

[25(Q X

Q2)]

(9-43)

yields

WQ Q
l

Q2
2Q 1 Q

(Qi

2
)

from which

Even if this condiFrequently the circuits are designed with Qi = Q 2

is not true, ordinarily Qi does not differ too markedly from Q 2 and
.

tion
it is

possible to

assume that

RADIO ELECTRONICS

222

[Chap. 9

Subject to this approximation, Eq. (9-44) becomes

<9 -45)

m,
which becomes,

in the

manner
S

The value

of

Eq. (9-46) with

of representation of Fig. 9-7,

%Va

VQ1Q2 =

the gain at either peak 2 mai

The

(9-40).

Km =

result

-j(gm QiQ 2

is

simply
(9-46)

obtained by combining

is

V R1R2)
2(1

+ j Va* -

1)

or

V R1R2)

= KfaQiQ

(9-47)

This shows that for the overcoupled case the maximum gain is the same
= 1, and at resonance 8 = 0.
The gain at the dip, at the frequency co can be found readily by setting
in Eq. (9-40).
S =
The result is

as that for critical coupling a

K min =

j(gm QiQ 2

V R1R2)

_^ q2

or

Kmln

g m Q,Q 2

VRJI2

(9_48)

YT~^

It follows from Eqs. (9-46) and (9-47) that increased coupling increases
the frequency separation of two peaks but does not change their amplitudes.
If the coupling is very large, then the approximation made in

Eq. (9-40)

is

no longer

valid.

The

effect of the factor co/co

in this equa-

is to increase the lower frequency maximum

frequency maximum.
The corresponding dependence of the current in the primary winding
on the frequency is of some interest. It is obtained directly by solving
Eq. (9-34) for Ii. The results are illustrated in Fig. 9-8. For the case
when the coefficient of coupling k is small, the secondary circuit has little
effect in the primary, and the resulting response is essentially that of
the simple resonant circuit. When k = k c the reflected resistance from

and decrease the higher

tion

the secondary decreases the current at resonance. On each side of

resonance, the reflected reactance is either inductive (below resonance)
or capacitive (above resonance).
A point is reached on each side of
resonance when the reflected reactances cancel the primary reactances,

which are capacitive below resonance and inductive above resonance.

As a result, the current shows peaks. For values k > k c the double
peak in the primary circuit becomes more pronounced.
,

TUNED POTENTIAL AMPLIFIERS

Sec. 9-3]

223

For the case where the primary and secondary Q values are not the
same, Aiken 1 has shown that a somewhat higher value of k is needed to
produce double peaks in the secondary. The flattest selectivity curve

may

be shown to occur when

1

(Qi

2 \Q
2\Q

although the mid-band gain

When Eq.
The

(9-49)

is satisfied,

is

not a

Q\

(9-49)

Qj
Qi

'

t
2

maximum under

the circuit

is

these conditions.

said to be transitionally coupled.

transitional value of coupling coefficient

is,

by Eqs.

(9-49)

and

(9-38),

(9-50)

For a coupling
divides into

than this value, the selectivity curve

For values less than this, the curve has a single

coefficient larger

two peaks.

peak.

Fig. 9-8. The primary current of a double-tuned amplifier, corresponding to the

conditions of Fig. 9-7.

For different primary and secondary Q values, but with Li = L 2 = L,

and for fc greater than the transitional coupling coefficient k Aiken has
shown that the frequency spread between peaks is
t

Aco

If

Ri

= R2 =

^(cooM) 2

it

R\
(9-51)

R, this becomes

Aw = ^ V(woM) 2
from which

R\

- R

follows that

coq

~ Vi^oM)
uoL

- R =
2

Vk 2

kl

(9-52)

RADIO ELECTRONICS

224

The

dip at resonance

when

k exceeds k t

is

[Chap. 9

a function of the ratio k/k c

have the form illustrated in Fig. 9-9. The

Aiken shows
response
remains above the center value is
which
the
over
bandwidth
this relation to

a/2

A/.

compare the gain of the double-tuned circuit with a

having the same Q. The gains of the two circuits at
resonance are given by Eqs. (9-10), and (9-41) for optimum value of a,
It is interesting to

single-tuned circuit

and are

Single-tuned direct-coupled:

Double tuned (with a

K
It is

res

observed that for

amplifiers are identical

and

-j0.5g m Q 2 R

critical
if it is

= gm woLQ

res

identical coils)

j0.5g m u>oLQ

coupling the gains at resonance of the two

of the

assumed that the tuning capacitance

k/kc
Fig. 9-9.

The

effect of the ratio of actual to critical coupling in a

individual tuned circuit in the double-tuned circuit

is

double-tuned

circuit.

one-half the tuning

capacitance for the single-tuned case.

Despite the fact that the response characteristics are optimum under
coupling conditions, the transformers in narrow-band doubletuned amplifiers are usually undercoupled slightly. This is done in
order that the frequency alignment of the tuned circuits may be made
critical

easier, since,

with undercoupled stages, each stage can be adjusted sepamaximum response at the specified frequency. If over-

rately to give

coupled circuits

exist,

owing to the interactions between

resulting double peak, this alignment

coupling case

is

is

more

critical.

coils

The

and the
critical

likewise difficult to align.

The bandwidth of the amplifier, under optimum conditions a = 1 and

with equal primary and secondary values of Q, is readily calculated.
Under these conditions Eq. (9-42) becomes

n =0 ,

TUNED POTENTIAL AMPLIFIERS

Sec. 9-4]

225

and the magnitude becomes

by

Since,

(9-54)

Vl +

K,res

4S 4 Q 4

bandwidth gives a measure

definition, the

spread over which the gain remains within 3 db of the

of the

frequency

maximum

value,

then

45 4 Q 4

from which

it

follows that the

bandwidth

B =

25/o

is

(9-55)

comparison of this result with Eq. (9-16) for the single-tuned stage
shows that the 3-db band width of the double-tuned circuit is 1.414 times
that of the single-tuned stage.

Cascaded Tuned Amplifiers. It is frequently necessary to incormore than one stage of amplification in a given amplifier.
Although such a practice provides a higher gain, this higher gain is
accompanied by a narrower bandwidth than for the single stage. The
situation here is sensibly the same as that which was considered in Sec.
9-4.

porate

Analytic expressions for the effect of cascading identical amplifiers

6-6.

method of analysis.
n single-tuned stages in cascade. The gain of such an
n-stage amplifier becomes, from Eq. (9-15),
are readily possible, following the previous

Consider

first

(9-56)

To

find the corresponding

[1

bandwidth,
(25Q e ) 2 ]'" 2

it is

noted that

V2

so that

28Q e =
But the bandwidth

is

V2

'"

given by
(9-57)

This

may be expressed

in

terms

of the

bandwidth

of the single stage B\, in

the form

B ln =

B,

V2

1/n

-~T

(9-58)

RADIO ELECTRONICS

226

[Chap. 9

Table 6-1 gives the bandwidth reduction function \/2 Un 1. It is

two stages in cascade have a bandwidth that is
only 0.64 times that of a single stage. To maintain a given bandwidth,

seen, for example, that

it is

accordingly necessary that the

number

as the

of the individual stages

be decreased

of stages is increased.

corresponding expression

is

possible for the double-tuned amplifier.

For such an n-stage amplifier, with critical coupling a = 1 and equal

primary and secondary values of Q, the relative gain becomes, from Eq.
(9-54),

\KreJ
It follows

from

The bandwidth

may be

45 4 Q 4 )" /2

this that

of the w-stage amplifier then has the

B 2n =
which

(1

25/

yj-

form

written in terms of one-stage bandwidth as

B in = B v^
2

'"

(9-59)

The band-width reduction factor is tabulated in Table 9-1. For a

two-stage double-tuned amplifier with the coils critically coupled, the
bandwidth

is

0.802 times that of the single-stage amplifier.

TABLE

Note that

9-1

THE DOUBLE-TUNED-AMPLIFIER BANDWIDTH REDUCTION

FACTOR FOR a = 1
\Z2V - 1
n
1

2
3
4
5
6
7

8
this reduction

is

1.00
0.802
0.713
0.659
0.622
0.592
0.568
0.548

considerably less than the corresponding reduction of the

two-stage single-tuned amplifier.

This arises from the fact that the

curve of the double-tuned amplifier has
steeper sides than that of the single-tuned circuit and with successive
stages drops away less rapidly than for the single-tuned case.
In paramplification

or selectivity

Sec. 9-5]

TUNED POTENTIAL AMPLIFIERS

227

an ideal amplifier with a rectangular response curve would show

no bandwidth reduction with the addition of successive stages.
9-5. Gain -Bandwidth Product.*
It is of interest to tabulate the gain
at resonance of the three amplifier circuits that have been studied.
These
ticular,

follow:

K
K

= gm oioLQ e
MQ
res = g m
K, = j0.5gm QiQ 2 s/ R\R 2

Single-tuned direct-coupled:
Single-tuned transformer-coupled:

Double-tuned:

These expressions

may be

res

<j}

interpreted as showing that the gain in each

case has the form

K =
s

gmZ

(9-60)

where gm is the transconductance of the tube and Z is the effective

impedance of the load. Moreover, the foregoing analyses for these
amplifiers show that the bandwidth in each case varies inversely with the
effective Q of the tuned circuit.
Clearly, therefore, the higher gains are
accompanied by a decreasing bandwidth.
The gain-bandwidth product of the single-tuned direct-coupled ampliThere results
fier is obtained by combining Eq. (9-10) with Eq. (9-16).

K B

res

which

may

g m w<>LQ e

^ = gm

be written in the form

"F

2irC

In the limit where the capacitance C is due only to interelectrode capacitances, the limiting gain-bandwidth product is

K ~

= 2,(0,

(9 " 61)

c,o

is the same value as found in Sec. 6-5 for the

The gain-bandwidth product of the double-tuned

and

combining Eq. (9-41) with Eq. (9-55) and

K B=
res

Q Q 2 VRJl*
1

gm

V2

V Q1Q2R1R2

Svivo^

amplifier.
is

found by

is

= 7=

RC

amplifier

VQ1Q2
7T-

(9 " 62)

This expression shows that the gain-bandwidth product of the doubletuned amplifier is y/2 as great as that for the single-tuned circuit. That
*

Refer to Sec. 6-5.

RADIO ELECTRONICS

228

by

[Chap. 9

tube input and output capacitances by the use of the

double-tuned circuit, there is an increase in the gain-bandwidth product.
9-6. Stagger Tuning. 2
If it is desired to build a wide-band high-gain
amplifier, one procedure is to use either single-tuned or double-tuned
circuits which have been heavily loaded so as to increase the bandwidth.
is,

splitting the

The gain per

stage

is

correspondingly reduced, by virtue of the constant

gain-bandwidth product. The use of a cascaded chain of stages will

provide for the desired gain. For example, a particular amplifier comprising nine cascaded single-tuned stages each having a 6 Mc bandwidth
has an over-all bandwidth of 1.7 Mc. A nine cascaded chain of doubletuned amplifiers, each also of 6 Mc bandwidth, yields an over-all bandwidth of 3.2 Mc. Generally, for a specified gain and bandwidth the
double-tuned cascaded amplifier is preferred, since fewer tubes are often
(2)f

L-L,

-5-4-3-2-1
Fig. 9-10.

The adjustments

-,

.,/fesijltanr

X res oonse

of frequency of a stagger-tuned pair.

possible, and also since the pass-band characteristics of the double-tuned

cascaded chain are more favorable, falling more rapidly outside the pass
band. From practical considerations, the double-tuned stages are more
difficult to align, and they also are more sensitive to variations in tube
capacitance and coil inductance than the single-tuned circuits.
A means is available for achieving the large bandwidth and other
characteristics of double-tuned circuits by using single-tuned circuits.
This consists in taking two single-tuned circuits of a certain bandwidth,

and

displacing, or "staggering," their resonance peaks by an amount

equal to their bandwidth. The resultant staggered pair will have a bandwidth that is \/2 times as great as that of each of the individual single-

tuned

making up the pair; the over-all selectivity function will be

form with that of a single-stage double-tuned system. The

circuits

identical in

general situation

An

is

illustrated in Fig. 9-10.

analytic expression

is

readily obtained for the over-all character-

the stagger-tuned pair.

If it is noted that the general selectivity

function of the single-tuned direct-coupled circuit is, from Eq. (9-14),
istics of

TUNED POTENTIAL AMPLIFIERS

1
K
1
1 + j28Q
K
1 + jx

Sec. 9-6]

rea

229

and the bandwidth between the 3-db points

B =

2 8/o

is,

from Eq.

(9-16),

then the corresponding selectivity functions of the two circuits are,

respectively,

+ x+
j(fi

1)

+ j(x -

i)

and

\W
By

multiplying the selectivity functions together, there results

1

The magnitude

x2

2jx

of the resulting function is

(k~)i {k~) s ~ V4Tx~< ~

vT^Hp

new frequency

(9 " 63)

uo and where Q
comparison of this
expression with Eq. (9-54) for the double-tuned circuit shows that the
forms of the variation are identical.
It is of some interest to compare the gain-bandwidth products of the
following: two synchronously tuned stages, a stagger-tuned pair, two
synchronously tuned double-tuned stages. The results are found to be,
Thus, not only are the
respectively, 0M3K 2 B U 0.707K 2 B h 1.13K 2 Bi.
resultant gain-bandwidth products higher for the more complicated
coupling systems, but the selectivity is also better.
The advantage of stagger-tuned amplifiers, and the principle may be
extended to stagger triples (and to stagger n-uples in general), lies in
This
the fact that simple single-tuned circuits are used throughout.
makes the alignment of the stages relatively easy, especially if stagger
triples or higher were to be used, since no interaction exists among tuning
elements of the several stages. To attempt a triple-tuned single-stage
coupling network proves an almost impossible practical tuning task. A
practical disadvantage of the stagger-tuned circuit exists which makes
the double-tuned circuit preferable and often almost necessary for 60 Mc
and above. The input impedance of an amplifier stage at these high
frequencies is low and may be no more than several thousand ohms.

where

is

So is

the value of

the value of

Qe

referred to the

for each circuit referred to

coo.

RADIO ELECTRONICS

230

[Chap. 9

This input loading of the tuned circuit of the previous stage may make it
almost impossible to effect stagger tuning for a prescribed over-all bandwidth, whereas this loading affects the double-tuned response characteristics to

a lesser degree, with a consequent less stringent restriction on the

operation.

As noted, the
triples

and staggered
In this case a centered single-

principle of stagger tuning can be extended,

have been used

in radar receivers.

tuned circuit of relative bandwidth 2 and relative resonant gain

is
combined with two single-tuned circuits, each of relative bandwidth 1
and relative gain 1, staggered so that their resonance peaks are + "\/3/2
from the band center. The resulting sensitivity function is of the form

K
K
r

(9-64)

Vl +

This selectivity function has the same form as that for an optimally

flat

triple-tuned circuit.
9-7.

The

Parallel- or

Twin-T

Circuit. 3

The

works
-\fr

plifiers

use of

RLC

resonant net-

for achieving band-pass

am-

with high selectivity has

been considered in some detail in

Such amplifiers are not suitable for use at
the low frequencies, since it is difficult to obtain high-Q coils at the
Fig. 9-11. The parallel-T circuit.
low frequencies, and moreover the
circuits become rather bulky at these lower frequencies.
In fact, a tuned
circuit which has a high Q at, say, 30 cps would be extremely difficult to
build.
Fortunately, a number of RC networks possess frequency-selective
They
properties like those of resonant and band-pass filter sections.
have, as a result, found widespread application.
The parallel-, or twin-T,
circuit is one of this type and will be examined in some detail.

R
-AAWV

Fig. 9-12.

The form

-A/WW-

The

parallel-T circuit

the foregoing sections.

drawn

for a junction analysis.

be analyzed is illustrated in Fig.

not the most general choice of parameter, it is a form
which has received widespread use. To analyze this network, the
current-source form of the network is used, and the circuit is rearranged
The output potential can be written in determinantal
as in Fig. 9-12.
9-11.

of the parallel-T circuit to

While

this

is

TUNED POTENTIAL AMPLIFIERS

Sec. 9-7]

form by inspection.

It is

+ jun
-G

2(0

E =
2

GEi

C)

-j<*c

(9-65)

G
G + juC

+ jwtfC)

2(0

-G

-jcoC
'

Ei

Now

determinants yields

of these

(S+>c)

-J<C

The expansion

231

a n

u 2n 2 R 2 C 2

R C + j2(n +
2

(9-66)

l)a>RC

define the quantity

1
COO

and combine with the above.

E2

(coo)

(co/co

(9-67)

nRC

This gives

2
)

+3

(percent)

coo

which assumes the form

(n-t)

20

15

2(n 2

o/coo

+3-

(co/coo)

Band*vid+hB
1)

10

(9-68)

plot showing the variation of

a function of

co/co

for

1 is

(J

as

given

Note that the ratios

E2/E1 are the same for equal values
co/coo and coo/ co.
in Fig. 9-13.

should be observed that at resonance E 2 = 0, and the network may

It

0.6

0.S

1.0

1.2

0)/O)

Fig. 9-13.

The

and w/wo

for

1.4

1.6

>-

relation between
2 /E,
the parallel-T circuit.

be terminated in any impedance Z across the terminals without any effect.

finite terminating Z will greatly affect the frequency-response

However, a

characteristics of the network, except at the resonant value, although the

For example, if
comparable with the network parameters, the off-resonance

general characteristics remain substantially unchanged.

the load

is

RADIO ELECTRONICS

232
response

is

seriously affected, the selectivity

[Chap. 9

poor,

is

and the curve

is

no

longer symmetrical about the resonant point.

To

find the

bandwidth

of the

network, write

(9-69)

Equation (9-68) then becomes

But the bandwidth is defined by the requirement that the amplitude

by 3 db over the frequency range. This occurs when

ratio

fall

n2

1
*

fib

or

%
But

for

to

it is

(o>

(9-71)

O!0

^ _

Hence

2 4-

O)

tog

seen that the bandwidth

B =

2A<o

coq)(co

is

tio)

given by

= ^ =

(9-72)

fi 6

As already noted, if the network is loaded, the sharpness of the null

point will be affected, and the bandwidth or the Q of the network will be
correspondingly changed. It is of importance therefore that the input
impedance of the network be ascertained. This is readily accomplished
by noting from Fig. 9-12 that the total current from source Ei is made up
two components, that toward junction
Evidently
of

which

Ii

Ei(G

i!

Ei

7
Zl

_ Ei _
~ 17 "

and that toward junction

+ jo>C)

3.

(9-73)

is

so that

R
_
~
+jcR

At the resonant frequency nwCR

Zl

1,

= ITTI =

and

H
:

r^o>cR

for

I (1 - &

{g -

u>

Sec. 9-7]

TUNED POTENTIAL AMPLIFIERS

233

In fact, Zi does not vary rapidly in the neighborhood of resonance, and

this expression may be used for Zi for ordinary calculations.
Attention

network is symmetrical, so that Z 2 = Zi.

For the normal circuit parameters used, the output impedance is fairly
large.
Consequently a load of several megohms might still constitute an
appreciable load on the network. This fact will dictate both the location
of the grid resistor when such a network is used in an amplifier and also
the form of the amplifier circuit. Moreover, the network is inherently a
rejection device, and its use in a cascade circuit with the network in
either the input or the output of an amplifier will provide a rejection, or
"notch," type of device. By combining the network in appropriately
chosen feedback circuits, a frequency-selective band-pass amplifier may
be achieved.
The simplest form is sketched in Fig. 9-14. Loading of the
is

also called to fact that the

Fig. 9-14.

simple form of parallel-T band-pass amplifier.

output of the network is avoided by injecting the signal in the cathode

To avoid loading of the input, more elaborate circuits must be

circuit.

used.

Three circuits incorporating such parallel-T networks in negative feedback amplifiers are given in Fig. 9-15.
The output response characteristics of such amplifiers are of considerable interest.
Observe that in these circuits g =
at the resonant frequency and (J ^
for au other frequencies.
Consequently feedback
occurs at all frequencies except at the frequency w at which (J = 0.
Thus the circuits have maximum gain at co and, owing to the feedback,
This means that the circuit
the gain falls at all other frequencies.
attenuates all frequencies except co
so that a "band-pass" amplifier
,

does
If

will

exist.

the nominal gain of the amplifier

K, then with feedback the form

is

be

K/ =
1

(9-76)
ffK

234

RADIO ELECTRONICS

(c)

Cascode

Fig. 9-15. Practical parallel-T feedback band-pass amplifiers.

[Chap. 9

TUNED POTENTIAL AMPLIFIERS

Sec. 9-7]

Now

combine

with Eq. (9-70)

this expression

The

235

result

is

K, =

1-K-

-J

n2

Then

rr

K/ =

n*+l\

.2

17

from which

it

0-77)

o^nr~Li

follows that

11

K| 2

/ 2 n2

\S2

the effective

It is of interest to define

RLC circuit.

comparison with an

nM-jV

/2

(9-78)
1

of this circuit, as a basis

At the half-power points the gain

of

ratio

must be l/\^2, whence

Upon

solving for

ti b

\a b

-K +

from

K is negative and

\Q b

1,

which

is

tt

But by Eq.

greater than

2(n

|1

+ lY

- KP ^_

when K

/ 2 w2

ft<b

Also,

this expression, there results

[1

Ordinarily

/ 2 n2

<9

fl
*

1.

Then when

|1

K| 2

'79

>

>5> 2,

X)/n

K|

the usual case,

+R

2(n 2

'

l)/n

"

(9-80)

(9-72)

Then

Q =

.f^w
+ l)/n

2{vr

(9-81)

RADIO ELECTRONICS

236

The use

of a

[Chap. 9

pentode in a single-tube parallel-T feedback

ordinarily have a nominal gain of 100 or

more

circuit will

at audio frequencies.

Consequently such an amplifier with a parallel-T loop becomes equivalent

Q of the order of 25 or more, for n = 1. A
triode amplifier will, because of the lower gain, give a lower effective Q.
However, this may be ample for many applications. Of course the cascode circuit of Fig. 9- 15c has a high gain with triodes and will yield
to a resonant circuit with a

appreciable
9-8.

values.

The Bridged-T and Wien-bridge

discussed in the foregoing section

is

The

Circuits.

parallel-T circuit

only one of a number of different

TABLE 9-2
FREQUENCY-SELECTIVE RC NETWORKS
Bridged

1.

R
\R,

Bridged

w^'
2R&

O)/o)0

+j

11.

RC

-3 In +
2

E,

<n
S2>

3.

wher e Q =
~R

_L
1

2.

V2
LC

+ 4ra -

for

Wien bridge

RC

may be used to yield band-pass

tuned resonant circuits. Among
several forms) and the Wien-bridge networks

frequency-selective networks which

characteristics

somewhat

like the

others, 4 the bridged-T (in

RLC

have the desired characteristics. The final results are tabulated

without proof in Table 9-2 (see Probs. 9-24 to 9-26).
also

TUNED POTENTIAL AMPLIFIERS

237

REFERENCES
Aiken, C. B., Proc. IRE, 25, 230 (February), 672 (June) (1937).
Wallman, H.,
Radiation Lab. Rept. 524 (Feb. 23, 1944).
Wallman, H., Electronics, 21, 100 (May, 1948).
Stanton, L,, Proc. IRE, 34, 447 (1946).

1.

MIT

2.

3.

Massachusetts Institute of Technology, Radiation Laboratory Series, vol.

18, McGraw-Hill Book Company, Inc., New York, 1948.
As a general reference, consult:
Sturley, K. R., "Radio Receiver Design," pt. I, John Wiley & Sons, Inc.,
New York, 1943.

4.

5.

PROBLEMS
= 250
A 6SJ7 pentode is used in a certain class A r-f amplifier, with
E = 100 volts, and E = 3 volts. At these conditions the tube
parameters are approximately g m = 1,600 jtmhos, r p = 1.2 X 10 6 ohms. A
9-1.

volts,

CC 2

cc \

single-tuned load consists of a 1-mh coil in parallel with a 100-fifii capacitor.

The resonant Q of the load is 200.
a. Determine the potential gain of the stage at the resonant frequency.
b. Determine the potential gain of the stage 10 kc above and below resonance.
9-2. In a single-tuned direct-coupled amplifier stage using a 6SJ7 tube that is
tuned to 1,100 kc, it is found that the bandwidth is 18 kc. Determine the Q of

the circuit.
9-3. It is discussed in the text that the response of the single-tuned directcoupled or transformer-coupled amplifier for small deviation 5 and high Q is
given by either Eq. (9-14) or Eq. (9-29).
a. Calculate the error in each case for Q = 3, 8 very small.
b.

for Q large, 5 = 0.1.

single-tuned circuit employing a

Repeat

9-4.

6SK7 tube feeds a diode detector.

The tank is tuned to /o = 1 Mc,
6SK7 that rv = 10 6 ohms, gm = 2,000

The equivalent output

circuit is illustrated.

L =

Assume

0.5 mh, Q = 60.

jumhos.
Determine:
a.
6.

Gain at resonance.
Gain at resonance

for the

v
if

detector

.^fx
is

removed.
c.

Bandwidth with and without the

f"-^

\S>

*f

< Diode detector

% extent /cJSO*

*
detector circuit.
9-5. A direct-coupled single-tuned amplifier has a band width of 150 kc and a
resonant shunt impedance of 50 kilohms. What must be the value of the shunting resistance across the tank if the gain is to be constant within 10 per cent over

the 150-kc band?

9-6. A direct-coupled single-tuned amplifier has a bandwidth of 50 kc, when
C = 25 utii. Calculate the bandwidth under the following conditions:
by
o. C is increased to 100 MMf, and the resonant frequency is kept constant
appropriately changing the number of turns on the coil.
b. C is increased to 100 jujuf, but L is maintained constant.
changing the
c. C is maintained at 25 \ifxl, but L is increased by a factor of 4 by
position of the slug-tuning core.

RADIO ELECTRONICS

238

[Chap. 9

A single-tuned direct-coupled stage has a Q = 100 when tuned to 800 kc.

equal signals are fed to the grid, one of which is 50 cps off resonance and the
other of which is 5,000 cps off resonance. What will the amplitude ratio be in
the output of the amplifier?
9-8. A direct-coupled single-tuned amplifier is to have a bandwidth of 200 kc
at 4.7 Mc. If the total capacitance is 25 yu/uf:
a. Calculate the maximum impedance and the value of L.
b. Over what frequency band is the total phase shift through this amplifier
less than 30 deg?
9-9. The circuit of a cascode band-pass amplifier (often called the Wallman
circuit) is shown in the accompanying diagram.
Zi and Z 2 are parallel resonant
circuits tuned to the same resonant frequency.
9-7.

Two

a. Find an expression
and l/r p2 where Gx and
,

for
<? 2

Y 2 ,i D assuming that g m2 is much larger than Gi, G 2

are the shunt conductances of the tuned circuits at
,

resonance.
b. Subject to these same conditions, find an expression for the over-all gain
of the amplifier, and show that it is the same as that of a single grounded-cathode
stage with a parallel resonant interstage network [Eq. (9-13)].
The advantage of such a cascode amplifier over that of a single-stage circuit
utilizing a pentode is that it provides the low noise level of a triode with the
operating characteristics of a pentode. Triodes may be used because the Miller
effects are negligible owing to the very low potential gain of the first stage.
Also,
the cathode lead inductance in the second tube is unimportant because of the low

impedance level.
9-10. Repeat Prob. 9-8
for

optimum

9-11.

10

Mc

for the transformer-coupled single-tuned amplifier,

coupling.

C =

double-tuned circuit with

for the critically coupled stage.

primary and secondary windings are

12 /i/if has a bandwidth of 1

Determine the value of L and Q

Mc
if

at

both

identical.

6SK7 tube is critically

455 kc and has a bandwidth of 12 kc. The total primary
C and total secondary C are each 26 pui. The coils and loading are the same.
a. Calculate the values of shunt resistance of each circuit, L and M.
b. Calculate the mid-frequency gain.
Choose g m = 1,500 yumhos.
c. If the resonant frequency and the bandwidth are maintained constant,
calculate the mid-frequency gain as C is varied from 10 to 100 nn(.
9-13. A 6SK7 double-tuned circuit comprises two identical 200-/ih coils, with
Q = 80, which are tuned to 500 kc.
9-12.

coupled.

single-stage double-tuned amplifier using a

It operates at

Calculate the critical coefficient of coupling.

Calculate and plot the gain of the stage as the mutual inductance is varied
from zero to twice the critical value.
9-14. Determine the proper design for the winding of an i-f transformer with
L\ = L2 and each winding tuned to resonance by a capacitance of 100 /a/uf. The
a.
b.

TUNED POTENTIAL AMPLIFIERS

239

secondary potential is not to fall below 0.88 of the peak value in a 10-kc band,
centered at 465 kc. Find k, L h L 2 Qi, Qn and the secondary potential, with 1 volt,
465 kc to the primary. Assume critical coupling.
9-15. A 6SJ7 is used in a double-tuned circuit which feeds a diode detector and
automatic-gain-control circuit. The significant portion of the circuit is shown.
The rms potential across the secondary feeding the detector-automatic-gaincontrol circuits must be 10 volts.
,

Diode detector and age

equivalent load 200*

i-f=4S6 kc

What
What

the coefficient of coupling?

the value of e g to give the required output?
9-16. The i-f amplifier section in a radar receiver consists of four identical
stages, each of the form illustrated.
The maximum over-all gain of the four
stages is 10,000, and the gain at 28.5 and 31.5 Mc is 7,070.
6AC7 tubes are used,
with g m = 9,000 ^mhos, C< = 11 ju/xf, <7 = 5 iiitf.
a.
6.

is

is

Calculate the value of the plate-load resistance.

Calculate the wiring capacitance.
c. Calculate the value of the inductance of the coil.
9-17. A six-stage single-tuned i-f amplifier using 6AC7 tubes has a maximum
over-all gain of 4,100 and an over-all bandwidth of 6.0 Mc.
If the over-all gain
had to be obtained from four stages instead of six stages, what would have been
the bandwidth?
9-18. A six-stage single-tuned amplifier using 6AC7 tubes has a maximum
over-all gain of 530,000 and an over-all bandwidth of 2 Mc.
a. If it is found that the over-all bandwidth need not be greater than 1.5 Mc,
what would be the corresponding over-all gain by an appropriate change in the
value of the effective shunt resistance of each stage?
6. If the original over-all gain of 530,000 had been obtained from four stages
instead of six, calculate the over-all bandwidth that would result.
9-19. Suppose that three identical stages having the characteristics of Prob.
9-2 are connected in cascade.
Calculate and plot a curve of relative gain in
decibels vs. frequency.
Carry out the calculations to frequencies at which the
gain is down at least 75 db below the optimum value.
9-20. A three-stage direct-coupled single-tuned amplifier is used in a broadcast receiver.
A three-ganged 165-jUjuf capacitor is used to tune the receiver
over the range from 550 to 1,650 kc. The loading is chosen to give a minimum
bandwidth of 10 kc.
o.
6.

RADIO ELECTRONICS

240

[Chap. 9

a. Determine the variation of bandwidth as the receiver is tuned over the entire
range, assuming that Q remains constant.
b. Repeat for the case where C is fixed at 100 yu/uf and L is varied.
9-21. Refer to a tube manual, and prepare a table of the merit of the following

tubes: 6AB7, 6AC7, 6AG7, 6AK5, 6C5, 6K7, 6L6, 6SF5, 6SJ7, 6SK7, 6V6, 6Y6.
9-22. A two-stage direct-coupled single-tuned amplifier using 6AC7 tubes

operates at 60 Mc and is to have a 2-Mc bandwidth.

a. What gain is possible if both stages are tuned to the same frequency?
Assume that the shunt capacitance is 25 ;ujuf.
6. If the stages are to be stagger-tuned to be critically flat, what gain is
possible?
9-23. Show that, by choosing the three single-tuned stages in the manner
discussed in the text to yield a staggered triple, the relative response function has
the form given by Eq. (9-62). Sketch the individual response characteristics
and that of the resultant staggered triple.
9-24. Analyze the bridged-T network, given as circuit 1 in Table 9-2, in the
manner of Sec. 9-7. Verify the data given in Table 9-2 for this network.
9-25. Repeat Prob. 9-24 for the bridged-T network, given as circuit 2 in Table
9-2.

9-26. Repeat Prob. 9-24 for the Wien-bridge network, given as circuit 3 in

Table

9-2.

The network shown is used in a simple potential-feedback circuit.

Calculate the transfer function (3 of the network, when R\Ci = RiC 2
Calculate and plot on a decibels vs. log / scale the relative gain of the

9-27.
a.
b.

amplifier.

9-28. Repeat Prob. 9-27 for the network shown.

9-29. The two amplifiers of Probs. 9-27 and 9-28 are connected in cascade.
Calculate and plot the gain of the resulting two-stage band-pass amplifier.

CHAPTER

10

TUNED POWER AMPLIFIERS

10-1. Introduction.
In common with the operation of the classes of
tube circuits being studied, it is the function of an r-f power amplifier to
convert d-c power from the power supply into r-f power.
Owing to the

amounts

power that

of

may be involved, it is essential that this conversion

be effected at the highest possible efficiency.

Essentially, therefore, the

power amplifier may be regarded as a power converter, as contrasted with

the r-f and i-f potential amplifiers that are used to raise a potential level.
The settings of the r-f power amare chosen to ensure a high
conversion efficiency.
plifier

The

basic circuit of a tuned

amplifier

is

power

substantially that of the

sihgle-tuned direct-coupled type dis-

cussed in Sec. 9-1.

The

differences are in the

magnitude

essential

the grid-bias supply potential

of
cc ,

Fig. 10-1. Schematic diagram of a tuned

amplifier.

power

the corresponding value of the grid

of power involved
A schematic diagram
given in Fig. 10-1.
Owing to the negative bias on the tube, which is adjusted approximately

input signal
of a

eg

and the amount

tuned power amplifier

is

to plate-current cutoff in the class B amplifier and which is adjusted

beyond plate-current cutoff in the class C amplifier, harmonic currents
are generated in the plate which are comparable in amplitude with the
fundamental component. However, if the Q of the tuned plate circuit
has a value of 10 or more, the impedance of the tank circuit to the second
or higher harmonics will be very low.
As a result, the higher-harmonic
potentials across the tank will be very small compared with the fundamental potential. That is, the effect of the harmonic generation in the
tube plate current is largely suppressed by the tuned plate load.
But the requirement that the Q of the tank circuit must be high in order
to suppress harmonics in the output imposes a limitation on the fre-

quency-response characteristics of the amplifier, since then the gain is

constant only over a very narrow band of frequencies.
Consequently

such amplifiers are confined in their operation to narrow frequency bands.

241

RADIO ELECTRONICS

242

IChap. 10

In fact, as will be discussed in some detail, the class B amplifier may be

used to amplify a narrow band of frequencies of differing amplitudes,
whereas the class C amplifier is confined to a narrow band of frequencies
of constant amplitudes.
Despite these severe restrictions, both classes
used in restricted applications, the class B
r-f carrier wave, the class C amplifier as a
frequency multiplier or as a source for the production of an a-m carrier
wave.
of amplifier are extensively

amplifier to amplify

an a-m

Tank

10-2. Properties of the

The tuned

cuit.

diagram
Fig.

tank

10-2.

circuit.

is

drawn

is

tively to the plate tank,

typical coupling network

is

10-1

as

a simple parallel resonant circuit.

Ordinarily the load is coupled induc-

tuned-amplifier

typical

Fig.

of

Cir-

plate load in the

that shown in Fig. 10-2.

assumed to be so adjusted that l/2x

s/Li.Ci.,

The

and a more

capacitor

C2

the resonant frequency

secondary circuit, is equal to the operating frequency of the ampliBecause of the resonance in the secondary circuit, only a resistive
component R'L = oiM 2/{R" + R 2 ) is reflected into the primary of the
tuned circuit. The equivalent circuit then becomes that shown in
of the

fier.

Fig. 10-3.
If

the characteristics of the tank

circuit

were

ideal,

the impedance at

resonance would be resistive and

Fig.

10-3.

The equivalent

circuits

of

Fig. 10-2.

equal to the shunt resistance Ro of

the resulting network. The impedance would be zero at any of theharmonic frequencies. That is, the impedance would be

= Ro
=

Z(cDo)

Z(ncoo)

These ideal conditions do not prevail

(10-1)

2,3, 4,

in practice,

to achieve relatively low impedance for Z(no>o).

although

it is

To examine

possible

this, refer

to Eq. (9-6) for the impedance function of the simple tuned circuit,

Z ~ RlQ '
At resonance
Z(

to

= w and
,

= RlQ

Note, however, that,

if

Equation

0.

-J^) = r lQ

(l

Q =

Z(

+ 8- j(l/Q)
+ d + jQS(2 + 5)
1

10,

(10-2) reduces to

then

= RlQ X
2

(10-2)

1.005/ -5.7

A an "'

TUNED POWER AMPLIFIERS

Sec. 10-2]

which shows that the impedance of the tank circuit

and is given by

Now
co

Z(2co

Q =

and Eq.

= R L Q>

_^

i-j-L
= R
ltLV
LQ
H.5Q ~
TTJT

ance

When

- 25

QQ +

h5Q

r+Y25 Q*

)
'

R^J^Q^-J^^

dO-6)

harmonic to the fundamental-frequency imped-

Z(wo)
fact,

(10-4)

then
Z(2o>o)

In

10 this reduces to

ratio of the second

is

= Q

(10-2) reduces to

Z(2.)

The

consider the situation at the second-harmonic frequency.

2coo, 5

For

(10-3)

follows that

it

Ro = RlQ 2 = cooLiQ

essentially resistive

= R = RlQ 2

Z(coo)

Under these conditions

is

243

= RlQK\/1.5Q)
=
RlQ 2

under the extreme conditions when Ipi

J_
1.5Q
=

I p i, the relative power

ratio is

P L2

I*pl

/| 2

Re Zjcop)
Re Z(2 )

RlQ'IU

RlQ'IU^I +

where Re denotes "the real part

=
2.25Q 2 )

With Q =

of."

^=

*U

10, this

n2 j

reduces to

900

"L2

power

Clearly, therefore, the second-harmonic

is

negligible

under these

conditions.

Obviously, there will be losses in the tank circuit owing to the resistive

component

of the coils,

to the load

is

P'l

and the power

= (Qhi)

lost in the

PL =
The

and perhaps the capacitor.

coW

R>

tank circuit

(QIPl y

Rl

circuit transfer efficiency,

R2

The power

delivered

R'l

(10-7)

R + R%

is

^fV ^f^)

(10-8)

which

is

denned as the ratio

of the

power

RADIO ELECTRONICS

244

[Chap. 10

delivered to the load to that supplied to the tank circuit,

An

P
plf

possible

Vm

P L ~ Pt
X 100%

100% =

and informative form

by writing it as follows:

_ power delivered

to secondary

power delivered to load

power delivered to secondary

where 771 is associated with the first ratio and

second ratio. These may be written as
IIR'l

IfiR,

The

expression for
uiaLi/Ri

I\{R 2

r\\

may

woL,/(R'L

R +
1

=
R'l)

is

associated with the

R'L

R'l

R +
2

R'l

be written in the following forms:

Ri)

171

where

tj 2

R'l

+ R'L

Similarly

(10-9)

for the circuit transfer efficiency

a +to primary
power delivered

7,2

given by

'

lP X

interesting

is

is

Q
= Q 01

il

Qoil

~ 0^

/in ir

>

(1(MOa)

Q n = woLn/Rx is the unloaded Q of the primary coil at resonance

= u<Jji/{R'l + Ri) is the loaded Q of the primary circuit at

Qoiz,

resonance, including the reflected resistance

secondary in the primary circuit

In an entirely similar way, the expression for 172 may be written in the form
of the

^
where Qo 2

=
=

uoLi/

is

o>oL 2 / (R 2

- %2

the unloaded

(10-10b)

Q of the secondary coil at resonance

Q of the secondary coil at resonance but without any effect of the primary
circuit on the secondary
The complete expression for the circuit transfer efficiency becomes
Q02L

R'l) is the loaded

For high circuit transfer efficiency, the loaded values Qou, and Q 02L
must be low, and the unloaded values Q i and Q02 should be high. Ordinarily the loaded Q's must be 10 or greater in order to provide for a low
harmonic content in the output. The unloaded Q's are subject to purely
practical limitations; the possible values depend on the power output, the
character of construction of the coil, and the frequency of operation.

Sec. 10-3]

TUNED POWER AMPLIFIERS

Typical values for

coils of

245

conventional design vary somewhat as follows

range from 500 to 1,500 kc:

for frequencies in the

~ 100-200 for low-power coils

~ 500-800 for high-power

Unloaded Q

coils

10-3. Choice of Ql-

influence the choice of

It is of

some

interest to

examine the factors which

Several of the factors have already been

considered, but for completeness these will also be included in the tabula-

The

tion below.

following conditions prevail for low Ql-

3.

High circuit transfer efficiency

Broader bandwidth.
Higher harmonic components.

4.

Greater

1.

2.

L/C

77.

ratio.

has been considered in considerable detail in Sec. 10-2. Factor

This must be adequate to pass
2 relates to the width of the pass band.
the desired frequency band but must attenuate the frequencies outside
A measure of the response is obtained from Eq.
the specified band.

Factor

(10-2),

which becomes, for frequencies near resonance

(0/co)[l

Z(coo)

CI
^

0-121

(<>/)*]

Factor 3 was discussed in some detail in Sec. 10-2, where it was shown that
When Q L is low, the
is small if Q L is fairly high.
harmonic output is not negligible and might result in troublesome harthe harmonic output

monic potentials in the circuit.

Factor 4 is examined through Eq. (10-4) for the lowest Q L for a specified
R this demands that the L/C ratio must be high. The highest L/C
ratio exists when C is a minimum, which, in the extreme, is the tube plus
If a capacitor is used, it should be relatively
stray wiring capacitances.
;

In any design considerations Q L

by the allowable harmonic content and by power considerNormally, as already discussed, Q L will range from 10 to 20.
ations.
The unloaded Q u is determined by requiring that the circuit transfer
efficiency should be high, perhaps 90 per cent, at the lower powers and
With Q L and Q u known, the circuit
should be higher for high powers.
small, in parallel with a large inductor.

is

established

constants can be determined.

Example.
is

Evaluate the approximate circuit constants of a tank circuit which

72-ohm load at 2 Mc from an a-c supply of 2,000 volts.
Choose Ql = 12; i\ = 90 per cent. Also given, R'[ = 72 ohms,

to deliver 500 watts to a

Solution.

P" =
a.

500 watts.

Power input

p = ^? =
1

556 watts

RADIO ELECTRONICS

246
b.

From

expression (10-11)

Q" =
c.

[Chap. 10

u=

120

Since

coaLi

u>qL\I\I\

then
.

/l

Also

=
d.

To

find

M,

12

556

=3 33amp
-

^oobX

IS"

3.33

^ = 2^00^000 -

X ^ hen ^

132 4
-

10

- 12

farad

note that

Mh = IJt'l

P =

I\R'l

Hence

=
IiU

lib)

y/500
3.33
e.

72

=
10 6

Loaded

556

woo =

lO- 6 henry

' 277

amp

R
=

Loaded Ro =
g.

4.53

Current /
*' 1

/.

Unloaded

10-4. Class

7,220

ohms

R
12

Unloaded

R =

B Tuned

Amplifiers.

7,220

72,200

ohms

Considerations regarding the actual

choice of tube will be given in Sec. 10-16.

Transmitters

high- or low-impedance triodes, tetrodes, or pentodes.

that the plate-circuit efficiency,

i.e.,

may employ

It will

be found

the ability of the tube to convert

power from the supply into a-c power, is not particularly dependent
on the type of tube that is used. This fact will become clearer in the light
of subsequent discussions.
Under class B operation, the grid-bias supply potential Ecc in Fig. 10-1
is made negative by an amount sufficient to reduce the plate current to
zero for zero signal potential e B
If the dynamic characteristic of the
amplifier is linear over the range of operation, then for sinusoidal input
d-c

signal potential the current will consist of half-wave rectified pulses.

Sec. 10-4]

TUNED POWER AMPLIFIERS

The construction

for

247

deducing the output waveshape

is

sketched in

Fig. 10-4.

important that it be recognized that Fig. 10-4 represents an idealwhich depends upon a linear dynamic curve. This is not
completely true, although, in the analysis to follow, it will be assumed
If the dynamic curve is not linear,
that the linear relation does apply.
then a graphical solution must be used in order to determine the shape of
the plate-current curve and the linear class B analysis is not valid.
To find the operating path of an amplifier with a tuned load, a special
It is

ized picture

is required, since the conditions are different from those of

an amplifier with a pure resistance load. This is so because of the interrelation of a number of factors and the different manner of operation of
the circuit. Among the important factors that must be considered are

construction

Fig. 10-4.

The output waveshape from a

class

stage, with a linear

dynamic curve.

the allowable plate dissipation of the tube, the Q of the circuit, the effective shunt resistance of the tank circuit, the grid driving potential, the

shape of the plate-current wave, and the corresponding harmonic components in the plate current. Ordinarily a method of successive approximations is necessary in which a given set of conditions is assumed and a
If a consistent solution is not found, a second trial
calculation is made.
must be made. This procedure must be continued until a consistent
solution

is

found.

Although the determination

of the operating

the linear analytical solution to follow, the

path

method

not essential for

be discussed here,

is

will

permit a check on the validity of the linear assumptions.

a general method and will also be used later in the discusThe details of the construction are
sion of the tuned class C amplifier.
since

it

will

Moreover,

it is

illustrated in Fig. 10-5.

To
is

find the operating path,

sinusoidal

when the

it is

assumed that the plate-potential swing

grid input signal

is

sinusoidal.

Also, as a starting

RADIO ELECTRONICS

248

[Chap. 10

The
point, it is assumed that e b mm is approximately 10 per cent of E bb
value of e c msJC must not be allowed to reach an instantaneous positive
potential that is higher than the plate potential eb min ; otherwise the curThis may cause serious damrent to the grid will increase very rapidly.
.

Even if no damage results, the increasing grid current

accompanied by a decreasing plate current, and in consequence the
analysis will no longer be valid owing to the resulting nonlinearity of the
dynamic curve. With the indicated choice of conditions, the analysis
can be completed, and a calculation can be made of the following: the
d-c power from the plate-supply source, the a-c power output to the load,
age to the tube.
is

P/ate
current

wave

r
60

*~

120

cot

180

y-

r Grid

Sni!^\poferrHal

Plate potential

e,

swing

t>min

Fig. 10-5.

The

construction for determining the plate-current waveshape graphically

from the plate characteristics.

and the plate

dissipation.

If

the plate dissipation

is

within the rating

of the tube, then the resulting calculations will indicate the adjustments
of the circuit

parameters that are necessary to achieve the indicated

results.

The

specific

procedure

is

the following (refer to Fig. 10-5)

Select

any

particular instantaneous grid potential e c such as that corresponding to

,

the point
tial e b

by

c.

Determine the corresponding instantaneous plate potenA b at the same phase angle in the operating

locating the point

By

up to

intersection with the curve for the

on the operating path will be located.
Other points are determined in a similar manner. For class B operation,
the operating path should be approximately linear and should intersect
cycle.

projecting

selected grid potential, the point

Ebb approximately.
determine the shape of the plate-current pulse as a function

the plate-potential axis at

To

its

of the

TUNED POWER AMPLIFIERS

Sec. 10-5]

249

phase angle, the current corresponding to each point A on the operating

path is plotted as a function of the appropriate phase angle. The corre-

sponding plate-current pulse is plotted in Fig. 10-5 as

of Fig. 10-6 illustrate the important

waveshapes

The curves

of the amplifier.

10-5. Analytic Solution of

Class

(ib,wt).

An

Amplifier. 1

Tuned

analytic

solution of the tuned class

B am-

based on finding an analytic

form for the tube characteristics.
From Eq. (2-14), the general relationship between the plate current
plifier is

and the plate and grid potentials

of the form
ib

k (e c

ec

is

+- >

found that for power

paramthe plate current is of the form

Actually,

it is

triodes over a wide range of

eters

% =
which

may

is,

class

be written in the more complete form

h This

The important waveshapes in a

tuned amplifier.

Fig. 10-6.

k (e c

of course,

simply the

g m (e c

first

(10-13)

term in the Taylor expansion for the

current.

The instantaneous

potentials are of the form

ec
eb

But

since the current

is

= Ecc + E m cos cot

= Em Eplm cos wt

zero

when the

+-

ec

grid signal

(10-14)

is

zero, then, for

% =

0,

which requires that

E +
cc

or, for cutoff,

E
E = cc

ao-i5)

RADIO ELECTRONICS

250

(EE

By combining Eqs. (10-13) and

becomes

E + E gm

(10-14), the expression for plate current

cos ut

cc

-\

(I

which

is

g m (^E gm cos mt

flW

[Chap. 10

(Egm -

\
cos

cat

fi

cos co^

cos

(10-16)

tot

written in the form

4 =

Ibm cos

cot

4 =
where

/i

<

ost

<

<

coi

<

(10-17)

#m

The average value

of

the plate current

?r

P*

' rf

is

10 " 18 )

or

/ir

Also,

by Fourier

x/2
'

1^

(10-19)

ir

analysis, the amplitude of the

of the plate current

cos w /d(i)

Jo

fundamental component

is

1
= - /P* 4
tJo

cos

cof

d( w <)

(10-20)

or
2

= ~

/1

T-

But

at resonance Z(co

I''

-Ko is resistive,

Ep\m

j-,

It follows

from

Rolbm
2

and the fundamental-frequency

is

(10-22) with (10-21)

Upln

IplmRo

and

(10-22)

(10-17),

Ro
I ei
_ -2-9m\U<,m

Ep

_i_

Bp lm

^?

m\

this that
TP

(10-21)

<Z

potential difference across the load

Combining Eq.

2
Itm COS ut d((0i)

JO

TP

TUNED POWER AMPLIFIERS

Sec. 10-5]

251

or

E lm =
which

rms value

yields, for the

Ro

(10 " 23)

Ro+lr,,

fundamental-frequency component

of the

of current, the expression

Also,

from Eqs. (10-21) and (10-24),

_ h, =

The gain

2V2

7T

of the amplifier is

7T

^+

2?V

given in Eq. (10-23) and

K= -

/to
is

2^o

(10 " 26)

The d-c power input to the plate circuit, which is equal to the average
power furnished by the plate supply when the d-c power dissipated in the
plate load resistance

is

negligible, is given

by

2tt

Endh

2*J

d{u>f)

This becomes
Pbb

The

a-c

and

is

power output

of

= E i-

importance

d(cti)

= Ebb Ib

(10-27)

that at the fundamental frequency

is

given by
e L i P d(ut)

2x Jo
JO
which becomes
2t
1
Pl = 7T \[
Ztt

Jo

Ep

m COS wt I p i m cos

d{u>t)

alt

Pl = E pl Ipl = I^Ro
The

plate-circuit efficiency,

which

is

the ratio of

vP

r^X

Vp

Ep ilp
_
Ebb {2 V2/tt)Ip1

VP

100% =

Pac

to

hb , is

X 100%

78.5

(10-28)

2V2

Ep _
em>
i

X
4

E p i
Ebb
t

(10-29)

RADIO ELECTRONICS

252

The

plate dissipation

[Chap. 10

given by

is

2ir

etib d{o>t)

2tt

Jo

or

Pp =

2'

(Bw

i Jo

which becomes, by virtue

It is of

some

= E* J&
~ Pl

l)

(10 " 30)

Eqs. (10-27) to (10-29),

of

PP =

mum

~
~

(1

Vp)P

(10-31)

interest to calculate the results corresponding to the opti-

conditions

ec

m **

eb

mm

For

this condition

^c.max

eb

Ebb

Eg m
Eplm

cc ~\~

(10-32)

from which

Egm
By

~\~

Ep

Ecc

Efa

Eqs. (10-14) and (10-23), this yields

p
&gm

-u
T

v n
&(/m

Zrp

_ ^bb
P

FT

+
i

Ebb

,
>

/to

or

^~ _
The corresponding

2r p

(1 " 33)

l)i2.

expressions for the fundamental-frequency component

and the d-c components

of current are, respectively,

1)

\/2

(10-34)

2r p

2r?

+ J +lTa

(m

l)^o

and
/

g^M +

The corresponding values

of the

l)

optimum

jP m ,

calculated from these expressions for I v \ and

plate-circuit efficiency is

Poc

(10-35)
o

and

The

lb-

t\

are readily

expression for the

found to be
'Ebbbi

1)

2r p

'

12

'

2rp

1)

l)#o

(jt

1)#

which reduces to
tq k

B(M

or

nn, m

TUNED POWER AMPLIFIERS

Sec. 10-6]

253

Ordinarily the plate dissipation will be a fixed rating of the amplifier

and is the limiting factor on the output power. The appropriate value
of Ro is then specified, since all aspects of the circuit may be expressed in
terms of it. To examine this, note that

Pp

may

which

Pp =

Elb -{n

expression for
"

1)

+(fi

IpiRo

-R
1)R<>

2r p

y/2

0*

l)i2o.

may
,

be rearranged and yields the following quadratic

from which i2 may be evaluated:

_El(2_\\

4r p

be written as

This expression

Rl

Ebblb

PvV

10-6. Analysis of Class

4r j

Ro

(m

+ 1)

Amplifiers.

_ El irp
P p o* + 1)*

An

(10-37)

analysis of the operation

tuned class C amplifier can be made on the basis of the assumption

of a linear tube characteristic, essentially as an extension of the method
This analysis is considerably complicated by the fact that
of Sec. 10-4. 2
Ecc is no longer the single value chosen to yield a zero current for zero
Moreover, it is no longer valid to
excitation but is now a parameter.
assume that the operating characteristic is linear. Hence, although such
a linear-tube-characteristic analysis is possible, it is a poor approximation.
It does have the advantage over other methods of giving an explicit
of the

solution for the

optimum operating

mate nature, other methods are

conditions.

To see that the operating path is not linear,

is

again employed.

The only

Owing

to

its

approxi-

preferred.

the construction of Fig. 10-5

differences that exist arise because the

Ecc is adjusted beyond the cutoff value. With such values of

and with the appropriately increased value of grid driving potential,
the results have the form illustrated in Fig. 10-7. The curves of Fig. 10-8
illustrate the important waveshapes in such an amplifier.
grid bias

cc

comparison of these curves with those of Fig. 10-6 indicates that in

C amplifier the plate current consists of pulses the duration of
which is less than 180 deg of the cycle. Also, it is not possible, in general,
to derive easily an analytic expression for the shape of the plate-current
the class

pulse.

progress can be made in finding an approximate analytical soluthe curves of Fig. 10-7 are idealized. The idealization made is in
the assumption of linear curves, as illustrated in Fig. 10-9. This approximation permits the operating path to be represented by two straight-line

Some

tion

if

segments.

It is

now

possible to write

an expression for the plate-current

RADIO ELECTRONICS

254

Fig. 10-8.

The waveshapes

[Chap. 10

at various points in the tuned amplifier.

TUNED POWER AMPLIFIERS

Sec. 10-6]

255

cut

The

Fig. 10-9.

This

pulse.

is

conditions in an idealized class

amplifier,

given by the relation

4 = k[E plm

cos ut

(E p i m

Ed )]

for

ib

>

(10-38)

where, by definition, for the condition of zero current

- =
"ft

ut b

i
1

COS

Ei
^
vim
m Ed

-^O-fe)

plm

Note that the maximum tube current

Ib.max

With the shape

is

given by

kEd

(10-40)

known,

of the current pulse

plate-circuit information.

(io - 39)

The average value

it is

possible to

compute

of the plate-current pulse

is
1

P*.
d(at)

which

is

given by the relation

h =

h/2

fc

T Jo

[E plm cos ut

(E plm

- Ed )]

d(ut)

This integrates to the value

^Epim

sin

(Epi m

Ed)

(10-41)

RADIO ELECTRONICS

256

Similarly, the amplitude of the

[Chap. 10

fundamental component

is

given by

the integral
1
P*
~

Ip\m

if

which

may

ih

cos wt d(ut)

Jo

be written in the form

Ip\m

2k

6b/2

[E plm cos

co<

Jo

(Ep im

Ed)] cos ut d(wt)

This integrates to the value

_2k Eplm
'

(6 h

sin 8 b )

7T

(Eplm

- Ed

sin

(10-42)

with this analysis and obtain exprespower transferred to the load, the plate dissipation in the
tube, the power supplied by the plate power supply, and the plate-circuit
It is quite possible to continue

sions for the

manner analogous to that for the class B amplifier. Hownoted that the construction of Fig. 10-9 is necessary in order to
deduce the operating path before the approximate operating path may be
obtained.
The results will be in error consequently, owing to the approximations.
Moreover, once the construction of Fig. 10-9 is available, a
semigraphical solution may be effected directly without the approximations involved in the foregoing.
Because of this, the above method of
analysis will not be continued, but the semigraphical method will be
efficiency, in a

ever,

it is

discussed in detail.
is called to the fact that, with the class C amplifier, there
be no output for small grid signals, since the plate current is zero
Consequently, the output potential is not proportional to the input potential, and these amplifiers cannot be used where such a linear relation must
be maintained. They are used extensively for amplifying a signal of
They are also used extensively in radio communicafixed amplitude.
This latter
tions as either low-level or high-level modulation stages.
application will be examined in detail in Chap. 12.
When the amplifier
is biased to class B operation, a linear relation between the output and
input potentials does exist and such amplifiers find extensive use in those
The most important applicaapplications requiring this characteristic.
tion is to increase the power level of a modulated carrier wave.
10-7. Semigraphical Analysis of Class C Amplifiers.
Before carrying
out the details of the analysis, attention is called to a second method of
obtaining the operating path of a tuned power amplifier.
This makes
use of the fact that the operating line appears as a straight line on the
constant-current (eb,e c ) characteristics of the tube.
These constantcurrent tube characteristics are available for transmitting-type tubes
and are provided for this particular purpose.

Attention

will

TUNED POWER AMPLIFIERS

Sec. 10-7]

To

verify that the

dynamic

characteristic

stant-current characteristics, use

is

made

is

257

a straight line on the con-

and

of Eqs. (10-13) for the grid

plate potentials, viz.,

E + E gm cos wt
E p m cos cot
cc

(10-43)

is valid when the Q of the tank

combine these expressions by writing

This latter expression

Now

greater.

cc

~ E gm

E
eb

Ebb

E%\ m

Epl m

circuit is 10 or

cos wt
cos at

Adding these expressions gives

ec

This

may

eh

E p\m

_|_

Egm

cc

__|_

be written in the form

ec

Eam
LJ

E p lm
^

eb

+ E0M +
'

which

is

Ebb

Eplm

'

gm

Rplm

J-*

771

(10-44)

Abb

the slope-intercept form of the equation of a straight

line.

The

results are illustrated in Fig. 10-10.

s-

-= const

Plate current (ma)

(O-cut-offbias)

Fig. 10-10.

The operating line on the constant-current curves of

In order to establish the range of operation,

the end points of the region of operation.
specifying Ebb,

a,

ec

m!>JC ,

necessary to specify

Ordinarily this

is

done by

quantities which are determined from con-

economy, power output desired, efficiency, and tube

The manner of this dependence will be investigated below.

siderations
ratings.

eb,

it is

a power tube.

of

RADIO ELECTRONICS

258

With

[Chap. 10

these factors specified, the operating characteristics of the amplifier

are obtained from the curves in the

manner

illustrated in Fig. 10-11.

I Locus of cons fanfe*

Fig. 10-11.

10-8. Grid

The operating

/ee

characteristics of a class

and Plate Currents

amplifier.

Amplifiers. 3

In order to
obtain a numerical solution of the operational features of the amplifier,
such as power output, efficiency, grid driving power, and plate dissipain Class

average and rms values of the grid and plate currents are
These must be deduced from the plate- and grid-current pulses
as obtained from the curves, as discussed above.
It is well to examine this matter before considering a detailed analysis of the amplifier
tion, the

required.

operation.

An inspection of Figs. 10-7 and 10-8 shows that the plate- and gridcurrent pulses possess zero-axis symmetry.
Consequently, these recurring waves may be represented by a Fourier series involving only cosine
may

In particular, the plate- and grid-current pulses

sented analytically by series of the form
terms.

% =

lb

Ic

ic

The average

+
+

Ip\m COS Wt
Iglm COS Wt

+
+

Ip 2m COS 20>t
Ig 2m COS 2ut

or d-c value of the plate current

ib

is

+
+

be repre-

,,v

given by the integral

d(wt)

which becomes, by virtue of the zero-axis symmetry and the fact that
conduction proceeds over the angle d b
,

0b/2
it,

d{oit)

(10-46)

Sec. 10-8]

TUNED POWER AMPLIFIERS

259

This integral expresses the area under the plate-current pulse. Since,
however, an analytic expression for the current pulse is not available,
recourse is had to any of the available methods of numerical integration,
e.g., through the use of a planimeter; by dividing the base of the wave into
equal parts, approximating the

mean

ordinates of the resulting rectangles,

and then summing the areas of these rectangles; or through the use of
other methods devised for numerical integration.

The

details of the second

method

Suppose that Fig. 10-12

is the current waveform,
certain
features of which are to be examined. Suppose that the half
7r
3jr
cot
2
4
recurrence period is divided into n
Fig. 10-12. Current waveform and its
equal parts; hence each division is
approximate representation.
=
180/n deg long. Since the
v/n
current flow will proceed for less than 90 deg in each half period, and
taking account of the symmetry, the integral for I b is then given with good
approximation by the expression
are given.

4(0)

h=

(10-47)
k = 1,2,3,.

where u(kir/n) denotes the value of the current at the angle kv/n.
The average value of the grid current is found in a similar manner from
It is
the graph of the grid-current pulse.

2x Jo
which has the form

= i

(10-48)

T Jo

where 8C denotes the grid-current conduction angle.

approximate calculation, this becomes
Ic

In terms of the

(10-49)
*

1,2.3,...

of the fundamental-harmonic component of the plate

obtained from considerations of the general Fourier series
This leads to the form
representation of the current.

The amplitude

current

is

2T
1

I pim

= -

[
I

T J0

ib

cos

u>t

diost)

RADIO ELECTRONICS

260

which

may

[Chap. 10

be written, in view of the existing symmetry, in the form

Ipim

cos

ib

cot

d(wt)

(10-50)

TV

This integral may be expressed as a summation by the approximate

methods that have been employed above. This becomes
4(0) cos

+
,

lf

fkA
\n)

tb

kir

C0S

(10-51)

^.

The amplitude

fundamental-harmonic component

of the

way

obtained in the same

plate current.
It is given by
current

is

Iglm

as the corresponding

COS

ic

T Jo

of the grid

component

of

d(cot)

Cot

which reduces to the form

Iglm

re ' /2

cos

ic

cot

d(wt)

(10-52)

Jo

In general, the grid current flows for a relatively small portion of the cycle
neighborhood of 8 C = 0. But the value of cos cot does not appreci-

in the

Then approximately

ably differ from unity during this interval.

Iglm

from which

rc/2

IT

JO

ic d(cot)

follows that

it

Iglm

27,

(10-53)

not necessary to plot the grid- and plate-current wavemay be taken directly from the curve of
Fig. 10-11 and combined in a table like Table 10-1 to yield the desired

In general,

it is

forms, since the information

results.

10-9.

Power Considerations in Class C Amplifiers. A number of the

same as those considered in Sec. 10-3 for the class B ampli-

results are the

Here too the d-c power input to the plate circuit, which is equal
power furnished by the plate supply when the d-c power
dissipated in the plate load resistance is negligible, is given by
fier.

to the average

P =
bb

The a-c power output

quency and is given by

if *.L

^ Jo

jr-

Zw Jo

E bbi

of

e i P d{cot)

d(cot)

importance

Pl

2*

= Enh

(10-54)

that at the fundamental fre-

is

2'

f
/

Jo

Ep

m cos

cotI p i m

cos

oit

d(ut)

TUNED POWER AMPLIFIERS

Sec. 10-9]

which

261

is

Pl =
The

plate-circuit efficiency

(10-55)

is

= 15L X 100% =

Vp

The

X 100%

plate-circuit efficiency depends, of course,

E plm

since, for a specified Ebb,

dependence

may

dictated

is

by

eb.mi*.

(10-56)

on the value

of e&.min,

calculation of this

be accomplished, using the results of Sec. 10-6.

TABLE

The

10-1

ANALYSIS OF CLASS B AND CLASS C TUNED AMPLIFIER

Ebb

Tube

6c. max

^h.max
Gc.max
tc.max

it

et

cos 6k

ib{B k )

icW

ib(8h)

length of line
2

eb

PQ
3

cos 8k

cos 0k

h _

n [

f+Z

t't(0)

COS

n
general form of the relationship

is

best presented graphically, as in Fig.

which shows the plate-circuit efficiency vs. the plate-current

conduction angle d b with Epim/Eu, as a parameter. It might be noted
that typical values for class C operation are 8 b in the range 120 to 150 deg,
80
with corresponding plate-circuit efficiencies approximately from t) p
10-13,

to 60 per cent.

RADIO ELECTRONICS

262

The power

dissipated in the plate of the tube

P' =-1
2tt

is

(En,

ebib d(b)t)

Zir
27i

lo

[Chap. 10

given by
e L )ib d(wt)

Jo

which reduces to

Pp = E bh - Epl Ipl =

Pbb

By combining

this

- Pl

(10-57)

with Eq. (10-56), there results

Pp =

(1

(10-58)

Vp)P

This expression shows that the plate dissipation decreases as the output
power increases, for a given plate
95

power input.

The average grid power supplied

by the driving source is given by

90

85

>
9

80

If,
2wJo

e g i c d(wt)

This reduces, under the assumption

is at its maxi-

0.9

that the grid potential

3 75
0.8
-2
a.

mum

value when the grid current

flows

and does not vary appreciably

70

during this interval, to

65

P = Egm ~
g

which

60
80

140
Plate current angle

100

120

Eam I

6^

10-13. Approximate plate-circuit

efficiency for different angles of current
flow. (After A. W. Ladner and C. R. Stoner,

The

Fig.

d(ut)

is

180

160

ic

results of

(10-59)

Thomas have shown

4

that the grid driving power is given

more accurately by the expression

"Short Wave Wireless Communication,"

Sons, Inc., New
chap. 10, John Wiley
York, 1950.)

&

somewhat

P g = E gm I

better approximation

^0.85

0.16 cos

0.9E gm I c
is

given by Maling,

(10-60)

for triodes

(10-61)

I ^0.81

The average

0.20 cos

grid dissipation

is

P =

for tetrodes

and pentodes

given by the expression

f
/

2*
e cl "

rf (

TUNED POWER AMPLIFIERS

Sec. 10-9]

This

may

be written as
Pc

(?cc

iw Jo

= E h + E gm I
ec

But the
battery

cc

(10-62)

term gives a measure of the amount of power that the grid

absorbing from the input driving source, since
cc

^^

first
is

P =
and

263

is

inherently negative.

cc i c

= E

d(f)

(10-63)

Hence the power dissipated

in the grid

circuit is

= Pa

P,

(10-64)

Example. In order to illustrate the calculations for a typical transmitting tube,

consider the following specific problem: A type 806 triode having the constant-

pt

Cc nsfant

ops

>

urrent

06

haracter sties

Tri

7.7

^0.6
)0

8 X>C\J200^16 D0~-

2C

Plate potent

a\

3200

2800

)0
\4

^0.J

'

Plate clirrent,
'es

<?./,

Scale

cm.

Fig. 10-14. Constant-current characteristics of an 806 triode.

current characteristics shown in Fig. 10-14

the following conditions:

E =
et.

2,500 volts

1.0

is

used as a class

Ecc

= -500
=

volts

755 volts

amplifier,

under

RADIO ELECTRONICS

264

[Chap. 10

Determine the following:

a. Power supplied by the plate power supply.
b. A-c power output.
c. Plate-circuit efficiency.
d. Plate dissipation.
e. Grid driving power.
(Note: The details of the solution are given in Table 10-2.)

TABLE 10-2
ANALYSIS OF CLASS C AMPLIFIER

Ebb =
fib.min
6c.

Tube 806
Egm = 755
Ei m = 2,245

E = -500

2,500

cc

10

e b ,min

= 255

max

Icaux

185

ma

deg

cos

8t

COS

1.0

0k

ma

10

20

30

40

50

60

70

80

90

0.985

0.94

27.8 27.4

0.86

26.1

PQ =

24.1

Bb

0.64 0.50 0.34 0.17 0.00

21.3

800

750

640

410

ie(k)

185

170

120

55

12

825

788

710

|555

314

120

cm

27.8

0.76

825

cos 9 k

825

of line

ib(8k)

ib(Ok)

/b.mi

255

n = 18
Length

k
9k,

c.max

17.9

13.9

9.5

4.8

0.0

150

96

h = Vi S a2 H + 2,750) = 176 ma
h = Hs( 18 % + 357) = 25 ma
(

5
= H( 82 A

Pbh =

2,500

2,245

Pl =

=
p =
P =
V

35

X
X

K4

(l

0.9

+ 2,463) =
176

= 440

319 ma
watts

319

watts

X 100% = 81%
X 440 = 83.5 watts
X 755 X 0.025 = 17 watts

0.81)

10-10. Design Considerations for Class

Amplifiers.

The

analysis

based on the assumption that the locus of the operating

point of the tube characteristic is known. Frequently, however, the
engineering design carries with it the requirement for the selection of the
tube and the selection of the operating conditions that govern the locus
presented above

is

A numto give a high plate-circuit efficiency, and other specified results.

ber of factors are important in such a design, and it is desirable to examine
the influence of these.

The important
class

factors that are involved in the engineering design of a

amplifier are the following

TUNED POWER AMPLIFIERS

Sec. 10-10]
1.

The peak space

usually controlled

is

demanded

current that should be

by the values
peak-space-current demand is given by
This

I8,max

I &,max

of e b

~\~

265

mm and

ec

of a given tube.

m , since the total

I cram

y(^6,minj^c,max)

The minimum potential to which the plate falls, e b min

The maximum value of the instantaneous grid potential, e c ,mi.
4. The angle of plate-current flow, 6 b
5. The angle of grid-current flow, 6 C
6. The plate supply potential, E^.
The influence of each of these factors is considered in some detail.
Item 1. In so far as the total space current that may be safely drawn
in a vacuum tube is concerned, it is limited by the allowable emission
from the cathode, if saturation current may be drawn from the tube.
Although it might not be too unreasonable to draw emission saturation
2.

3.

current on the current peaks in a tube that

is provided with a pure-tungunwise to drive a tube with either a thoriated-tungsten

or an oxide-coated cathode to such extremes.
Reasonable figures for
the average emitter are
Tungsten filament
approximately 100 per cent of total emission

sten filament,

it is

Thoriated-tungsten

current.

s,

m , x from 15 to 35 per cent of the total emission

current.

Oxide-coated cathode

from 10 to 20 per cent

I,, m **

of the total emis-

sion current.

Items 2 and 3. The optimum values of ef,, m i and e c ma x will be such that
Moreover,
the total allowable peak space current will not be exceeded.
,

must be so chosen that the maximum plate current

This requires that the tube must not be driven so hard
that it operates in the region of rapidly falling plate current.
Such a
condition is avoided by keeping e b mlu > e c max
However, high platetheir relative values

occurs at wt

0.

circuit efficiency results

power

it

is

when

required that

Item

4.

9b

ec

m .

although for low grid driving

Typical values of the ratio

to 2.

The range over which

duction angle

e c mai
,

>

e b mi

e&.min/ecm.* usually range from

e b ,mm

plate conduction occurs,

influences both the average current I b

harmonic current amplitude Iv \ m

current amplitude,

it is

For a large value

desirable that d b be made
.

i.e.,

the con-

and the

first-

of the first-harmonic
large.

However,

in

order to provide a high value of plate-circuit efficiency, small values of 6 b

are indicated.
plate efficiency

Consequently,

it

and power output.

is

necessary to compromise between

Typical values for class

operation,

RADIO ELECTRONICS

266

[Chap. 10

as already discussed, are 8 b in the range from 120 to 150 deg, with corresponding plate efficiencies tj from about 80 to 60 per cent (see Fig. 10-13).

With the choice of I s e 6 mill e c ma x, and 8 b specified, the other operating

It is desired, therefore, to examine the relaconditions are established.
tion that expresses the grid bias, E cc and also the grid conduction angle
,

To find an expression for E cc it is

8C
noted that the plate current becomes zero when wt = 8 b/2. At this
point, the grid signal is given by
,

in terms of the fixed parameters.

= E gm

eg

cos

ost

= E gm

cos

(10-65)

2"

This follows from

e b /n = 0.
at this point it is necessary that e c
the fact that the plate current may be written by an expression of the
e b /n must be zero.
By
e b /fi) and, for i b to be zero, e c
form ib = f(e c

But

virtue of this

Egm
But

8
^+E +

cos

cc

Ehb - Eplm

cos

8 \
^ =
)

since

Eg m E cc
= E bb E v
~\~

@c,w&x

eb
it

follows that
(e c max
,

cc )

cos

db
^ +E +
cc

1
-

E
bb

(Ebb ~~

eb,mia)

cos

from which

~\~

cc

angle of grid flow

becomes zero when

from which

it

wt

f~

cos (8 b /2)
cos
(0 6 /2) /
J

The

Cc, max

ft

readily determined, since the grid current

is

8 c /2.

ec

= E gm

At

this point

cos

+E =
cc

follows that

cos=-!^
where

cc

is

(10-66)
1

(10-67)

obtained from Eq. (10-66).

10-11. Grid Bias.

ever, the bias potential

bination in the grid

foregoing mathematical discussion assumed

The
is

Often, howcc was constant in magnitude.

obtained by means of a resistor-capacitor com-

that the grid bias potential

line, in

choice of grid resistance

R g is

the average grid current I c

the

manner illustrated in Fig. 10-15. The

by the required bias potential and

dictated

This

is

frequently referred to as grid-leak

bias.

It

might be thought that I c would be a

definite value for a given

peak

TUNED POWER AMPLIFIERS

Sec. 10-12]

driving potential

E gm

with the result that the grid resistance would be

As a

firmly established.

R g are accompanied
Ic with the result that, for fixed Egm

practical fact, variations of

by an almost inverse variation

267

of

remains sensibly constant. It is desirable, therefore,

that the largest R g possible be used, with stable amplifier operation.
This follows from the fact that the loss in the grid resistor is due to the
But for a given negative bias, I c R g is constant,
heating, or c R g loss.
Consequently, by increasing R g I c is
and Ic varies inversely with R g
reduced, and the corresponding loss is
the potential

cc

reduced.

The

grid driving

power

of the order of 5 to 10

is

usually

per cent of the

the amplifier P L)
operated within its
designed frequency limits. When operated above the normal frequency
a-c

power output

when the tube

of

is

limits of the tube,

the grid driving

power increases rapidly, owing in some

measure to increased dielectric losses,

^
am

Fig. 10-15.

g rid

P lifier

The use of a grid resistor

capacitor for biasing the

but principally because of transit-time loading. This latter factor is disA limit is thereby set to the h-f limit

cussed at some length in Sec. 5-8.

of the tube.

and Amplifier

has been noted on

p depends upon the
Moreover, the plate-current conducplate-current conduction angle 8 b
tion angle depends upon the grid bias and the magnitude of the grid
driving potential, more negative values of E cc and higher E gm being accom10-12. Grid Potential

Linearity.

several occasions that the plate-circuit efficiency

It

-q

The general character of the variation of

panied by smaller values of 6 b
output current and plate-circuit efficiency as a function of input grid
.

potential

is

shown

in Fig. 10-16.

These curves show that the a-c component of current Ip i, and the
efficiency -q p increase with increasing values of E g over a wide range of E g
A saturation value is reached beyond which there is no essential change,
except that the grid current, and so the grid driving power, continue to
increase.
An interesting fact is that the situation remains roughly the
same whether fixed bias or grid-leak bias is used. With grid-leak bias,
however, the input power rises to larger values than with fixed bias.
This is so because an increased E g tends to result in a higher I c but this
in turn causes an increase in E cc
Hence, for a given output power, a
larger E gm is required, with a corresponding less linear relationship
between Eg and I pl
Clearly, overdriving the amplifier merely results in
Underdriving leads to a
a high power dissipation in the grid circuit.
reduced amplifier output and efficiency.
.

RADIO ELECTRONICS

268

[Chap. 10

The question of a linear relation between Ip i and Eg is of considerable

importance when the tube is used for grid-circuit amplitude modulation.
This matter will be discussed in Chap. 12.
160-r

Grid driving po+en+ial,

Eg~*~

Fig. 10-16. The effect of varying the input grid potential on several of the important
amplifier factors.

The semigraphical method

10-13. Approximate Analytic Solution. 6

discussed in the foregoing sections permits a relatively accurate analysis

This analysis requires that

of the performance of a class C amplifier.
the instantaneous plate and grid potentials and currents should be
obtained and plotted and then from these curves should be derived such
loss, the power output, and the grid driving

information as the plate

power.

This method, while

it

possesses the virtue of yielding a relatively

accurate solution, does have the disadvantage of being fairly laborious.

An approximate calculation of the performance can be obtained without
recourse to the point-by-point analysis.

This approximate calculation

takes advantage of the fact that the total space current can be expressed
quite accurately by an expression of the form

is

= %

The constant a
a

is

in Figs.

ic

k (e e

+-

for e c

+^>

(10-68)

1 to %.
The value of
889A triode, the characteristics of which are
2-18 and 2-19. The accuracy of the results that follow

found to be

shown

ordinarily lies in the range from

1.0 for the

under this approximation will probably be well within the accuracy with
which tube circuit conditions are known and within the reasonable variations of individual tubes from the average of the group.
Evidently,
of current

when the

current

is

specified

by

this relationship, the pulses

have a form that may be analytically expressed.

However,

TUNED POWER AMPLIFIERS

Sec. 10-13]

269

the current pulses so specified are for the space current,

plate current.

and not for the

Consequently the method yields information regarding

By making reasonable assumptions, informaobtained regarding the grid-current pulses. The difference
between these two must then be the plate current. Since the grid current
is ordinarily a small fraction of the total space current, then an error in
the choice of the grid-current pulse will not introduce a very large error
the space-current pulses.
tion is also

in the resulting plate current.

To examine

the situation in some detail, use

tions for the grid

and plate potentials

made

is

of the

Eqs. (10-14)],

[see

known

rela-

viz.,

= E cc + E gm cos cot
" E.
= Em
^iplm COS

ec
e&

Ctlt

with

Egm

Now

Ebb

Ct

^6,min

i.e., within the limits from 6b/2

be written in the form

within the limits of current flow,

may

to 8b/2, the space current

(e c ,mx

which

is

i3

c,max

Eplm

(e e

cos ut

cc )

m + ^)

But the maximum value

cos wt

E^
+E +

Ebb

^&,min

cc

+ (e +
cc

(1

cos ut)

cos

cot

which occurs when cos

of this expression,

(10- 69)

u>t

1,

is

= hi e

4-

e6 mi

Hence the

ratio of the current at

'

(10-70)

any instant to the maximum value

is

given by

But the value

E +
cc

cos wt

Ebb/ m

(1

COS

(10-71)

cot)

provide for a conduction angle 8 b is

combining Eqs. (10-66) and (10-71), the cur-

of the bias potential to

By

given by Eq. (10-66).

rent ratio becomes

'

cos

cot

+
,

cos

(0,,/2)
, n /n ;

cos (e b /2)

(1

cos ut)

which reduces to the form

is

I,,

cos (0 b /2) cos

cos (d b /2) - 1

cot

(10-72)

RADIO ELECTRONICS

270
This expression

may be

[Chap. 10

used to provide the fractional average or d-c

component of the space current and any of the fractional harmoniccomponent values. In particular, the fractional average or d-c component of the space current is given by the expression
d(wt)
2tt

jo

which reduces to the form

Bb/2

t Jo

cos (6b/2) cos wt

cos (d h /2)

d{ut)

(10-73)

In a similar way, the fundamental component has the form

C2ir

1
T

which

may

cos ut

d{oit)

be written as
o

T
-tslm

Is.nmx

fDb/2
/

T J

cos {0b/2) cos ut

cos {6 b /2) - 1

cos ut

d{o>t)

(10-74)

These expressions, for given values of a, are functions of 6b and they may
be conveniently expressed in graphical form. The curves of Fig. 10-17
,

Fig. 10-17. Curves giving the relation of d-c and peak fundamental-frequency component of current as a function of the angle of current flow and the peak amplitude.

give the relation of the d-c and fundamental-frequency components of the

space-current pulse as a function of the angle of flow

amplitude

/,,.*,

with a as a parameter.

db

and the peak

TUNED POWER AMPLIFIERS

Sec. 10-13]

To

271

find the corresponding values of the plate current, the grid current

approximated by assuming an analytic form for the equation of the

As the grid current is usually a small part of the total
space current, a reasonable choice for the grid current will provide good
is

grid current.

An

results for the plate current.

expression that represents the grid

current with good approximation is*

ic

k'

(e c

>

for e c

(10-75)

Consequently, by following a parallel development to that above, there

results

cos (8 c /2)

cos

cos (0 c /2)

Ic.

u>t

(10-76)
1

from which the d-c and fundamental-frequency components become

cos (0 c /2) cos ut
cos (6 c /2) - 1

e c /2

1
TT

d(wt)

(10-77)

cos ut d{ut)

(10-78)

and
c/2

cos (8 c /2)

cos (8 c /2)

10

cos wt

These expressions are functions of 6 C and have the same graphical form
as the space-current components.
Consequently the curves of Fig. 10-17
are also a valid representation of the grid-current components.
Example. To illustrate the methods of Sec. 10-13, the example in Sec. 10-9 is
repeated according to the methods of Sec. 10-13, and the corresponding results will
be compared.

Peak space current: /, = Ib ,mx

To

find the grid current, choose

185

1,010

ma

then
48

96

Fig. 10-17

Hence the currents

of

0.14

hi.

=
=

26
48

=
=

0.14
0.26

X
X

185
185

find the plate current, the space current

Maling

suitable for

0.26

importance are

I cXm

825

755

To

500

and from

finds that the

exponent 2

is

must be

suitable for

most tetrodes and pentodes.

ma
ma

most

calculated.

triodes,

To do

and a value

1.4

this,

seems

RADIO ELECTRONICS

272
choose a

[Chap. 10

%, and
378

Oj,

60

COS"

6b

755

120

Then, from Fig. 10-17,

=
from which

The

it

8,

0.35

max

follows that
/,

I.im

0.19
0.35

X
X

1,010
1,010

=
=

353

ma
ma

166

192

plate-current components are then

h=
Ipi m

The

/.i

0.19

max

s,

I,

353

I c = 192 - 26
48 = 305 ma

ma

significant quantities are

Pn =
PL =
P =
Pg =
i?p =

X 10~ 3 = 415 watts

3
10(2,500 - 255) = 342 watts
2 x 305 X
415 - 342 = 73 watts
755 X 0.26 X 0.9 = 17.6 watts
84 Mis X 100 = 82.3%

2,500

166

more accurate calculation would require a logarithmic plot of Eq. (10-68) for
the particular tube and a determination from this of the exponent a. However,
the use of the approximate methods would not be justified under these circumstances in general since the effort involved would be comparable with that in
applying the methods of Sec. 10-8.
10-14. Push-Pull

Tuned

illustrated in Fig. 10-18.

The push-pull tuned amplifier is

be observed that this circuit possesses

Amplifiers.
It will

C-

Fig. 10-18. Push-pull tuned amplifier.

the expected form, with the tuned circuits in the input and output. The
and grid tanks are excited by current pulses which have the wave-

plate

forms illustrated in Fig.

10-8, except that current pulses occur

each half

cycle instead of the alternate half cycle shown, alternate pulses being

inverted.

TUNED POWER AMPLIFIERS

Sec. 10-15]

The main

273

between single-tube and push-pull operation are

assume that each tube, relative to the mid-point
or d-c level, has its associated L and C tank circuit.
When connected in
push-pull, the tank elements are then essentially in series, with the total
L being doubled (assuming no magnetic coupling exists between these
coils), and the total C being halved.
This reasoning is not generally
correct, since mutual coupling between the coils does exist.
But it does
make plausible the fact that the tank circuit Q is roughly doubled. The
circuit might involve some readjustment if a given Q is desired, but this
is

differences

For

of interest.

this,

generally easy to achieve.

Relative to the harmonic content, the situation here is no different

from that which exists in the untuned push-pull amplifier. The correctly
adjusted push-pull amplifier automatically eliminates all even harmonics,
and the second harmonic, which would normally be the most troublesome,
disappears.
This condition is true quite independently of the value of
The third-harmonic component is still present. If the analysis of
Q.
Sec. 10-9 or 10-13 is extended to an evaluation of the third-harmonic
components, it is found that, for 6b ~ 120 deg and Q L = 10, the ratio
of third harmonic to fundamental is less than 0.01.
In fact, with a
push-pull circuit, satisfactory operation with Q L as low as 5 is feasible.
10-15. Neutralization.
When a triode is used as a power amplifier in
the circuit of Fig. 10-1, some feedback between the grid circuit and the
plate circuit exists through the interelectrode capacitance C gp
This feedback is usually positive and tends to cause the stage to oscillate. It may
be balanced out by one of several methods of neutralization. Tetrodes
and pentodes usually do not require neutralization, as the capacitance
.

Cgp is

amount of feedback
very simple and consists
in providing feedback through external circuits in an amount equal to that
through Cgp but in opposite phase.
is

usually small enough in these tubes so that the

The

very small.

principle of neutralization

is

Two common methods

The

of neutralization are illustrated in Fig. 10-19.

and the second is called plate

In Fig. 10-19a the plate is connected through the adjustable neutralizing capacitance Cn to a point in the grid circuit which has a
potential of phase opposite to that of the grid.
In Fig. 10-19& the grid is
connected to a point in the plate circuit at which the potential is of opposite phase to that of the plate.
In these circuits the feedback through
first

is

called grid neutralization,

neutralization.

Cgp has been neutralized, although a degree of dissymmetry relative to

ground exists. This fact is important, since Cgp is not the only capacitance through which feedback may occur, and these methods of neutralization will not generally overcome such other capacitance effects.
A symmetrical circuit with ground as datum provides the best possible
arrangement for complete neutralization, since it is possible to balance

RADIO ELECTRONICS

274

[Chap. 10

capacitance effects to ground. Such a neutralizing method was developed by C. S. Franklin and has the form shown in Fig. 10-20.
The diagrams of Fig. 10-21 show the neutralized push-pull tuned
amplifier, together with the balancing circuit.
Observe that the balancing circuit is symmetrical with respect to ground.

Grid

Plate

Plate

Grid

(a) Grid neutralization

Fig. 10-19.

10-16.

(b) Plate neutralization

Diagrammatic and schematic sketches

Tubes as Tuned Class C

compare the triode as a tuned

of grid

Amplifiers.

class

and plate neutralization.

It is of

some

interest to

amplifier with the tetrode and

According to Eq. (10-61) the grid driving power of the triode

higher than that of the tetrode and pentode under the same conditions

pentode.
is

Fig. 10-20.

The Franklin symmetrical

of grid driving potential,

neutralization method.

average current, and grid conduction angle.

However, such a comparison

is

inconclusive in

itself,

since ordinarily

these factors are not the same.

Consider
factor

fi

first,

therefore, the influence of changing the amplification

of the tube.

This comparison

reference to the operation of

two

is

best accomplished

by

direct

similar triodes with different values of n.

TUNED POWER AMPLIFIERS

Sec. 10-16]

275

The actual operating conditions for two groups of similar triodes of different n are contained in Table 10-3.
It might appear initially that the low-ju tube, owing to the lower
cutoff value, and so a lower operating bias, might be accompanied by a

P2
Fig. 10-21.

The

neutralized push-pull amplifier, and the neutralizing circuit.

low excitation power.

higher

e c m>1
,

As a

But the low-^

triode requires a larger

result of the higher e c max

slight decrease in the

conduction angle

dc

and the lower

When

all

E gm

and a

there

is

TABLE 10-3
COMPARISON OF LOW- AND HIGH- M POWER TRIODES*
250TH

304TL

3,000

37
3,000

2,000

20
2,000

-350

-150

-300

-200

335

333
90
395
32

600
85
480
36

600

250TL
14

D-c grid potential, volts

D-c grid current, ma
Peak r-f grid input potential

45

720
29

Grid dissipation, watts

Plate power input, watts
Plate dissipation, watts

Courtesy of Eitel-McCullough,

Inc.,

12

15

19

11

1,000
250
750

1,000
250
750

1,200
300
900

San Bruno,

Calif.

factors are taken

304TH

125
325
39

1,200
300
900

RADIO ELECTRONICS

276

into account, the average grid current /

Although

lower for the lower-^ tube.

is

Egm is higher for the low-p tube,

current, the grid driving

low-/*

[Chap. 10

and the

power is
As a

high-/* tubes.

owing to the lower average grid

approximately the same for both the

result, therefore, triodes of the high-/*

type are to be preferred, since lower excitation potentials are required.

In fact, the tendency in the design of transmitting triodes is toward the
higher values of

Consider

now

p.

a tetrode as a tuned class

by the screen

screening of the plate

amplifier.

Owing

to the

grid, the control-grid current is

not

dependent on the changing plate potential e b as it is in the triode. As a

consequence, full plate potential may be used without excessive excitation
power; that is, e c m ai may exceed e&, m i without excessive grid driving
power. This applies also for the pentode and represents one of the
advantages of the tetrode and pentode over the triode.
A second advantage of the tetrode and pentode results from the smaller
grid-plate capacitance than in the triode, with the great advantage that
neutralization is unnecessary.
The double screening effect of the screen
and suppressor grids gives the pentode preference over the tetrode in most
,

cases.

10-17. Parasitic Oscillations.

some neutralization scheme

will

As discussed in Sec. 10-15, the use of

prevent a power amplifier from oscillat-

feedback through the grid-plate capacitance is

While this overcomes the tendency of the amplifier to
oscillate at the fundamental frequency, the possibility remains for the
system to oscillate at a variety of frequencies, the tank circuits comprising
wiring or fixed inductances, with stray or wiring capacitances. These
frequencies bear no relation to the fundamental or any of the harmonics
Parasitic oscillaof the amplifier and are known as parasitic oscillations.
tions are more likely to occur in amplifiers which incorporate tubes in
parallel, and in such cases it is common to include "suppressor" resistors
ing, since the effect of the

largely canceled.

in the plate leads (and also in the screen leads,

may be suspected in an
Among these are a
conditions.

Parasitic oscillations
of different

if

pentodes are used).

amplifier under a

number

high plate current, grid

currents which appear unusual, general instability, and peculiar transient

behavior.

If

the parasitic oscillation occurs in the a-f range, or if these

an audio rate, transformers and iron-core

oscillations cause blocking at

chokes

may

"sing."

Of course, in this latter case, if a receiver is tuned

any audio variation will appear as an audio

to the parasitic frequency,

signal.

The suppression of parasitic oscillations, whether these be of a low

frequency or of a very high frequency, is often a difficult problem, since
it is not easy to ascertain the circuit elements which cause the parasitic
tank circuit. Often a rearrangement of circuit elements will eliminate

Sec. 10-18]

TUNED POWER AMPLIFIERS

the oscillations.

Often appropriately connected resistors or shunting

277

capacitors will be sufficient.

10-18. Grounded-grid Amplifier.

grounded-grid

circuit,

Fig. 10-22.

in the

manner

triode

may

be connected in a

illustrated in Fig.

10-22.

It is

grounded-grid tuned power amplifier.

observed that the grounded grid forms a screen between the input and the
output circuits, quite like the screen grid in a tetrode or a pentode. As a
result, the tendency to oscillate is considerably reduced.
If oscillations

do occur, the necessary neutralizing capacitors are considerably smaller

than in the conventional grounded-cathode circuit.

The curves

of Fig.

10-23

show

the approximate waveforms which

appear in a class B grounded-grid

tuned amplifier. Note that the

t -*

input excitation potential and the

output plate potential are in phase,

although, as usual, the grid poten-

and the plate potential are 180

Observe also
of phase.
that the input and output circuits
are in series, and therefore the excitation source supplies some power
to the output.
As a result, the exciter must have a somewhat higher
power capacity than that required
for the more conventional circuit.
Some of this added power is transtial

deg out

Fig. 10-23. The important waveforms in

the grounded-grid tuned class B amplifier.

ferred to the plate circuit.

When a grounded-grid circuit is to be modulated, it is necessary to

modulate both the exciter and the output circuit. This follows because
some of the exciter power appears in the output.

RADIO ELECTRONICS

278

10-19. Frequency Multiplication.

There

is

[Chap. 10

frequently need for a sys-

tem to provide frequency multiplication, in order to furnish a frequency

which is some multiple of a given driving frequency. Such a system as
this often permits high frequencies with less difficulty

to generate

and amplify the high frequencies

larly the case

when

than in attempting
This is particu-

directly.

crystal-controlled systems are used, since crystals

become extremely thin and consequently are

both mechanically and electrically. By using a crystal of low
frequency and then employing frequency multipliers, effective crystalcontrolled h-f systems are possible.
The use of tuned amplifiers for frequency multiplication follows from
the fact that the current pulses in the class C amplifier, with a sinusoidal
for the high frequencies
fragile

Fig. 10-24.

The waveshapes at important points in a tuned class C frequency multiplier.

driving potential, are nonsinusoidal in character and so contain harmonic

components of the fundamental frequency. This fact was made use of in

evaluating the output characteristics of the tuned amplifier [see Eq.
In addition to the fundamental
(10-50)] and was illustrated in Fig. 10-8.
component, which is of importance in the tuned amplifier, the second-,
third-, or higher-harmonic components are of importance in frequency
multipliers.
In this latter case, the output tank is tuned to the desiredharmonic component, depending on the order of frequency multiplication
desired.

The output from a given amplifier at the various harmonics will depend
on the bias used on the amplifier. The higher the harmonic required, the
greater must be the negative bias, and correspondingly the higher must
be the grid drive that is used. For best results, the tank circuit is
energized for one-half cycle of the harmonic output for each primary cycle.

TUNED POWER AMPLIFIERS

Sec. 10-19]

279

Thus

if a third harmonic is required, the output circuit is energized once

every third half cycle, and the bias should be so adjusted that the plate-

current conduction angle

somewhat as

nearly 36

is

illustrated in Fig. 10-24.

60 deg.

The

situation

is

Clearly, of course, the higher the

harmonic, the narrower will be the tube current pulse. Since the available
drive is usually limited, this means that, with increasing harmonic multiplication, the smaller will be the output.
Because of this, frequency
multiplication higher than the fourth harmonic with a single stage is
usually impractical.

Higher values

of multiplication will use

two

or

more

stages in cascade.

Harmonic amplifiers may incorporate triodes, tetrodes, or pentodes,

although triodes are more often used in the higher-power stages.
When
triodes are used, neutralization

Frequency multiplication
vibrator circuits, as

it is

is

is

necessary.

very

common through

the use of multi-

possible to synchronize the input of one multi-

Such multivibrator

vibrator with the output of the previous stage.

permit stable multiplication as high as 9 or 10. If a sinusoidal

output is desired from a multivibrator multiplying chain, it is necessary
only to incorporate a tuned circuit in the output to act as a filter and
circuits

thereby extract the desired sinusoidal component.

For the low-power

applications, the use of multivibrator multiplying chains has largely

supplanted the tuned-circuit frequency multipliers.

REFERENCES
1.

2.

3.

4.
5.
6.

W. L., "Communication Engineering," 2d ed., pp. 582-590,

McGraw-Hill Book Company, Inc., New York, 1937.
Everitt, W. L., "Communication Engineering," 2d ed., pp. 565-594,
McGraw-Hill Book Company, Inc., New York, 1937.
Massachusetts Institute of Technology Staff, "Applied Electronics," chap.
X, John Wiley & Sons, Inc., New York, 1943.
Cruft Laboratory, War Training Staff, "Electronic Circuits and Tubes,"
chap. XIV, McGraw-Hill Book Company, Inc., New York, 1947.
Mouromtseff, I. E., and H. N. Kozanowski, Proc. IRE, 23, 752 (1935).
Thomas, H. P., Proc. IRE, 21, 1134 (1933).
Maling, H. F., private communication.
Terman, F. E., and J. H. Ferns, Proc. IRE, 22, 359 (1934).
Terman, F. E., and W. C. Roake, Proc. IRE, 24, 620 (1936).
Everitt,

PROBLEMS
10-1.

type 800 tube

operating curves for

is

Em =

power amplifier. Find the

and the following three conditions of grid

to be used in a tuned

1,000

characteristics:

E = -55
cc

volts

= -95
= -135

Egm =170
=
=

210
250

volts

peak

RADIO ELECTRONICS

280

Would any

[Chap. 10

of the indicated operating conditions yield class

10-2. The type 800 tube is operated as an

conditions under the following conditions:

r-f

power amplifier under

D-c plate potential

D-c grid potential
Peak r-f grid potential

class

000 volts

55

volts

170 volts
15

5 700

Calculate the following for

a. Power output.
6. Plate dissipation.
c.

Choose

operation?

et,,

mia

1.5e c

ohms

m i:

Plate-circuit efficiency.

The impedance of the tuned circuit at resonance.

10-3. A type 833A triode has the following maximum

d.

a class

ratings for operation as

r-f amplifier.

4,000 volts

D-c plate potential

D-c grid potential
Peak r-f grid potential

120

Plate dissipation

35
15,800 M mhos

n
gm
It

is

of 16.0

volts

180 volts
400 watts

planned to use this tube as a class B amplifier with a grid-signal frequency

Mc, a plate supply of 4,000 volts, and a value of e&, mi /e c mal equal to 2.0.
.

Find the following:

a. A-c power output.
b. Current by the plate power supply.
c.

d.
e.

Impedance

of

tuned

circuit at resonance.

L and C in the tuned circuit if the loaded resonant Q is 12.

g. What value of R should be used if the maximum allowable plate dissipation
400 watts and
m a = e ,max?
10-4. A class C amplifier uses an 851 tube and operates under the following
/.

is

Plate dissipation.
Plate efficiency.

The values

of

ei,,

conditions:

D-c plate potential

D-c grid potential
Peak r-f grid potential

2,500 volts

250

Plate dissipation

Shunt resistance
D-c grid current

of

tank

circuit

0. 10

Calculate the following:

Output power.

a.

A-c plate potential.

b.

c.

D-c plate current.

d.

Plate-circuit efficiency.

e.

Grid driving power.

/.

Grid dissipation

Assume that

eb.min

volts

450 volts
550 watts
550 ohms

ec

m ,x.

amp

TUNED POWER AMPLIFIERS

A

10-5.

class

amplifier uses

281

an 852 tube and operates under the following

conditions

D-c plate potential

D-c grid bias
Peak r-f signal
D-c plate current
D-c grid current
Fundamental component

3,000 volts

600

15
120

of plate current

Calculate the following, assuming

a.

volts

850 volts
85 ma

e(,,

mn
i

ma
ma

peak

e c .m*x'-

Output power.

b.

Plate-circuit efficiency.

c.

Grid driving power.

Specify the elements of the plate

d. The amplifier is to operate at 1,500 kc.
tank.
Choose Q = 23.
10-6. The typical operating conditions for the type 893 A-R transmitting
triode when used as a class C r-f power amplifier or oscillator are

D-c plate potential

D-c grid bias
Peak r-f grid signal
D-c plate current
D-c grid current
Grid driving power
Fundamental component

Minimum

18 000 volts
,

1,000

volts

1,630 volts
3.6 amp
.21

of plate current. ...

value of plate potential

et.min

amp

approx.

340 watts approx.

6.25 amp peak
1,000 volts

Calculate the following:

Power output.
b. The inductance required in the tank circuit.
Assume that the effective
Q of the tank circuit = 5, resonant frequency = 1 Mc.
c. The required capacitance to tune to 1 Mc.
d. The circulating current in the tank circuit.
e. The grid driving power.
Compare this result with that shown in the
tabulation above. Explain any discrepancy.
/. The power input to the plate circuit.
a.

g.

The

10-7.

plate-circuit efficiency.

type 851 triode

is

used as a class

amplifier.

The operating

tions are to be

D-c plate potential

D-c grid potential
Peak r-f grid potential

2,500 volts

250
450

Ratio ei.min/ecm,!

Determine the following:

D-c plate supply power.

a.
c.

Plate-circuit efficiency.

b.

A-c output power,

d.

Plate dissipation.

Grid driving power.

10-8. Repeat Prob. 10-7 using the method of Sec. 10-13.
e.

volts

condi-

RADIO ELECTRONICS

282
10-9.

class

amplifier

is

[Chap. 10

operated under the following conditions:

D-c plate potential

D-c grid bias
Peak r-f grid

3 000 volts
,

200

volts

360 volts
2

fih.min/^c.max

2.2 amp
120 deg

Peak space current

Conduction angle
Ratio

D-c grid current to peak space current

D-c plate current to peak space current
Peak plate a-c current to peak space current
Frequency
Loaded Q

0. 15

0.21
0.37
2
12

Mc

Calculate the following:

a.

D-c plate current.

b.

Plate-circuit

power input.

Plate dissipation.
d. Plate efficiency.
e. Grid driving power.
/. Load impedance.
y. Tank-circuit capacitance.
g. Tank-circuit inductance.
Plot the appropriate
10-10. Refer to the illustrative example in Sec. 10-9.
data to determine the idealized class C amplifier characteristic illustrated in
Fig. 10-9.
Compare the values of
and I p i m deduced from Eqs. (10-41) and
c.

(10-42) with those given in Table 10-2.

10-11. Show the general character of the construction (like Fig. 10-7) for
determining the operating features of a class C frequency tripler. In this circuit
the output tank is tuned to a frequency that is three times the frequency of the
grid driving source.
What can be said about the plate conduction angle?

CHAPTER

11

OSCILLATORS
11-1. Introduction.

ined in Chap.
in

an amplifier

The

It is there

7.
if

feedback in amplifiers were exampositive feedback is to be avoided

to be achieved.
On the other hand, if the

effects of

shown that

stability is

provided with a sufficient amount of regenerative feedback, the

vacuum-tube circuit will serve as a generator of periodically varying
waves. This output may be sinusoidal, with a high degree of purity of
waveform; it may be an essentially square wave, hence being of high
harmonic content; or it may be a periodically recurring nonsinusoidal
waveshape.
circuit is

large variety of feedback circuits

which

differ considerably in detail

are available for the production of self-sustained oscillations.

possess certain features which are

common

to

all.

These

In each case a circuit

through which is fed back into the input circuit a certain fraction
output and in such a phase and of such an amplitude that selfexcitation results.
In the usual class of tube, the feedback from the
output to the input circuit is accomplished externally to the tube itself
exists

of the

by means

of coupling networks.
In certain special types of tubes, e.g.,
a klystron of the reflex type and a magnetron of the running-wave type,
the feedback is accomplished through the electron beam itself.
Never-

even these can be represented by equivalent circuits which have

common to all feedback oscillators.
Conventional self-excited oscillators ordinarily operate as class C
devices, although class A oscillators are possible and will be discussed
theless

features that are

below.

It is desirable to consider the class

oscillator first.

It is

important to keep clearly in mind the fact that the theory of class C
oscillators is of necessity only an approximate one owing to the nonlinear
character of the region of operation of such devices.
Therefore, this
theory, while contributing materially to an understanding of the operation of the device, must be recognized as a limited solution of the oscillator
problem. But by supplementing the theory with practical design and
operating data a generally satisfactory understanding is possible.
Oscillators may assume a variety of forms, depending upon the application.
The application will dictate the frequency range, the necessary
283

RADIO ELECTRONICS

284
frequency
trols,

stability,

[Chap. 11

the output power, the character of the tuning con-

For example, relatively

and other items.

1-f

power

oscillators are

often used for induction heating, and such units in excess of 50

available commercially.

The question

of

frequency stability

kw

is of

are

rela-

minor importance in this application. Similarly, the higher frequency dielectric heaters are also relatively simple power oscillators, with
little real need for frequency stability.
The oscillator which is part of the superheterodyne receiver has a
Such an oscillator must be simple in design
different set of requirements.
but at the same time must be quite stable. It must also be of low power
and small size and must be capable of tuning over a relatively wide frequency range.
The oscillator which is to be amplified to serve as the carrier of a
radio transmitter must be extremely stable, of low power, and of a single
tively

fixed frequency.

In this case temperature-controlled crystal oscillators

and amplifiers to set the desired frequency and

are used, with multipliers

power

level.

11-2. Conditions for Self -excitation. 1

must be

fulfilled for oscillations

refer to Fig. 11-1.

To ascertain the conditions that

to be sustained in a vacuum-tube circuit,

In Fig. 11-la, the circuit

is

supposed to be open at

-oA-

Coupling

Coupling
network

network

(a) Block diagram

Fig. 11-1.

A vacuum-tube

point A, as shown.

circuit

(b) Equivalent circuit

with coupling between the output and input

Suppose that a potential

E9

is

impressed on the

grid of the tube, without regard to the source of this potential.

output current
circuit.

Now

circuit.

The

given by g m E g for an assumed linear operation of the

because of the current I to the input of the coupling netis

work, a certain potential will appear across the output terminals of this
coupling network. If this return potential is equal in magnitude and
phase to the original potential E, the circuit may be connected at point A

and the system

will continue to operate.

the basis of this discussion the criterion for oscillation may be
given in terms of the total gain, both amplitude and phase, around the

On

loop consisting of the input through the amplifier, and then through the
coupling network.
Oscillations will occur if this total loop gain is unity

OSCILLATORS

Sec. 11-2]

and the phase change through the loop

multiple of

is

285
either zero or

an

integral

2ir.

To examine

first the
network, which is denoted by Zt, and
which is defined as the ratio of the output potential to the input current.
From the diagram of Fig. 11-16, this is seen to be Z T = E/I. Consequently, for oscillations to occur, it is necessary that

transfer

the criterion for oscillation analytically, consider

impedance

of the coupling

or
gmZ T

But

= -E,

(v g

since the output potential of the

to the input potential

by the

E2

and

so,

vacuum-tube amplifier

is

related

gain, then

= KE

fl

combining with the above,

Therefore, for oscillations to be sustained,

it is

necessary that

(11-1)

It

should be noted that both the transfer impedance Z T and the driving-

point impedance of the coupling network

Z (denned

are
2 /I)
involved in this expression, the latter appearing in the expression for the
Equation (11-1) is one form of the Barkhausen criterion for
gain.
as

oscillations.

parallel

expression for the conditions for sustained oscillations

involving only the impedance

potential

2 is

is

readily possible.

Thus noting that the

given by IZ,

(gm E g

+ ~)z = -E2

This becomes

rt +^)z=

KE

which reduces to

9mZ

(k +

^)

= ~1

(11 - 2)

RADIO ELECTRONICS

286
This

may

also be written in the

h.
This

is

[Chap. 11

form

-k

; =

another form of the Barkhausen criterion for oscillations.

should be noted that the ratio E/E 2 is a measure of the fraction

of the output potential that is fed back into the input circuit through
the coupling network. This ratio is denoted as (J, and
It

+- =

(11-4)

-ff

Note, by combining Eqs. (11-3) and (11-4), that the feedback ratio of the
is related to the gain of the amplifier circuit by the simple relation

network

?K =
which shows that the total loop gain
was discussed in Sec. 7-5.

is

(11-5)

unity.

This condition for

oscilla-

tion

The condition K(J = 1 is the minimum case for oscillation and provides
no margin for changes which might occur either in the tube or in the circuit components of the oscillator.
Ordinarily practical considerations
dictate that K(5 be slightly greater than unity in order to allow for incidental variations in the tube and circuit parameters. A factor of safety
of 5 per cent is usually adequate under most circumstances to ensure
oscillation,

although often the factor actually used

is

much

higher than

this.

11-3. Influence of Transconductance g m

The criteria for oscillation
given in Eqs. (11-1), (11-3), and (11-4) are valid only for the linear region
of the tube characteristics, since it is only for this region that the current.

source equivalent circuit

is valid.
Despite this limitation, the expressions
be extended with significant results over the nonlinear portion of
the tube characteristics.
Refer first to Eq. (11-1). As a first approximation, K/n, for constant
/x, depends directly on
Factors other than g m are external to the tube
gm
and are independent of the potential and current magnitudes. Clearly,
g m is the only factor in the expression that will depend on the portion
of the tube characteristics that is used.
Likewise, in Eqs. (11-3) and
(11-4), if it is assumed that n remains substantially constant over wide
excursions of signal amplitude, and this is a reasonable assumption, and
since
involves
and vPJ or gm then here, too, gm is the only factor
that may vary during operations.
The transconductance g m which is the slope of the e c -% curve, is not
constant for large changes of input grid potential. In fact, it is precisely
this variable character of g m which accounts for the successful operation of

may

fj.

OSCILLATORS

Sec. 11-3]

287

For a given set of circuit parameters, the oscillations will

assumed by g m is such that the conditional equations [Eqs. (11-1), (H-3), and (11-4)] are satisfied, when sustained oscillaIf g m cannot assume a sufficiently large value for these
tions will result.
equations to be satisfied, then the potential fed back from the output
to the input circuit is insufficient for maintaining the oscillations and
they will die out. If g m were too large, the potential fed back would be
greater than that required for the oscillations just to be sustained and the amplitude would continue to increase.
oscillators.

build

up

until the value

If the average transconductance g m is

defined as the slope of the line connecting

the

two extreme points on the transfer

appropriate to the input

characteristic

signal, as illustrated in Fig. 11-2,

-m =

- h
E

,min

The Barkhausen
become

gUiZr ^1 +

=r

3 \g m Z

1\
)

<

>

=1

sustained oscillations

>

growing oscillations

decaying oscillations

(11-7)

y-J

K
\g m z

An

sustained oscillations in modified form

average transconductance g m

for

criteria

Fig. 11-2. Sketch for defining the

(11-6)
c

approximate relation between the average transconductance gm as

and the mutual conductance of the tube g m is posThis is done by assuming that the transfer curve is linear, as

defined in Eq. (11-6)

sible.

illustrated in Fig. 11-3.

is

It follows for the usual class

condition which

illustrated that

gm

and

(11-8)

2E

also that
gm

The

=
-E,

ratio of these expressions gives

(H-9)
c

RADIO ELECTRONICS

288

Fig. 11-3. Idealized linear transfer characteristic of a

[Chap. 11

vacuum tube

in

an

oscillator

circuit.

8h
is

This expression may be written in terms of the plate conduction angle

To do this requires the expression for the plate-current pulses, which
seen to be

% =
But the

+E

g m (-E g0

plate current

c , mtx

ib

- Egm + Egm
when

Eg0 +

cos

f or i b

cot)

(11-11)

This leads directly to

4-

= E gm (l -

cos wt b )

(11-12)

from which
-E g

j?c,m i

?Tp

By

combining

_ sm

= sm Ji

from

this that g m
(

6b

sin' -r

(11-13)

is

proportional to g m but otherwise depends

If, for example, 6 b = 120 deg,

E g0 + E m^)/E gm
c,

(11-14)

dm

only on the ratio

~0

with Eq. (11-10), there results

this expression

9m

It follows

then from the expression gm /g m = 0.25.

11-4. Alternative Expression for Average Transconductance. 2 A
slightly different expression for the transconductance to be used in the

OSCILLATORS

Sec. 11-4]

Barkhausen

criterion

In this case the procedure assumes,

system operates as a class C device. Also,

possible.

is

as in the foregoing, that the

it is

289

based on the fact that oscillation

the fundamental frequency

when the loop gain at

the assumed condition for

sustained

is

That

unity.

is

is,

oscillation is that

(5K

fundamental frequency

at

(11-15)

(5 is the network feedback factor, and K is the amplifier

fundamental frequency.
The assumption is now made that it is possible to define an effective
transconductance g m such that the gain of the amplifier K at the fundamental frequency is given by a relation of the usual type, namely,

where, as before,

gain, at the

K =
despite the fact that the system

is

(11-16)

g~JL

operating under class

conditions.

An approximate

expression for g m in terms of the g m of the tube is

possible for a transfer characteristic of specified shape.
Suppose that

the transfer curve

find a relation
is

defined

is

linear, as illustrated in Fig. 11-3.

between the transconductance

by the

Iv \m

and

gm

k*

(n-17)

illustrated, g m is given

2
7T,

rT
ib

cos wt dt

But the plate-current pulse has the known form given by Eq.
Therefore the required expression

_
-

2
7p

which

it is

Iplm

T
J-pim

by Eq. (11-9). To find the

necessary to evaluate the fundamental component
This is given by the Fourier expression

For the linear curve

of current,

It is desired to

relation

gZ

desired relation,

of the tube,

gm [

E g0 + E m ^ E gm + E gm
c

(11-11).

is

cos ut] cos

a>t

dt

(11-18)

This expression integrates to

Ejnt
9m

1 plm

sin 2ut\
2a,

tb
\

)\-

which becomes
IP im

<ch

sin

2&

(11-19)

Therefore by Eq. (11-17)

(jm

gm

u>k __ sin 2wfe

ir

2t

(11-20)

RADIO ELECTRONICS

290

In terms of the conduction angle

0&
__ sin db

6t

2x

But

[Chap. 11

(11-21)

2tt

for the transfer curve as denned, the plate conduction angle

is

defined through Eq. (11-12), namely,

=
6

wt b

cos

"

(11-22)

&gm

Equation (11-21) shows that the effective slope g m is proportional to the

normal slope gm but otherwise depends only on the ratio

EgQ

4" 2? Cl nia%

E gm
The dependence

on the same factors as in the prior calculadifferent manner.

As a comparison with Eq. (11-14), again assume that db = 120 deg.
For the present case g m /g m = 0.195, a value slightly less than that given
before.
However, since both forms (11-14) and (11-21) are only approximations, either may be used with roughly comparable results.
11-5. Amplitude and Frequency Conditions. Since in Eqs. (11-7) the
quantities K, Z, and Z T are complex in general, the conditions for selfsustained oscillations require that both the real and the imaginary parts
of the expressions separately and simultaneously satisfy the appropriate
conditions.
Refer specifically to the third equation of (11-7). Here the
conditions to be satisfied are
in this case

tion in Sec. 11-3, but in a

is

somewhat

-ts + 9-(-i)
Im
where Re and

G5

(11-23)

= Im
(-e)
^)

Im denote the real and imaginary parts,

respectively.

The

equivalent pair of expressions are

I
+ -1=1K
l^z
A

(11-24)

The first of the two sets of conditions contains a great deal of information
concerning the amplitude of the oscillations and specifies, in fact, a value
of g m or g m
tions.

and

in consequence determines the amplitude of the oscilla-

The second

about the frequency

of the

two

sets of conditions contains information

of oscillation.

OSCILLATORS

Sec. 11-6]

An

examination of the

first of

291

Eqs. (11-24) reveals the following general

information: Since n appears in the denominator, then for large

fi

there

an almost 1 1 correspondence between gm and K. Since, however, gm

will vary over a range from zero to some finite value, then any condition
that makes Z large (and K therefore reaches a constant value) would
permit most easily the production of sustained oscillations. That is, sustained oscillations are favored by circuits for which Z > r p
An antiresonant circuit possesses a high impedance at the resonant frequency, and
the impedance drops rapidly as the frequency moves off resonance.
Such
a circuit is suitable to provide sustained oscillations and also provides
good frequency stability. However, certain nonresonant frequencyselective circuits are also used satisfactorily in oscillator circuits.
Circuits of both the resonant and nonresonant frequency-selective network
types will be studied in detail.
11-6. Fixed Bias and Starting Characteristics.
Before examining
particular types of oscillator circuits, it is well to examine the effect of the
is

Fig. 11-4. A feedback oscillator biased to the left of cutoff,

secants for determining the average transconductance.

and the corresponding

on the operating features of the feedback circuit. In particular,

suppose that the bias of the circuit, as indicated in Fig. 11-2, is set to a
value beyond the cutoff of the tube. A number of possible operatgrid bias

Evidently the initial value

oscillations cannot build up
since the circuit conditions correspond to decaying oscillations.
Suppose that a potential e 2 is applied, whether from an external source
or produced by a transient phenomenon.
If the mean value of g mi is
small, the conditions required by Eqs. (11-7) for sustained oscillations
may still not be satisfied and the oscillations may die out. If the signal,
ing conditions are illustrated in Fig. 11-4.
of Qm

9mi

is

zero,

and Eqs.

(11-7)

show that

RADIO ELECTRONICS

292
say

e3,

appears on the grid and

to grow, the amplitude of

tained oscillations are

if

e gi will

this

is

[Chap. 11

sufficiently large for oscillations

increase until the conditions for sus-

fulfilled.

Clearly, for an oscillator biased near to or

beyond

cutoff, the circuit

not be self-starting. However, as is evident by comparing the results

illustrated in Fig. 11-4 with those in Fig. 11-2, the amplitude of the oscilThis results in an
lations is larger in the heavily biased oscillator.
will

increased efficiency, a feature that

oscillator

although

Also, since

it is

it is

may

be very desirable in a power

usually of small consequence in low-power sources.

quite easy to induce oscillations, the overbiased condition

not objectionable from this point of view.

11-7. Grid-resistor-Grid-capacitor Biasing Circuits.
The use of a
grid-resistor and grid-capacitor combination, as illustrated in Fig. 11-5,

is

Fig. 11-5. An oscillator with grid-resistor and grid-capacitor biasing.

conditions are sketched.

The build-up

not only allows the self-starting feature but also provides for an operating
bias at or beyond cutoff.
The operation of the circuit is illustrated in
Fig. 11-5 and is essentially the following: When the circuit is first placed
in operation, the grid bias is zero and the operating point is high on the
characteristic, where the value of g m is large.
The third of the criteria
[Eqs.

and growing

oscillations occur and continue to

the positive portion of the swing, the grid
potential becomes positive, thus charging the capacitor.
The time constant of the grid resistor and capacitor is such that a substantially steady
(11-7)] applies,

increase in amplitude.

On

is maintained.
This bias displaces the operating point to the
as illustrated, with consequent increasing amplitude of oscillations.

bias

amplitude
condition

left,

The

of the oscillations continues to increase until

is

reached between this

an equilibrium
amplitude and the consequent bias E c
.

OSCILLATORS

Sec. 11-8]

293

The magnitude of the bias may be controlled to some extent by the proper
choice of R g and C B although, as discussed in Sec. 10-9, the values of R g
and C g are not critical. They are generally determined experimentally.
As illustrated, the amplitude of oscillation will be such as to allow
,

a small grid current to flow during the positive peaks of the cycle.

It

is

which serves to charge the grid capacitor. In

fact, the variation in grid current can be used as an indication that oscillations have been established and also as a rough indication of the
this small grid current

amplitude.
If

R Cg

the time constant

is

too large, the bias potential across

adjusts itself slowly to sudden changes in the amplitude of oscillation.

Cg
If

adjustment is so slow that the oscillations can die out before

the bias potential can change appreciably, then with sudden changes in
the amplitude the action is very much as though fixed bias were used.
As a result, it is possible that the oscillations will die out. Hence a
possible condition is one in which the oscillations first build up in amplitude to the equilibrium value. Any slight irregularity that tends to
reduce the amplitude of the oscillations will cause the oscillations to die
Once the grid
out owing to the substantially steady bias that exists.
capacitor discharges through the grid resistor and the bias reduces
sufficiently, the oscillations will again build up, until the above process
This intermittent operation can be overcome by decreasrepeats itself.
ing the time constant R g C.
For stability to exist, it is necessary that
the bias reduce as the amplitude of oscillations decreases.
11-8. Tuned-plate Oscillator.
The tuned-plate oscillator is one in
which an antiresonant circuit is connected directly in the plate circuit
this rate of

of the

tion

vacuum
being

tube, the grid excita-

supplied

by

inductive

The

coupling to this plate circuit.

complete circuit has the form illustrated in Fig. 11-6.

It should be
specifically noted that this is just the
circuit of a tuned class C amplifier, but
with the circuit providing its own grid
excitation.
Consequently the analy_
_,
.
Fig- 11-6. Tuned-plate oscillator,
.\,
j
, ,
V.
ses of Chap. 10 of the tuned class C
amplifier apply for the tuned-plate oscillator except that the grid driving
power reduces the total available power output. Owing to this slight
difference, power tubes carry a single manufacturer's rating as an r-f
,

power amplifier and oscillator.

In order to examine certain of the properties

of the oscillator the condi-

tional equation for sustained oscillations [Eqs. (11-7)] will be examined.

For simplicity,

it

will

be assumed that the grid current

may

be neglected.

RADIO ELECTRONICS

294

The

grid excitation

[Chap. 11

then simply

is

E g = jaMI L = Z MlL
The

impedance

transfer

is

given by the relation

Zt

= -E. = _ Zfc
Ip

But

since IxZl

IC

7
z

ZC

'

ii

ic

ZmIl

_
" "

u+

ZmZ
- - zrrz;
c

it(z t /z c )

<n -26)

Also, the quantity

where
_

Zr,Z c

Zi

The

(11-28)
c

conditional equation for sustained oscillations

in the

form

then be written

ZjfZ c

ZL
which

may

rp

(Z L Z C /Z L

Zc)

(11-29)

is

gm Z M Z c

(Z L

Zc)

ZiZ

"p

or

gZ* = B +i(L - ^) "i

By

+^

(H-30)

equating the real and the imaginary terms, there results

g m ^-

= R

amplitude condition
(11-31)

wL

and

The

first of

wC

77

phase condition

coCVj,

these equations

may

9m

be written in the form

n RC

txM

(11-32)

specifies the average of g m (or g m ) and which, through Eq. (11-14)

This expression proor (11-21), specifies the plate conduction angle 8b.

which

vides information, at least in principle, concerning the amplitude of the

oscillations.

The second equation becomes

2

=zM i+ f)

(n -33)

OSCILLATORS

Sec. 11-9]

the quantity

If

co

is

defined

by the

295

relation
1

wo

Eq. (11-33) becomes

(11-34)

This equation shows that the frequency of oscillation will be approximately the resonant frequency of the circuit, the factor involving the
However, the frequency of oscillation will
ratio R/r P being small.
always be slightly higher than the resonant value. Clearly, the tube
plays only a minor part in determining the frequency of oscillation, the
external circuit elements exercising the main control.
In fact, the
influence of the tube on the frequency becomes less as the shunt resistance
of the antiresonant circuit increases or, correspondingly, as the series
resistance in the tank decreases.
If circuits of

very low dissipation

are provided, the oscillator has a

very high degree of

diagram

sinor

stability.

B2

of the circuit in

steady oscillating state may be

drawn; this applies for the fundamental frequency. In the diagram
its

(Fig. 11-7), the sinors are

not drawn

owing to the different orders

of magnitude that usually exist
among the currents and potentials,
to scale

flEg

Fig. 11-7. The sinor diagram of a tunedplate oscillator.

Also, angles are exaggerated for

clarity.

Under most circumstances the angle a = tan -1 (uL/R) = tan -1 Q will

be very nearly equal to 90 deg, and the feedback angle (90 a), that is,
the angle between E 2 and E g will be very small.
This means that the
,

feedback occurs substantially with 180 deg phase displacement, so that

a decreasing plate potential reflects itself as an increasing potential on
is also a phase shift of 180 deg through the tube, the
around the complete loop is zero or 2x, as already discussed.
11-9. Other Oscillator Circuits.
A variety of vacuum-tube feedback
oscillator circuits exist, each of which possesses some special characteristics.
The basic circuits and the coupling networks of the more important types of oscillators (see Prob. 11-1 for the amplitude and frequency
equations) are contained in Fig. 11-8.
In each of these circuits operation

the grid.

phase

is

Since there

shift

essentially class C, the essential difference

among them being

in the

coupling network.

Each

of these circuits

provides an antiresonant circuit of some type,

RADIO ELECTRONICS

296

[Chap. 11

with either inductive or conductive coupling between the output and

input circuits. This does not imply that only circuits which possess
an antiresonant circuit will operate successfully as an oscillator. In
fact, circuits in which the feedback is accomplished through resistance

Tuned grid oscilla+or

L2

R2

-rAvCv-/W!P-i-r7W" -j\A/w-r1

c
Hartley

Tuned grid tuned plate

Fig. 11-8.

The

circuits

and coupling networks

of the

more important

oscillators.

and capacitance networks will be examined in some detail. However, the

above networks do possess a feature that is common to all feedback
oscillator circuits; they all provide a 180-deg phase shift between the
output and input circuit so as to satisfy the necessary condition for
regenerative feedback to exist.

OSCILLATORS

Sec. 11-10]

A special

word

297

desirable about the tuned-grid-tuned-plate oscillator.

is

This oscillator depends for

its

operation on the feedback that will be

C gp In this circuit the plate

tank resonant frequency is slightly higher than the resonant frequency of
the grid tank.
This causes the plate circuit to be inductive, and a negative input resistance results which overcomes the grid-circuit losses and
thus allows oscillations to occur.
possible through the grid-plate capacitance

11-10. Stabilization of

Feedback

Oscillators. 3

Vacuum-tube

oscilla-

with variations in any operating

characteristic that involves either the tube or the circuit parameters.
For example, a change in temperature may cause a change in the inductance and capacitance of the tank elements and may also cause a change
tors will suffer changes in frequency

in the grid-cathode

and the plate-cathode interelectrode capacitances.

Also, a change in plate potential will result in changes in the interelectrode

A change in the coupled load causes a change in the shunt

with a consequent change in frequency. Although these
factors have been neglected in the explicit discussion given above of the
various oscillator circuits, they do play a part in determining the frequency, since they will contribute to a variation of the tube or circuit
capacitances.

resistance,

constants of the coupling network.

A relationship can be obtained which serves as a guide in estimating

the relative frequency stability of an oscillator, the comparative frequency
stability of several oscillators, or the influence of factors which have only
a slight effect on the frequency of the oscillator.

The desired relationship may be found from Eq. (11-5), viz., K(J = 1,
which is the condition for oscillation. When oscillations exist, the amplitude |K(J| and the total circuit phase shift must have such values as to
Suppose that the phase shift
satisfy the basic condition for oscillation.
through the tube does not vary appreciably with frequency, for very
This is generally a valid assumption. Any
small changes in frequency.
variation in frequency must be accompanied by a change in the phase of
the $ network which is such that the total phase of the circuit remains
substantially constant; otherwise oscillations will cease.

variation of phase with frequency dd/doi

frequency stability

by the

defined

of

the system.

In

value of
infinite,

As an

the

is

Sf

rate of

the frequency stability

fact,

wo
etco

00

The

evidently related to the

Sf

is

relation

S/

where

is

mean frequency
is

(11-35)

of the oscillator.

Clearly, the larger the

the system. In the limit as S/ becomes

completely frequency-stable.

the more stable

the oscillator

~i

aw/coo

is

indication of the significance of this stability criterion, suppose

RADIO ELECTRONICS

298

[Chap. 11

that a variation occurs in some element of the circuit, other than the
network, which is supposed to control the frequency of the system.
(3
If the requisite phase condition implied by Eq. (11-5) were initially satisfied at the particular frequency of oscillation, then when the change
in the element occurs, the circuital phase condition will, in general,

As a

longer be satisfied.

result,

the frequency
is satisfied once again.

no

of the oscillator will shift

until the circuital phase condition

If

the

($

net-

work produces a very large phase shift for a small frequency change, i.e.,
circuital
if S f is large, then the required frequency shift to restore the
be very small.
measures may be taken in order to improve the
This would include the careful choice of the
of an oscillator.
and capacitor, either with negligible temperature coefficients
such temperature variation that a change in one is counteran opposite change in the other. Any changes that might result

phase condition

A number
stability

inductor
or with

acted by

will

of corrective

Fig. 11-9. Electron-coupled oscillator.

from changes in the plate potential can be overcome by the use

of ade-

quately regulated sources.

The effect of changes in the load impedance on the frequency may be

eliminated by using an amplifier to separate the load from the oscillator.
This system is called a master-oscillator power-amplifier arrangement and
usually abbreviated MOPA.
The oscillator and power amplifier can be combined into a single tube
by using a tetrode or a pentode. Such an oscillator is called an electronA typical electron-coupled oscillator circuit employing
coupled oscillator.
Here the cathode, grid 1, and grid 2
a pentode is illustrated in Fig. 11-9.
are operated as a conventional Hartley oscillator, grid 2 acting as the
is

ordinary anode in a triode. The current to grid 2 is small, but it is

maintain the oscillations. The main part of the space

sufficient to

current serves to produce the power in the load impedance.

current

is

controlled

plate current

is

by the

oscillator portion of the tube,

The

plate

but since the

substantially independent of the plate potential, except

at the very low plate potentials, there

is

a very

little

reaction between the

output circuit and the oscillator section of the tube.

In such electron-coupled oscillators it is found that increasing the plate
potential causes the frequency to decrease slightly, whereas increasing
Hence,
the screen potential causes the frequency to increase slightly.

OSCILLATORS

Sec. 11-10]

299

by obtaining the screen potential from a potential divider, as shown in

Fig. 11-9, and by locating the screen tap at the proper point (and this is
determined experimentally), it is possible to make the frequency substantially independent of the plate supply potential.
The effects of a varying plate resistance rv on the frequency can be
materially reduced through the use of resistance stabilization.
In this, a

R s is added between the plate of the tube and the tank circuit.
This added resistance serves to make the total effective resistance in the
plate circuit so high that changes in the plate resistance of the tube have
resistance

very

little effect

on the frequency.

The

resistance also serves as a con-

venient means of controlling the feed-

back and hence the amplitude

oscillations.

of the

It is ordinarily desirable

that the resistance be

made

so high

that the oscillations will just barely

start.

Llewellyn 3 has shown that the frequency of oscillation can be made to

approach the resonant frequency of
the tuned circuit by inserting suitable

The basic

Fig. 11-10.

reactances in series with the grid or

circuit of

imped-

ance stabilization.

with the plate, or both. This might

be called impedance stabilization. It follows from the equivalent circuit
of an oscillator shown in Fig. 11-10 that

[lEg

=
=
=

+ 01212 + Pl3l3
+ P22I2 + 92313
031^1 + 932^2 + 933^3
Qllll

(11-36)

P21I1

where

9u =
912
913

rp

Zi

-f-

Z5

= 921= (Zi
= 931 Zm

92:

Zm)

Zo
923
933

=
=

Zi
932
rg

-j-

Z2

Zs

2Zm

= (Z + Z M )

(11-37)

Z2

Zi

Also

E = hr g

(11-38)

ff

Equation (11-38) is combined with Eq. (11-36), and for oscillation the
determinant of the coefficients of the currents must vanish. If it is
assumed that the circuits external to the tube are relatively loss-free,
only r p and r g are resistive all the other terms are reactive.
The resulting
expansion is a complex function. The expressions that result by equating
the real and the imaginary terms are
;

Xo{r P (X 2

+X +
4)

rg

{X l

+X

B )}

= Mjrtir, from the

reals

and

+ ^(X, + X M )(X + X M
+ I M)V, + (X + X M Yr p

(Xj

(11-39)

RADIO ELECTRONICS
X [r pr - (X, + X )(X + Xt)] - 2X M (X + X M )(X + X M
= -XoXji - {X + x M)\x + x - (x + x M nx + x

[Chap. 11

300

If in

4)

from the imaginaries.

Eq. (11-40)

and

)
(

t)

11-40)

have such values as to

satisfy the condition

2Xm(Xi

+ Xm){X +

Xm)

(Xx

+ x M nx +

x<)

(x 2

+ Xm)\X! + x

t)

terms not containing Z then the resonant frequency

is exactly that to cause
to become zero and to remain so independently
of r p r g and y..
That is, the frequency of oscillation is exactly the seriesresonant frequency of the tuned circuit.
As a particular example, consider a platestabilized Hartley oscillator, as shown in Fig.

which contains

all

The condition
comes, by writing X 4 =
11-11.

Fig.

11-11.

plate-stabi-

Xi = 2o>M

lized Hartley oscillator.

which requires that

By

be negative.

X = 6

and

since, for

M
U+M

for

stabilization

be-

0,

oL *

\Z7Tm)

coLi

setting

u>C s

0,
1

dU +L + 2M)
2

then

+ U + 2M
L
2M L +
+
+ M)
Li

C6 = C

MY
(U +
\L 2

M
M

For the ideal case in which the effective tank circuit Q is extremely high,
the compensation is perfect, and the frequency is independent of the tube
potentials.
In the actual case, the compensating reactances must be
experimentally,
and the compensation, although not perfect,
adjusted
in independence of frequency from
substantial
improvement
represents a
tube variations.
Several other cases, with

all

power sources omitted, are

illustrated in

Fig. 11-12.

11-11. Crystal Oscillators. 4

The frequency

stability of

an

oscillator

can be made very high by

circuits.
Such crystals, which are sections cut from a quartz crystal in
such a way that the flat sides are perpendicular to an electrical axis,
when stressed or compressed along this axis, are accompanied by the
appearance of electric charges on the surface of the crystal. Conversely,
when such crystals are placed in an alternating electric field, they are set

utilizing piezoelectric crystals as antiresonant

OSCILLATORS

Sec. 11-12]

301

into mechanical vibration.

If the applied electrical frequency is very
near to that which produces mechanical resonance, the amplitude of the
vibrations will be very large.

A
as

vibrating crystal can be replaced

shown

In this

in Fig. 11-13.

the crystal and

tion L, C,

Cm

electrical circuit,

represents the capacitance of

its

mounting when

represents the electrical equivalent of the vibrational

and

mass that

is

it is

characteristics of the material.

crystal

by an equivalent

circuit,

is

not oscillating the series combina;

the electrical equivalent of the

C is the electrical equivalent of

effective in vibration,

Fig. 11-12. Impedance-stabilized oscillators.

R is the electrical equivalent of the coefficient

L and C are in series resonance at the frequency

the crystal compliance, and

of friction.

The values

of

of mechanical resonance.

One

of the

most common types

illustrated in Fig. 11-14.

circuit, it is

When

seen that the oscillator

grid type, the crystal

of crystal-controlled

the crystal
is

is

replaced

oscillators is
its

equivalent

essentially of the tuned-plate tuned-

making up the tuned-grid

Fig. 11-13. The equivalent electrical network of a vibrating quartz crystal.

by

Fig.

11-14.

circuit.

Owing

to the

crystal-controlled

oscil-

lator.

extremely high Q of the equivalent circuit of the crystal, which may be

100 times as high as that of a conventional electrical circuit, the crystal
can oscillate only over a very narrow frequency range. As a result, the
frequency stability of such an oscillator is very high. When the temperature of the crystal is maintained constant, the frequency drift can be

made

less

than

11-12. Class
sible

part in 10 6

Oscillator. 6

by operating an

Considerable stabilization is made posA instead of a class C device,

oscillator as a class

RADIO ELECTRONICS

302

for this eliminates the grid current

[Chap. 11

and any nonlinear

effects resulting

Moreover, the output waveshape from a class A oscillator will

be sinusoidal, with a high degree of purity of waveform. Owing to the
manner of its operation, the oscillating frequency is determined by the
resonant elements.
However, since the self-regulating amplitude-control
feature of the nonlinear tube characteristic is no longer being employed,
other methods must be provided in order to stabilize the amplitude of the

from

it.

Fig. 11-15.

output.

One way

linear-stabilized oscillator.

for providing for linear stabilized operation

In this

trated in Fig. 11-15.

the output of which

is

circuit,

the output

is

coupled to a

used to control the d-c bias of the

is illus-

rectifier,

oscillator.

If

the gain of the amplifier before rectification is large, thus yielding a large
A circuit
d-c output, the oscillator will operate as a linear amplifier.

showing the details

of

such an amplitude-stabilized oscillator

is

given in

Fig. 11-16.

Amplitude stabilization may be effected by providing an amplitudenetwork to control the output of the oscillator. Such a method

sensitive

Fig. 11-16.
is

employed

in the

The

circuit of a linear-stabilized oscillator.

Wien-bridge

oscillator,

which

described in the next

is

section.

11-13. Resistance -Capacitance Oscillators.

form

of coupling net-

work that has been used extensively in relatively 1-f oscillators is given in
Fig. 11-17.
The phase shift through such a network as this is a fairly
sensitive function of the frequency, and such RC oscillators possess good
frequency stability. However, such a simple network will not provide

OSCILLATORS

Sec. 11-13]

303

a large phase shift between the input and output terminals, and it is
necessary to incorporate a second vacuum tube in the circuit in order to

The

provide an additional 180-deg phase shift.

the manner of the previous circuits, is

shown

circuit,

when drawn

in

in Fig. 11-18.

may be

analyzed in a direct
of Eqs. (11-7).
Refer to Fig. 11-19, which shows the comThis circuit

manner by an application

11-17.

Fig.

An BC

-vwVVW- -Ifcoupling network for an

plete coupling network, in

box " which provides a gain
of this

oscil-

which the amplifier

is

replaced

by a "black

diagram shows that

may

KjvE'j

KjvIZj,

be written in the form

Ei
if it is

coupled

K and a phase shift of 180 deg. An inspection

Ei

This

An BC

Fig. 11-18.
lator.

oscillator.

assumed that the input impedance to the amplifier

is

very high.

Phase

E2

reversing
amplifier

'

gainKN
Fig. 11-19.

The

transfer

The complete coupling

impedance

RC

oscillator.

the network becomes

of

Zr

But

circuit of the

Ei

K;vIZ a

I,

Il

as

I(Z

fl

Z f) =

1L

RL

Zg

then
ZiT

Also,

it is

Z-L

Z,

+R

(11-41)
L

Rl

noted that the impedance looking into the feedback network

R L (Z + Z f
RL + Z g + Zf
g

is

(11-42)

RADIO ELECTRONICS

304

The term

+ K //x which appears in Eqs.

rp

(11-7)

[Chap. 11

becomes

^ RL + Z + Z
g

Equation (11-7) for sustained

becomes

oscillations

-^XN Z R L tARl + Zg+

may

This

^=

Rl(Zs

(11-44)

be written in the form

gZK N R L rp =
By

R
+

known

including in this expression the

^+

Z,)

(11-45)

expressions of

Z g and Z f

namely,

L' ~
and equating

real

rP

B t

first

'

cod

and imaginary terms, two expressions

R L R2 +

gm K N R Lrp R2 =
The

i(l/C s )

(r,

rp

Rl) [RiRi

RL +

(r p

-^jA

Rl) {Ri

They are

result.

+R +
2

of these expressions yields the following expression for the

frequency
1

CiC,

The second

m
^

RiR *

+1+

Rl/tJ

expression yields
1

K.nRlt p R2

rP

RL +

(rp

+ R L (b + R + M*)
)

(11-48)

Ordinarily the circuit constants are chosen: Ri

(11-47)

R2

R',

C\

C2

C;

Rl,
...

(r>+

^ _1
C R
2

RRl

= -

(11-49)

and

^-

3R

^+

h (h + 1)

(11 - 50)

OSCILLATORS

Sec. 11-13]

305

be observed from Eq. (11-50) that the coupling amplifier K N

(J network must indeed provide phase reversal,
as
Moreover, for the values of rp and RL which
indicated in Fig. 11-19.
might ordinarily be used, the magnitude of the gain of this coupling
amplifier might be of the order of unity or less.
What is implied by
this expression, therefore, is that the phase shift through the passive
elements of the coupling network, composed of C\R\ in series with the
parallel combination of C 2 and R 2 is
deg, a fact that is readily verified
by simple network considerations, and that the amplifier circuit is
required in order to provide the necessary 180 deg.
From purely practical considerations since the amplifier JL N would normally provide a gain
considerably in excess of that required, negative feedback is ordinarily
It will

which

is

part of the

It

Fig. 11-20.

The Wien-bridge

oscillator.

included in both vacuum-tube circuits, both to reduce the gain and to

improve the waveform. The manner of achieving feedback is illustrated
in Fig. 11-20.

known
The

Because

of

the form of the network, this

RC

oscillator is

as a Wien-bridge oscillator.

modification

shown introduces

certain significant consequences

In particular, for a given R t and tube
adjusted so that the bridge is balanced, the grid-cathode

into the operation of the circuit.

Tl,

if

potential

is

is

zero.

That

is,

the Wien-bridge output (in this case, chosen

as the grid-cathode potential)

is zero at the balance point but is different

from zero for all frequencies in the neighborhood of balance. Clearly,
there will be no feedback at the balance point, and the system will not

Note that, for the bridge itself (neglecting the loading of R*

by tube Tl), the conditions for balance and the frequency at balance are
oscillate.

RADIO ELECTRONICS

306

which become, for

i?i

= R =
2

R, C\

=2

= C2 =
COo

Thus, for these conditions, the circuit

C, as chosen

above

is

[Chap. 11

a frequency-selective feedback

For the circuit to act as an oscillator,

the ratio R 3 /Ri must be chosen greater than 2.
Continuous variation of frequency of this oscillator is accomplished
by varying simultaneously the two capacitors C\ and C 2 these being
variable air capacitors.
Changes in frequency range are accomplished
by switching into the circuit different values for the two resistors Ri and
rejection amplifier (see Sec. 9-8).

Variable resistors could, in principle, be used for the continuous

but it is more difficult to build resistors which track with the same
precision possible with variable air capacitors.
2.

control,

Owing to the limitations imposed on variable air capacitors, to attain

low frequencies requires that large resistances be used. A practical limit
exists to the size of R, in one instance because of blocking that might
occur in Tl (note that one R is the
grid resistance of this stage) and
second

because

problem
stray
Fig. 11-21. The Wien-bridge oscillator
with amplitude stabilization.

of

of shielding

60-cps

power supply.
the size of

is

the

potentials

increasing

the grid against

from the

practical limit to

perhaps 10 megohms,

so that low frequencies of the order of

10 cps

may

At the high frequencies, with the

R, the loading of the phase-inverting amplifier
the loading does not stop the oscillations, it will

readily be achieved.

consequent reduction
increases.

Even

if

of

amplitude of the oscillations with changes of

frequency range.
A very ingenious modification of this circuit, 6 which serves to stabilize
the amplitude against range switching and against aging of tubes,
replaces R t by a tungsten-filament lamp, ordinarily a 115-volt 3-watt
lamp. The revised circuit is given in Fig. 11-21. The effect of providing
the tungsten-filament lamp, which possesses a positive temperature
coefficient of resistance, is to provide a system which automatically
changes the feedback factor of the coupling network in such a direction
as to keep the total gain more nearly constant as the gain of Tl and T2
affect the stability of the

varies because of loading or other variations in the circuit.

Suppose

that there were, for whatever reason, an increase of current through the

lamp.

The lamp

resistance will increase, thereby increasing the feed-

OSCILLATORS

Sec. 11-13]

307

back factor and reducing the gain. The gain change will be such as to
maintain the current through the lamp almost unchanged.
The current through the tungsten lamp R t consists of three components; (1) the d-c component through Tl, which for class A operation is
essentially constant; (2) the a-c component through Tl due to the a-c
potential on the grid; (3) the a-c component of opposite phase through
R z to the cathode. The resultant a-c current through R 4 is comparable
with, and perhaps larger than, the d-c component, with the result that
the value of the resistance of i? 4 will be controlled to a very large extent
by the amplitude of the oscillations. Note, of course, that for adequate
control of the amplitude the thermal time constant of the tungsten lamp
1(
I

Fig. 11-22.

The bridged-T

or twin-T oscillator.

must be large compared with the period of the oscillations; otherwise an

amplitude drift may occur.
The parallel-T and the bridged-T networks which were examined in
Chap. 9 together with the Wien-bridge as coupling networks for frequency-selective amplifiers may also be used in oscillator circuits, 7 in
much the same way as the Wien bridge. However, as noted in Chap. 9
these networks are essentially rejection circuits at the balance frequency.
To effect the necessary inversion and the proper freedom from loading,
In this circuit, the inversion is
the circuit of Fig. 11-22 may be used.
derived from the resistor combination

T2 serving principally
That is, the cathode of

Ri, tube

to provide a low-impedance driving source.

tube Tl

In these circuits
is always at a higher a-c potential than the grid.
Ri must be larger than R 3
It is possible to view the foregoing circuits as well as the Wien-bridge
circuit as providing two potentials to the input of tube Tl, one of which
is of such phase as to constitute regenerative feedback and thereby
.

increases the gain over that in the absence of feedback, while the other

RADIO ELECTRONICS

308

[Chap. II

The resultant magnitudes of the two

frequency are such that degeneration
occurs at all frequencies except for a small range in the neighborhood of
the frequency of oscillation. It is only at these frequencies that oscillaintroduces degenerative feedback.

feedback potentials as a function

of

tion will occur.

From practical considerations, the Wien-bridge oscillator is preferred

to the other circuits, because of the ease of frequency control and variaand because fewer variable

circuit elements are required for control.

A simple one-tube phase-shift*

RC networks alone to provide

the requisite 180-deg phase difference

tion,

oscillator is possible

which incorporates

between input and output potentials. The circuit is illustrated in Fig.

11-23.
The operation may best be understood by considering that each
L section, which consists of a C and R
combination, shifts the phase by 60 deg
at the frequency co.
This explanation asct c2 C3
sumes that there is no loading by one RC
K T K T
combination on the others and that there
VFy
[
^
s n0 loa ding by the coupling networks on
1
I
..

the
Fig.

11-23.

phase-shift oscil-

lator

load resistance.

plate

The use

of

three such sections will shift the phase of

the output by a total of 180 deg relative to

the input, and

provide the

this,

2ir shift

together with the 180-deg shift by the tube, will

for successful operation.

Under these assumptions,

follows that

it

tan

= \/3 =

oRC

or
1

V3RC
A

(11-51)

more exact calculation, under the assumption that the amplifier loadby the phase-shifting network may be neglected, yields the relation

ing

Vq RC

(11-52)

would be possible, of course, to obtain the required 180-deg phase shift

with more than three sections, but there is no particular advantage in
It

doing

so.

As might be surmised, there

is

an appreciable attenuation

in potential

through the network. At the frequency of oscillation, it

is found that (5 = J^9For oscillations to be possible, i.e., to fulfill the
condition |(JK| = 1, the amplifier gain must be at least 29.
Consequently, either a high-ju triode or a pentode must be employed in such a
in progressing

circuit.

OSCILLATORS

Sec. 11-14]

309

may be varied by changing the value

feedback network. However, care in the
change of the element is important; otherwise the impedance looking into
the phase-shifting network, and its phase, may change to such an extent
The frequency

of

any

of the oscillator

of the elements in the

1 relation may no longer be satisfied, and oscillations will

For small variations in frequency, a single element may be
changed. For wide variations in frequency, the three capacitors should
be varied simultaneously. The three resistors could also be varied
simultaneously, but this would cause a serious change in impedance,
with a consequent effect on the amplifier gain and the possibility of discontinued oscillation. It is quite possible, of course, to remove the
Such a change
restriction that the values of all R and of all C be equal.
will complicate the matter of securing variable frequency operation.
11-14. Negative -resistance Oscillators.
One may consider the foredemonstration
of the fact that
going analyses of feedback oscillators as a

that the K(5

cease.

Fig. 11-24. A negative-resistance tetrode circuit, and the plate characteristics, showing
the region of negative plate resistance.

it is

possible to devise circuits containing

power

generated is sufficient to

to provide the

power that

is

vacuum tubes

overcome the

in

which the
and also

losses of the circuit

transferred to an external circuit.

If

the

represented by a certain
then
one might consider the
circuit,
equivalent resistance in the plate
such
a magnitude as just
resistance
of
tube as representing a negative
oscillations
in the circuit
The
terms.
to overcome the total dissipative
total loading or dissipation within the circuit

is

then be sustained at the stable level required by the variations of the

negative-resistance properties of the circuit.
If one is able to find a device that possesses a negative resistance, i.e.,
a device in which a positive increment of current through it is accompanied by a negative potential increment across it, then this can be used
will

to neutralize the positive resistance representing the total dissipation.

Such negative-resistance devices do exist, the simple tetrode operating

with a plate potential below that of the screen being a common example.
The connections of such a device and the plate characteristics which show
the region of negative plate resistance are

circuit that exhibits

an

shown

in Fig. 11-24.

effective negative resistance

and

at the

same

RADIO ELECTRONICS

310

[Chap. 11

time avoids the objectionable features of secondary emission is illustrated

1 1-25.
A pentode is operated with a plate potential that is lower
than the screen potential, and the suppressor grid is maintained slightly
negative relative to the cathode. Since the plate is at a low positive
potential, it does not exert much force on the electrons and under these

in Fig.

conditions the suppressor grid repels most of the electrons that

manage

to get past the screen grid, with a resulting higher screen current.
If
tive,

the suppressor potential is increased slightly, i.e., is made less negathen there will be less repelling action by the suppressor and more

The electrons
by the suppressor and returned to

plate current will flow at the expense of the screen current.

that were previously being repelled

the screen will

current.

now

pass to the plate, with a consequent reduced screen

if the screen potential is increased by the same

Note that even

potential as that applied to the suppressor grid the net effect

is still

reduction of the screen-grid current.

That

is,

if

to increase

the screen current were

somewhat with the

in-

crease of screen potential, the decrease in screen current owing to

the action of the suppressor grid

so
is

Fig. 11-25. A pentode circuit that exhibits

a negative output resistance.

the terminals

AB

much

is

greater that the net effect

a reduction of the screen current.

Therefore, with the circuit shown,

there is a decrease of current through

with an increase in potential across these terminals,

with a consequent negative resistance.

To analyze the circuit, 9 it is assumed that the change in screen current
is a linear function of the changes in the suppressor-grid and screen-grid
That
potentials and also that the suppressor-grid current is negligible.
is, it is

assumed that
AiC 2

The

g 3 2 Ae c3

Ae c2

r2

factor g 32 has the dimensions of a conductance and

is

such that

g 3 2 Ae c3 gives a measure of the influence of a change in current i c2 due to a

change in potential of the suppressor grid. Note from the foregoing discussion that g 3 2 is inherently negative since a positive Ae c3 is accompanied
by a negative AiC 2. The factor re2 is a measure of the change in i c2 due to

a change in

e C 2.

assumed that g 3i and rg2 remain constant over the range of operaand by noting that with a large C and R a change in potential Ae c2

If it is

tion,

appears on the suppressor as a change Ae c3 then

,

OSCILLATORS

Sec. 11-14]

The input

resistance

between points

A and B

Ae c2
Af c2

which

negative

is

r
1

>

the operation of the circuit of such negative resistances as

AB of the two circuits shown in Fig.

11-26.

negative-resistance, or dynatron,
Fig. 11-26.
transitron, oscillator.

The equivalent and

manner

in the

then

032^2

part of an oscillator, suppose that a tank circuit

Fig. 11-27.

is

when
-gztfgz

To examine

311

coupled to the terminals

circuits

may be drawn

and a negative-transconductance, or

simplified circuit of the negative-resistance oscillator

of Fig. 11-1.

cuit simplifies to that

is

These

Since the feedback potential

shown

in Fig. 11-27.

may

This

is

zero, the cir-

be drawn as a

simple coupled circuit, in the form illustrated in Fig. 11-28.

To

evaluate the characteristics of the cirapply Kirchhoff's law to the two-loop

network. This yields
cuit,

R
L

Tp

i
-

H
Cp)

Cp

(11-53)

+ R+Lv +
{

To solve for t 2

^v)

Fig. 11-28. The basic equivalent circuit of a negativeresistance oscillator.

the current through the inductance and load, the following

differential equation

must be evaluated:

W + (R
\l +
d 2i 2

1 \ di 2
7jc) Tt

(R + rp\
+ \-^cf)

*2

(11 " 54)

assumed that rpt which is inherently negative, remains substanconstant over the range of operation, this equation may be solved

If it is

tially

12

RADIO ELECTRONICS

312

[Chap. 11

directly to give, for the oscillatory case,

= Ae
where

and

2\L^r r cJ

The

6 are constants.

sin

( wi

expression for

form, although with different values for

and

0.

(11-55)

has exactly the same

The angular frequency

i\

of oscillation is

R +

_1(R, J_Y

rp

\ r^LC

\L

r p CJ

or
rp

+R
4

Under the assumed

(11-56)

r p <7/

oscillatory conditions, the expression for i 2 indi-

cates that the amplitude of the oscillations

may

decrease, remain constant,

or increase, depending

upon the expo-

nential term in the expression.

term (R/L)

(l/r p C)

is

the

If

positive, then

the oscillations which might have been

started in
fall

(l/r p C)

critical

amplitude

of

is

will ultimately

the quantity (R/L)

negative, the oscillations will

quantity (R/L) + (l/r p C) equals zero,

the exponential factor is unity and the

value L/CR.

the oscillations remains constant.

If

tend to increase in amplitude with time.

For the critical case for which the

Fig. 11-29. The amplitude of oscillation increases until the value of rr

assumes the

any manner

to zero.

this condition,

(11-57)

RC

"

and the corresponding frequency

For

is

=4K!WM'-)
Such negative-resistance oscillators
same manner as the normal feedback

are self-regulating in

<->

much

the

Thus, owing to the

variation of the negative-resistance characteristic of the tube circuit that
is used, if the quantity (R/L)
(l/r p C) were negative, thus allowing
oscillators.

for continually increasing amplitude of oscillations, these oscillations

would increase

until the region of operation extended to the point

where

OSCILLATORS

Sec. 11-14]

rp

= L/RC, when

313

the build-up condition would cease.

These condi-

It should be noted from the diagram

that even with an assumed sinusoidal output potential, and this is not a
required condition, the output current will be nonsinusoidal.

tions are illustrated in Fig. 11-29.

REFERENCES
Barkhausen, H., "Lehrbuch der Elektronenrohren," vol. Ill, S. Hirzel
Verlag, Leipzig, 1935.
Massachusetts Institute of Technology Staff, "Applied Electronics,"
chap. XI, John Wiley & Sons, Inc., New York, 1943.

1.

Dammers, B.

and H. vanSuchtelen, "Application

Valve in Radio Receivers and Amplifiers," N. V. Phillips
Gloelampenfabriken, Eindhoven, Netherlands, 1950.
Terman, F. E., Electronics, 6, 190 (1933).
Llewellyn, F. B., Proc. IRE, 19, 2063 (1931).
Jefferson, H., Wireless Eng., 22, 384 (1945).
For further details, see:
Terman, F. E., "Radio Engineering," 3d ed., Sec. 8-3, McGraw-Hill Book
Company, Inc., New York, 1947.
Arguimbau, L. B., "Vacuum Tube Circuits," p. 320, John Wiley & Sons,
Inc., New York, 1948.
Terman, F. E., R. R. Buss, W. R. Hewlett, and F. C. Cahill, Proc. IRE,
24, 649 (1939).

2.

G., J. Haantjes, J. Otte,

of the Electronic

3.

4.

5.

6.

88 (September, 1950).
Ginzton, E. L., and L. M. Hollingsworth, Proc. IRE, 29, 43 (1941).
9. Herold, E. W., Proc. IRE, 23, 1201 (1935).
10. As a general reference, see:
Edson, W. A., "Vacuum Tube Oscillators," John Wiley & Sons, Inc.,
7. Sulzer, P. G., Electronics, 23,
8.

New

York, 1953.

PROBLEMS
Show

11-1.

that the amplitude and frequency of the oscillators illustrated in

Fig. 11-8 are the following

Tuned

grid:

fiRL g C

1
9<

Hartley:

_ ^ C(Rr + R^jU+U + 2M)

9m

(Li

+ M) (Li + M)

Colpitts:

CO

m
g.

nRjd + C 2 )
- Ci/O

L(ix

11-2. Two identical triodes are connected in a Franklin oscillator.

mine, in terms of the circuit parameters:

Deter-

RADIO ELECTRONICS

314

The

a.

expression for the critical value of the resistance

[Chap. 11

Rn

at which oscilla-

tions will just begin.

The frequency

b.

of oscillation.

Assume that the power absorbed in the tuned circuit, which determines its Q,
be represented by an equivalent resistance in shunt with the inductance and

may

capacitance.
11-3. Two identical triodes are connected in the oscillator circuit shown.

Determine
a.
b.

The resonant frequency oj

The minimum values of g m assuming
.

under identical conditions,

that the tubes are being operated

for oscillations to be maintained.

R -200a
L'OJh
C'.Olftf

Rgl=Rg2 '1Mtl

6SN7

11-4. Obtain
oscillator

shown

an expression

11-5. Using the 806 tube

Chap.

10,

illustrated
*

for the operating frequency of the cathode-coupled

in the diagram.*

whose constant-current characteristics are given

calculate the performance

when

it

is

on page 315.

M. G. Crosby,

Electronics, 19, 136

(May, 1946).

in

used in the oscillator circuit

OSCILLATORS

315

^06
.....

./

11-6.

120 -L

&L m2J9mh

jn 3000*

type 852 triode has the following ratings as an r-f power amplifier and
down conditions without modulation per tube)

oscillator (key

D-c plate potential

D-c grid potential
Peak r-f grid potential
D-c plate current
D-c grid current
Driving power
Power output

The tube

3 000 volts
,

600

volts

850 volts
85 ma
15 ma approx.
12 watts approx.
165 watts approx.

operated under rated conditions in a tuned-plate oscillator, operat1 Mc.

Determine the following:
Grid dissipation.
6. Plate dissipation.
Power output.
d. Oscillator efficiency.
is

ing at a frequency of
a.
c.
e.

/.
g.

Resonant impedance of tank circuit.

Grid leak required.
Mutual inductance between grid and plate

Assume that the maximum

coils.

equal to the minimum plate potential.

Neglect the leakage inductances and resistances of the grid and plate coils,
and assume that the tank circuit has Q = 20.
11-7. A type 806 triode when used as an r-f power amplifier has the following
grid potential

is

ratings:

D-c
D-c
D-c
D-c

plate potential

2,500 volts

grid potential

500

plate current

195
25

grid current

Driving power
Grid resistor
Power output

This tube

volts

ma
ma

17 watts
20 000 ohms
370 watts
,

connected as a Hartley oscillator and is operated under the condiThe tank tuning capacitor is 250 /i/tf; the resonant frequency is
2 Mc; the loaded Q is 23.5.
Determine the following:
a. The inductance, and resistance of the tank circuit.
is

tions specified.

b.

The power output.

How

far from the bottom of the tank coil is the cathode connection?
The oscillator efficiency.
Assume e c max = eb mla
11-8. An 833A transmitting tube has characteristics that may be represented
approximately by the equations
c.

d.

ib
tb

=
=

10- 4 (25ec

eb )

amp

for (25e
for (25e c

+e >
+e <
b)

b)

RADIO ELECTRONICS

316

[Chap. 11

This tube is to be operated as a power oscillator in the circuit shown in the accomCalculate the following:
panying figure. Assume that ej, m n = e t mi
b. A-c power output.
a. Power input to the plate circuit,
When calculating the grid signal potential, neglect the grid driving power.
.

C=/ISfiMf

11-9. Typical constants of a crystal are

R =

1,500

ohms

L =

250 henrys

C =

0.04 n/tf

Cm =

8 ixnf

Q of the crystal.
Calculate the series- and parallel-resonant frequencies.
What is the percentage change in the series and parallel frequencies

a. Calculate the
b.
c.
is

if

Cm

doubled?
11-10. Crystals

trolled oscillators.*

may be

used in what are called series-resonant crystal-conDiscuss the operating features of the accompanying figures.

11-11. Discuss the operation of the Pierce oscillator illustrated. Observe

that the crystal is the only resonant element in the circuit, the crystal in this case
operating as a parallel-resonant circuit.

-k-

IRFC

11'12.

o Mc.

An

amplitude-controlled oscillator uses a 6J5 tube. It operates at

of the network and rectifier when the oscillator tube is

measurement

Grid

Rectifier

and
fitter

Osc///a/for
* F. Butler, Wireless Eng., 23, 157 (1946)

OSCILLATORS

317

removed indicates that, with an input of E v 8 volts rms, the potentials are
E, = 0.4 rms, E cc = 11.4 volts d-c. What is the value of R for the bias to
adjust itself to 7.5 volts?
11-13. Refer to the tuned-plate oscillator of Sec. 11-8.
Neglect grid loading, and evaluate the frequency stability S/ at the value of coo where 6 = ir/2.
Express the final results to show the dependence on the Q of the resonant circuit.
11-14. Deduce an expression for the frequency stability of the phase-shift
oscillator discussed in Sec. 11-13.
11-15. Evaluate S/ at the resonant value of a twin-T oscillator.
Neglect grid
loading.
11-16. Consider the coupling

network

illustrated,

which

is

for use in a phase-

shift oscillator.

a.
6.

work

Derive an expression for the transfer function ($ of this network.

Determine the frequency coo at which the total phase shift through the netis 7T.

Evaluate the expression for g at co

d. Determine the input impedance at co = w
11-17. o. Find an expression for the frequency stability
c.

Sf

of the circuit of

Prob. 11-16.

Suppose that the loading of one RC combination on the previous one is

phase shift is three times that of each section. Compute S/ for this. Compare with the results under (a).
11-18. Consider the phase-shift oscillator shown.
a. Evaluate the frequency of oscillation of this circuit.
b. Find the relationship among parameters for which the amplitude is independent of the frequency of oscillation.
6,

negligible, so that the total

r
11-19.

page 318.

cathode-coupled Wien-bridge oscillator is shown in the diagram on

critical value of Ri at which oscillations will just start.

Determine the

RADIO ELECTRONICS

318

[Chap. 11

The Meacham bridge-stabilized oscillator* is illustrated. Here Rt is a

the resistance of which varies with the amplitude of oscillation, although
the thermal capacity is sufficiently large so that the temperature remains substantially constant over the cycle of oscillation.
11-20.

resistor,

b.

If

Q =

= Ko(l
form
jS) in the
Derive an expression for S/.
10 6 of the resonant circuit (this is possible with a crystal), Ko = 100,

a. Suppose that the gain

immediate neighborhood of

0.1, calculate

of the amplifier has the

co

<S/.

Calculate the frequency shift A/// due to a 0.1-rad variation in the phase
angle of the amplifier.
11-21. The plate characteristic of a type 24A tube connected as a tetrode is
given in the figure. The d-c potential Eu. is adjusted to 45 volts. A parallelresonant circuit tuned to 1 Mc is used, with C = 250 /iyuf.
c.

Determine the minimum value of R for which oscillations will be sustained.

Plot the oscillation amplitude as a function of R.
Plot the current waveshape for maximum oscillation amplitude.
j.

A.

Meacham,

Proc.

IRE,

26, 1278 (1938).

OSCILLATORS
11-22. Discuss the operation of the
crystal-controlled oscillators* illustrated.

319

negative-differential-transeonductance
Note that here, as for the Pierce

oscillator (see Prob. 11-11), the crystal is the

only resonant element in the

circuit.

I'H
Hi"
11-23. Discuss the operation of the transitron oscillators illustrated.

Ebbl Ebbl

Ebb2

Ebbl

Ebb2

>

rp
11-24. Consider the two-tube oscillator circuit shown.
Show that for Ri
may be negaR
Ri the apparent resistance between terminals
c
conditions.
The circuit is supposed to operate under class
tive.

and

R +

AB

>

* S.

Bernstein, Electronics, 26, 198 (February, 1953).

CHAPTER

12

AMPLITUDE MODULATION
12-1. Introduction.

There are two major reasons

for transmitting

intelligence at a relatively high frequency level: (1) transmission

radiation

is

practicable at the high frequencies, and (2)

it is

by

possible to

transmit a number of messages simultaneously without interference if

the frequency level is different for each message. Of course, any complete system of conveying intelligence from one point to another

must be
capable of reproducing the intelligence, as represented by the amplitude,
or loudness, and the frequency, after transmission.
The process of altering the frequency level of the intelligence

The

is

broadly known as modulation.

inverse process, in which the intelligence

ated wave,

By

is

known

is

extracted from the radi-

as demodulation or detection.

definition 1 modulation

is

the process of producing a

wave some

characteristic of which varies as a function of the instantaneous value of

another wave called the modulating wave. The modulating wave is

usually the signal, the modulated wave being the h-f carrier wave, which
has been altered in a manner to carry the intelligence.
Consider a wave which may be represented analytically by the expression
e

where

is

the time.

= A

If either

A,

cos
co,

(ait

or 6

is

6)

varied according to

(12-1)

some func-

tion of the instantaneous value of a modulating wave, then this expression

modulated wave. It is possible, in fact, to produce a

which all three parameters vary simultaneously. However, in
each of the modulating methods that are important practically, only one
of these parameters is varied, and in commercial transmitters great care
is taken to avoid the use of more than one type of modulation.
In amplitude modulation, the amplitude A is varied in accordance
with the modulating wave, while co and 6 remain constant. In frequency
modulation, the frequency co is varied, and both A and 6 remain constant.
In phase modulation, the phase 6 is varied, while A and co remain constant.
It should perhaps be mentioned that phase modulation is not
of much practical importance in itself, but as will be shown later, it may
be used as an intermediate step in achieving frequency modulation.
will represent the

wave

in

320

AMPLITUDE MODULATION

Sec. 12-2]

12-2. Characteristics of Amplitude Modulation.

As

321
indicated, ampli-

produced by varying the magnitude of the carrier in

accordance with the amplitude and frequency of the modulating source.
tude modulation

is

Let the signal potential be designated as

= Em

em

and

let

cos

mt

03

(12-2)

the unmodulated carrier be written as

= E

ec

cos {w c t

(12-3)

6)

The carrier frequency co c is usually much greater than the signal frequency
m and is chosen at the designated frequency level desired for the transmission.
The resulting modulated wave has the form
oi

The amplitude

= (E

E +

k a E m cos

o3

m t) cos

uj,

(12-4)

k a Em cos

o> m t expresses the sinusoidal variawhere the proportionality factor k a

determines the maximum variation in amplitude for a given modulating
In this expression the arbitrary constant phase 6 has been
signal E m
chosen as zero, since it plays no part in the modulating process.
In examining this wave in detail, the expression is written in the form

factor

tion of the amplitude of the wave,

which
e

The

is

= E

cos w c t

+ ~^-

ma

known

is

ma =
and 100ma is the

c (l

+m

cos

um t)

cos

wc t

(12-5)

then expanded to the form

factor

tion.

= E

cos (

w m )t

Vh^s cos

(o> c

u>

m )t

(12-6)

as the modulation index,

k a E

(12-7)

~E7

-percentage

modula-

sketch of Eq. (12-5) has

the form shown in Fig. 12-1.

The expanded

expression of Eq.

(12-6) indicates the frequency spec-

trum of the modulated wave. The Fig. 12-1. The modulating signal and a
modulated carrier.
first term is of carrier frequency oi c
the second has the frequency w c + co m and is called the upper sideband.
Its
frequency is equal to the sum of the carrier and the signal frequencies.
The third term has the frequency co c wm which is equal to the difference between the carrier and the signal frequency. This component is
;

known

as the lower sideband.

modulated wave

is

plot of the frequency spectrum of the

illustrated in Fig. 12-2.

Modulation

may

therefore

RADIO ELECTRONICS

322

[Chap. 12

be viewed as the process in which signal information, which is specified

relative to a zero-frequency reference, is shifted on the frequency scale
At the same
so that it is specified with respect to the carrier level co,;.
time, the intelligence appears symmetrically disposed relative to co c
It should be emphasized that
Relative
Carrier Upper the foregoing is not a mathema.

amplitude

Lower
side

*T

side

.1.

band

tical

band

hag*

_i

fiction,

means

of

the

extract

Hi

spectrum.

as

it

is

possible

appropriate

filters

frequencies

In

fact,

in

by
to

the

the features

with
one or more of the frequencies in
the spectrum suppressed will be

of transmission of intelligence
12-2.
Frequency spectrum
Fig.
sinusoidally modulated wave.

of

examined below.

simple sinor representation of Eq. (12-6) by means of a pair of

rotating conjugate sinors in the complex plane is readily possible. A
little

thought

convince the reader that the three sinors in Fig. 12-3

will

represent the three terms of the equation and that the resultant sinor

does exhibit the properties of the

a-m wave.
In general, the modulating sigis not sinusoidal but is a comSince, however, this
plex wave.

nal

complex wave

may

be represented

by a Fourier series if the wave is

periodic or by a Fourier integral if
it

is

Fig. 12-3.

The

sinor representation of an

a-m wave.

nonperiodic, the modulated

wave possesses a frequency spectrum which is more complex

than that illustrated. But each frequency in the modulating signal
produces a pair of side frequencies in the frequency spectrum. Then
carrier

Relative

response
g(co)

Fig. 12-4.

The frequency spectrum

a signal with frequencies in the band

of a

g(co) will

complex wave.

yield a frequency spectrum

with a band of sidebands

~2 S(^o

u)

and

symmetrically disposed about the carrier wc

have the form

illustrated in Fig. 12-4.

Such a spectrum would

Sec. 12-3]

AMPLITUDE MODULATION

323

The frequency-shifting property of modulation is not limited to shifting

an audio-frequency to a higher position in the frequency spectrum. It
is possible to shift an h-f signal up or down in the frequency scale, and
both processes are important. Since such frequency shifting or frequency
changing does not involve directly the intelligence to be transmitted,
frequency changing is classed, not as a modulation process, but rather as a
detection process.

Amplitude Modulation. In the ideal case the ampliwhich in general will actually be the signal waveshape rather than a single constant amplitude, must be faithfully imposed
as the amplitude envelope of the carrier frequency, over the full range of
amplitude variations. Moreover, the signal waveshape must be maintained over the entire frequency band involved in the intelligence, from
the rather narrow band required for carrier telegraphy, to the broader
band required for commercial broadcasting, to the very broad band
12-3. Circuits for

tude

in Eq. (12-1),

required for television.

The degree of linearity, i.e., the degree to which the signal waveshape
and the envelope of the modulated carrier must be the same, is determined
by the character of the transmission. For commercial broadcasting,
it is required that the intelligence be transmitted with a minimum of
distortion over the specified audio spectrum, and over a wide degree of
modulation. While some consideration is given to the matter of efficiency of a modulated transmitter, ordinarily the principal consideration
It is possible to design highly efficient
is linearity rather than efficiency.
transmitters in which the power is varied with the modulation requirements, although in most cases the transmitters operate rather inefficiently, and at an average modulation of less than 50 per cent.
For commercial telephony, the requirements imposed are quite differThe frequency band is narrower, and
ent from those for broadcasting.
Power efficiency is
the frequency response need not be strictly linear.
important, and it is essential that a high modulation index be maintained
in order that the signal/noise ratio at the receiver

available carrier power.

Many

be optimum for the

small transmitters operate in

much

this

same way.
might appear reasonable to modulate the oscillator, and certain
be considered could be used in such an application.
However, the discussion of the problem of frequency stability of an
oscillator in Chap. 11 showed that the frequency does depend on the load,
and a varying load would in most cases be accompanied by a varying
frequency. The amount of frequency modulation so introduced is greater
than can be tolerated, although the modulation can be so introduced as
Because of this,
to have no adverse effect on the oscillating circuit.
It

of the circuits to

never modulated in regular broadcast practice.

Amplitude modulation is ordinarily accomplished in the power-ampli-

oscillators are

RADIO ELECTRONICS

324

[Chap. 12

may

chain and

be carried out at low level or at high level. In lowmodulation is accomplished at relatively low power
level, and the power level is then raised until the desired level is reached.
Since the carrier is modulated, the amplifier stages must maintain a linear
relationship between input and output over the entire signal-frequency
range, to avoid what is called sideband clipping, and over the entire range
Class B tuned power amplifiers are used in this applicaof modulation.
A system of this type for high-power
tion, as discussed in Chap. 10.
applications therefore comprises a low-power modulator and a chain of
The modulator has relatively
class B tuned-power-amplifier stages.
little effect on the over-all efficiency of the chain of relatively lowfier

level modulation, the

efficiency class

amplifier stages.

R-f
oscillator

Audio

Audio
input

amplifier

A-f signal

Modulator

Power
amplifier

Modulated
r-f output

Linear

detector
Fig. 12-5.

feedback scheme for improving the linearity of a modulating system.

In the case of high-level modulators, the unmodulated carrier

fied

by means

of relatively high-efficiency class

amplifiers,

and

is

ampli-

linearity

unimportant, owing to the constant signal level. The modulator,

which is simply an audio amplifier, must now provide the full audio
power component of the transmitter, since the modulating signal must be
made available at high power. This requires considerable power ampliThis system comprises, therefore,
fication of the modulating signal.
highly efficient class C amplifiers, but a large and inefficient modulator,
as regards over-all efficiency, there is little to choose between the lowlevel and the high-level methods for a given power output, and both
is

systems are widely used.

As will be discussed, an amplifier can be modulated by applying the
modulating signal in the grid, plate, or cathode circuits, and in the case
The point of
of pentodes either of the available grids may be used.
application of the modulating signal often determines the name or type
of the modulation.
The linearity of modulation can be improved by the application of
inverse feedback.
This is accomplished by means of a system such as
In this circuit the output of
that illustrated in block form in Fig. 12-5.

AMPLITUDE MODULATION

Sec. 12-4]

325

the linear detector (see Chap. 13 for a discussion of the linear detector),
which is assumed to be a faithful reproduction of the output from the
modulation system, is an audio-frequency signal which, in the ideal case,
Any difference between them serves
is identical with the audio input.
as a correction to reduce the difference, thereby improving the resulting
linearity.
It is important, of course, that audio-frequency or r-f envelope
phase shifts be minimized through the feedback loop, as well as that
proper amplitude relationships be maintained around the loop.
12-4. Square-law or Small-signal Modulation.
Amplitude modulation may be produced by impressing two sinusoidal potentials of different
frequencies in a nonlinear circuit.
The van der Bijl modulator is one of
the earliest devices embodying methods of this type, but it is seldom used
at present.
This modulator depends for its operation on the curvature
of the transfer characteristic of the tube.
A circuit of this modulator is

_L Modulated oufput

~pe 'Ec CUma coscom t)cos a>c t

Carrier

Fig. 12-6.

The

The

ModuW/ng

circuit of the

van der

Bijl

modulator.

and modulation-frequency potentials

and the
modulated wave appears across the tank circuit in the plate circuit of
the tube. The tank is tuned to the carrier frequency and must be
sufficiently broad to include all the important sidebands, usually about
+ 10 kc for normal broadcast purposes; otherwise sideband clipping
occurs.
The operation of the modulator is made clear in the sketches
given in Fig. 12-6.

carrier-

are both applied in the grid circuit of a triode or multigrid tube,

of Fig. 12-7.

To examine the modulation process

transfer curve

is

analytically,

it is

supposed that the

parabolic over the range of operation, so that the a-c

may be related to the input grid potential by the

terms of the series expansion

plate current

ip

where

ai

and a 2 are constants.

e

and

it

follows,

= Em

by combining

atfg

The

+ a el

two

(12-8)

excitation potential e g

+E

first

is of

the form

cos u c t

(12-9)

this expression with

Eq. (12-8), that the

cos

oi

mt

RADIO ELECTRONICS

326

[Chap. 12

Grid circuit
Total grid signal

Plate current

Modulated output

Low frequency
componentfeuppressed,

The waveshapes

Fig. 12-7.

plate current
ip

at various points in the

van der

a,\E m cos

ip

may

= a\E

modulator,

is

tis

mt

aiE c cos

u> c t

a 2 E^ cos 2

o>

mt

+
This

Bijl

+
a2

a 2 E\ cos 2

o> c t

E Em

co

cos

m t cos

co c t

be written as

+ a E m E cos
+ u m )t + a Em E
+ B^k + ^k cos 2w +
+

cos w c t

+ ai m cos Umt

(a> c

cos

^ ^

(a> c

cos

w m)<

2^

(12-10)

co c 5>> u> m and also that those frequencies which are not in
the neighborhood of u c are eliminated by the use of tuned circuits, the

Assuming that

only potentials which appear across the output are produced by

ip

= aiE

cos

oj c t

a 2 E m E c cos

(co c

o>

m )t
a 2 Em E c cos

(co c

co

m )t

(12-11)

AMPLITUDE MODULATION

Sec. 12-5]

This

may

327

be written in the form

ip

from which

a-iEc

II H

cos

os

mt^ cos wj

seen that the modulation index

it is

is

(12-12)

given by

(12-13)
Ol

The amount

of

modulated output available without appreciable disis not great, and the

tortion in such a modulator as here considered

efficiency is low.

Owing

to the fact that the plate circuit contains a parallel tuned rather

than a pure resistance load, the foregoing analysis

rect

is

not completely cor-

this despite the fact that the load is purely resistive at or near

By

taking this matter into account, Carson 2 has shown that

a substantial increase in output occurs with the substitution of the resonant load. However, other methods provide better modulation characresonance.

teristics

and have displaced the low-level method here discussed.

The use of a balanced modulator, 3 which

somewhat like the push-pull amplifier, auto-

12-5. Balanced Modulators.

possesses an appearance

matically eliminates either the carrier or the modulating frequency, as

well as many of the intermodulation frequencies.
The circuit of the

balanced modulator

is

given in Fig. 12-8.

A balanced

Fig. 12-8.

It is

assumed that the two tubes are identical and that the

symmetrical.
of the

modulator.

The tube

characteristics

may

circuit is

be expressed by equations

form
iP i
iP 2

But the input

=
=

aie g i

a x eg i

+
+

a&\ x

(12-14)

a^e.a 2

potentials have the form

eg i
e g2

= E cos w
= Z? cos

ct

+ Em cos m
+ Em cos

o> c t

u>

u>

(12-15)

mt

RADIO ELECTRONICS

328

TheD the currents

= aiE

become

in the plate circuits

[Chap. 12

+ a E\ cos
+ a El cos m + a E Em cos (u + um
+ a E Em COS
= a E cos
+ a\E m cos u m + a E\ cos a
+ m)t
+ a^E^ cos u m a E E m cos

ip i

cos

cc c t

a-iEm cos
2

wmt

ca

(tO c

oo c t

(co c

a E Em
2

03m)t

(12-16)

ct

)t

iP 2

ca c t

co

cos

(co c

m )<

But the potential induced in the secondary of the coupling network

given approximately by
e

Then the output

e

2o> c

Ma E
1

+
which

is

of the

cos ('

2w c Ma 2 Ec Em

may

+
1

M d(i

v\

i Pi )

is

(12-17)

dt

form

!)

cos

(co c

m )<

|j

cos

be written as
e

2<j) c

Ma

+ E m sin

co

m i!^ sin

co c

(12-18)

In certain applications it is found advantageous to use a balanced

modulator in such a manner as to eliminate the carrier. This is readily
accomplished by interchanging the sources fm and / in the diagram.

Fig. 12-9. Balanced modulators for producing suppressed-carrier modulation.

this is done, the output contains frequencies oi c + co m and co c u m

with the carrier term oi c missing. Such modulated waves with carrier
missing can be transmitted, as discussed in Chap. 1. The demodulation
problem is more complicated than when the w c term is present. This will
be discussed in Chap. 13. Two circuits of balanced modulators which
suppress the carrier are illustrated in Fig. 12-9.
In these circuits, the
LC circuit is approximately in resonance for all frequencies in the neighborhood of the carrier frequency co c
12-6. Single-sideband Suppressed-carrier Modulation.
Suppose that
the suppressed-carrier output from a balanced modulator were passed

When

AMPLITUDE MODULATION

Sec. 12-7]

through a suitable

filter

network so that one

329

of the sidebands

was

The resulting wave would be a single-sideband suppressedcarrier signal.

The demodulation of such a wave is feasible, but, because

eliminated.

fundamental limitations, such modulated waves are limited in their

The system is used extensively
in point-to-point commercial communications.
In the case of signals for which the modulating signal frequency fm
is small, the construction of narrow band-pass filters with sharp cutoff
is difficult.
A method for producing the desired results by phase shifting
rather than by means of filters has been developed 4 by the Bell Telephone
of

application to commercial telephony.

The

Laboratories.

essentials of this

method are indicated schematically

in Fig. 12-10.

Ec costoc i:

cosa,m. cosojc t

Modulator

Em cosa>,t
Add

Em Ec cosfwc-ojm )t

90 phase lag

ECc sinwc t
90phase lag

\
\Em Smu>mt
Modulator

Em En sin<Ainlslnco
-

Fig. 12-10.

12-7.

ct

modulator for single-sideband suppressed-carrier

Shunt Modulator.

The

signals.

Heising, or shunt, modulator

was

one of the early devices for producing amplitude modulation.

also

The

modulator is illustrated in Fig. 12-11. Observe that the

with T2 is just a simple tuned amplifier which is biased
to operate under class C conditions.
The carrier signal is applied to the
grid of this amplifier.
The circuit associated with Tl is a simple untuned
power amplifier with a load consisting of the a-f choke and T2 in parallel,
the bias E cci being so adjusted that this circuit operates under class A
conditions at the modulating frequencies fm
The effect of the amplifier
Tl is to produce a varying plate potential to tube circuit T2. Hence
in so far as T2 is concerned, the carrier signal e c is impressed in the grid
circuit, and the modulating signal e m appears in the plate circuit.
In
essence, therefore, T2 is operating as a plate-modulated class C amplifier.
A detailed analysis of this circuit is not warranted, owing to its limited
practical importance.
Such an analysis is relatively straightforward
if it is noted that the effect of Tl is to cause the plate potential of T2 to
be of the form e h = Ew, + E h cos co m t. In so far as Tl is concerned, this
tube functions as a class A amplifier which operates into an almost pure
circuit of this

circuit associated

330

RADIO ELECTRONICS

[Chap. 12

which consists of the high reactance of the a-f choke

shunted by the effective low resistance of the r-f stage.
Such a modulating system as this can produce a 100 per cent modulated
Since the audio amplifier Tl serves principally to alter the
carrier.
effective plate potential to the r-f amplifier, and since a substantial
portion of the power from the plate supply is absorbed by the Tl circuit,
the system is quite inefficient. For this reason, it is suitable only for
resistance load,

relatively low-level modulation.

yModulated
output

Fig. 12-11.

The

Heising, or shunt, modulator.

The

general principle of modulation disvery important, particularly if the carrier

and modulating potentials are large, or if the operation occurs near cutoff.
Such a situation exists when a class C amplifier is series-modulated,
whether this modulation is applied in the plate circuit, the grid circuit,
the cathode circuit, or the suppressor circuit, if a tetrode or pentode is
If a carrier potential were introduced into the grid of a class C
used.
amplifier, and if the modulating potential were introduced into the plate
12-8. Linear Modulation.

cussed in the foregoing section

is

then because w c ^> oi m the variation in potential in the plate cirby the signal could be considered to be the equivalent of a relaIf the relationship between
tively slow change in plate supply potential.
the output tank current is a linear function of the plate potential, for
constant grid excitation, the output potential (which is a linear function
of the tank current) would have the desired modulated characteristics.
With the proper design and adjustment, the modulation characteristic
of a class C amplifier is such that 100 per cent modulation with distortion
as low as 2 per cent in a plate-modulated amplifier and as low as 5 per
The distortion can be
cent in a linear grid amplifier may be attained.
reduced below these values by the use of inverse feedback in the circuit.
circuit,

cuit caused

AMPLITUDE MODULATION

Sec. 12-9]

331

12-9. Plate-modulated Class C Amplifier.

The basic
modulated class C amplifier is given in Fig. 12-12. It

that

it is

essentially the circuit of the class

circuit of a platewill

be observed

amplifier except for the

introduction of the source of modulating potential in the plate circuit.

The modulation

characteristic of such a plate-modulated amplifier is the

plot of the tank current as a function of the plate supply potential.

Fig. 12-14.

The

conditions during plate modulation.

ideal plate-modulation characteristic is illustrated in Fig. 12-13a.

erally the situation

which shows the

is

more nearly

effect of

The

Gen-

like that illustrated in Fig. 12-136,

tank impedance.

The more

nearly straight the

the less the modulation distortion. A high value of tank impedance

yields the smaller distortion, but the power output is also smaller.
If the frequency of the modulating signal is low compared with the

lines,

carrier frequency, the

impedance of the tank circuit at the modulating

Consequently the properties of the circuit

frequency will be negligible.

RADIO ELECTRONICS

332

by the modulating frequency, and the plate

shown even when e M

are not appreciably affected

and tank currents

[Chap. 12

will follow the characteristics

varies at the modulation frequency.

Suppose therefore that the carrier potential of angular frequency co c is

modulated by a modulating potential of frequency co m
The behavior of
.

the circuit

To
The

is

that illustrated in Fig. 12-14.

analyze the modulation process,

eg

and
If

it will

be supposed that

u> c

y> u m

grid-circuit potentials are, respectively,

ec

= E gm
E gm

cos w c t

cos

+E

oi c t

(12-19)
cc

the modulating potential in the plate circuit

= Emm

em

is

written in the form

wm t

cos

(12-20)

the resulting relatively slowly varying plate potential has the form

This

may

Eu,

+ Emm

cos w m t

(12-21)

um t)

(12-22)

be written in the form

= #(1

where m, the modulation index,

nt cos

given by the ratio

is

m= E

(12-23)

It should be noted from the curves of Fig. 12-13 that the rms tank
current and the d-c plate current are related to the plate potential by
expressions of the form

It

h=
Ibb

By

krebb

k b e bb

(12-24)

k b Ebb

combining Eqs. (12-24) with Eq. (12-22),

it

follows that

= krEbbil + m cos w m t)
= hEbb(l + m cos wm t)
h = iw>(l + m cos w m t)

It
Ib

Also, corresponding to the

current

iT

will

be chosen

may

current, the instantaneous tank

of the
it

This

rms tank
form

(12-25)

V2 I T

sin wj,

(12-26)

be expressed in the form

cos

oi

m t) sin

03 c t

(12-27)

AMPLITUDE MODULATION

Sec. 12-9]

But the output potential

ex.

333

that appears across the tank circuit

is

given

with good approximation by

El =

X l It/90

jXdT =

(12-28)

Therefore
et

V2 XJt sin (u + 90)

V2 X L k T E (l + m cos

=
=

et

bb

The corresponding plate-cathode

en

(12-29)

o) c t

is

ez,

is

eb

Also,

m t) cos

potential of the tube

eb

which

oi

+m

EbbQ-

\/2

and

Sees. 10-2

from the discussion in

u m t)(l

cos

X L k T cos u

c t)

(12-30)

10-3, the plate current is given

the instantaneous plate current has the form

= V2
~Q-

iP i

Likewise,

it

k T E bb (l

+m

cos w m t) cos w c t

follows from the curve of Fig. 12-13

h=

hb{l

+m

cos

(12-31)

and Eq.

(12-25)

w m t)

(12-32)

The above information may be used to analyze the performance of the

modulated amplifier.
The average power supplied by the d-c plate power source is
rT

'

Pbb

7ir
J

m JO

Ebbl b dt

where Tm is the period of the modulating cycle.

written as
Pbb

7fr

(12-33)

This expression

may

be

Tn

J. Jo

EbJail

+ m cos w m

t)

dt

This integrates to

Pm =

M.6 = k El

(12-34)

The power input by the modulating source

is

"Tm

Pm

rri

lm JO

QmJ-b dt

This becomes

Pm =

rT-

jr

m JO

Emm

cos

cc

m t 7w,(l

+ m cos w m

t)

dt

RADIO ELECTRONICS

334

[Chap. 12

which integrates to the form

Pm = E mm I bb

= P bb -y-

(12-35)

Clearly, for 100 per cent sinusoidal modulation, the modulating source

must

much power

deliver one-half as

as the d-c plate

power supply.

This requires, of course, that the modulating source must be an amplifier

of large

The

power capacity for a large power output.

power output at the tank circuit is given by the expression

a-c

PL =

r- e L i

i
J-

7fr

m J0

P i dt

rTm

jfr
1

m JO

i pl

dt

(12-36)

This becomes
1

pL =

--

2R

(^^j

+m

(1

cos

wjy

cos 2 w c t dt

Consideration must be given to the fact that averaging over the audio
is being carried out, during which time the term of carrier frequency
It is implicitly assumed, therefore,
is varying rapidly, since u c ^> a> m
that the net contribution to the average (the area under the appropriate
cycle

= (1 + cos 2w e t)/2 term is the factor

the trigonometric term at carrier frequency undergoing many cycles during the
slow audio-cycle period, yielding a negligible contribution.
By performing the indicated integrations,
curve)

This

by the

may

cos 2 w c t

be reduced to the form

PL =

^P

bb

(l+~)

(12-37)

may

be concluded from this that the d-c plate power supply furnishes
wave and the modulating amplifier
furnishes the power to produce the sidebands in the output.
The plate-circuit efficiency of the modulated amplifier is given by the
It

the power to produce the carrier

expression
V

= P

(12-38)

which becomes
V

or

Roty P(l
Q*k b P(l

+ m /2)
+ m /2)
2
2

AMPLITUDE MODULATION

Sec. 12-9]

This result shows that the plate-circuit efficiency

Therefore one

degree of modulation.

may

is

335
independent

of the

from

calculate the efficiency

considerations of the unmodulated amplifier as a simple class

To find an expression for the power dissipated

device.

in the plate of the tube,

it is

evident that

Pp
which

may

Pbb

+ Pm

Pl

be written in the form

Pv =
Consider the results

P(l

V)

(l

+ ^)

when the modulation index

(12-40)

m is

zero.

The

fore-

going become, in this case,

P bb = hEl
Pm =

By comparing

_
~

Rakl

Wh

these expressions with the corresponding expressions

m is not zero, it is observed that the addition

when

modulation increases the

This requires that a given tube when operated under
plate dissipation.
modulated conditions must be operated with a reduced plate potential
and current, if a specified maximum allowable plate dissipation is not
That is, since the input power to the plate circuit is
to be exceeded.
for
a given plate-circuit efficiency p the losses are
then,
Pm,
Pbb +
increased.
For a specified maximum allowable plate dissipation P p it
is then necessary to reduce Pbb + P m
In order to design the modulating amplifier, a knowledge of the effective impedance across the secondary terminals of the output transformer
This will permit a specification of the turns
of this amplifier is needed.
ratio of the modulating transformer in order to reflect the optimum value
This impedof impedance into the plate circuit of the modulating tubes.
ance is readily obtained by observing that the plate impedance is substantially resistive and must be given by
of

r\

P^m

Pm

Emm/%

Ebb

E mm I b {m/2)

_ p

C19AO\

Kb

This shows that the effective impedance is independent of the modulation.

The plate-modulated amplifier is used extensively in radio transmitters.
It has the advantage that modulation without excessive distortion is
Also, it operates
possible in practice by reasonably simple methods.
at high efficiency

and

is

relatively easy of adjustment.

It

has the

dis-

RADIO ELECTRONICS

336

advantage that a comparatively large amount

frequency

is

required.

The

of

resulting cost of the

[Chap. 12

power at the modulating

heavy and bulky modu-

equipment is sometimes greater than that of other methods. It

might be of interest to know that the two modulating transformers at
the 500-kw radio station
weigh approximately 19 tons each
and that an audio choke weighs
lating

WLW

12 tons. 6

12-10. Grid-bias

Modulation.

Amplitude modulation may be

Pig. 12-15

grid-bias

modulated

class

ac-

by

complished
modulating

connecting the
amplifier.
source in the grid
instead of the plate circuit,
The basic circuit of such a grid-bias modulated amplifier is illustrated in Fig. 12-15. Typical linearity curves of

such an amplifier for several different conditions are illustrated in the

The general character of the operation is illustrated
curves of Fig. 12-16.
graphically in Fig. 12-17.
Small grid drive

ccc

Fig. 12-16. Typical linearity curves of a grid-bias

different conditions of load.

modulated

class

amplifier under

Fig. 12-17. Conditions for 100 per cent modulation in a grid-bias modulated class

amplifier.

To
Sec.

analyze the operation, the procedure

is

essentially parallel to that of

12-9 for the plate-modulated amplifier.

modulation characteristic
is chosen of the form

is

linear
e

and that

= E om

cos

u3 c t

co c

It is

a>

assumed that the

The

carrier signal

(12-43)

AMPLITUDE MODULATION

Sec. 12-10]

and the modulating potential

is

of

the form

^ mm
The

COS

= E gm

(Al

+ Emm

cos u c t

mt

(12-44)

form

total grid-cathode potential has the

ec

337

cos

umt

+E

(12-45)

the slowly varying grid component being specified as

= E mm

e cc

Over the
current

if

linear range of operation, the analytic

form

k T (e cc

for the

rms tank

Eo)

(12-47)

the instantaneous value of the tank current

iT

this

(12-46)

cc

is

It
Also,

+E

cos w m t

may

= y/2

It sin u c t

sin

chosen as

is

be written as
iT

= y/2

iT

k T (e cc

uct

or

V2 k T (E mm

But when the modulation

iT

which

may

is

cos

wmt

zero, the

= \/2

k T (E cc

+E -

Eo) sin w c t

cc

(12-48)

tank current has the form

Eo) sin w c t

be written as
ir

where

I Tm

=
=

Irm sin wc t

-E

k T (E cc

(12-49)
)

Note that, when the modulation exists, the value

peak of the modulating cycle

of the

tank current

at the
is

I'Tm

V^2 kr{Emm

+E
cc

Eq)
(12-50)

The

conditions are best examined

graphically
Clearly,
is

as

in

12-18.

Fig.

the degree of modulation

seen to be

m
which

may

1't

Fl G 12 " 18 The tank current during

grid-bias modulation.
-

II

(12-51)

be written as

V2 k T (Emm + E - Eo) - V2 k T (E
V% k T (E K - Eo)
Cc

Eo)

RADIO ELECTRONICS

338

[Chap. 12

which reduces to

m=
By combining

E~E

Eq. (12-52) with Eq. (12-48), the instantaneous tank

current assumes the form

iT

which

= a/2

k T [(E cc

E )m

cos u m t

(E cc

a )]

sin

co c <

is

iT

= y/2 kT (E - E

In an entirely parallel way,

+ m cos om

)(l

cc

t)

sin

(12-53)

<a c t

possible to derive for the plate circuit the

it is

expressions

h=

Ibb

which

may

be written as

h(E cc - E

h=
=
The

+m

IbbiX

){1

+m

cos oij)
~

COS m<)

potential across the tank circuit

is

= jXJ T

Ex,

which

- Eg
jo)

k h (e
Kb\Hi cc

is

Ez,

= jX L k T {E cc - E

)(l

+m

cos o>J)

(12-56)

Therefore the instantaneous potential across the tank circuit

cos

(ii

m t) sin (w c t

is

given by

90)

or
eL

= y/2X L k T {E cc

Eo)(l

+m

cos

ta

m t) cos u c t

(12-57)

now possible to complete the analysis of the circuit.

Expressions for the various important values of power in the circuit
are readily obtained.
The average power input to the plate circuit by the
It is

d-c plate supply, over the modulating cycle,

is

T"

Pbb

1
7jT~
I m

f
JO
I

EbiJi, dt

This becomes, by Eq. (12-55),

Pbb

rT -

TfT
J-

m J

EbJ bb {\ + m

cos

cc

m t) dt

which integrates to
Pbb

EiJbi

(12-58)

AMPLITUDE MODULATION

Sec. 12-10]

339

Since this result shows no dependence on the modulation index m,

it

power

of

follows that the plate supply

independent of the degree

is

modulation.

The average output

Pl =

at the tank circuit

1

TFT
1

is

given by
T

Tm

e L i pl dt

Roili dt

jfr

im JO

m JO

which becomes, by Eq. (12-53),

Pl =

tjt

Tm

=^
Q

JO

- #

(E cc

(1

+m

*U) 2

cos

sin 2

<*J,

dt

The problem that exists here in examining the effect of the term of carrier
frequency when integrating over the modulating cycle is the same as that
discussed in connection with the comparable integration of Eq. (12-36).

The

Similar reasoning allows the present integral to be evaluated.

integral yields
Rok$>

This shows that the

The

(E cc

r-f

- E

(l

PT R (l +

(12-59)

power increases with percentage modulation.

modulated amplifier is given by

plate-circuit efficiency of the

= Pl =

(Rok T2 /Q 2 )(E cc

E bb k

bi

-
(E -

cc

+m

(1

/2)

which becomes

where

is

the effective series resistance in the tank circuit.

This

expression shows that the plate efficiency increases as the modulation

modulation index ( = 1) it is
be made as
cc
This requires that the load impedance of the modularge as possible.
lated amplifier be so adjusted that the peak amplitude of the output wave
index increases.

To

realize the highest efficiency for a given

necessary that the quantity

is

E E

only slightly less than the d-c plate supply.

The

plate dissipation of the tube

Pp =
which

given by the expression

Pbb

Pl

is

PP =
This

is

may

kE bb (E cc

- E

- Rk

(E cc

- E

(l

(12-61)

be written in the form

Pp = EbJb

- IIR

+ ~)

(12-62)

RADIO ELECTRONICS

340

[Chap. 12

be noted that the plate dissipation decreases as the percentage

modulation increases. Clearly, therefore, the plate dissipation is a
maximum when the amplifier is unmodulated.
Grid-bias modulation has the advantage that only a small amount of
modulating power is required. However, the carrier power that is
obtainable from the amplifier is approximately one-quarter of that from
the same tube when operating as a simple class C amplifier. This is so
because the peak power of the modulated amplifier corresponds to class
C operation, and with a completely modulated wave the peak current is
twice the unmodulated current and the corresponding peak power is four
times the unmodulated, or carrier, power. Also, the plate efficiency
during the unmodulated intervals is approximately one-half the efficiency
obtained with simple class C operation. This results from the fact that,
It should

if

the amplifier

of the

is

so adjusted that the plate potential

modulation

across the load

is

cycle, then,

halved.

when

there

is

is

small at the crest

no modulation, the

potential

This results in a high potential across the

tube, with a corresponding large plate loss.

As a

result, the plate-circuit

modulation is of the order of 34 per cent.

During 100 per cent modulation by a sinusoidal signal, the r-f power
output increases by a factor of J-*!, and the plate-circuit efficiency increases
to approximately 51 per cent.
It is interesting to compare the operating features of a plate and a gridThese are
bias modulated amplifier.
efficiency in the absence of

Grid-bias modulated amplifier:

Relatively low plate efficiency.

Low power

output in proportion to the capabilities of the modulated

tube.

Low grid-modulating power.

Plate-modulated amplifier
Large power output in proportion to the power capabilities of the
modulated tube.
Large modulator power.
In consequence, the over-all

efficiency, considering

both the modulating

and the modulated tube capabilities, is roughly the same. The choice
between the two methods of modulation is largely one of convenience,
since both methods of modulation will give sensibly 100 per cent modulated waves with low distortion, although it is more difficult to achieve a
linear modulating characteristic with control-grid modulation than with

other modulation methods.

The circuit adjustments are more difficult
with grid-bias modulation, as they are sensitive to variations in the carrier
exciting potential, the plate supply potential, and the magnitude of the
tuned load impedance. Despite these difficulties, control-grid modulation is used in high-power television transmission, since a plate modulator

AMPLITUDE MODULATION

Sec. 12-12]

would be very complicated,

required for this

TV

in

view of the width

of the

Control-grid modulation

service.

341
frequency band
generally used

is

in the last r-f stage.

A modulated output wave is produced

introduced into the cathode of the amplifier.

12-11. Cathode Modulation.

if

the modulating potential

is

The

It
basic circuit for cathode modulation is given in Fig. 12-19.
should be observed that in such a circuit the modulating potential appears
Consequently, it is to be anticiin both the plate and the grid circuits.

pated that the characteristics are a combination of those of the platemodulated and the grid-bias modulated amplifiers, with a plate efficiency
and modulating power requirements intermediate between the corre-

sponding requirements for the plate-modulated and the grid-bias modulated amplifiers.

Fig. 12-19.

The proportion

The

basic circuit for cathode modulation.

of plate relative to grid-bias

by increasing the modulating potential

(by increasing

g ).

By

modulation can be increased

Emm and making the grid bias large

reducing the grid-bias potential, a smaller modu-

have a greater influence in the grid circuit, and the

system becomes more nearly of the grid-bias modulated type.
The plate-circuit efficiency of such a cathode-modulated amplifier
should lie between the roughly 40 per cent of the grid-bias modulated

lating potential will

and the roughly 70 per cent of the plate-modulated amplifier.

power requirements of the modulator will lie between the
roughly 5 per cent of the output power for the grid-bias modulated
amplifier to the 50 per cent of the output power for the plate-modulated
The conditions which prevail will depend on the relative
amplifier.
amplifier

Also, the

potentials at various points in the amplifier, the percentage modulation,

and the degree

An

of linearity of operation.

interesting point to be noted

is

that the carrier excitation potential

E gm is approximately half that required to operate the circuit as a simple

class C amplifier to full output, and the grid bias is usually much greater
than the normal class C bias. This means, of course, that the amplifier
is

being only lightly driven.

12-12. Other

tode

is

Methods

of

Modulating a Class C Amplifier.

pressor-grid circuit.

may

If a

pen-

be applied in the supThe general characteristics of such a suppressor-

employed, the modulation potential

RADIO ELECTRONICS

342

[Chap. 12

modulated class C amplifier are similar to those for control-grid

However, adjustment of the amplifier is simpler. Fairlylinear modulation up to 100 per cent may be obtained.
Suppressor-grid modulation possesses the unusual feature that the
modulation characteristic lies wholly within the region of negative supgrid

modulation.

The

pressor potential.

suppressor-grid current

substantially zero;

is

hence the modulation power is zero.

Modulation may be accomplished by injecting the modulation in the
screen circuit of a tetrode or pentode.
Some power is required from the
modulating source, and 100 per cent modulation usually cannot be
achieved without some distortion.
12-13. Cross-modulation Distortion.
One of the most undesirable
forms of distortion which may occur in r-f amplifiers is known as "cross
modulation." This effect may occur if a powerful transmitter is in the
vicinity of a receiver and arises if the signal due to the transmitter at the
grid of the first tube is sufficient to modulate other signals over a substantial tuning range of the receiver.
It also occurs over a limited range
when an unwanted modulated signal is applied to the grid of the first
tube along with the desired signal. If the curvature of the dynamic
characteristic contains a third-order term in its series representation, the
undesired modulation is transferred to the desired carrier.
Suppose that the dynamic characteristic of the tube may be written
in the form
ip

and suppose that the

eg

= Em

a,ie e

a 2 e|

a 3 e*

a 4 e*

(12-63)

signal applied to the grid circuit has the

cos Udt

+E

c2 (l

+m

cos

u m t) cos wc2 t

form
(12-64)

avoid unnecessary complication, the desired carrier E ci is written

without modulation, although in general this carrier will be modulated.
No loss of generality results by doing this. Consider the third-order
term in the series. This is

To

a z el

a 3 [E cl cos u cl t

+E

ci (l

+ m cos

u>

m t) cos w c2 <] 3

which contains the following term, upon expanding the expression:

SazE c i cos
This

may

+m

cos

co

m <) cos w c2 t] 2

be written as

3a 5 2?c i cos

and

co c ii[J c2 (l

this contains a

1
oi c it

yEli

component

cos

2o> o2 t

(1

+m

cos

iM

m t)

AMPLITUDE MODULATION

Sec. 12-13]

343

Observe that the desired carrier is now modulated by the undesired

modulation and its second harmonic.
To reduce such cross modulation requires that the selectivity of the
total input circuit to the first r-f stage must be high.
Of course, by
reducing the higher factors a 3 CM, etc., in the power-series expansion
of the tube characteristic will also result in a decreased cross-modulation
,

effect.

REFERENCES
1.

Standards on Transmitters and Antennas, Institute of Radio Engineers,

1933.

J. R., Proc. IRE, 9, 243 (1921).

Peterson, E., and C. R. Keith, Bell System Tech. J., 7, 131 (1928).
4. Black, H. S., "Modulation Theory," D. Van Nostrand Company, Inc.,
New York, 1953.
5. Chambers, J. A., L. F. Jones, G. W. Fyler, R. H. Williamson, E. A. Leach,
2.

Carson,

3.

6.

and J. A. Hutcheson, Proc. IRE, 22, 1151 (1934).

As general references, see:
Massachusetts Institute of Technology Staff, "Applied Electronics," 1st ed.,
sees. 12-4 to 12-6, John Wiley & Sons, Inc., New York, 1943.
Cruft Laboratory, War Training Staff, "Electronic Circuits and Tubes,"
chap. XV, sees. 15-18, 15-19, McGraw-Hill Book Company, Inc. New York,
1947.

PROBLEMS
12-1.

The equation
e

a.
b.

What
What

(15

of a

modulated wave

10 sin 3,000*

is

8 cos 10,0000 cos

frequencies are contained in the modulated

is the amplitude of each?

2ir

10 6 f

wave?

12-2. Carry out the analysis to show that amplitude modulation results in a
square-law circuit when the carrier potential is applied in the grid circuit and the
modulating potential is applied in the plate circuit.
12-3. A shunt modulator using a bridge connection which requires no transformer is illustrated.*
a. Derive an expression for the signal to the load R
b. If Wc = a, what is the output?
t.

D. G. Tucker, Electronic Eng., 22, 139 (1950).

RADIO ELECTRONICS

344

12-4. Prove that the carrier term

is

[Chap. 12

missing in the balanced modulator of Fig.

12-9.

The balanced modulator

12-5.

In this

circuit,

of Fig. 12-9a is used for carrier suppression.

the potentials applied are the following:

= E

ec
em

with

o> c

com

and with

Em

Em

= 0.5E

C.

cos
cos

wc t
o>

mt

Assume that the transconductance

gm

varies linearly with grid potential.

a.

Obtain an expression for the output potential from the modulator.

6.

Plot the envelope of this potential.

12-6. Suppose that a band-pass

filter is

connected in the output of Prob. 12-5

of such characteristics that the lower sidebands are eliminated.

a. What is the expression for the resulting output?

Plot the envelope of this wave.

An ideal diode for which rv = 1,000 ohms in the forward direction and
There are applied to this
r p = oo in the inverse direction is used as a modulator.
circuit the two potentials
6.

12-7.

= Ec

cos

us c t

+ Em

cos

o>m and with E m = 0.5E C

Determine the amplitude of the component

with w c ^>
a.

wmt

of current of angular frequency

The effect of the diode is to permit conduction for one-half of each

modulating cycle. The effect may therefore be represented as a square wave with
unit amplitude during half of each cycle of w m and zero for the other half cycle.
oi c

co.

Hint:

b. Repeat for the component of frequency u> c

co m
12-8. Repeat Prob. 12-7 for the case where the single diode
.

is

replaced by

four diodes connected in a bridge circuit.

12-9. A type 851 power triode operates with a bias of 300 volts and with a
peak r-f signal of 525 volts. The load impedance R = 1,500 ohms. Determine
and plot the a-c plate potential across the tank as a function of the d-c plate
supply potential, for the following values of plate potential: Ey, = 500, 1,500,
2,500, 3,500 volts.

12-10. Repeat Prob. 12-9 for R = 1,000 ohms.

12-11. An 851 triode is used in a plate-modulated class C amplifier. It operates with a bias of 300 volts, a peak r-f signal of 525 volts, and a plate supply
When e^, = Ebb the peak plate swing is 1,750 volts. DeterEbb = 2,000 volts.
mine the plate-current waveforms at the values of eu, = 1,000, 2,000, 3,000 volts.
12-12. A type 891 r-f power triode has the following ratings as a class C oscillator for telegraphy:
,

D-c plate potential

D-c grid potential
D-c plate current
D-c grid current
Grid driving power

10 000 volts
,

Power output
Peak r-f grid potential
If

a.

When

the only limiting factor, determine the corresponding

C telephony, allowing for 100 per cent plate modulaplate-modulated 100 per cent, determine:

the plate dissipation

is

ratings of the tube for class

tion.

000 volts
45 amp
0.105 amp
310 watts
10 kw
2 900 volts
,

The audio power

required.

AMPLITUDE MODULATION
6.
c.

The impedance
Power output.

345

offered to the audio source.

d. Plate-circuit efficiency.

12-13. An 852 transmitting triode has the following ratings as a plate-modulated r-f amplifier under carrier conditions that allow for 100 per cent modulation:

D-c plate potential

D-c grid potential
D-c plate current
D-c grid current
Peak r-f grid potential
Grid driving power
Power output

The

amplifier

is

2,000 volts

500
67

ma

30 ma
750 volts
23 watts
75 watts

sinusoidally plate-modulated 75 per cent.

6.

Audio power required.

Impedance offered to the audio source.

c.

Plate efficiency.

a.

volts

Determine:

Average plate dissipation.

e. Grid dissipation at the tube terminal.
If the amplifier were unmodulated, what would be the maximum allowable
r-f power output, assuming that the plate dissipation is the limiting factor, and
that the plate-circuit efficiency remains constant?
12-14. The results on a plate-modulated class C amplifier are given in the
figure.
Suppose that this modulated amplifier is operated at 1,600 volts d-c,
d.

10

Type SIC

>

'It

Q
0.6
0.5

0.4
0.3

^
o

0.2
O.I'

400 800

1200 1600 2000 2400 2800 3200 3600

with 75 per cent modulation. Assume a constant plate efficiency of 60 per cent.
Calculate the following:
a. The power supplied by the d-c plate source.
b. The power supplied by the audio source.

The r-f carrier power.

The r-f sideband power
e. The resistance of the class C amplifier to the modulating source.
/. The maximum and minimum instantaneous voltage between
cathode when the carrier is unmodulated.
c.

d.

g.

Repeat

(J) for

m=

0.75.

plate

and

RADIO ELECTRONICS

346

[Chap. 12

12-15. The essential elements of a plate-modulated class C amplifier are illustrated in the diagram. The tubes are operated under the following conditions:

6N7

843 tube:
fe
7b

1.2

Mc

= 350 volts
= 30 ma d-c

d-c

tube:

En, = 300 volts d-c

Effective plate-plate resistance 8,000

ohms

843 tank circuit at 1.2 Mc is 10,000 ohms.

843 tank is 15.
The carrier output is to be 5 watts 100 per cent modulated.

Impedance

of

of

Effective

Calculate the following:

a.
b.
c.

The plate-circuit efficiency of the 843.

The power required from the 6N7.
The load impedance presented by the 843

to the secondary of the modulating

transformer.
d. The transformer ratio that should be used.
e. If the output is to be down 1 db 5,000 cycles off resonance, what must the
effective Q of the 843 tank circuit be?
/. The value of C.
g. The approximate value of the r-f choke RFC.
h. With a loaded Q of 15, calculate the value of L and C of the 843 tank circuit.
i. If the actual resistance in the 843 tank is 5 ohms, how much resistance is
coupled into this tank from its load?
j. What is the efficiency of power transfer from the tank to the load?

12-16.

An 806 tube is to be used as the plate-modulated class C power amplifier

It operates from a 2, 500- volt power supply.

Carry out the
design of this and the associated transformer-coupled class B modulator. The
design must fulfill the following specifications:
a. The plate dissipation is not to exceed 150 watts.
6. The transmitter is to be plate-modulated 70 per cent.
c. The distortion in the modulating envelope is not to exceed 10 per cent.
d. The transmitter frequency is 2 Mc.
e. The grid bias may be obtained with grid leak, fixed bias, or a combination of
both.
/. Specify the plate supply potential, grid bias, and excitation potential.
of a transmitter.

g.

h.

The output

is to feed a 76-ohm antenna.

Specify the primary inductance, secondary inductance, tuning capacitance,

AMPLITUDE MODULATION
and mutual inductance.

Assume that the unloaded Q

of the coils

347
is

300 and

that the loaded Q is 12.

i. Give the modulation-transformer requirements.
Be sure that any assumptions that are made are clearly stated.
12-17. The characteristics of an 802 tube when used as a grid-bias modulated
Calculate for to = 1.0 the following:
class C amplifier are given in the sketch.
a. The power supplied by the d-c source.

The r-f carrier power.

The sideband power.
d. The plate efficiency.
e. The plate dissipation.
/. The maximum and minimum instantaneous
plate when the carrier is unmodulated.
b.
c.

g.

Repeat

h.

What

(/) for

is

m =

potential between cathode

and

0.75.

the amplitude of the a-f signal to achieve the desired degree of

modulation?
i.

j.

What
What

Choose

is
is

the a-f driving power?

the grid-cathode power loss?

RL =

12.6

ohms;

Q =

sou-

26.5.

A+SO

Gria vo/ts

1.0

0.9

At
0.8

a 0.7
E

"OS
I

0.5

0.4

0.3

0.2
0.1

-175
-125
-150
-75
-100
Instantaneous a-f grid volfs

-50

12-18. Consider a class C amplifier that is modulated 100 per cent. Assume
the output power to be 1,000 watts. Assume reasonable values of plate-circuit
efficiency for each of the tubes involved, and calculate (1) the total power required
by the modulated tube, (2) the total power required by the modulator tubes,
and (3) the over-all plate-circuit efficiency when the amplifier is:
a. Plate-modulated, using a class A modulator.
6. Plate-modulated, using a class B modulator.
c. Grid-bias modulated.
Repeat the calculations for zero per cent modulation.

CHAPTER

13

DEMODULATION
13-1. Introduction.

When

a radiated modulated carrier signal reaches

the receiving point, the signal, or intelligence, must be extracted therefrom. The process by which the signal is recovered from the modulated
is broadly known as demodulation or detection.
Before considering methods of detection, it is desirable that the
As already
processes of modulation and demodulation be reviewed.

wave

discussed in Chap. 12, the process of modulation may be considered to

be one in which the signal frequencies co m which are specified relative to
,

the zero-frequency reference level, are shifted upward on the frequency

scale and the sidebands are symmetrically disposed about the carrier

frequency

a> c .

This frequency shifting

is

accomplished by mixing the

signal-frequency group, centered about the zero frequency, with the

carrier frequency in appropriate multiplying circuits.

nonlinear characteristic

is

essential in the

The

existence of a

modulator in order to

the mixing or multiplication of the two waves.

In the process of demodulation the signal spectrum, which

is

effect

centered

about co c is shifted downward on the frequency scale so that it is centered,

once again, relative to the zero-frequency level, thus returning it to its
This frequency shifting is accomplished by
original frequency position.
mixing the signal-frequency group which is centered about the carrier
frequency co c with the carrier frequency co c in appropriate multiplying
The existence of a nonlinear characteristic in the demodulator
circuits.
is essential in order to effect the mixing or multiplication of the two waves.
Observe, therefore, that both the modulating and the demodulating
processes involve frequency shifting; both frequency shifts are by an
amount u c and both processes are accomplished in circuits which possess
,

nonlinear characteristics, in order to effect multiplication of the waves.

In fact, similar circuits are used in certain cases for both processes,
although certain essential differences exist. In the modulating process
the carrier signal is generated in one channel, and this is combined in the
modulator with the audio signal, which has been generated in a second
348

DEMODULATION

Sec. 13-2]

channel.

349

In the demodulating process, the required carrier wave is

modulated carrier, and no separate

ordinarily contained in the incoming

carrier-generating circuit

is

necessary.

from the incoming wave, as

is

Of course,

if

the carrier

is

missing

true in suppressed-carrier transmission,

is

necessary that a separate locally generated carrier be

made

it

available

the original signal frequencies are to be extracted. This matter will

be discussed later.
It is customary to consider frequency changing, which is the process
of eliminating the original carrier from the modulated signal and subIn essence, therefore,
stituting for it a new carrier, as demodulation.
frequency shifting in which the signal frequency does not play a direct
part is also classified as demodulation.
13-2. Plate Detection.
Detection is possible when a modulated potential is applied to the grid of a tube which is biased to the nonlinear portion
The general features of such plate detecof its transfer characteristic.
The output curve
tion are made evident by an examination of Fig. 13-1.
if

eg *EC (I+m sin co t)sin cuc t

Fig. 13-1. Sketch showing the operation of a square-law detector.

clearly

shows the presence

component that varies

Of course the h-f carrier component
readily accomplished by using an appropri-

in the output of a

at the modulating frequency rate.

must be eliminated, but

ately placed low-pass

To examine

this is

filter in

the circuit.

the operation of the circuit analytically,

it

will

that the transfer curve with respect to the operating point

be supposed
is a simple

square-law characteristic of the form

(13-1)

Suppose that the signal

e

is

an a-m wave

= E

c (l

+m

of

the form

cos u m f) cos w c t

(13-2)

RADIO ELECTRONICS

350

Then the output current

+ m cos m
aE
= ~ (1 + m cos u m
=

ip

aEl(\

cc

t)

the terms

will contain

cos 2

[Chap. 13

o> c t

+ m cos

o>

t)

mt

(l

+ 2 cos 2<a^ +

cos

2co c t)

_4 + zL4
2"

cos

cos

&Uc + um )t +

cos 2o> +
+
4

-5- cos (2co c

+ _ cos

2w m )t

ct

(2co c

2oi m l

cos

(2co c

w m )t

2a> m )t

(13-3)

number of steady comnumber of components of

Clearly, there will be included in the output a

ponents, a term of modulating frequency, a

sums and differences of the carrier and side freand a number of components of twice the carrier and side

frequencies equal to the

quencies,

frequencies.

network is used which attenuates all components except those in the neighborhood of the modulating frequency
co m
The terms that appear in the output will then be
Suppose that a

selective

ii

= amEl

cos a m t

+ ^ cos

term plus one

which consists

of the desired

frequency.

the second-harmonic amplitude

If

say, 10 per cent of the fundamental,

it is

of

(13-4)

2w m t

second harmonic of this

is

to be kept smaller than,

necessary that the modulation

index be less than 0.4 for sinusoidal modulation. Despite this limitation,
such detectors have been used extensively with generally satisfactory
results, probably because the average modulation of the ordinary radio
program is of the order of 40 per cent.
It is of interest and significance in subsequent work to examine those
terms in the neighborhood of the second harmonic of the carrier 2a> c
although these results are not of importance at this particular point.
The terms are
,

aEl

[0 + tO

cos

2oi c t

cos

+m

(2o) c

cos

(2co c

2u m )t

+
^-

oi

m )t

cos

+m
(2co c

cos (2o c --

2w m )t

um )t

(13-5)

DEMODULATION

Sec. 13-3]
If

m is small,
ii

which

351

this equation can be written as

aE

[cos

2u c t

-f-

cos (2 c

u m )t

+ m cos

(2a> c

<o

m )t]

is
l2

= aE*

(1

2to cos u m t) cos 2u c f

(13-6)

if one were to use a tuned circuit at the output of a square-law

which is tuned to 2co c the output would be of second harmonic of
the carrier but the modulation frequency would be unchanged.
Frequency doublers of this type find extensive use in h-f operations.
13-3. Diode Detection.
Diode detectors have almost completely sup-

Therefore,

circuit

planted other types of detectors in

home

radio-receiver use.

They

operate very satisfactorily, although the modulated input to the detector

Fig. 13-2.

The

circuit

and the operational

in excess

must be large

may

characteristics of

of 1 or 2 volts.

result because of the nonlinear

an average diode detector.

Otherwise excessive distortion

of the diode cir-

dynamic curve

Since the diode detector possesses characteristics in common

with ordinary rectification, it is possible to present a qualitative discussion of its operation on this basis.
There are two important types of operation of a diode detector. One
cuit.

and the other is envelope detection. The

average detection is illustrated in Fig. 13-2. Also illustrated
It will be
in this figure is the character of input and output waves.
observed that this is just a simple diode rectifier without a filter in the
is

known

as average detection,

circuit for

output. The application of a modulated wave to the circuit yields an

output the average value of which contains the modulating frequency.
It follows from simple rectifier theory (refer to Chap. 3) that the output
potential

is

of the

form

E" =

^o^^
+

ir(,/t

rv )

+ m cos aj)

(13-7)

RADIO ELECTRONICS

352

The use

[Chap. 13

of appropriate circuits permits the modulating-frequency

com-

ponent to be extracted from the output.

The circuit for envelope detection is illustrated as Fig. 13-3. It will
be observed that such circuits are essentially those of a simple diode
The operrectifier with a capacitor filter.

^X

Xj

yC'

f)

same as that

^ |D

Fig. 13-3.
detection.

ation of this circuit

substantially the

is

for the simple rectifier, with

the additional fact that the a-c input potenWith the proper
tial varies in amplitude.

circuit for envelope

filter capacitor, the waveforms

n pU t potential, the output potential,
and the tube current are roughly those shown in Fig. 13-4. Since the
carrier frequency is much greater than the modulating frequency, the

choice of the

jagged appearance of the curves is considerably worse than is actually

the case. In effect, therefore, the output potential of the detector follows
the envelope of the modulated input wave and has the waveform of the
modulating potential.
Output potential

Modulated
input potential

"a

aed

mmtlillilith

x\\\f

Fig. 13-4. Waveforms of the input potential, output potential, and plate current in a
diode-detector circuit.

13-4. Analysis

of

Diode Detection. 1

An

analytic

gives certain of the properties of the diode detector

is

expression that

possible

if

several

reasonable assumptions are made. In particular, it will be assumed

that the d-c potential across the diode load, which arises from the tube
current during the portion of the cycle when the input potential exceeds
the potential across the output capacitor, will remain constant over the
It will also be assumed that the static charactercarrier-frequency cycle.

the diode is linear. The conditions that apply during one cycle
impressed modulated r-f signal wave are as illustrated in Fig. 13-5.
The equation of the assumed characteristic is

istic of

of the

n =
4 =
But the instantaneous

= g*.

eb

>0

eb

<

potential applied to the diode

(13 _ g)

is

given by the

DEMODULATION

Sec. 13-4]

353

equation

= E

eb

which

c (l

+m

cos u m t) cos w c t

Ea

written for convenience as

is

eb

E' cos w c t

Ea

(13-9)

where E'

Ea

is

is the instantaneous amplitude of the modulated carrier and

the average value of the rectifier carrier signal and appears across

Ec fl+ma cosafa t)cos

Fig. 13-5.

the capacitor.

The

=
=

ib

shift

of a diode detector.

plate current will be of the

ib

The

The approximate action

ojc t

g P (E' cos

o) c t

Ea

form

>
<

eb

eb

(13-10)

from one equation to the other occurs at the angle

that angle defined

8,

which

is

by

E' cos u c t

The instantaneous

current

ib

will

E' cos 6

= Ea

contain a

number

(13-11)
of

harmonic com-

ponents, but those of interest are the d-c or average component and thn
a-c term at the driving frequency.
The d-c component of ib is given by

Ib

which

is,

*k Jo
I

because of the symmetry that

1
IT

This

may

dd

exists,

f
Jo

be written as

h=

1
i"

Jo

g p (E' cos 9

~ Ea

dd

(13-12)

RADIO ELECTRONICS

354

[Chap. 13

which becomes

h=^

- Ea 6)

(E' sin 6

(13-13)

7T

This becomes, by Eq. (13-11),

h = ^ '(sin 9 -

cos

61

(13-14)

0)

TT

Also,

it is initially

assumed that

Ea =
By

IbR

combining Eqs. (13-14) and (13-15), there

T

It

Ea =

= ^

E' cos

g p E'

li

results

(sin 9

cos 9)

(13-15)

IT

from which

= -

(tan

(13-16)

61)

This expression shows that there is a direct functional relationship

between the operating angle 6 and the ratio of the plate resistance of the
diode to the external resistance.
of r p /R

The

is

very

An explicit expression for as a function

may be given graphically.

but the information

difficult,

detection efficiency of the diode detector

of the average value of the load potential

That

is

defined as the ratio

to the

peak

a-c input E'.

is,

This

Ea

may

7j

cos e

(13-17)

be written, by Eq. (13-15), as

n

so that

= |? =

(sin

e cos 0)

(13-18)

appears as a function of the ratio R/r p

Clearly, therefore,

the detection efficiency for a specified choice of diode and load,

i.e.,

for

obtained from a plot of Eq. (13-18), which is

contained in Fig. 13-6. Observe that since r p /R is a constant of the
circuit, the conduction angle 9 is likewise a constant for the circuit.

a specified ratio rp /R,

is

This means, of course, that

or the amplitude E c

6 is

independent

of

the degree of modulation

The detection efficiency is readily computed for the case when the
by-pass or output capacitor is not used, i.e., for the average detector.
In this case, the output potential is given by Eq. (13-7) and is

E =

~~Td~\
x{K

rp )

E '( 1 + m

cos

"J)

(13-19)

DEMODULATION

Sec. 13-4]

when the peak value

is

W
The

detection efficiency

is

= E

c (l

+m

(13-20)

,(lf^

detection efficiency for this case

mum

cos ooj)

the ratio of Eq. (13-19) to (13-20) and

'

The

355

is

<

13 - 21)

seen to approach 1/x as a maxi-

is

value.

1000

Fig. 13-6. Various important terms in the analysis of diode detectors.

I.Or

50

10

500

100

1000'

R/rP
Fig. 13-7. Variation of the detection efficiency with a>RC as a parameter.

It is possible to

compute 2 the values

of detection efficiency for values

the value specified as ooRC = oo in Fig. 13-7, as specified

in Eq. (13-18), and the value o>RC = 0, the value specified in Eq. (13-21).
of

C between

The curve

for

uRC =

is

also

shown

in Fig.

13-7.

The

designer's

RADIO ELECTRONICS

356

[Chap. 13

is to choose as large a value of C as possible, consistent with no

diagonal clipping (see Sec. 13-5). The other curves of Fig. 13-7 will be

problem

discussed later.

Another quantity of importance in detector operation is the power

absorbed by the detector, or the power loss in the diode circuit. To
evaluate this requires a knowledge of the fundamental-frequency component of the current. The maximum value of the current is given by
the Fourier coefficient
2*

-1

Ipim

cos

it

cot

d(wt)

JO

IT

This becomes, writing a as the parameter of integration,

I vim

= - (

g P (E' cos a

him =

- Ea

cos a da

(13-22)

which integrates to

or

The power input

to the diode and

p =

ei

fJ

>>

its

load

E'

may

(0

is,

sin 6 cos 0)

(13-24)

defined

is

by the

relation

7T

^2P

(13 " 2o)

- sinT^sT)

g p (6

be written as

r p(i

-z

Re

That

is

Li

effective resistance of the diode circuit

which

(13-23)

0)

dt

Lt

Re

cos

sin

P =
The

(B

^-z
sin

(13-26)

cos 5

the effective resistance in parallel with the capacitor due to

the loss in the diode circuit

equal to

is

/3rp .

A plot

of

/3

is

also contained

in Fig. 13-6.

By combining

Eqs. (13-26) with (13-16), an expression for the equivais possible.

This is

lent resistance that shunts the diode input circuit

R
This expression

is

rp

tan

rp

sin

cos

plotted as a function of R/r p in Fig. 13-6,

DEMODULATION

Sec. 13-5]

357

For the case when is high, the equivalent input resistance reduces to
Noting that for high 6 is small,
17

a simple form.

77

sin e

cos

0=1

P
2

and
Re

R
Hence the

cos

sin
6

cos

sin 9 cos 9

(13-27)

2,

becomes R/2rj for large 77. Since

than one-half the load resistance.
13-5. Distortion in Diode Detectors.
There are two sources of distortion in a simple diode detector.
One results from the curvature in the
tube characteristic, making the efficiency of rectification vary according
to the amplitude of the envelope.
This source of distortion may be
minimized by making the load resistance large compared with the diode
plate resistance and by making the amplitude of the carrier envelope
applied to the diode reasonably large.
Under practical conditions, when
the detection efficiency exceeds 80 per cent, the distortion from this source
is of the order of 2 per cent for a completely modulated wave.
With
small signals the distortion may reach as high as 25 per cent for a completely modulated wave when the signal potential is a fraction of a volt.
The second source of distortion arises from the fact that the potential
across the capacitor in the output can die away only as fast as the charge
can leak off through the load resistor. Hence, unless the time constant
of this circuit is properly chosen, clipping may result during the troughs
of the modulated signal.
If the h-f variations of the output
of modu/afed rave
potential are to be small, the time
77

1,

then

effective input resistance

is

slightly greater

constant of the load circuit

RC must

be large compared with the period

of
the
carrier-frequency
cycle.
if this value is made too
the output potential cannot

However,
large,

decay as rapidly as the envelope

decreases,

and clipping occurs.

The

Fig. 13-8. Diagonal clipping in a diode

detector when the load-circuit time
constant is too large.

conditions discussed are illustrated in Fig. 13-8.

To ascertain the maximum allowable value of the time constant, it

should be noted that this value must be such as to permit the capacitor
to discharge at the same rate as the decrease of the modulation envelope.
This

may

be estimated in the following manner. 3

The most unfavorable

condition occurs at the highest modulation frequency w m that the detector

is

designed to handle and

is

that for which the equation of the envelope

= E

c (l

+m

cos caj)

is

(13-28)

RADIO ELECTRONICS

358

At any

particular time

to,

[Chap. 13

the value and the slope of the modulation

envelope are

= E

c (l

-a)

If

t a)

(13-29)

m mE c sin

co

mt

the potential across the capacitor equals the modulation potential at

the time

U,

= Ec (l

ea

and

it

The

+ m cos <

(13-30)

decays thereafter according to the exponential expression

ea

initial rate of

change

(^f)t

To

+ m cos u m

eri0-v-*> /BO

(13-31)

is

=
= ~
WJ e" ~ WJ ^

+ m cos Wmt ^

(13-32)

avoid the diagonal clipping illustrated in Fig. 13-8, the capacitor

must be less than the value of the envelope for time t > to, and

potential

the slope of

must be

ea

than that of the envelope at

less

to-

This

requires that

m cos wm

(1 -f

^ wmmE

t )

sin

wmt

or
1

m sin m
+ m cos w m
o>

RC

/10 .

to

tt,

For the initial rate of decay of the capacitor potential to be greater

than the rate of decay of the envelope potential, it is necessary that
^

RC >

0>n,

m sin w m
+ m cos

to
co

m<

But the most severe condition on the RC constant is that for which the
is a maximum.
To find this, consider the expression

fraction

d
dt 1

m sin u m
+ m cos

to

oj OT jf

This yields
cos u m to
sin

umto

= m
= y/l m

from which

J_ >
RC
If this

JUL

\A - m

(13-34)
2

is satisfied, the output potential follows the waveform of

According to this equation, as the modulation approaches

equation

the envelope.

co

DEMODULATION

Sec. 13-6]

359

100 per cent, the required time constant approaches zero.

Consequently,

at 100 per cent modulation, the output potential contains the carrier as
well as the modulating frequency.

By taking into account a number of factors that were neglected in the

above analysis, such as the impedance of the source supplying the modulated potential, the results must be modified somewhat.
Experimentally
it has been found that the amount of harmonic generation is not excessive
for sound reproduction if

^
This equation expresses the relation

co

mm

among

(13-35)

the circuit parameters which

permits the detector to follow the modulation envelope.

13-6. Diodes with Complex Load Impedance.
A detector stage

normally provided with a

filter in

the output in order to prevent any

is

r-f

1AGC
Fig. 13-9. Diode-detector circuit.

potential from reaching the subsequent stages, since only the d-c

and

modulation-frequency terms are desired. The d-c component that is

developed across the diode is ordinarily used for the purpose of automatic
gain control (see Sec. 13-8) the a-c component is fed to the audio ampli;

fiers

to provide the a-c signals at a potential

and power

level dictated

by

the output device.

To

achieve the requisite filtering necessitates the addition of circuit

elements, with the result that the impedance that the complex load offers
to the diode at the modulation frequency
offered to the d-c
filter,

which

is

component.

is

quite different from that

typical circuit, which includes the r-f

usually a simple II-type resistance-capacitance

the coupling network to the

filter,

and

audio stage, is illustrated in Fig. 13-9.

Observe from this circuit that the a-c impedance at the modulating
frequency is smaller than the d-c resistance, owing to the shunting effect
of the filter capacitors

and

first

also because of the fact that the grid resistor

of the next stage parallels the load resistor.

Because

of this, the proper-

the diode rectifier are not precisely those discussed in Sec. 13-5.
Consider the situation for a sinusoidal modulation envelope to the

ties of

A sinusoidal component of current at the

modulation frequency and a d-c component of current due to the carrier

input of the diode circuit.

RADIO ELECTRONICS

360

[Chap. 13

level will exist in the diode circuit.

It is necessary that the negative

should never exceed the d-c component; otherwise the diode will cut off, with an a-c waveform distortion.

peak

The

component

of the a-c

waveform

a-c

have the negative peaks clipped if this happens.

be used as a starting point for obtaining sigthe present connection. It follows from Eq. (13-15)
will

Equation (13-15)
nificant results in

of current

may

that the ratio

Eac

Peak

Edc

which

may

(13-36)

(13-37)

be written as

Peak I ac \Z
IdcR

In this expression Z m is impedance of the total diode output circuit at the

modulating frequency and R is the total diode output-circuit resistance
at the d-c or zero-frequency level.
Now from the discussion given above,
the maximum value that m may have and still avoid negative clipping is
that for which the peak I ac current is just equal to the d-c component
Clearly, the maximum value of m without negative clipping is then
Id c
-

_
Figure 13-9

is

\Z m

_ impedance

of load to

co

resistance of load to d-c

examined

in the light of Eq. (13-36).

For the usual

choice of circuit parameters (see Fig. 13-13 for a detailed circuit) the

reactance of Ci
of

C2

is

is

very low.

high at the modulating frequencies, and the reactance

Approximately, therefore, for this circuit

Ri

^_
Tallin

The

Ri

R2R3
R2

4-

R2

actual choice of the parameters Ri,

between detection

efficiency

and the

Rs

2)

and

is

a compromise

effect of possible nonlinearity of the

diode.
It is clear from the discussion in
from the discussion leading to Eq. (13-38) that the

13-7. Rectification Characteristics.

Sec. 13-5

and

signal output

which

also

from the diode detector

consists of a d-c potential

Ea

the average value of the rectified carrier signal, and this appears
across the capacitor C, and an a-c term of modulating frequency, this
is

being a measure of the amplitude of the envelope of the incoming moduThus one might consider C as an effective by-pass for
carrier-frequency currents, with two potential components across the
lated carrier.

output, the d-c and the a-c terms.

the amplitude of the

r-f potential,

The quantitative relationships among

the average rectified current, and the

DEMODULATION

Sec. 13-7]

average plate potential

361

Ea are contained in the rectification characteristic

and the results on a

the a-c potential
circuit
shown,
In
the
6H6
potential
E
while
the
bias
maintained
constant,
E is
a which simulates
current is
rectified
and
the
resistor,
is
varied,
the
load
the drop across
of

the diode.

circuit for obtaining these curves

are illustrated in Fig. 13-10.

read on the microammeter.

Indicated on this characteristic are a series of load lines, which correspond to different static load resistances R. These are drawn on the
curves in the customary way. Their use permits one to obtain the
output potential as the carrier potential varies. Thus, by plotting the
current as a function of time as the carrier amplitude varies because of

-40

-30

-20

Ea (d-c
Fig. 13-10. The rectification characteristic of a
determination.

the modulation,

it is

6H6

diode,

-10
volts)

and the

circuit

possible to determine both the output

used for

and the

its

dis-

tortion.
For example, if the unmodulated carrier amplitude is 10 volts
rms, with the modulation causing the amplitude to vary between 5 and
15 volts rms, with, say, R = 100 kilohms, the load current will vary
between approximately 60 and 180 /xa, with the output potential varying

between 6 and 18

Attention

is

volts.

The

detection efficiency in this case

is

called to the fact that, in addition to the several static

drawn on the curves of Fig. 13-10, there is also shown

an "a-c load line." The line shown represents a 0.25-megohm static load
that is shunted through a coupling capacitor by a second 0.25-megohm
At the
load, which might be the grid resistor of the first audio amplifier.
audio frequencies, the actual load is the parallel combination of these two
load lines which are

RADIO ELECTRONICS

362

[Chap. 13

resistors.
The circuit of Fig. 13-13 indicates a static load of 250 kilohms
with a dynamic load consisting of the series combination of a 50-kilohm
resistance and a 250-kilohm resistor paralleled by two 500-kilohm
resistors.
The a-c load line illustrated has been drawn for an effective
value of the unmodulated signal of 10 volts. Observe that the load line
does not pass through the point (0,0). Consequently, if the operation is

no current, severe distortion may result. The

curves of Fig. 13-11 show the char-

carried to the region of

'*
l

'shunting

acter of the variation of the distor-

RL = 0.5

tion with per cent modulation.

13-8. Automatic

Gain

Control.

The average amplitude of the modulated carrier wave that reaches the
detector stage will depend upon a

number

of factors, including the field

strength at the receiver of the station to which the receiver is tuned,

20
60
40
80
100
Per cent modulaf ion
Fig. 13-11. Variation of distortion with
per cent modulation.
{From F. Langford-Smith, "Radiotron Designer's Handbook," 1st ed., chap. 18, Amalgamated
Wireless Valve Company Ply. Ltd.,
Sydney, Australia, 1941.)

and the propagation conditions between the transmitter and receiver.

It is desirable, therefore, to incor-

porate some means within the receiver for maintaining the average

modulated carrier amplitude at the detector at a constant level so as to

avoid the effects of fading. Such an automatic gain control (age) (often
called automatic volume control, avc) will automatically vary the gain
of the r-f or

i-f stages to yield a substantially constant level at the detecFigure 13-12 shows a block diagram of a receiver incorporating
automatic gain control.

tor.

W
R-famplifier

Defector

converter

a-f amplifier

i-f amplifier

AGC

|}^^

Low pass
filter

Fig. 13-12. Block diagram of a receiver incorporating automatic gain control.

The details of a circuit in which a diode is used both as a linear detector

and to supply a d-c potential for automatic-gain-control purposes are
In this circuit the use of a separate isolating
d) permits the extraction of a d-c potential
that is proportional to the average modulated carrier level. Also, the
d-c potential across the diode load resistance is blocked by the use of a
capacitor-and-resistor combination (R 2 d), the a-f component being
illustrated in Fig. 13-13.
resistor

and capacitor filter (R lt

made

available for the following a-f amplification stages.

DEMODULATION

Sec. 13-8]

363

network RC\ is made

amplitude
correspondlong enough
that
the
automatic-gaining to the modulation, but short enough so
control potential varies with the average amplitude of the carrier, drop-

The time constant

of the automatic-gain-control

to average out the variations in carrier

ping

off

as the carrier fades

and increasing as the

T
X

e_

carrier

becomes stronger.

0.0O025

t-Ht

50 k % 0.00025

500*

AGC*

250*
-^VWV

-"iAAMi

A-f signal

c.
/VW-

_
Fig. 13-13.

500*

diode detector and automatic-gain-control circuit.

used to vary the bias, and hence the transconductance,

amplifier tubes, to maintain a substantially constant
In this way, fluctuations in the average amplitude of the modulevel.
The r-f and i-f
lated carrier delivered to the diode are greatly reduced.
amplifier tubes that have automatic gain control applied to them should
This potential
of the r-f

and

is

i-f

be of the remote-cutoff or variable-mu type; otherwise the system may

be very critical of adjustment, owing to the extremely marked sensitivity
of transconductance with bias of the sharp-cutoff type of tubes.

T
DelayedAGC

%
Fig. 13-14.

t-Awv-f

005

o
o>
o

diode detector and delayed automatic-gain-control circuit.

If it is desired to

have the automatic gain control operate only after

minimum, so that the reception

by the automatic-gain-control system,

the carrier strength reaches a specified

of

weak

signals will not be affected

a biased or delayed automatic-gain-control circuit

may be

cuit with delayed automatic-gain-control potential is

In this circuit the automatic-gain-control potential

shown
is

used.

cir-

in Fig. 13-14.

obtained from a

RADIO ELECTRONICS

364

[Chap. 13

Also, the automatic-gain-control potential

separate diode.

obtained

is

of the previous stage in order that

a potential that
across the cathode resistor of the triode element of the duodiode triode
may be used as the reference level, below which no automatic gain control

from the output

is

applied.

13-9. Superregenerative Detector.

If the quality of reproduction of
a receiver is not critical, and high sensitivity is required from relatively
simple apparatus, the superregenerative receiver possesses considerable
merit.
In such a receiver, regenerative feedback is incorporated in the
The amount of regeneration that is provided sets the
detector circuit.

detector at the point of oscillation, and the smallest signal

of initiating the oscillations.

Once the

oscillations

is

have been

capable

initiated,

they will increase in amplitude until the nonlinear characteristic of the

tube sets the amplitude level, exactly as for the simple oscillator. Moreover, the oscillations will continue to exist, unless a means is provided
The effect of the feedback is to produce a negafor controlling them.
tive input conductance, precisely in the manner of the discussion of
With sufficient negative conductance, oscillations of the type
Sec. 11-14.
discussed in Sec. 11-14 will occur, whence the rate of rise of the oscillation
amplitude, which is exponential, is inversely proportional to the induct-

ance of the tuning

circuit.

Also, the rate of rise

is

found to be propor-

tional to the amplitude of the initiating signal.

To make such a regenerative detector useful, it is necessary to interrupt the oscillations periodically and reset the detector to its sensitive
position.
The usual method of obtaining the desired control is to add an
In this way the circuit
a-c potential to the plate supply of the detector.
can be made to yield negative conductance for at least a part of the half
cycle during which the a-c potential adds to the plate potential, though
The
it has a positive conductance during the remainder of the period.
a-c potential is produced by an auxiliary oscillator, which is known as
the quench oscillator. In the simplest case, the detector may be made
The
self-setting by adjusting the circuit to be of the self-blocking type.
Observe
general character of the operation is illustrated in Fig. 13-15.
that the pulses of oscillation repeat at the quench frequency. A typical
superregenerative detector is shown in Fig. 13-16, in which Tl is the
detector and

When

T2

is

the quench oscillator.

the circuit

is

so adjusted that the

quench

oscillator stops the

they reach the optimum value determined

by the tube characteristics, it is said to be operating in the linear mode.
This is the operation that is illustrated in Fig. 13-15. If the growth of
oscillations continues until they are limited by the tube curvature before
the quench oscillator causes them to decay, the system is said to be operatThe basis for this name is found in the
ing in the logarithmic mode.

growth

of oscillations before

DEMODULATION

Sec. 13-9]

365

above a certain minimum value will produce the same

pulse amplitude, but the area under the pulse envelope is proportional to

fact that all signals

the logarithm of the signal amplitude.

Fig. 13-15.

The

general character of the action of the superregenerative detector.

Suppose that the incoming signal is modulated. The rate of rise of

oscillations is proportional to the amplitude of the initiating signal, and
if the quench frequency is considerably higher than the modulation frequency, the amplitude of the pulses will follow the modulation envelope

Superregenerative

Quench

oscillator
detector
Fig. 13-16. The elements of a superregenerative receiver.

The need for the high quench frequency arises from the
pulses are of varying amplitude and so introduce
quenched
fact that the
It is necessary to choose the quench frequency
additional sidebands.
than the modulating frequency if the added
greater
times
to be several
with the original sidebands and thereby
interfere
to
are
not
sidebands
of the signal.

RADIO ELECTRONICS

366

[Chap. 13

produce distortion in the rectified output. The quench frequency is

ordinarily chosen to be five to fifty times the modulation frequency.
Owing to the sensitive state of the detector, noise potentials due to
thermal agitation and shot effect will also initiate the oscillations and
produce pulses at the quench frequency. Since the initiating noise
As a result,
signals fluctuate, there will be a noise signal in the output.
the noise level of the receiver is high. In addition, external noise which is
received with the signal will alter the value of the total signal at the
The amplitudes of the
instant when the oscillations begin to build up.
pulses are thereby affected, with a consequent noise in the output.
The selectivity of the superregenerative receiver is low, owing to the
loading of the input tuned circuit of the detector by the low conductance.
Moreover, the selectivity cannot be increased by the addition of r-f
amplification because the regenerative feedback from the detector will
cause shock excitation of the added tuned circuits. For short-wave
applications this lack of selectivity is actually desirable, both because the
available frequency band is great and tuning is necessary over a very wide
band, and because many transmitters might not be very stable.
13-10. Suppressed-carrier Demodulation. Some general comments
concerning suppressed-carrier transmission were given in Sec. 1-5. Also,
it was shown in Sec. 12-3 that the output from a balanced modulator
which was arranged to suppress the carrier contained the sidebands of an

a-m

carrier.

Clearly, the transmission of suppressed-carrier signals

is

accomplished at high efficiency since the greater part of the power of an

a-m wave is in its carrier, and with its suppression a saving in power is
effected.

The

process of demodulation for suppressed-carrier transmission proves

to be rather difficult.

The production

of a carrier at

low power in the

it and the
It is important

receiver which will maintain the proper phase relation between

sidebands over a period of time is extremely difficult.

to examine the effects of phase shifts between the carrier and the sidebands in order to evaluate the suppressed-carrier system of transmission.
The situation may best be examined by reference to the sinor representation of the process.

Consider the particular case when the reintroduced carrier is 90 deg

its required phase position for the exact reproduction of the
modulated carrier. Also, it is supposed that the carrier amplitude is
large compared with the sideband amplitudes, as it would be if the modulation index were small.
The situation is illustrated graphically in Fig.
13-17 for several positions during the modulating cycle.
An examination

away from

of these figures

shows that the resultant potential E R does not vary

However, Er does undergo

appreciably from the reintroduced carrier.

periodic phase variations with respect to

c,

since,

during one half cycle

DEMODULATION

Sec. 13-11]

367

Under these
and during the other half cycle E R lags Ec
frequency
constant
in
but varies
conditions the resulting wave is no longer
frequency
variation
is accomat the modulation-frequency rate, and the
panied by a cyclic change in amplitude.
The effect of phase shifting of the carrier of an a-m wave is the reduction of the modulation index and the introduction of a phase or frequency
modulation. When the carrier phase shift is 90 deg, the amplitude

ER

leads

c,

modulation may almost disappear

and the phase modulation that reIn fact,
sults may be considerable.
this latter case is the

method for pro-

ducing a p-m wave as one step in the

production of a f-m output in the
Armstrong system of frequency modulation,

Chap.
If

which

will

be discussed in

14.

the reintroduced carrier

is of

from the original

carrier, there will be a beating with
the sidebands and only a transient
different frequency

condition will prevail.

For the proper reproduction

of

the original signal, the carrier phase

must be maintained,

relative to those

of the sidebands, to within a fraction

of a cycle.

mitted,

If a pilot carrier is trans-

control

circuits

might be

possible to achieve the required fre-

quency and phase stability between

the pilot and the locally generated
If no pilot carrier is availcarrier.

Fig. 13-17. Diagrams showing the

instantaneous amplitude of a reintroduced carrier which is 90 deg out of
phase with the sidebands.

then the frequency stability of

both the transmitting and receiving
oscillators must be controllable with a very high degree of precision.
This latter condition cannot be achieved even with present-day techConsequently, completely suppressed carrier transmission is not
niques.
able,

practical.
If a supnetwork or

13-11. Single-sideband Suppressed-carrier Demodulation.

pressed-carrier signal were passed through a suitable filter

through appropriate circuits so that one of the sidebands was eliminated,

The demodulation
the resulting wave would be a single-sideband signal.
of such single-sideband signals proves to be feasible, and such a communiActually, of course, the supprescation system is used commercially.

RADIO ELECTRONICS

368
sion of the sideband

is

[Chap. 13

accomplished at the transmitter, so that only a

single-sideband suppressed-carrier signal

is

radiated.

For a given

signal

strength at the receiver, this system has the same power economy as the
suppressed-carrier system.
In addition, it effects a saving in the band-

width required in the r-f spectrum.

In the case of a pure tone, the reintroduction of a carrier uc to the
signal of frequency w c + co m (assuming that the upper sideband co c + oi m
is transmitted) will return the frequency w m in the output.
However,
it is necessary that the amplitude of the reintroduced carrier must be
properly chosen if the envelope of the reconstructed signal is to be a
Moreover, the phase of the reintroduced
replica of the original signal.
carrier and that of the sideband are important, since the resultant amplitude will depend on the relative phases of these component waves which
are to be combined.
If the signal is a complex wave, as is usually the case, the effects of
the frequency and the phase of the reintroduced carrier become more
The general effects of an inadequate
difficult to discuss qualitatively.
carrier amplitude and incorrect carrier phase are a distortion of the
resulting amplitude, owing to the different components contained in the
In general, the phase of
signal being affected by differing amounts.
the different frequencies will be shifted by different amounts. Also,
the amplitudes may be differently affected.
The shape of the resulting envelope depends on the relative phases of
the different components of the signal hence the resulting waveshape will
vary with the phase of the reintroduced carrier. However, since the
human ear recognizes only amplitude and frequency, distortion due to
phase will not be recognized. Consequently such single-sideband sup;

pressed-carrier operation

In fact,

is

generally satisfactory for speech transmission.

the reintroduced carrier is altered in frequency, the pitch of the

signal will change, because the signal frequency will not be translated
A translation of several cycles from the
to the proper zero position.
zero

is

if

not particularly objectionable.

While a translation

of several cycles might not influence too seriously

a speech signal, it will seriously affect the transmission of music. This
follows from the fact that, for music, the fundamental and its harmonics
are important in presenting the tonal qualities of each instrument.
If
the frequency and amplitude of the reproduced harmonic components are
altered, the character of each instrument will be lost.
Because of its limitations, single-sideband suppressed-carrier transmission is limited in its application to commercial telephony. It is used
extensively in point-to-point commercial communication. 4
13-12. Mixers and Converters.
A superheterodyne receiver (see Sec.
1-8 and also Prob. 13-8) incorporates a mixing element, a device in which

DEMODULATION

Sec. 13-13]

369

combined with the signal from a local

from one frequency to another.
The use of such a device permits the carrier level of any signal to be
shifted to a preset i-f value and then to provide i-f amplification to bring
the incoming modulated signal

is

oscillator in order to shift the carrier level

the potential level to the 10 volts or so desired at the input of the diode
detector.

The use

of the

superheterodyne

vides a higher selectivity than a tuned

is

widespread because it proprimarily because of the

r-f circuit,

use of double-tuned amplifiers.

As was noted
or converter

is

more may

in Sec. 7-2, the equivalent noise resistance of a

Owing

relatively high.

mixer

to this, a noise potential of

1 fiv

on the grid. Moreover, since the second detector

potential level is perhaps 10 volts, the maximum gain that can be used
between the mixer grid and i-f amplifier output is then determined by
For example, if a signal/noise
the allowable signal/noise potential ratio.
ratio of 100 is required, the largest useful mixer-amplifier gain is about
Since some gain occurs in the mixer and is roughly 25 per cent of
10 5
or

exist

that of the

i-f

amplifier stage,

two

i-f

amplifier stages will provide all the

requisite gain.
If

a separate local or beating oscillator tube

the combining

is

accomplished

is

used, the tube in which

is

called a mixer.

If

a multigrid tube

is

used to serve as both the local oscillator and the mixing element simulIn both cases the effectiveness
taneously, it is referred to as a converter.
with which mixing is accomplished, i.e., the ratio of the i-f current in the

output to the signal potential input to the circuit, is an important quanThis quantity is called the conversion transconductance and is, by

tity.

definition,

=
~

dib, i-f

E'

de c

(13-39)

r-f

It should be noted that this quantity is quite different from the mutual
conductance of the tube.
13-13. Square-law Conversion.
To examine certain of the aspects of
the conversion process, it will be supposed that the output from the local
oscillator is combined with the modulated carrier potential, and this combined potential is supposed impressed directly on the grid of a square-law
If the tube characteristic is represented by an expression of the
tube.
form

(13-40)

ae\

then with the application of the potential

eg

= E

c (l

+m

cos

oo

m f) sin u c t

sin

o>o<

(13-41)

RADIO ELECTRONICS

370

[Chap. 13

the following terms will appear in the output:

sin
+ El sin
+
+ wm
-.[ El sin
m EJ sin
u m )t + 2E E sin u sin u
|
+ mEl sm e< sin ( + w m)t + wifi sin sin
mE
E(,
sin
sin
(o + u m )t + mE E sin w sin
+
m
E
+ sin
+ u m )t sin ( - w m )t
2

co c <

co

(a> c

-]

'

(co c

ct

oj <

(co c

co c i

)t

w m )<

(<.

a>,)2

(a) c

(13-42)

By the use of frequency-selective circuits, all terms will be eliminated

except those having frequencies in the neighborhood of co o> c = u,.
Thus
i

there will remain in the output the following:

= a[E Ea
c

cos

out

+ %mE E<i cos

c

(w<

w m )t

+ HmE E
c

which

may

cos

( Wi

oi

m )t]

(13-43)

be written as
i

= aE E
c

+m

(l

cos w m <) cos

(13-44)

wrf

That is, the only signal that can get through the tuning circuits which have
been tuned to the i-f frequency and which have a bandwidth sufficiently
wide to accommodate the a-f spread is essentially the modulation amplitude at the i-f frequency.
As noted in Sec. 13-1, frequency changing is classed as detection, whence
the converter

is

frequently called the "first detector."

a detector only in the sense that

it

Actually it is
permits obtaining output frequencies

that are different from the input frequencies

the frequency from the

The problem
the input

r-f

of

r-f level

to the

i-f

or, rather,

that

it

shifts

level.

keeping the difference between the local oscillator and

frequencies constant as one varies the antenna tuning from

a position corresponding to one end of the band (say 550 kc) to one
corresponding to the other end of the band (say 1,600 kc) is not a simple

matter

if

one wishes to adjust a single control.

It requires careful con-

struction of the variable capacitors and the choice of constants so that

they "track" together.

the entire band.

In general, perfect tracking

Ordinarily provision

is

is

not possible over

possible for ensuring perfect

tracking at only three specific points.

The errors over the intervening
ranges are not great enough to throw the beat frequency out of the i-f
pass band.
13-14. Generalized Conversion Theory. 5
it

was assumed that both the broadcast

signal

In the foregoing discussion

and the

local oscillator sig-

nal were impressed on the grid of a square-law amplifier.

circumstances the conversion

is

distortionless.

To

Under these

reduce the interaction

Sec. 13-14]

between the two

modern

DEMODULATION
coupling

circuits, electron

371

is

ordinarily

Consequently the only coupling

practice.

is

employed

in

that through the

electron stream.

Under the assumption that the

signal potential

is

small and that the

local oscillator potential is large, the signal-electrode

transconductance

be considered as a function of the oscillator potential only. Then

the signal-electrode plate transconductance g m may be considered as
varying periodically at the oscillator frequency. The situation is then

may

Fig. 13-18. Illustrating the variation of plate-grid transconductance of a converter

tube with a large oscillator potential on the grid.

somewhat

gm

When

Because of the periodic variation

be represented by a Fourier series of the form

as illustrated in Fig. 13-18.

that occurs in g m this

a small signal

~2

is

may

bi cos

b 2 cos 2

co <

(13-45)

applied to the tube, the resulting a-c plate current

has the form

iP

This

may

gmE' COS

(13-46)

0> c t

be written in the form

ip

E' cos w c t

b n cos nuatj

or

ip

^ E' cos

oi c t

E'

cos w c t cos uot

b n cos ncoot cos

co c

(13-47)

RADIO ELECTRONICS

372

For a

circuit

tuned to the frequency

co

ui c ,

the

[Chap. 13

i-f

frequency, the output

is

ip,i-f

^ E'

cos

(coo

whence the conversion transconductance

Upon combining

gc

(13-48)

out

is

E'

(13-49)

known form

this with the

E' cos

9c

61

u c )t

for 6 1; there results

Q m cos wot d(w4)

2^

(13-50)

This expression indicates that the value of g c depends on the magnitude of

the mutual conductance at the operat-

way

ing point and also on the

conductance

is

varied

by

that this

the local

oscil-

lator potential.

The continued development

(13-50)

is

possible

of

Eq.

by examining curves

showing typical variations

of the plate-

grid transconductance with the oscil-

These are given

found that in most
6 UrVeS f r
Practical cases, the (g m ,e c ) curves can be
tube
represented with reasonable approximation over the operating range by an expression of the form
lator bias potential.

Oscillator bias volts

Fig.

13-19.

in Fig. 13-19.

transconduc
different types of

Typical

gm

g m o (l

It is

with 0.5

<

Suppose that the

(see Fig. 13-18 for notation).

<

1.5

(13-51)

oscillator potential is of

the form
ec

Eom cos u

(13-52)

Equation (13-50) then becomes

2g m0

2x

For convenience,

(l

"

this expression is written in the

gc

where

f(d)

fiom COS Cdo<V

cos
)
jjr

" (l
ZQmO
2g m0
/1
I
f
/
X^ m ,mi JO \

II
A
d(W)

,
cos
(13-53)

form

}igm,^J(6)

Eom cos
7om

U>A

mt\
I

(13-54)

cos

co i

d(a>o<)

DEMODULATION

Sec. 13-14]

But

373

since

(1

Eom

_i_

(13-55)

^E~a /

then
f(e)

The function
such results

['

= ?

f(6)

W0</^)1
t^t

COS
(Eton
v

V JO

jOm/J^A

cos

,/
,
f\
co < rf(o;o<)

(13-56)

can be evaluated for different values

of a.

Plots of

are given in Fig. 13-20.

.6

.5

t-4

.3

40

20

60

80

100

140

120

160

180

e
Fig. 13-20. Plot of the function /(0) for various values of a.

Observe from these curves that the values of f(d) for a given value of 6
do not vary markedly from each other for different values of a. The
maximum value of conversion transconductance occurs in the neighborhood of 90 deg. Specifically, for a = 0, which occurs for constant
gm

<7m,.x

over the

full

operating range, f(6)

1

9c

2
-

in

2/ir,

0.32gv m

For a linear transconductance variation (a

quadratic transfer characteristic, the

optimum

1),
is

and so
(13-57)

which corresponds to a
approximately 120 deg,

which case
go

= KsWmax X

0.53

0.27fir m

m,x

(13-58)

Finally, for the transconductance variation specified by a = 0.5, which

corresponds to a space-charge (three-halves power) variation, the optimum

RADIO ELECTRONICS

374
is

[Chap. 13

approximately 105 deg, in which case

gc

0.57

(13-59)

0.29^ m maI
,

Note that all values of g c /g m ,* are roughly the same, so that, for conduction angles 90 < 6 < 180 deg, the conversion-transconductance ratio
varies over the range from 0.25 to 0.30.
If an exact determination of g c for a specific tube is required, this
can be accomplished by a numerical evaluation
given by Eq. (13-50).

of the Fourier coefficient

REFERENCES
W. L., "Communication Engineering," 2d
McGraw-Hill Book Company, Inc., New York, 1937.

427-433,

1.

Everitt,

2.

Marique,

3.

Terman, F. E., and N. R. Morgan, Proc. IRE, 18, 2160 (1930).

Nichols, H. W., Elec. Commun., 2, 11 (1923).
Peterson, E., and F. B. Llewellyn, Proc. IRE, 18, 38 (1930).
Herold, E. W., Proc. IRE, 30, 84 (1942).
Peterson, L. C, and F. B. Llewellyn, Proc. IRE, 33, 458 (1945).
Rothe, H., and W. Kleen, " Elektronenrohren als Anfangsstufen Verstarker," Akademische Verlagsgesellschaft, m.b.H., Leipzig, 1948.

4.
5.

6.

J.,

ed.,

pp.

Wireless Eng., 12, 17 (1935).

PROBLEMS
13-1. A sinusoidal potential is applied to the average detector of Fig. 13-2, with
Ei = 100 kilohms. Determine the variation of the rectified output potential
with applied signal.
13-2. The envelope detector of Fig. 13-3 has

Ri

The impressed
e

100 kilohms

potential

is

2*-

12 cos

1.5

C=

and

X \W +

cos %r

200

fifxi

1.51

Write an expression for the instantaneous potential across C.

10 8 i

Neglect tube drop

in the diode.

The
13-3.
carrier is sinusoidally modulated 80 per cent by a 7.5-kc signal.
signal input to the envelope detector of Fig. 13-3 with Ri = 100 kilohms has a
What should be the maximum value of C
carrier component of 12 volts peak.
no distortion? Neglect the tube resistance.
13-4. The input to the detector of Prob. 13-3, with

for

C=

100

Ri
I

100 kilohms

megohm

lOOO/isec

a pulsed r-f signal, the carrier frequency being 5 Mc.

the potential across R,.

is

T~|

R =

Sketch the waveform of

DEMODULATION
13-5. In the circuit of Fig. 13-3, the diode efficiency

375
is

0.90.

Calculate and

plot as a function of the modulating frequency w m the maximum degree of

modulation of the input signal e for distortionless rectification.
13-6. Suppose that the d-c load resistance of a peak diode detector is 250
= 0.90.
kilohms and
a. Calculate and plot the maximum degree of modulation of a signal without
negative peak clipping as a function of the ratio of a-c to d-c load impedances
t)

between

for ratios

and

0.5.

the signal is 100 per cent modulated, plot the approximate distortion
under these conditions.
13-7. Determine from the rectification characteristics of a 6H6 diode the
largest permissible modulation without clipping if the d-c impedance of the load
is 100 kilohms and the modulation frequency impedance is 50 kilohms.
If the
effective modulation is 70 per cent with a carrier of 10 volts rms, determine the
approximate percentage distortion.
13-8. A superheterodyne as indicated in the diagram in block form gives for an
unmodulated carrier-frequency signal of 5m v input to the r-f amplifier a detector
d.

If

output of 5

6SK7

6SA7

R-f amp

Converter

volts.

If

6H6

6SK7

6SK7
l-f amp

automatic gain control

is

Defector

used on all tubes, what must the

output? Assume that the full

carrier level be to give a 5- volt output; a 10- volt

d-c output potential is used for automatic gain control.

13-9.
receiver is provided with a square-law detector,

which

is

represented

by the expression
i

ke 1

Calculate the form of the output, and specify whether or not an intelligible a-f
signal results when the following are applied to the input:
a. One sideband is eliminated from the transmitted wave.
6. The carrier is eliminated from the transmitted wave.
c. One sideband and the carrier are eliminated from the transmitted wave.
13-10. A single-sideband transmitted wave is appljed to a square-law detector
for

which
iP

0.5(6

e)

ma

If the input consists of a supplied carrier of 2.5 volts

amplitude
13-11.

peak and the sideband

volt peak, calculate the signal current in the output.

plot of the curve of
6L7 pentagrid tube is used as a converter.

is 1

1500

RADIO ELECTRONICS

376

[Chap. 13

transconductance as a function of grid 3 bias is sketched in the

A beating oscillator signal of 24 volts peak, which
causes a bias of 20 volts to be developed, is applied to grid 3. The incoming
signal of 100 /xv is applied to grid 1, which is maintained at a bias of 6 volts.
a. Evaluate the conversion transconductance under these conditions by a
plate-grid

diagram

for this problem.

graphical solution.
b.

Compare with the

Sec. 13-15.

results obtained

by applying the appropriate

results of

CHAPTER

14

FREQUENCY MODULATION AND DETECTION

14-1. Introduction.
Chapter 12 was confined to a discussion of amplitude modulation, in which the amplitude of the transmitted wave is
altered in a manner dictated by the amplitude and frequency characteristics of the signal.
It was there pointed out that in addition to such
amplitude variations it is possible to effect changes in the frequency or in
the phase of the transmitted signal.
In particular, however, if intelligence is to be transmitted, it is essential that the two features that
characterize intelligence, viz., amplitude, or loudness, and frequency,
must be available.
In frequency modulation the transmitting frequency is varied by an
amount depending on the signal amplitude, and the signal frequency
determines the rate at which the variation takes place. In phase modulation the phase of the transmitted wave is shifted by an amount that
depends on the signal amplitude, and the rate at which this shift occurs
is made proportional to the signal frequency.
In general, any system
that can transmit the two aspects of information required for the intelligence could serve as an acceptable system of communication. A variety

have been devised which are satisfactory and which

and p-m systems. Certain
aspects of these systems have been discussed in Chap. 1.
It should be particularly noted that the amplitude of the oscillations
in the f-m system is not involved in the actual process of transmitting
intelligence.
Consequently, it is possible to make the system insensitive
to any a-m disturbances.
This is a particularly desirable feature since
atmospheric and man-made disturbances are largely amplitude-modulated.
Owing to the difference in character between a-m and f-m signals,
it is possible to separate and extract the signal from the interference.
14-2. Basic Characteristics of Frequency Modulation.
To examine
graphically the fundamental principles of frequency modulation, suppose
that a telegraph dot and dash are applied to an a-m and to an f-m system.
The results have the forms illustrated in Fig. 14-1. For the a-m system,
the frequency of oscillation remains constant, but the amplitude is zero
or a constant, depending upon the time in the cycle.
In the f-m system,
of pulse systems

possess certain advantages over the a-m, f-m,

377

RADIO ELECTRONICS

378

[Chap. 14

the amplitude remains constant, but the frequency changes from a value
/ 2 to a value /i or f3 when the signal is applied.
If the applied signal is sinusoidal and of frequency fm the effect produced in an a-m system has the form illustrated in Fig. 14-26 and the
,

-if

Pulse

AM

FM

Fig. 14-1.

The primary

features of

a-m and f-m waves.

produced in an f-m system has the form illustrated in Fig. 14-2c.

Phase and Frequency. The general expression
for an unmodulated carrier wave is given by

effect

14-3. Instantaneous

ec

= Ec

sin {u c t

In this expression the period of the wave

+
is

(14-1)

0)

given by

T = \ =
u
j

The quantity

(ji c t

is

the function.

written as
tential at

by

<p(t),

The output

of a sinusoidally

is

related to the phase

(14-2)

<p

the phase

If

is

the value of the po-

any instant

is

represented

the expression

(c)

modulated a-m and f-m transmitter.

the total instantaneous "phase"

of

Fig. 14-2.

Ec Sin v

14_3 )

frequency

by the expression
d<p

(14-4)
dt

Sec. 14-4]

FREQUENCY MODULATION AND DETECTION

379

This expression agrees with the usual definitions of frequency, and in the

unmodulated case

^ (wd +

01

= wc

6)

14-4. Frequency Modulation.

Frequency modulation is produced by
varying the instantaneous frequency of a carrier by an amount that is
proportional to the amplitude of the modulating signal and at a rate

The amplitude

given by the frequency of the modulating source.

assumed to remain constant

modulating signal has the form

in the process.

carrier is

= Em

em

the f-m

oi

mt

of

the

if

the

(14-5)

oi c

k f E m cos

u>

Em

for a given signal strength

mt

(14-6)

maximum

proportionality factor k f determines the

quency

is.

wave has an instantaneous frequency given by the expression

oi(t)

The

cos

That

To determine the expression for the f-m wave, use

This requires that

from which

it

(lip

01

+
,

01 c

variation in fre-

7-T

cos

Kfhj m

01

is

made

of

Eq. (14-4).

mt

follows that
<p(t)

kf

oi

1^

dt

which yields the expression

<p(t)

The initial phase

oi c t

sin

oi

mt

what

do

follows, for

it

(14-7)

plays no part in the

Thus, for the f-m wave,

e

= E

sin

(w

The instantaneous frequency

/
which has a

01m

6 is neglected in

modulating process.

maximum

w2t

et

of

the f-m

jc
fc

+
'

kf

sin

w m tj

wave
cos

'2tt

oi

(14-8)

is

mt

(14-9)

value of

/-=/
and a minimum value

kf

+ */!=
2tt

(14-10)

(14-11)

of
/min

/c

fc/

|f

RADIO ELECTRONICS

380

The maximum swing

frequency from
denoted by
It

of the

frequency deviation and

is

By analogy with amplitude

mean value

its

*/

(14-12)

|f

modulation, the modulation index

_ = fd =
m
f
T

<^d

called the

is

is

-/.=/.- /- =

n=

[Chap. 14

is

defined as

i?m

,
k
f

,-.

,m

(14-13)

fc

Also, the ratio of fd to the modulating frequency fm

called the deviation

is

and has the value

ratio

OT/

^= ^

(14-14)

fc/

In terms of these factors, the expression for the f-m wave assumes the

form

= E

sin

(co c

8 sin

w m t)

F-M Wave.

(14-15)

To examine

Frequency Spectrum of
f-m wave, it is necessary to expand the expression [Eq. (14-15)] that
This is done as follows:
represents the f-m wave.
14-5.

the spectrum

of the

Use

is

2? c [sin

now made

co c t

cos (5 sin u m t)

cos w c t sin (5 sin w m t)]

(14-16)

of the following expansions,

cos (5 sin

co ro <)

Jo(&)

Ji n {8) cos 2nv> m t

(14-17)
sin (5 sin

os

m t)

</2n+i(8) sin

(2n

l)co m <

n=

where the function /(5) is the Bessel function

order n. The f-m waveform then becomes
e

= Ec

sin

w c t[J

+E
which

may

(S)

+ 2J (S) cos 4o)J +

+ 2/ (8) sin 3co m +

2J 2 (5) cos 2u m t

cos w42Ji(5) sin

co

m<

kind and of

of the first

(14-18)

be written in the form

e

= J

+
+
+
+

(8)E C sin

J 2 (5)c [sin
J 3 (8)Ec [sm

(co c

+
+
+

m )t

<a

- w m )i]
- 2ai m )<]
3co m )/]
(<o

sin (w c

2co m )<

sin

3o> m )<

sin

(co c

(14-19)

x cos y

cos x sin

(co c

where use has been made

sin

co c t

J i($)E c [sm (w c

2/

of the trigonometric expansions

= H[sm
= H[sin

(x
(a;

+
+

y)

sin (x

2/)

sin

(a;

y)]
y)]

(14.20)

Sec. 14-5]

The

FREQUENCY MODULATION AND DETECTION

Bessel function

" 2^1

J n (8)

2(2n

is

denned by the

2)

2(4) (2n
5

2(4) (6) (2w

381

series

2)(2n

4)

2)(2ra

4)(2

6)

(14-21)

from Eq. (14-19) that the spectrum of the f-m wave consists
a carrier and an infinite number of sidebands all of whose amplitudes
Graphs of several of these functions
are various-order Bessel functions.
It will be noticed that J (5) has a root at
are contained in Fig. 14-3.
It follows

of

BESSEL FUNCTIONS OF THE FIRST KIND Jn (6)

i*0

-0.6

Fig. 14-3. Bessel functions of the

first

kind.

This means that the carrier will vanish when the frequency
This fact provides
deviation is equal to 2.4 times the audio frequency.
deviation
since
the zero point of
a method for measuring the frequency
radio
receiver.
selective
the carrier can be observed by a
A list of the roots of various Bessel functions is given in Table 14-1.

about

2.40.

TABLE 14-1
ROOTS OF /()
n =
2 4048
5 520

8 654
11 792

14 931
18 071
21 212

3.882
7.016
10.173
13.323
16.470
19.616
22.760

5.135
8.417
11.620
14.796
17.960
21.117
24.270

6.379
9.760
13.017
16.224
19.410
22.583
25.749

7.586
11.064
14.373
17.616
20.827
24.018
27.200

8.780
12.339
15.700
18.982
22.220
25.431
28.628

RADIO ELECTRONICS

382

[Chap. 14

examine a particular situation in some detail. Conis modulated +75 kc by an a-f signal of 7.5 kc.
The corresponding deviation ratio is 10.0. The important Bessel funcA
tion that occurs in the expression for such an f-m wave is J(10.0).
Notice that
plot of </(10.0) as a function of n is given in Fig. 14-4.
It is instructive to

sider

an

r-f

signal that

JnOOh

-0.3 L

Fig. 14-4.

plot of /(10.0) as a function of n.

0A"

0.2

Fig. 14-5.

Jn

(10.0) falls off

The

spectral distribution in

toward zero rapidly

for

amplitudes are significant out to about

14

7.5

an f-m wave with

14,

Notice that this

for sign,

as dictated

is

It

is

10.

105 kc
is

given in Fig.

just a plot of the spectral lines, without regard

by the Bessel-function

separation between individual sidebands

quency.

greater than 10, but that the

which corresponds to

plot of the spectrum corresponding to Eq. (14-19)

14-5.

is

plot of Fig.

14-4.

7.5 kc, the modulating

The
fre-

evident from this plot that the total bandwidth necessary

210 kc.
In Fig. 14-6 is illustrated a series
of plots for constant modulating frequency fm but for various values of
frequency-deviation ratio 5. The increasing number of sidebands with
to include

Two

all significant

sidebands

is

similar plots are also included.

Sec. 14-5]

FREQUENCY MODULATION AND DETECTION

Likewise, the disappearance of the carrier

clearly seen.

5 is

is

383

plainly

seen.

The

plots of Fig. 14-7

show the spectra

for a constant frequency devia-

tion but for various modulation frequencies.

the total bandwidth required to include

somewhat with increasing deviation

tion fd

the

number

of significant

should be observed that

sidebands decreases

For a given frequency devia-

ratio.

except for the very small value of

bands are contained within the range jd

It

all significant

5,

almost

The curve

all significant side-

of Fig. 14-8 indicates

sidebands (those with amplitudes exceeding

6=^0.1

*" 3

..lllllll,

0.5

4.0

lll,l .111,,.

i,

8-

J 1,

.,1

lil.l.ll l.l.llll...
1>-<i-4

15

'O-0^\\

.lllll.

2.0

"A

,|||||,

-4

,-OJd

ll,

l.ll.l.l .l.li.illi,...

..llll.ll.ll.l.l.ll.ll.llll,.

O-rf-J

11,1, hi ll.l.ll.ll.llllll...

--- <v

Fig. 14-6.
wd,

and

The

fixed

o>

spectral distribution in an f-m

wave

for different values of

8,

different

per cent of the largest sideband component) in an f-m spectrum for

different values of

n and

5.

In particular,

if 5

5,

then n must be about

8 for J(S) to be negligible compared with unity.

Some very important information

is

contained in Fig.

appreciate this, consider the present Federal Communications

sion

(FCC) regulations on frequency modulation.

14-8.

To

Commis-

These regulations

specify:

frequency deviation, fd = 75 kc.

B = 200 kc (including a 25-kc band at each

1.

Maximum

2.

Allowable bandwidth,

end).
3.

Frequency

stability of carrier,

+2

kc.

RADIO ELECTRONICS

384

[Chap. 14

Remembering that the frequency deviation is related to the amplitude

and frequency of the modulating potential according to the relation
wd = kf Em
then for that amplitude which provides a frequency deviation of 75 kc
the bandwidth requirements increase with decreasing fm
But from Fig.
.

_l

as

i.o

2.0
,

3.0

..(III III..

5.0

...lll.l. I.lli...
*l Y*>m
W-aJd--^
I

10.0

llll-ll-ll

20.0

l-ll-llll.

llllll.lllilhlilnlililillilill.lll.llllli..

240

l lllli,ll.,l, l l.li.l.lJ,l,l.l, l l.l l ,l,ll..llllM..

U-a>d~ *J
Fig. 14-7.

and

The

different

spectral distribution in

am

an f-m wave

for different values of fixed

oid

14-8 it is clear that all of the significant sidebands or energy is contained

within fd for values of 8 = 10 or higher.
For S lower than 10, the number
of sidebands outside of fd begins to increase rather seriously.
This occurs
for frequencies of 7.5 or higher.

Note, however, that with ordinary

Sec. 14-6]

FREQUENCY MODULATION AND DETECTION

385

broadcasts the amplitude of the frequencies in the range of 7.5 kc and

is considerably smaller than with the frequencies in the middle or

higher

lower registers.

Clearly, therefore,

Em

for these higher frequencies

is

and hence the total active fd is

materially less than the allowable 75 kc and the system is operating within
a narrower bandwidth than might otherwise be required. In fact, the
bandwidth is so much less than that allowed that it is found convenient
to include weighting or preemphasis networks in the transmitter.
These
have such properties that they tend to accentuate the higher frequencies.
smaller than for the mid-frequency band,

500

100

50

10

I
I

10

20 30

50

500

100

1000

Note that n may

Fig. 14-8. Values of n for a given value of S to make J n (S) < 0.01.
assume only integral values. {From L. Mautner, " Mathematics for Radio Engineers,"
Pitman Publishing Corporation, New York, 1947.)

In this way, the higher-frequency components, which ordinarily do not

much energy, are expanded beyond their natural level. This
proves to be a desirable practice in that it tends to give an improved
contain

signal/noise ratio, for the noise generated within the tube circuits

is

uniformly distributed over the frequency band. Of course, for fidelity in

reproduction, the receiver must have networks which deemphasize the

incoming signal to yield the proper energy distribution. More will be

said about this when the circuits for effecting preemphasis and deemphasis

are discussed.

14-6.

Phase Modulation.

Phase modulation

is

produced by varying

the instantaneous phase of the carrier at a rate that is proportional to

the modulating frequency and by an amount that is proportional to the
strength of this modulating signal.

The amplitude

of

the carrier remains

RADIO ELECTRONICS

386

unaltered in this process.

If the

modulating signal has the form

em

= Em

the instantaneous phase of the

wave

=
where k p

is

0o

[Chap. 14

sin

um t

given by the expression

is

k p E m sin w m t

(14-22)

a proportionality factor that determines the

maximum

varia-

tion in phase for a given signal strength.

To

find the analytic expression for the

fact that the instantaneous phase

is

(pit)

Equation

= u

ct

= E

sin

(co c i

is

made

k p E m sin w m t

when

(14-3) for the present case,

e

p-m wave, use

of the

given by

written in

(14-23)
full,

becomes

k v E m sin u m t)

(14-24)

The constant phase

ing process.

O is taken as zero, as it plays no part in the modulatThis expression is written in the form

= E

where the

maximum

sin (wJL

deviation in phase
9i

comparison

6d sin

u m t)

(14-25)

is

k p Em

(14-26)

Eq. (14-25) with Eq. (14-15) for the f-m wave indiidentical.
Consequently the entire discussion of the spectral distribution of the energy contained in an f-m wave
can be extended to p-m waves. Therefore the frequency spectrum of a
p-m wave having a maximum phase deviation of, say, 10 rad will be
identical in form with the frequency spectrum of an f-m wave having a
cates that the

of

two forms are

deviation ratio of 10.

There is one very significant difference between the f-m and the p-m
waves, however. This difference is contained in the form of the phase
deviation 8d and the deviation ratio 8 that appears in Eqs. (14-25) and

The

(14-15), respectively.

differences

lie

in the definitions of d d

and

5,

namely,
8d
5

k p Em

for

p-m waves

k/Em

=
=

tor

i-m waves

03m

p-m waves the phase deviation depends only on the amplitude

modulating signal, and all modulating frequencies of equal Em will
possess equal values of d d independently of the frequency o> OT
As a
result, the spectral distribution will be the same in each case, although
the separation of the spectral lines will depend on the modulating frequency. In particular, if it is supposed that the maximum phase deviaClearly, for

of the

FREQUENCY MODULATION AND DETECTION

Sec. 14-7]

tion of a particular

wave

is

387

5 rad, there will then be approximately 8

components present. If the modulating frequency

then the bandwidth is 2 X 8 X 5,000 = 80,000 cps. If the

significant sideband

5 kc,

is

modulating frequency for the 5-rad maximum deviation is 50 cps, then

the bandwidth is 2 X 8 X 50 = 800 cps.
In the case of frequency modulation, if the value of E m is such that the
deviation ratio is 5 = 5 for a modulating frequency of 5 kc, then for an
equal E m at 50 cps the corresponding deviation ratio is 500. The resulting spectral distribution in these

two cases

will

ent, there being 8 significant sidebands for S

be altogether differand there being in

5,

= 500. The bandwidth is

= 80 kc under the first conditions and is approximately
= 50 kc in the second case.

excess of 500 significant sidebands for 5

X
X

2
2

500

5,000

50

Owing to the simple

that

it

difference in

form between

8d

and

5, it

might appear

p-m waves to produce f-m waves. This

arrange, by means of appropriate circuits,

should be possible to use

would be possible

if

one could

to cause the apparent phase deviation to vary inversely as the modulating

Such circuits are possible, and the Armstrong method of

producing frequency modulation operates on this principle. The details
of this method will be discussed below.
14-7. F-M Transmitters
Reactance-tube Types. A variety of methods for the production of frequency modulation exist, although they do
not all enjoy very great flexibility. In principle at least, the most direct
way of producing an f-m wave is to alter the capacitance in the tank
This might conceivably be done by incorporatcircuit of an oscillator.
ing a capacitor microphone as part of the tank capacitor in an oscillator
A considerably more satisfactory method, and one which accomcircuit.
frequency.

plishes the

same

result in substantially the

same way,

is

to incorporate a

reactance tube in the tank circuit.

A reactance-tube circuit is an electron-tube circuit which is so designed

that the effective output-terminal impedance

inductive or capacitive depending

upon

is

largely reactive, either

manner

of connection, the
being varied by varying the potential on one
grid of the tube.
By incorporating such a circuit as part of the tank
circuit of an oscillator, the effective tank-circuit inductance or capacitance, and so the resonant frequency of the oscillator, may be varied

magnitude

its

of the reactance

electrical means.
Moreover, it will be shown below that the instantaneous carrier frequency of such a reactance-tube transmitter is directly
proportional to the instantaneous potential on the control electrode of
the reactance tube.
As a result, therefore, the instantaneous frequency

by

is directly proportional to the amplitude of the control

which is proportional to the modulating-signal amplitude.
block diagram of a simple reactance-tube f-m transmitter is shown

of the transmitter
signal,

RADIO ELECTRONICS

388

The

in Fig. 14-9.
circuit are

essential features of certain of the elements of the

examined below

some detail. In particular, the operation

and the operation of the preemphasis circuit,

in

of the reactance-tube circuit,

will

[Chap. 14

be examined analytically.

Modulating
signal

Emfm

Audio

amplifier (50-

fc =J.97"c
Preemphasis

Reactance
modulator

circuit

Oscillator

15000
cps)

fyS.91**

Buffer
amplifier

Fig. 14-9.

Fig. 14-10.

14-8.

Three

Tripler

fd '3./2" doublers fj25^

fd -7S'

The reactance tube and

its

Power
amplifier

fe-

te

simple reactance-tube f-m transmitter.

The Reactance Tube.

ance-tube circuit and

fc=9S.4 MC

its

equivalent circuit.

schematic diagram of a simple reactIt is

is given in Fig. 14-10.

equivalent circuit

desired to determine the effective output-terminal impedance of this

circuit.

This

is

Note from the equivalent

readily accomplished.

circuit

the following relations:

Ic

Xc

with
Also, the grid potential

E,

R - jX c

(14-27)

= J_
coC

is

(14-28)

The

plate current

is

Ei

mE,

which may be expressed as

RE,

Ei
rv

The

total current
Il

is

I,

rp

(14-29)

R jX c

then

Ei
Ic

R - jX c

Ei

/?Ei

R - jX c

(14-30)

FREQUENCY MODULATION AND DETECTION

Sec. 14-8]

The output-terminal admittance

Y = ^ =
which

may be

1_

the circuit

of

is

given by the relation

R
R - jXc

1
R - jX c + +

389

(14-31)

expressed in the form

Y =

R -

j(l/C)

rp

+
'

(l/g m )

(14-32)

- j(l/g m RCa)

This expression indicates that, in so far as the output circuit of the

is concerned, it may be represented by the circuit of
Fig. 14-11.
Clearly, if the impedances r p and R jXc are large
reactance tube

compared with
J_
gm

and

if

E,

1/uCR

is

J*

g m RCo>

large

compared

Fig. 14-11.

The equivalent output

circuit

of a eapacitive reactance tube.

with unity, the output-terminal

admittance becomes purely eapacitive and arises from an equivalent capacitor having a capacitance g m RC.
Since the output impedance of the reactance-tube circuit above may be
made to appear as a pure capacitance g m RC, then if the modulating signal is made to vary the g m of the tube, and this is readily accomplished
by applying the modulating signal to the grid of the tube, the effective
capacitance will then change with changes in grid potential.
The circuit
of such a reactance-tube f-m oscillator is given in Fig. 14-12.
Also

-6

C-Reac+ance tube

Fig. 14-12.

A reactance-tube

f-m

Tuned plate osc

oscillator,

-4

-2

Grid volts

and the curve showing the variation

of

with grid potential.

included

is

a curve showing the variation of g m of the tube with changes

in grid potential.

reactance tube

may

be connected to yield an effective inductance,

rather than an effective capacitance, across the output terminals.

Such

given in Fig. 14-13. By proceeding in the same general manner as for the eapacitive reactance-tube

circuit,

with

its electrical

equivalent,

is

RADIO ELECTRONICS

390

Fig. 14-13.
circuit, it

An

inductive reactance tube and

its

[Chap. 14

equivalent circuit.

can be shown that the equivalent output admittance has the

form

+1
r +

R +
Rg +

1/juC

which an equivalent circuit

form illustrated in Fig. 14-14.

for

T+

J_

+j

9m

(14-33)

9m

This equivalent circuit has the

exists.

By

oCR

the proper choice of the various

elements in the

circuit,

the circuit

reduces to the simple form, com-

an equivalent induct-

prising only

ance

CR/gm

reactance-tube oscillator that

incorporates an inductive reactFig. 14-14. The equivalent output circuit

of an inductive reactance tube.

ance-tube
Fig.

as

circuit

Hartley oscillator
14-15.

is

part of a

illustrated in

Also included

is

graph showing the variation of gm with changes in grid 3 potential.

An approximate expression for the variation of the frequency of a
reactance-tube oscillator as the potential on the control electrode is varied
is readily possible.
The transconductance may be expressed analytically
as a function of the potential of the control electrode.
Note from the

L- Reactance tube

-20

Hartley oscillator

-16

-12

-8

Cj (volts)

(a)

Fig. 14-15. An inductive-reactance-tube f-m oscillator,

variation of g m with grid 3 potential of the 6L7 heptode.

-4

(b)

and the curve showing the

Sec. 14-8]

FREQUENCY MODULATION AND DETECTION

391

curve that

= te

gm

But

+G

(14-34)

since

+ Em

= E cc

cos

wj

Em

cos

(14-35)

then
gm

+G +

G f^
ilO

J
iio

uj

For the capacitive reactance-tube

(14-36)

circuit,

the effective output capacitance

is

G =
e

If it is

gn

CR = G CR

(l

assumed that the frequency

+ J* +

|=

cos

of oscillation of the oscillator

(14-37)

is

that of

the tank circuit alone, then

=
2tt

The

(C

C.)

(14-38)
2tt

lator

VL
^L

Co

+ LdGoCR (l +

^ + |=

carrier frequency is evidently the value of the

when the modulating-signal

potential

is

zero.

cos

uj)

frequency of the
This is

oscil-

(14-39)

2^L
The frequency

ratio f/fc

is

CoEq

GoCR

given by
1

=
1

CRE m

LdG

+L

[C

(14-40)

cos u m t

E + GoCR(E + E

By

this expression

the

in the expansion, since the total

expanding
first term
then

by the binomial theorem and retaining only

Em

fc
This expression

may
/

/c (l

Ulf cos

frequency

cos a m t

C\E /GoCR

+E +E

be written in the form

cc )}

w m t)

cc

shift is small,

(u_ 41)

RADIO ELECTRONICS

392

where the modulation index

is

2 CoEo/G

J CC

As discussed

14-9. Preemphasis Circuits.

amount

E
CR +

mf =

[Chap. 14

(14 " 42)

So

in Sec.

14-4,

there

is

energy contained in the h-f portion of the audio

spectrum. As a result, the deviation at these high frequencies is far less
than the maximum allowable value of 75 kc. The corresponding bandwidth requirement is correspondingly less than the allowable 150-kc
In fact, the relative h-f amplitudes are so low that it is customary
total.
to include preemphasis networks in the circuit to accentuate the h-f
terms.
In this way the relative signal strength at these higher frequencies is improved relative to tube and circuit noise, which has a uniform
Of course, corresponding
distribution over the entire audio spectrum.
deemphasis must be incorporated in the receiver in order to bring the
relative amplitudes of all frequencies to their proper levels.
Preemphasis circuits are chosen to satisfy the equation
relatively small

of

5
Ex

(14-43)

Vl +

(i/co)

The value of &>i was originally chosen to be 1/toi = 100 jusec, but it is now
taken as 75 jusec. With such a preemphasis circuit the amplitude of a
2,100-cps signal is increased in the ratio \/2/l over the normal level, and
the relative amplitude of a 21-kc signal is increased in the ratio 10/1.
C'75/j./j.f

WW\

f--0
/h+ Vr-y

+r

1(

+r

EA
Two

Fig. 14-16.

Either an

Two

RL

different preemphasis circuits.

may

be used to accomplish preemphasis.

In the RL circuit,
readily found to be

circuit

different circuits are illustrated in Fig. 14-16.

the potential ratio

E _
writing wi

E /Ei
c

is

ju>L

E,

By

CR

or a

room

R+jwL
= R/L,

Vl +

this

CR

(14-44)

becomes
1

("i/")

the mathematical development becomes

circuit,

(14-45)

Vl +

Ei
Similarly, for the

/tan- 1 (-R/wL)

(R/uL)

Ei

R +

l/jo>C

(14-46)

Sec. 14-10]

which

FREQUENCY MODULATION AND DETECTION

is

Ei

By

393

writing

a>i

(14-47)

1/juCR

= \/CR\
1

Ex

(14-48)
j'coi/w

or

E.
Ei

VI +

(14-49)
2

F-M Transmitters. Just as in the

necessary that the average or carrier

14-10. Frequency Stabilization of

case of the

a-m

transmitter,

it is

frequency of an f-m transmitter be maintained very nearly constant, even

though the instantaneous frequency of the f-m transmitter varies with
the modulating signal.
When a reactance-tube modulator is used to
modulate the carrier, the carrier cannot be crystal-controlled and the
average frequency will depend to some extent on the temperature, the
tube characteristics, and the various potentials. Slight drifts in the
operating characteristics of the reactance tube or slight changes in any of
the circuit elements will be accompanied by an appreciable change in the
average frequency. It is possible to minimize the effects of the drift in
the reactance-tube characteristics by employing two such tubes in a
balanced connection. Nevertheless the stability is not sufficient without
employing some type of stabilization to maintain the carrier frequency
within the 2-kc deviation specified

Two

basically different

methods

lator are in present-day use.

by the

FCC

regulations.

a reactance-tube moduIn both cases a standard reference freof stabilizing

quency is provided by a crystal-controlled oscillator, and the fundamental

some subharmonic of the transmitter frequency is compared with this
reference frequency.
Deviations between the two serve to actuate control circuits which operate in such a manner as to reduce these deviations.
The RCA and Federal Telecommunications Laboratory schemes employ
somewhat similar all-electronic methods to effect the frequency stabilization, and the Bell Telephone Laboratories method employs a frequencysensitive servomechanism which drives a small motor to which is geared
the tuning capacitor, the direction of rotation of the motor being determined by the relative frequency of the transmitter and the reference
or

standard.

The RCA method of stabilizing a reactance-tube f-m modulator is

shown schematically in Fig. 14-17. In this circuit the frequency of the
1

from the center frequency of the f-m transmitter

amount, say 1 Mc. The two frequencies are mixed,

reference crystal differs

by some

definite

and the difference frequency

is

applied to a discriminator (the operation

RADIO ELECTRONICS

394

[Chap. 14

The

of the discriminator will be discussed in Sec. 14-16).

from the discriminator, which

is

a direct measure

d-c output

of the

difference

then applied to the grid of the reactance-tube modulator in

such a manner as to make the net control potential equal to zero when the
difference frequency is exactly 1 Mc.
Any frequency drifts can then be
Clearly, the method cannot yield perfect stability,
appreciably reduced.
for unless there is a slight frequency difference, no control potential is
However, the improvement that results is
applied to the modulator.
sufficient to maintain the frequency stability within the FCC regulations.
frequency,

is

Signal

Reactance

Variable

Multipliers

tube

frequency

and

modulator

Discriminator

RCA method

Fig. 14-17.

The Federal

oscillator

power amp.

Mixer

Crystal
oscillator

Output

of stabilizing a reactance-tube frequency modulator.

center-frequency-stabilization system

is

shown schemati-

In this system the frequencies of the crystal oscillator

and the master oscillator are each divided to a common frequency and are
then combined in a balanced phase detector. The d-c output potential,
which is a measure of the phase difference between the two oscillators, is
used to actuate the reactance-tube modulator in a manner to lock the
oscillator mean frequency to that of the crystal reference frequency.
This system maintains the center frequency constant within about 1 kc.
cally in Fig. 14-18.

Signal

Reactance

Variable

tube
modulator

frequency

and

oscillator

power amp.

Low

Output

Freq.

pass

Balanced

filter

phase
detector

Fig. 14-18.

Multipliers

divide/

Crystal
oscillator
Freq.
divider

The Federal center-frequency

stabilization system.

The schematic diagram of the Bell Laboratories method of stabilizing

the frequency of an f-m transmitter 2 is illustrated in Fig. 14-19. Here
the output is frequency-divided, and the resulting subharmonic is then
modulated by the output of the crystal-controlled oscillator in such a
manner

as to produce two-phase beat currents.

to operate a small synchronous motor to which

These currents are used

is

geared a tuning capaci-

which is part of the f-m oscillator circuit. If the subharmonic

remains in synchronism with the crystal reference, the motor does not
tor,

Sec. 14-11]

FREQUENCY MODULATION AND DETECTION

move.

the carrier frequency

If

drifts,

395

the armature rotates, the direction

of rotation being set so as to readjust the carrier to the correct frequency.

This method of stabilization proves to be very satisfactory and operates

over a very wide range of drifts to yield satisfactory results.

and

frequency

tube
Signal

Multipliers

Variable

Reactance
modulator

poweramp.

oscii/afor

Output

Geared
to tuning
Synchronous
condenser

motor

Fig. 14-19.

The

Frequency
divider

Stabilizing

Crystal
oscillator

modulator

of stabilizing a reactance-tube frequency

method

Bell Laboratories

modulator.

14-11.
shift

F-M Transmitters Armstrong System. The Armstrong phase-

method

of obtaining

frequency modulation incorporates a crystalConsequently this

controlled oscillator as a basic element of the system.

system has an intrinsic stability as good as that

additional frequency stabilization is required.

of its crystal,

and no

Before the specific features of this method of producing frequency

modulation are examined, it is pertinent to examine the spectrum of p-m
waves with small value of maximum phase deviation. For the particular
case where 6 d = 0.5, which is illustrated in Fig. 14-6, the significant terms
depend on the following values of Bessel functions,

=
=
=

(0.5)

JiCO.5)

J 2 (0.5)
J(0.5)

= E

0.0306

=0

and the modulated wave has the

which

0.9385
0.2423
for

explicit

sin (u c t

>

form

0.5 sin

u m t)

+ u m )t
2u m )t +

sin

(14-50)

is

= 0.9385

sin

u e t + 0.2423# c [sin
+ 0.0306S c [sin (co c

(co c

sin (w c
(co c

u>

m )t]

2a> m )t]

(14-51)

Note, however, that the second sideband components are quite small and
that the expression may be written approximately as

0.9385

sin

o, c t

+ % [sin

Clearly, this expression will be

(w c

oo

m)t

more accurate

sin

(co c

oo

m )t]\

for values of B d less

(14-52)

than

0.5.

RADIO ELECTRONICS

396

Now

consider the corresponding

e

which

may

a-m wave, having the form

+ ma

c (l

sin

um t)

sin

wct

(14-53)

be written as

= E

= E

[Chap. 14

sin

co c t

cos

(oi c

o}

m )t

cos (w c

u m )t] (14-54)

Note specifically that, if ma = dd, the only essential difference between the
a-m and the p-m waves is in the relative phase of the carrier and the sidebands. Evidently for small values of 6 d if the sidebands of the a-m
wave can be shifted by 90 deg with respect to the carrier, a p-m wave
,

results.
It is immaterial, of course, whether the phase of the carrier or
the phase of the sidebands is shifted in order to achieve the p-m waves.

The

process here discussed can be given graphically in a

quite illuminating.

Fig. 14-20.

It

was shown

The sinor representation

of

in Sec. 12-2 that

an

a-m wave.

manner that

is

amplitude modulation

Fig. 14-21. The sinor representation of a p-m wave.

could be represented by means of a sinor diagram. In this diagram, the

carrier potential is represented by a fixed sinor, and the sideband components are represented by two sinors which rotate in opposite directions.

This sinor representation

is

redrawn for convenience in Fig. 14-20.

a-m wave at any instant.

The

resultant sinor e represents the

A
is

corresponding sinor representation of the process of p-m production

Here, as shown in Eq. (14-52), the carrier must be shifted

possible.

by 90 deg relative to the sidebands. The resultant sinor diagram

then has the form shown in Fig. 14-21. It is evident from this diagram
Moreover, since dd is chosen to be small,
that a p-m wave does result.
the amplitude variations that result in this process are very small. For
Figure 14-22 shows the
large deviations B d distortion is introduced.
in phase

percentage distortion vs.

maximum

phase deviation which

results.

block diagram of an Armstrong type f-m transmitter is given in Fig.

14-23.
The essential features of the system are the following A stabilized
200-kc primary frequency oscillator is used to control the mean or carrier
:

frequency of the radiated wave. Part of this 200-kc signal is mixed in a

balanced modulator with a signal representing a frequency-distorted version of the audio signal, the predistorted signal being such that the ampli-

Sec. 14-11]

FREQUENCY MODULATION AND DETECTION

397

30
Fig. 14-22. Per cent distortion vs.
in

40

50

Qmax deg
phase deviation in degrees, which results

maximum

the Armstrong modulator.

Carrier

200

amplifier

CombininfyOQK
amplifier fd

6
doubters

I2.?cps

200*c

I2.8

90phase

MC

HE

KC

Buffer

Balanced

Frequency 99J

amplifier

modulator

converter 0.78"c ttripler

5doublers

MC
Power 95A MC
KC
amplifier
75"1?
7S

95.4

200 KC
Crystal

Pre-

Crystal

oscillator

emphasis

oscillator

Predistor ter

Em fm
a-f
amplifier

Signal
Fig. 14-23. Block diagram of an

tude

is

made

Armstrong f-m transmitter.

to vary inversely with its frequency.

balanced modulator

is

The output from the

the a-m sidebands with the carrier-frequency

component missing. The modulation products are shifted through 90 deg

phase and are then combined with the carrier in the combining buffer
amplifier.
The result is an f-m wave, which has been achieved from the
p-m wave, the phase deviation of which has been made to vary inversely
in

with the modulating frequency.

The

resulting frequency modulation

is

RADIO ELECTRONICS

398

multiplied in frequency until

[Chap. 14

brought to the desired frequency level for

it is

final amplification

and transmission.

14-12. Predistorter

Circuit.

function of the predistorter circuit

The
is

to

provide an output potential the ampli-

Pig. 14-24.

tude of which varies inversely with the

frequency of the input potential. A

predistorter circuit.

circuit

The

sults is given in Fig. 14-24.

which achieves the desired

re-

ratio of the output to the input poten-

given by

tial is

Ri

Ri/j<*C

Ri

R+

1/juC
Ri/juC
R, + 1/juC

+ jwCR,

R +

Ri
1

jaiCR
+ R + Ri

(14-55)

+jCR

1
l

or

E_c

Ex

R/Ri

+ joiCR

(14-56)

This becomes, for the specific circuit constants indicated on the diagram,
1

(14-57)

1.02 -f j2.51/

Note

particularly that for

are given within

all

frequencies in excess of 50 cps the results

per cent by the expression

Ex

(14-58)

j2.51/

14-13. F-M Transmitters The Phasitron.

A cutaway sketch of the
General Electric GL-2H21 phasitron is shown in Fig. 14-25. It consists
of a cathode, an electrostatic focus and deflection system, and an anode
structure.
The electrons that are drawn from the cathode surface to
the anode assembly are acted on by the focus elements to form a tapered,
thin-edged disk, whose axis is the cathode and whose focus is at anode 1

anode assembly. The deflection system consists of 36 rigidly

mounted elements whose active portions lie in a radial plane below the
Every
electron disk and a solid neutral plane located above the disk.
third deflector is connected together and to one phase of a three-phase
excitation source.
The three-phase potential source comprises a crystal
oscillator and phase-splitting network.
The action of the deflection system is such that portions of the electron
disk are deflected above or below the normal plane by the magnetic field
The appearance of
of the three-phase system to form a sinusoidal edge.
of the

Sec. 14-13]

the disk

is

FREQUENCY MODULATION AND DETECTION

clearly illustrated in Fig. 14-25.

The

disk

may

399

be considered

as rotating at a rate determined by the crystal oscillator.

Anode 1 is a cylinder with 24 holes

punched alternately above and below

the normal plane of the disk. Electrons striking the surface of the cylin-

der are collected

by

it,

while those

which pass through the holes are colFigure

lected by the solid anode 2.
14-26 shows a developed portion of
anode 1. The solid sine curve represents the edge of the electron disk at
the time the maximum number of
electrons passes through the openings
The dotted curve shows
to anode 2.
the situation one-half cycle later, and
almost no electrons pass through the
openings to anode 2. If, therefore, the
two anodes are connected to opposite
ends of a resonant circuit, the circuit
will be excited at the crystal driving
frequency and in a time-phase sense
that is determined by the phase of the
anode-current pulses.

Frequency modulation of the resonant anode circuit is produced by

phase modulation of the electron disk.
This is accomplished by applying the
audio signal to a solenoid which surrounds the phasitron. The axial magnetic field that is so produced causes
the electron disk to be advanced or
retarded about
zero-signal

its

axis relative to its

position.

Consequently

the phase of the oscillator

is

shifted,

with a resultant production of

p-m

Moreover, since the magnetic

field is produced by a solenoid which
is essentially a pure reactance at audio
waves.

Fig. 14-25. The General Electric

phasitron.

GL-

2H21

frequencies, then, for a constant po-

and hence the magnetic field that is produced,

vary inversely with the frequency of the impressed potential.
Clearly, therefore, the output from the oscillator is an f-m wave.

tential input, the current,

will

RADIO ELECTRONICS

400

[Chap. 14

schematic diagram of the phasitron f-m transmitter

is

given in Fig.

14-27.
Holes punched In Anode

Minimum current to Anode 2

Fig. 14-26.

Signal

Maximum current to Anode 2

developed portion of anode

1.

Audio
frequency
amplifier

/-phase to

J-phase

r-f amp.

network

xtal
oscillator

Output

Fig. 14-27.

The General

Electric phasitron f-m transmitter.

14-14. F-M Receivers.

The basic circuit of an f-m receiver is somewhat similar to that of an a-m receiver of the superheterodyne type.
However, there are a number of significant differences in the two receivers.
The required bandwidth in the f-m receiver is larger than that for a-m
reception, which requires that the frequency converter and the r-f and
the i-f amplifiers must be designed for this broader bandwidth. Also,

Input
88-/08 M'

R-f amp

88108

Mixer

13.7

i-

f amp
and

limiter

Audio
Discrimimn

frequency
amplifier

I
\

f^l3.7 Mc

Beat
frequency
osc.

Fig. 14-28. Block diagram of a typical f-m receiver.

the last i-f stage of the f-m receiver is operated as a limiter, thus eliminating any fluctuations in the amplitude of the i-f carrier, however produced.
The other outstanding difference is in the circuit used to demodulate the

f-m carrier. The f-m discriminator that is used to convert from frequency
modulation to amplitude modulation does not appear in an a-m receiver,
and also the detector, while it uses conventional diode circuits, operates

FREQUENCY MODULATION AND DETECTION

Sec. 14-16]

somewhat

differently in the

examined below.
A block diagram

f-m

of a typical

a receiver must provide a high

circuit.

f-m receiver
r-f

The operation
is

shown

401

of this will

in Fig. 14-28.

be

Such

gain in order to permit high sensitivity

with amplitude limitation. Also, it is necessary to use a relatively high

frequency in order to permit the necessary 225-kc bandwidth. In

i-f

addition, the high intermediate frequency has the feature that the

image

In particular, in the block

diagram shown for use in the range from 88 to 108 Mc, the image fresignals fall outside of the tuning range.

quencies
14-15.

lie

in the

The

band 115.4

Limiter.

to 135.4

Mc.

It is the function of the limiter to

remove any

amplitude modulation that might exist in the signal. These fluctuations

in the amplitude of the i-f carrier might have been produced either by

20 k

30*
J3+(SO-90 v)

10

20

30

40

Input, rms
Fig. 14-29.

The

circuit of a limiter,

and the general character

variations in the transmitting conditions or

static.

of the results.

by man-made

or natural

Such a circuit, which usually operates on the nonlinear portion

an output potential that is sensibly independent of the amplitude of the input potential. Such limiter action
is easily secured by operating the plate of a tube at a very low plate
potential, by using a high series grid resistor, by using a low screen potential, or by a combination of these three.
As ordinarily used, the last i-f stage of the f-m receiver is usually
operated at low screen and low plate potential, in a circuit of the type
The operation of the circuit depends
illustrated, to serve as the limiter.
on the characteristic of pentodes with low applied potentials. With
such low potentials the operation falls below the knee of the plate characteristic.
The plate current becomes independent of the amplitude of
The general character of the gain
the grid potential over wide ranges.
of its characteristic, provides

curve is also included in Fig. 14-29. Such a circuit as that illustrated

will saturate at about 10 volts input to the grid, and at this point the
stage has a gain of approximately
14-16.

The

generally used

is first

3.

In demodulating an f-m wave, the method

to convert from frequency modulation to amplitude

Discriminator.

RADIO ELECTRONICS

402

[Chap. 14

modulation and then to demodulate the amplitude modulation by conThe circuit that is used to effect this conversion is
ventional methods.
discriminator.
known as a
A variety of such circuits is possible, and
in
will
be
considered
some
detail.
several
form
discriminator
simplest
of
comprises an ordinary resonant
The
differs somewhat from the
that
is
tuned
to
a
frequency
that
circuit
or
carrier
frequency
of
the
f-m
signal.
An exact analysis of such
average
actually
quite
complicated
owing
to the fact that the frea network is
applied
signal
changing
one,
and
the problem is therequency of the
is a
requires a transimple
steady-state-frequency
one
but
really
fore not a
that
if
the bandwidth
analysis.
An
analysis
of
the
problem
shows
sient
large
audio
rate
at which the
tank
circuit
is
compared
with
the
of the
frequency of the applied potential is varied, the normal steady-state

Fig. 14-30.

methods

of

simple antiresonant circuit operating as a discriminator.

analysis

quasi-steady-state

may

method

be applied without serious error. If this

may not be used, the problem

of analysis

becomes exceedingly complicated.

The approximate action of such a
ically in Fig.

14-30.

slope detector is illustrated graph-

Clearly, as the carrier frequency fluctuates, the

current in the detuned circuit varies, increasing as the impressed frequency approaches the resonant frequency of the circuit, and decreasing

The
as the impressed frequency departs from the resonant frequency.
output from such a circuit is an a-m wave. However, since the side of
the simple resonance curve is not linear, the a-m output is distorted.
A second feature of such a simple circuit as an f-m detector is that it
That is, if the applied potential is doubled,
provides no a-m rejection.

This means, therefore, that any amplitude

is also doubled.
modulation that exists in the applied signal will also produce an effect
As will be seen later in Sec. 14-20, noise produces such
in the output.
an a-m effect, and in order to ensure noise-free performance a very good
Because of its
limiter is required ahead of this discriminator circuit.
several shortcomings, such a simple discriminator is not used.
the current

Sec. 14-16]

FREQUENCY MODULATION AND DETECTION

403

The linearity can be greatly improved by using two "off -tuned" or

"stagger-tuned" circuits instead of one and then choosing the difference
between the two outputs.
is

Such a stagger-tuned discriminator circuit 3

A is tuned to

In this discriminator, the input

illustrated in Fig. 14-31.

pE

Fbt input
J

^&

ui

c2 - - \r2

1
i
discriminator employing two stagger-tuned circuits.
<

Fig. 14-31.

the carrier frequency fc circuit B is tuned to a frequency that is somewhat

higher than fC) and circuit C is tuned to a frequency that is somewhat
lower than fc
The a-m output from such a circuit is without appreciable
,

owing to the linear

resulting characteristic around the
point fc
Such circuits suffer from
distortion,

the fact that reception

is

possible

at three points, corresponding to

each outer portion of the resonant

curves and also to the center or
desired

linear

operating region.

The response from such a

is

circuit

illustrated in Fig. 14-32.

As seen

in the diagram, the

Fig. 14-32. The a-m output from the discriminator of Fig. 14-31.

output from each circuit

is

passed through

The capacitors
Ci and C2 are equal and have negligible reactance at the carrier frequency. The resistances Ri and R 2 are equal and are quite large. The
a diode detector of the envelope or peak detection type.

C =50ju/uf

b'

L3 -2mh
Fig. 14-33.

center-tuned discriminator circuit.

d-c potential across C\R\ is a measure of the amplitude of the output

from circuit B, and the d-c potential across C2R2 is a measure of the amplitude of the output from circuit C. Also, the total output across a'b'
is then a measure of the difference between the outputs from circuits

RADIO ELECTRONICS

404

and C and has a form

[Chap. 14

curve

like the resultant

and

careful adjustment of the circuit constants,

e B of Fig. 14-32.

if

By

the frequency devia-

is limited to the range between /2 and fx, the rectified potential is an

approximately linear function of the impressed frequency.
Another commonly used type of discriminator circuit 4 is shown in Fig.

tion

14-33.

show that

It is possible to

this circuit is substantially a stagger-

tuned pair 5 and that the results illustrated in Fig. 14-32 also apply

for

this case.

limited analytic solution of the operation of the center-tuned dis-

criminator circuit

is

Mc

and

At the

2.

first
i-f

the series circuit comprising

frequency, assumed to be 4.3

in this circuit,

Xc =

X Ci
XLS
The

Consider

possible.

CL3C4 across terminals

=
=

10 12

2x X 4.3 X 10*
400 ohms
2tt

potential across

4.3
is

10 6

50

800 hmS

lO" 3

50 kilohms

then seen to be

J-X.L3

JK-X-a

+ XC

t)

Consider now the mutually coupled circuit. If the mutual inductance

small, the impedance coupled into the primary circuit is small and
approximately, therefore,
is

E12

r ^_

which becomes, for high-Q

Ri

+ jX L

coils,

E12
(14-60)

jX Ll
The

potential induced in the secondary

is

then

E sec = jwMl = +
If

Ei,

(14-61)

the loading effects of the diode rectifiers are neglected, then

- jX c &.
= + jXctM/Li
~ R2 + jX
+ jX L - jX C

R2

where

Xi =

coL 2

(14-62)

1
,,-

C0C2

Note now that the output d-c potential E a is proportional to the peak
of the envelope of E m and correspondingly the output potential E^ is
-

Sec. 14-16]

FREQUENCY MODULATION AND DETECTION

The

proportional to the peak of the envelope of E^.

potential

405

total output d-c

is

E a%> = E a +

Eob'

'

= Ea Eb

However, the a-c envelopes can be represented

and E 12 namely,

Eob

in

(14-63)

>

'o

terms of the potentials

Eoo

= E oc

Ei Ec

E& c

Ex

E12

E i2
(14-64)

-E c

i,

E12

Ei

Consider the situation when the instantaneous frequency equals the

At this frequency, the secondary circuit is resonant,

carrier frequency.

and the quantity

is

zero.

The

E^

potential

is

then given by the

The

lowing expression, obtained from Eqs. (14-62).

chosen arbitrarily.

M
The

potentials E^,

fol-

positive sign

is

(14-65)

and E^ then have the amplitudes and phase somewhat

accompanying sinor

as illustrated in the

diagram.

Note that since

Eao

Eio

and

Ea
'o

Eb'o

then

E' b.

When the instantaneous frequency is greater than the carrier

X is positive and Eqs. (14-62) may

the secondary reactance

in the

form
T,

El2

03C2L1

The corresponding

Ri

jXi

E 12 r/-e
-Pj0C2LI

frequency,

be written

(14-66)

diagram has the form shown.

sinor

Here, since

EaQ

>

Ebo

Ea >
B

Eb>

is

positive

Eao

then
'

and

it

follows that

E a %>
When

is

the instantaneous frequency

negative and Eqs. (14-62)

M
uCzLi

is less

than the carrier frequency,

become

v
Ri

3X1 =

M
3

EuY/8

(14-67)

RADIO ELECTRONICS

406

The corresponding

sinor diagram in this case

is

[Chap. 14

as shown.

<

Ebo

Ea <

E\,'

Eao

'o

EaV

is

In this case

negative

An analytical solution is also possible,

the details of which more or less duplicate the foregoing semiquantitative graphical solution.
This development proceeds from Eq. (14-62), which is written in the form

_ _

Eo6

X C2 k yX /Li
2

R + jX

Ei2

Also, noting that

R*

+ jX = R
2

2 (l

+j28Q 2 )

and
1

R
and choosing the positive

LoC 2R2

sign,

= Q2

then

kQ
3
12

\U + j28Q
1

(14-68)
2

Combine the foregoing with Eqs. (14-64), and recalling that the audio
output from a peak diode detector is related to the impressed modulated
signal through the detector efficiency (see Sec. 13-4), then directly

Ea =

kQ 2

'

7f

+j 2\L! 1 + j28Q En
2

Eb =

and

'

7?

2\L

+ j2SQ

(14-69)

En
2

Therefore, since

Ea

Ea
i

'b'

Eb'o

then

*-

kQ 2

^| 1+ 4\fer +

j2dQ 2

plot of this expression has the general

The

foregoing analysis

is

form

illustrated in Fig. 14-32.

subject to the very serious limitation that

the potential E 12 does not remain constant with frequency. As a result,

Eq. (14-70) does not represent too well the action over the entire range.

FREQUENCY MODULATION AND DETECTION

Sec. 14-16)

As a matter

by

of fact,

careful design the effect of a varying

serve to improve the performance of the discriminator,

linear range considerably

beyond that given by

The complete expression

407

E i2 may

by extending the

this expression.

for the discriminator

may

be obtained in

the following way: Refer to Fig. 14-34, which isolates the inductively

Pentode

2
The inductively coupled portion

Fig. 14-34.

The

coupled portion of the network.

b
of the center-tuned discriminator.

controlling equations of this net-

work are the following

4r I +

+3

o>C

/2i

+j

(L,

- jwMh =

i!

(14-71)

-juMh

Is

In the neighborhood of the resonant frequency

a>

+ j2dQ )li - jco MI =

-jcooML + R,(l +j'25Q )I =
Ri(l

Now

solve for

RR
1

Ii in

2 (1

terms of

I.

R\R

Dividing both sides by

+ j25Q

(14-72)

(mM)>

+ j25Q +
+ j28Q

(1

+j28Q

2)

11

the resulting expression

2 (l

)(l

j2SQ 2 )

2,

may be written as

result is

+j25<?i)(l +j26Q,)

+
(1

The

these

-j

cooC,

may

WQrQz

be written as
(14-73)

I:

2)

Hence
I:

from which

it

= -jQi

follows that

Ii >5> I.

l )(l

+ j2bQ
+ j28Q +
2

2)

WQiQ*'

(14-74)

Therefore
.

(14-75)

Ii

so that

El2

=
(1

+j28Q

1 )(l

+ j28Q
+ j28Q +
2

2)

VQiQ,

RJ.

(14-76)

RADIO ELECTRONICS

408

Combine

[Chap. 14

this expression with (14-68) to obtain Hob, namely,

kQ 2

E
(1

+j28Q

+ j28Q +

)(l

2)

k*QiQ t

RJ.

(14-77)

Equations (14-76) and (14-77) determine E 12 and Eoi quite accurately

for ordinary primary and secondary values of Q.
Then
It is customary to choose Qi = Q 2 = Q.
1

El2

(1

+ j28Q
+

+j2SQ) 2

kQ*

RJ
(14-78)

kQ
(1

For completeness, the

+J25QY +

kQ*

Ril

ratio Eoi/Ei 2 is noted,

ab

kQ

IL2

+ J 28Q

This latter expressions shows that the form given in Eq. (14-68) is not
However, the first of Eqs. (14-78)
altered by the more complete analysis.

Fig. 14-35. Response obtained from an overcoupled discriminator.

shows that
assumption
there

now

Ea =
'o

E i2

Ea

'b'

+ j28Q
Rtl
^ J3 22^L,l
\Li + j28Q/
+ j28Q) + kQ'Y
J28Q/
V"
\L~
kQ
+ j2SQ
RJ
+j28Qy + kQ X J 2^L l+j28Qj
1

'

17

(14-80)

(1

whence

not a constant but varies with frequency, under the

pentode current I. In place of Eqs. (14-69)

follows

(1

Ef

is

of a constant

finally
1

= vRl!
(1

(I

+ J28Q
+

+j25Qy

+jl2\Lil

kQ'

kQ

+J28Q

This expression gives a good representation of the discriminator curve.

FREQUENCY MODULATION AND DETECTION

Sec. 14-17]

409

complete discussion for the determination of the optimum values

kQ and -v/L 2 /Li will not be undertaken. Such an analysis shows that
kQ = 1 and y/L 2/Li = 1 are desirable values. If kQ is less than unity,
of

which

is

the condition for loose coupling, the linear portion of the disIf kQ is greater than unity, which
is relatively small.

criminator curve
is

the overcoupled condition, the results obtained have the form

illus-

trated in Fig. 14-35.

14-17.

The

The Ratio Detector. 6

ratio detector has

found extensive

use in f-m receivers, since a very good noise-free performance

is

possible

with relatively few components, and with no requirement for limiter

circuits.
To understand the basic operation of the ratio detector, refer
This is not the complete circuit of the device, but it does
to Fig. 14-36.

b
Fig. 14-36.

The

(\\)T2

b'

basic ratio-detector circuit.

permit a direct discussion of its operation. It should be noted that this

circuit is quite similar to the Foster-Seeley circuit of Fig. 14-33, except
that the diode Tl

is

reversed in polarity.

E = E volts at
across R is E- = E

Suppose that the potential developed across Ri

the center frequency.
volts.

The net

Seeley circuit.

The

potential

potential developed
3

is

2E,

instead of zero, as in the Foster-

is detuned slightly, the potential

AE, while that across R 2 changes to

the input signal

If

E +
potential E

across Ri changes to, say,

is

2E. Now, however, the

potential across R has changed by an amount AE and may be used
Observe that since E has not changed, the capacitor
as an audio output.
AE.

The net

remains

C may be made very

capacitance of 10

The

fii

large

and

is

usually an electrolytic capacitor with a

or more.

foregoing paragraph shows that the output potential E$ remains

It is now of interest to examine the

change of amplitude of the input signal which occurs at an
audio rate. The presence of the large capacitor C prevents E 3 from
varying over the audio cycle. Similarly, the potential E 2 cannot follow
variations in amplitude of the input signal.
Thus, since E 2 is independent of amplitude variations in the input, it is expected that changes in E 2

constant for changes in frequency.

effect of a

RADIO ELECTRONICS

410

[Chap. 14

be proportional only to the ratio of the magnitudes of the potentials

two diodes, which in turn depend on the frequency.
To show these results analytically, it is convenient to use certain of
the results of the analysis of Sec. 14-16.
By comparing Figs. 14-36 and
will

applied to the

is seen to lie in the polarities of E a > and

two circuits. However, in both cases E12 and Eoj are the same
and Eqs. (14-78) apply here. For the ratio detector

14-33, the essential difference

'

of the

as before,

Ea

'b>

= Ea

+E

ob -

'o

T)(\E

\Ebol)

and

Ea v =

fiRiI
(1

1 + J28Q
+ J28Q)* +

/!

fcQ 2

E_

\L]

.1
J

kQ

+j2SQ\

\U

kQ

\\

yjia +J2SQ

(14-82)
\)

This expression shows that the total output differential potential is a

function of the tube current /, and so of the input-signal amplitude.
In so far as the variation with frequency is concerned, this expression
varies very slowly,

and over the range

of operation

it

is

substantially

constant.

Now

examine the

ratio
.

Eg'

Eb

+J

'

-2

o
2

Ls

kQ
1

+ j28Q

(14-83)

kQ
1

+ j2dQ

This ratio is independent of /, and so of the input-signal amplitude.

Also over the range of operation the expression on the right approximates
to

const

E,ob'

where

is

a constant.

Ah

(14-84)

This expression shows that the ratio depends

directly on the frequency deviation

from the center frequency.

is shown in Fig.

The more complete form of the ratio detector

The changes that have been introduced in this

14-37.

circuit are designed to

overcome the tendency of the diodes to cutoff. This is precisely the

problem that exists for the simple peak diode detector, and as a result
the load resistors must be made quite low, the ordinary values ranging
from 2 to 10 kilohms. Actually a single load resistor R 3 is used, since the
diodes are essentially in

series.

should be noticed also that, because of the low load resistance Rz

with its consequent heavy primary loading, it is expedient to tap down
on the primary winding. While this produces a loss in secondary potenIt

Sec. 14-18]
tial,

FREQUENCY MODULATION AND DETECTION

411

the reduced loading more than overcomes this loss of potential.

While the change discussed effects a substantial improvement, a sudden

change in applied-signal amplitude will still result in diode cutoff.
The second change that is made is the addition of a second time conThe resistances Rt and R 2 are
stant, represented by R1C1 and i? 2 C 2
and
smaller
than
and
C 2 are essentially i-f by-pass
much
R
ordinarily
Ci
s
.

capacitances.

The

C 3 may now

d-c potential across

with a large reduction in input amplitude,

stantially reduced.

Because

and

discharge rapidly

cutoff likelihood

is

sub-

of this loading, the ability of the detector

to reject a large increase in input amplitude

is

slightly impaired.

In addition to the elimination of the need for the limiter with the
ratio detector, the small signal

fact that the limiter

Fio. 14-37.

detector

may

is

a-m

rejection

is

so improved, owing to the

eliminated, that the required input to the ratio

The

practical

form

of the ratio-detector circuit.

be considerably smaller than with the discriminator for

noise-free operation.

14-18.

The Bradley Detector. 7

tode which combines

The Bradley

detector

is

the function of limiter, discriminator,

a single hep-

and

amplifier.

It operates essentially on the locked-oscillator principle, a local oscillator

being forced to synchronize with the incoming f-m signal, the synchronization current being used as a measure of the frequency deviation of the
oscillator

from the mean or carrier

level.

typical Bradley detector circuit

is

illustrated in Fig. 14-38.

essential oscillator-circuit elements of this detector circuit are

The

redrawn

This circuit shows a simple

normally at the center i-f
operating
Colpitts oscillator circuit which is
tank circuit, the effect
second
tank
is
a
Coupled to the oscillator
level.
reactance
into the tank.
inductive
an
of which will be shown to couple
for detailed consideration in Fig.

14-39.

is determined by the two capaciinductance of the oscillator

and
effective
the
tors Ci and C\ in series
is contained in the plate
component
tank.
It will be shown that an a-f
by the signal.
controlled
moreover
that
its
amplitude
is
current and
complete analysis
the
signals
involved,
Owing to the amplitudes of the

The

resulting frequency of the oscillator

RADIO ELECTRONICS

412

[Chap. 14

would follow that employed in discussing the operation

of

an

oscillator.

Instead of a complete analysis, a simplified analysis is included here,

although this is sufficient to indicate the general aspects of the circuit
behavior.

Consider Fig. 14-40, which shows a portion of the tank circuit

A-f

Ebb
Fig. 14-38.

Fig. 14-39.

The

The Bradley

oscillator

detector.

and reactance varying winding.

e2

Fig. 14-40. Oscillator tank Li

of the oscillator

and coupled tank

circuit to the right of a-a.

Qtv-ws

and reactance winding.

circuit.

This

is

the portion of the

It is desired to evaluate the input

to this network.
If

the secondary circuit were absent, then evidently

Zi

jwLi

impedance

Sec. 14-18]

To
be

FREQUENCY MODULATION AND DETECTION

413

when the secondary circuit is coupled, use will

fundamental component of Ij> is in phase with
tank potential Ei but is controlled in amplitude by the

find the value of Zi

made

of the fact that the

the oscillator

This fact

signal.

be

will

Tentatively, therefore, I p

justified later.

is

written

= -gJLi

IP

(14-85)

Here, as for the oscillator, g m is a variable factor, and its value is dependent on the signal conditions.
It is assumed that the secondary is very heavily loaded so that the
secondary tank appears as a resistance R 2 over the range of operating
frequencies.

In practice

chosen small and provides a bandwidth

It will also be assumed that

i2 2 is

approximately six times the operating range.

the coupling

small and that

is

affected by the current

since the coupling

With

is

Ii.

I,,

is

large, so that the current I 2 is

This latter

is

made fairly tight in

not

actually a poor approximation,

order to provide good sensitivity.

these assumptions, then

E2

21P

gm RiEi\

Also

E2

(jmRl

juL 2

JC0L2

From

E,

the diagram

jwMl =

jojLJ.i

Combine equations

Ei

to get

from which

El (l

- fegM)

or

Ldi = Ei (l
The input impedance

is

Zl
is

gm

yj~

then given by

= j"Ll

which

=
pfr \ZL
/7-7f
gmRik
1 /L 2

(14 - 86)

written as
Zi

juLi

gm R

(14-87)

where

RADIO ELECTRONICS

414

The total oscillator tank capacitance (C and C\ in

The oscillator frequency is given approximately by
i

O)o

Ordinarily x

is

coo

[Chap. 14
series) is

denoted C.

Vcw(i -

x)

y/cu

small compared with unity, and, with good approximation,

-vkO-^g)

(14 - 88)

This expression shows that the oscillator frequency varies linearly with
g m since gm is the only variable term in the right-hand member.
According to Eqs. (14-87) only a reactance is reflected into the primary
circuit.
With conventional design, as is expected, a resistive term is also
With the proper adjustment of the circuits,
reflected into the primary.
the resistive term may be made invariant with the value of g m
The effect of the signal potential es that is applied to grid 3 is to be
examined. To do so, refer to Fig. 14-41, which shows a portion of the
complete circuit. Recall that it is the
function of grid 1 to control the cathode
,

current.
of the
plate.

Also, grid 3 controls the portion

cathode current which reaches the

That portion of the cathode current

which does not reach the plate is collected

by the screen. Recall also that, since grid
1 is

Fig. 14-41. The heptode in the

Bradley detector.

part of the oscillator circuit, the cath-

ode current w ju CO nsist of current pulses of

The waveshapes
e usuaj c j ass q variety.

at various points of the oscillator are precisely those

which

tuned amplifier and have the shapes illustrated in Fig.

important waves for the fundamental-harmonic component
class

and the

exist in the
10-8.

The

of current

grid driving potential are

ik

and

eg

= I km cos w
= E gm cos co

(14-89)
<

these two waves being in phase.

In so far as the current reaching the plate is concerned, this is affected

From a curve such as Fig. 14-15b the instanof grid 3.
taneous value of gm and so the effective g m of Eq. (11-14) or g m of Eq.

by the potential

depend on the
If these potentials are in
and the potential e g
As the relative phase varies, the
phase, the plate current is a maximum.
plate current also varies and becomes a minimum when the two potentials
(11-21),

and

in consequence the current to the plate, will

phase of the potential

es

are 180 deg out of phase.

FREQUENCY MODULATION AND DETECTION

Sec. 14-18]

Suppose now that the phase

mum.

of e s is

The corresponding value

frequency, from Eq. (14-88),

the phase of e s

is

a maxi-

maximum, and the oscillator

some particular minimum value. Let

of g m is a

at

is

such that the current

415

now

vary, with a resulting decrease in the plate current ip

reduced value of gm results in an increase in the oscillator frequency.

The
With a properly designed system, the phase

of the

system

will

vary until

the plate current, and so the value of g m will assume such a value that
the oscillator will synchronize, or lock in, with the signal frequency.
,

If

the signal frequency changes to a value beyond which the system can

"breakout" will occur.

Suppose that the circuit is properly designed and adjusted so that
lock-in does occur between the signal frequency and the oscillator.
It
still remains to show that the audio-modulating signal can be recovered.
To do this, examine the Fourier expression for the amplitude of the
fundamental component of plate current. This is given by the standard
follow,

expression
Ipim

Now

Tp

T
ip cos

<a t

dt

(14-90)

ip is of sufficiently short duration, then the

be approximately unity during the time when ip has
any important value. As a result, the integral is given, with good approximation, by the form
if

the current pulse

factor cos

cc t

will

Ipim

2
-j,

T
ipdt

(14-91)

J
form relating the d-c component

of

the plate-current pulse with the functional form, except for the factor

2.

But

it is

noted that this

is

just the

Therefore
/*,

now

0.57 plm

(14-92)

between the signal frequency

It is
follows:
d-c
plate
current,
as
By
Eq. (14-88) the oscillator
and the
(1)
If
remains
fixed
frequency is linear with g m
e
in peak value, which
(2)
g
since
not
vary,
then
varies
directly
does
does
as
ip [see Eq. (14-85)].
it
ei
gm
possible to argue a correlation

(3)

Idc varies linearly

with I plm [Eq. (14-92)].

Idc varies as

To make

Therefore

the frequency of

es

(14-93)

necessary only to include a

resistor in the plate circuit of the heptode, across which will appear the
practical use of this result,

it is

audio signal.

The

a-m

rejection in the Bradley detector

is readily underwith the incoming signal,

and suppose now that the amplitude of the input signal changes. The
plate current i p will change, since i p depends on the amplitude of the

stood.

process of

Suppose that the

oscillator is locked in

RADIO ELECTRONICS

416

on grid

signal e s

3.

When

be reached.

Such a change

new

frequency, and a

will

[Chap. 14

momentarily change the oscillator

e and e g will

equilibrium phase condition between

new

the

equilibrium condition

is

reached, the oscil-

be that of the signal. That is, the amplitude

variations are translated into phase variations of the oscillator, but not
into frequency variations.
Since the normal modulation produces many
kilocycles per second of frequency change, the phase variations due to
amplitude changes which occur at a relatively low frequency rate do
not produce a significant audio
lator frequency will

still

Anode

output.

Quadrature grid-

14-19.
Limiter grid-

The Gated-beam Tube

The

as a Limiter-Discriminator.

cross-section schematic of the gated-

beam
ment

tube, showing the arrangeof elements,

is

given in Fig.

In normal operation, the

plate and accelerator are operated
at positive potentials, with the lim14-42.

Accelerator-

Cathode

Fig. 14-42. Schematic of the gated-beam

tube showing arrangement of the

ments.

ele-

iter grid

and the quadrature

grid

beam current.
from the cathode are accelerated by the accelerexercises complete control of the electron beam
serving to control the

In this tube electrons

The

ator.

limiter grid

and determines the extent

structure.

The character

graphically in Fig. 14-43.

grid potential

is

this figure that when the limiterbeam reaches the plate without
grid is made more than a few volts

Observe from

positive, the electron

However,

difficulty.

of the electron drift out of the accelerator

of the control of the limiter grid is illustrated

if

the limiter

negative, cutoff of the plate current occurs.

feature of the design of the tube

is

that

when

the limiter grid

is

made

negative, this tends to produce a space-charge cloud in the accelerThis space-charge cloud causes a reflecator-limiter grid-shield region.
tion of the cathode current, which diverges and is collected by the
That is, very few of the electrons are able to return to the
accelerator.

cathode owing to the concentration of space charge and the divergence

of the

beam

As the

path.

electron

beam emerges from

the accelerator,

it is

converged by

the electron-lens system, comprising the accelerator structure and the

various grids, and is acted upon by the quadrature grid. The characteristics of

as

the quadrature grid are somewhat like those of the limiter grid,
The design here too is such that, with the appliin Fig. 14-44.

shown

cation of a negative potential to the quadrature grid, the electron beam

reflected away from the quadrature grid and returns to the outside

is

surface of the accelerator structure.

FREQUENCY MODULATION AND DETECTION

Sec. 14-19]

417

'00
'olts

_Eb *130 volts I

Equad" 3 volts 1
1

TP

>l+s
1

55

40

Q.

ol

15
20
10
5
Limi+er grid voltage in vol+s

-10

25

30

Fig. 14-43. Plate current vs. limiter grid potential, with accelerator potential as

parameter.
1

E*'6.3 volts

Eb '6L

E,,^ +3vo Irs

volts

+1

-1

-1.5-

-10

10

Quadrature

20

30

40

grid voltage in volts

Fig. 14-44. Plate current vs. quadrature grid potential, with limiter grid potential as a

parameter.

to the control characteristics of the limiter and quadrature

appears that both exert nearly step-function control over the
plate current.
In effect, therefore, each grid acts like a potential-conIf either or
trolled gate.
If both gates are open, plate current exists.

Owing

grids, it

both

of the gates are closed, the flow of plate current is blocked.

RADIO ELECTRONICS

418

[Chap. 14

When

used as a limiter-discriminator, the circuit is substantially the

The cathode resistor R k is so adjusted that the
self-bias sets the plate current on the curved portion of the characteristic.
The i-f signal then alternately drives the tube from the condition of
following (Fig. 14-45).

plate-current cutoff to plate-current saturation.

The

result

is

essentially

a square-wave variation of beam-current density beyond the limiter grid.

As indicated in Fig. 14-45, a high-Q parallel-resonant circuit is connected to the quadrature grid. The circuit is tuned to the center freof the f-m wave.
The current pulses transfer some energy to the
quadrature-grid tank circuit, which produces a quadrature-grid potential,

quency

Audio

Anode
Quadrature grid
Limi-re r grid

SOOkn
I-f input

RFC
Accelerator

Rl*Plate load resistor

Cathode

Rd* Plate linearity resistor

Ci 'Integrating capacitor
Cc * Coupling- capacitor
Fig. 14-45. Circuit of the gated-beam limiter-discriminator.

which also acts on the beam current. Because a phase difference exists
between the limiter-grid potential and the quadrature-grid potential,
only a fraction of each square-wave beam-current pulse reaches the plate.

When

a center-frequency signal

is

applied to the limiter grid, the

potential developed on the quadrature grid lags the input signal potential

by approximately 90

deg.

pulse reaches the plate.

In this case, half of each square-wave current

When

the signal to the limiter grid

is

slightly

above the center frequency, the potential on the quadrature grid lags the
input signal by less than 90 deg. Now less than half the square-wave
current pulse reaches the plate.

Correspondingly, for a frequency

less

than the center frequency, the limiter grid -quadrature grid potentials are
more than 90 deg apart, and more than half the square-wave current pulse
reaches the plate. An essentially linear relationship exists between

Sec. 14-20]

FREQUENCY MODULATION AND DETECTION

419

frequency deviation and average plate current, within the desired range
Therefore, the modulation
of frequency deviation, as seen in Fig. 14-46.
content of an f-m signal is available as variations in the average plate
current.

Limiting is accomplished because the beam current is limited in passing

through the accelerator assembly. That is, if the minimum signal to
be received appears on the limiting grid with sufficient amplitude to
drive the tube from cutoff to saturation, the desired square-wave current
The only differences that might occur due to differpulses are produced.
ences in amplitude of the input signal will be in the slopes of the pulse
sides.
Consequently, limiting and a-m rejection are realized.

Frequency deviation

in

kiiocyc/es

Pig. 14-46. Frequency deviation vs. audio output potential.

14-20. Comparison of Modulation Systems.

certain significant factors of

It is desired to

a-m and f-m systems,

in order to

examine
compare

the two systems of modulation.

Amplitude-modulated transmission is the oldest method of radio transA considerable amount of

is still used very extensively.
f-m transmission exists, although it represents a relatively small part of
the total transmission. Amplitude modulation is used over a very wide
range of carrier frequencies, extending from the very low (say several
hundred kilocycles) to the ultra-high-frequency bands in the hundreds
of megacycles region.
While a-m transmission is inferior to f-m transmission when considered from the point of view of receiver noise and
certain transmitter considerations, it has not been displaced by f-m
transmission owing to many factors, a number of which will be examined,
mission and

RADIO ELECTRONICS

420

[Chap. 14

Consider first the transmitter output stages. The r-f amplitude of an

f-m transmitter is a constant, independent of the modulation index,
whereas the peak power will be four times the mean value in an a-m
transmitter with modulation index m = 1.
For a total fixed allowable

and for fixed plate-circuit efficiency, the output power

f-m transmitter with given output tubes will be twice that if a-m
employed.

plate dissipation
of the
is

The signal/noise ratio in the receiver is very important in the over-all

comparison. It will be found that the f-m system possesses distinct
advantages over the a-m system. Likewise, in the matter of interference
by two transmissions on the same carrier frequency, the interference is
far less with frequency modulation than with amplitude modulation.
To compare the relative effects of internal noise on f-m and a-m
systems, the usual approach is the following: The effect of a single
component

frequency is examined for each

then generalized to account for the complete
band of frequencies contained in the noise spectrum.
sinusoidal

system.

The

of noise of arbitrary

results are

Consider
2nfnc

first

carrier signal

an a-m system. The input

specified by the peak ampli-

tude

will

it

is

Also,
c and the angular frequency to c
be supposed that there is a single fre.

quency noise or interfering signal, specified by

ftnd
j t wiU be assume d that E c
En

Fig. 14-47. The combination

of signal and noise
an a-m

The

system.

resultant signal at the input of the receiver

is the combination of the two, and they combine in the manner illustrated graphically in Fig. 14-47. Analytically the

resultant signal

is

eB

But

since

oi

= En

+E

cos wt

cos

does not differ too markedly from

0>c

co r

(14-94)

it is

convenient to write

U>nc

(14-95)

(14-96)

Also, for convenience, write

E n = aE
where, as specified above, a <K

1.

Equation (14-94)

may

then be written

as
eB

which

may

= aE

cos

{u> c

+E

cos w c t

aE

sin

u nc )t

(14-97)

be recombined to the form

eR

Ec(l

a cos

u> ne t)

cos u c t

a>,.<

sin

wct

FREQUENCY MODULATION AND DETECTION

Sec. 14-20]

This

may

be written as

= E VTl

en

where

<p

a cos

co c t)

tan

_.1

a 2 sin w nc t cos (w c t

a sin

+ a cos

For the specified condition that a

en
<p

<p)

(14-98)

u> nc t

co c

As

421

1,

these latter equations reduce to

= Ec (l + a cos u nc t) cos (co < + <p)

= tan -1 (a sin co nc ) = a sin cc nc t
c

(14-99)

magnitude of the resultant signal E R

If this
through E c E n and back to E c + E n
passed through an ideal peak detector, the detected output is the
E n which varies at the rate

illustrated in Fig. 14-47, the

varies from
signal

is

value of

E + E

Limifing

' nc

circle

If

the peak detector

is

followed

by

an audio amplifier which has a flat

frequency response with a bandwidth
/2-/1, then those noise components
with frequencies near enough to the
^
"~"~,T^
,
, m
.
Fig. 14-48. The combination of signal
,
carrier frequency fc> so that the de.
and noise in an f m system
tected noise frequency / is within
the frequency range /2-/1, will produce an output noise signal.
Noise
components of equal amplitude within the amplifier frequency band will
produce equal outputs.
If the carrier is amplitude-modulated 100 per cent, the peak signal
output, which is again designated as E c appears in the peak detector
,

output.

Therefore, the signal/noise ratio for a single noise

component

is

given by

(100% a-m) =

(14-100)

Now

suppose that the same carrier and noise component are applied
The combination of potentials indicated in Fig. 14-47
and given by Eqs. (14-99) still applies. Now, however, Eqs. (14-99) are
to an f-m system.

written as the single expression

e

= E

(l

a cos w nc t) cos (w c t

a sin u nc t)

(14-101)

interest is in the phase between E c and E, rather than

amplitude variations, which are removed by the limiter. The
situation is now represented in Fig. 14-48 and by Eq. (14-101) the input
to the receiver is an f-m wave with deviation ratio a.
The phase displacement between E c and Er is a sin w nc t, as indicated in the figure.
But the time rate of change of this displacement represents a departure

Note that now our

in the

RADIO ELECTRONICS

422
in

frequency of

Er due

Hence, writing

to noise.

[Chap. 14

Aa> as this departure in

frequency, then
Aco

Because

of

= u nc a

w nc t

cos

(14-102)

the limiting, the only effect of the noise component which

Consequently,
is the frequency variation Aco.

reaches the f-m detector

the amplitude of the noise

fnc

directly proportional to fnc

is

and

is

zero

when

0.

The frequency deviation

is fd

signal/noise ratio for a single

and the

for 100 per cent modulation,

component

| (100% f-m)

of noise

is

k fi

(14-103)

comparison of this expression with Eq. (14-100) for the a-m case shows
But here, as for
is larger until / c becomes equal to fdthe a-m case, noise components producing frequency components greater
than / 2 are eliminated by the audio system. Hence only those noise
components which cause fnc to be equal
that this value

to / 2 are important.

comparison of the effect of noise

an a-m system with that in an f-m
system is portrayed effectively in Fig.
D
\F
14-49.
Consider an a-m receiver which
i2
is provided with an ideal i-f system
Clearly,
Fig. 14-49. A graphical representswhich has a bandwidth 2/2
tion of the effect of noise in a-m and
Dotn t h e i-f and the a-f amplifiers will
f-m systems.
remove any noise outputs which have
been generated by components outside of 2
Also from Eq. (14-100) the
signal/noise ratio is independent of /.
Thus the rectangle OABD represents the constant effect of noise for all frequencies for which f c < f2
In the case of an f-m receiver, according to Eq. (14-103), the relative
effect of noise increases linearly and reaches E n /E c when fnc = fd
This
is represented by the triangle OEF.
In this case, the i-f bandwidth
But the a-f system again removes noise components above
equals 2fd
in

/2

Hence the

resultant

is

represented by triangle

OCD.

important

It is

components in the range from/2 to/,i produce signals after

detection which are within the a-f band.
This situation arises from the
fact that the original i-f components were produced by the audio signal
within the bandwidth / 2
To compare the two systems quantitatively, attention is first called
to the fact that the noise is continuous over the audio band, whence an
infinite number of noise components are contained in the band +/2
centered about the carrier frequency fc
If E n (fnc ) dfnc denotes the peak
to note that

i-f

amplitude

of

the noise potential in the frequency range dfnc then the

,

Sec. 14-21]

FREQUENCY MODULATION AND DETECTION

total noise in the

fnc

Ji

and

a-m output

is

423

that contained in the frequency range

is

&(wn) =

= En

V //' El dSnc

Vh

(14-104)

In the f-m case, it will be assumed that the detector recovery constant is
K, which relates Aco of Eq. (14-102) to the actual output potential.

Then

K^J

(14-105)

nc

It follows therefore that

E N (i-m) =

^ j* 0|J KTn

dSnc

= K^f*-yjj

Suppose that 100 per cent modulation is assumed in each

a-m system is

(14-106)

case.

Then

the signal/noise ratio for the

i (a

"

m) =

Also, the signal/noise ratio for the

s
Consequently the ratio

(f'

m) =

of these

(14 " 107)

bTvT,
f-m case

is

given by

m.mhVWi
two quantities

S/N (a-m)

(14- I08)

is

Si

This result shows that, for a deviation ratio /d //2

improvement with an f-m system

For a large deviation

ratio,

is

or 4.8 db.

the signal/noise

75 kc, / 2 = 7.5 kc, whence

correspondingly higher and is 20 db.

say, for example,

improvement

is y/2>

1,

/<*

10,

the

very interesting and important feature of an f-m

This is the tendency of the f-m system to
ignore the weaker of two signals of nearly equal amplitude and equal or
nearly equal frequencies. To examine this matter, consider the desired
signal to have twice the amplitude of an unmodulated interfering one,
The desired signal may be
the average frequency difference being
14-21. Capture.

system

is

known

as capture.

modulated, but if /, > /m the conditions will be considered for a period

which is short compared with /m so that the frequency of the desired
modulated signal is essentially fixed. For added convenience, the situation will be examined at the time when the modulated signal is at the
instantaneous frequency /c which occurs at a zero point in the cycle
of the modulating frequency.
The situation discussed may be depicted
,

RADIO ELECTRONICS

424

The

graphically, as in Fig. 14-50.

for the discussion of noise

is

[Chap. 14

similarity of this figure to Fig. 14-48

evident.

Figure 14-50 is a representation of the desired signal plus the interfering

Also shown is the limiting circle, which specifies the output after
signal.
the limiter stage. Note that, owing to the interference, the resultant

However, as
the value of the signal E s
position
of
E
is
just
that of E
angular
average
long as Ei < E s the
B
change
the
average
frequency
of the
signal
not
does
and the interfering
signal

may

deviate

6 from

desired signal.

The phase

variation, or perturbation, occurs at the rate

ji.

But

if fi

is outside of the audio range, the variations in frequency of E R are not

In the ideal case, the detector responds only to E as long as
heard.

Fig. 14-50.

The

sinor representation of a desired signal plus an interfering signal.

Ei < E and as long as/< differs in frequency by more than/ 2 of the audio
system.
If the interfering-signal frequency /; is in the audible range, the phase
perturbation will produce a frequency deviation which

The equivalent frequency modulation that

is

produced

may
is

be heard.

small, since

Eq. (14-96)]. In any case, therefore, the

even if the signal and the interference are at
In practice, most f-m detectors will ignore
exactly the same frequency.

dd/dt

is

proportional to

interference

is

/< [see

negligible,

the interfering signal as long as the ratio of the desired- to interferingsignal amplitude

is

perhaps 3 or greater.

REFERENCES
1.

2.
3.

Crosby, M. G., RCA Rev., 5, 89 (1940).

Morrison, J. F., Proc. IRE, 28, 444 (1940).
Travis, C, Proc. IRE, 23, 1125 (1935).

D. F., and S. W. Seeley, Proc. IRE, 25, 289 (1937).

Arguimbau, L. B., "Vacuum Tube Circuits," pp. 486-494, John Wiley

4. Foster,
5.

&

Sons, Inc., New York, 1948.

Sturley, K. R., Wireless Eng., 21, 72 (1944).
6. Seeley, S. W., and J. Avins, RCA Rev., 8, 201 (1947).
7. Bradley, W. E., Electronics, 19, 88 (October, 1946).
Television Project Engineers, "Introduction to Television Receiver Design,"
Philco Corporation, Philadelphia, 1951.

FREQUENCY MODULATION AND DETECTION

425

PROBLEMS
14-1.
of the

Determine and plot the instantaneous frequency corresponding to each

phase functions in the

figure.

Sine

<P

HM

Parabola

*-

Write an
14-2. A 100-Mc f-m signal is modulated 75 kc at a 400-cps rate.
expression for the instantaneous potential if the signal amplitude is 10 volts and
sec.
if both the frequency and instantaneous magnitude are a maximum at t =
14-3. A wave is frequency-modulated at an audio rate of 5,000 cps.
If the
frequency deviation fp is 75 kc, plot the spectrum of the wave, including all significant sideband

components.

of a 15-kc wave causes a 75-kc frequency deviation of

Plot the spectrum, and calculate the bandwidth required to pass
all sidebands of appreciable magnitude.
b. Suppose that the amplitude is altered to give deviation ratios of 3 and 1.

14-4. a.

The amplitude

an f-m wave.

Repeat part a

for these

two

cases.

these results, estimate the value of the deviation ratio that may be
used and still be within the FCC limitations of +75 kc frequency band spread.
14-5. Consider the reactance-tube circuit shown in the diagram.
Show that
c.

From

by choosing L properly Z
tion.

Compare with Eq.

is

capacitive and

is

given by C/g m R without approxima-

(14-32).

14-6. Given the reactance-tube circuit illustrated at the top of page 426.
that if the reactance of C is negligible at the operating frequency, and if coL
the effective input impedance results from an inductance L/g m R.

Show
Ji>

R,

RADIO ELECTRONICS

426

14-7. Consider the reactance tube

this problem.

of

10

The

oscillator

in g m

is

shown

in the

diagram

for

Mc

with a frequency deviation

required to achieve the desired frequency

oscillator is to operate at 5

What change

kc.

and

[Chap. 14

modulation?

14-8.

Assume that between the

teristic of a

6SJ7 tube

may
gm

limits of

2 and 6

volts the (g m e c i) charac,

be represented by the expression

3,000

500e c

^mhos

is connected as a reactance modulator, as illustrated in the accompanying figure. It is desired to have a center or carrier frequency of 5 Mc and a frequency deviation of 7.5 kc. Determine the correct setting of the oscillator tank
capacitance and the required modulating potential.

This tube

14-9. Carry out the analysis to show that Eq. (14-33) does give the equivalent
output admittance of the inductive reactance circuit.
14-10. Obtain an approximate expression for the variation of the frequency
of an f-m oscillator as the potential on the grid of an inductive reactance tube is
varied.
Proceed in a manner analogous to that employed in the text in obtaining
the corresponding expression for a capacitive reactance tube.
14-11. The essential circuit of a balanced reactance tube is given in the accompanying figure (battery supplies have been omitted for convenience). Deduce
an expression for the effective output impedance of this circuit. In this calcu

lation,

FREQUENCY MODULATION AND DETECTION

Also, suppose that for a
choose Xi/Ri = X /R - a, r =

audio signal the

(g m

,e c )

relationship

is

the ideal curve

shown on page

427
given

391.

Oscillator

IT

14-12. Calculate and plot the potential ratio for the preemphasis or accentuator
shown in the diagram as a function of frequency. Plot the curve on semi-

circuit

logarithmic paper.
0.00/u.icf

-KJOOkn

\E2

14-13. Calculate and plot the potential ratio for the deemphasis circuit shown
accompanying figure as a function of frequency. Plot the curve on semilogarithmic paper.

in the

lOOkn
vwwa.

E2
'0.00/nfif\

14-14. Determine the maximum frequency deviation possible with the ArmThe
strong system of Fig. 14-22 if the distortion is to be less than 6 per cent.
phase varies at a 60-cps rate.
14-15. What must be the ratio of sideband to carrier potentials in the Armstrong system to produce a frequency deviation of +12.2 cps at the audio frequency of 400 cps?
14-16. Show by an analysis similar to that which leads to Eq. (13-6) for the
a-m case that the output of a frequency doubler to which an f-m signal is applied

the f-m signal centered about the second harmonic of the carrier.
14-17. An f-m wave of the form given in Eq. (14-15) is combined with a large
amplitude sine wave Eo sin u>4. Show that if both waves are applied to a rectifier, the output will contain the f-m wave shifted in the frequency scale.
The
14-18. a. An incoming f-m wave is deviated +75 kc at a 10-kc rate.
The signal is passed
local oscillator is so set that the i-f frequency is 13.7 Mc.

is

RADIO ELECTRONICS

428

[Chap. 14

through two stages of single tuned amplifiers. Discuss the effect of the tuned
on the important sidebands with respect to amplitude and phase.
b. What bandwidth should the amplifier possess if all sidebands of amplitude
greater than 0.01 per cent are to be passed, but with not more than a 30 per cent
reduction in amplitude of any sideband being permitted.
circuits

An

14-19.

antiresonant circuit consists of a capacitor of 65 /x/if and an inducand 16 ohms. It is to be used to receive a wave having a frequency
modulation of 1.5 kc. What should be the value of the carrier frequency? Estimate the percentage modulation of the output.
14-20. Suppose that a discriminator as shown in Fig. 14-30 comprises two circuits which have bandwidths of 200 kc and are tuned approximately to 4.7 Mc.
Plot the resultant discriminator characteristic for the following separation of the
resonant peaks: 150, 200, 250, 300 kc.
14-21. a. Plot the discriminator curve specified by Eq. (14-70) for

mh

tor of 0.4

Write 28Q = x, and plot as a function

b. Repeat for Eq. (14-81).

of x.

14-22. Use the curve of Prob. 14-216 to design

peak bandwidth of 400 kc at 15 Mc.
is

a discriminator

14-23. Show that, other things being equal, the output

down from that of the discriminator by a factor of 2.
14-24. Would it be more difficult to design a broad-band

having a peak-

of the ratio detector

ratio detector than a

discriminator of the same bandwidth? Explain.
14-25. Suppose that the dot on the secondary winding of the Bradley detector
of Fig. 14-38 were at the other terminal (i.e., that the mutual inductance is
Discuss the mechanism of pull-in to synchronization.
reversed).
14-26. Suppose that the i-f signal to grid 3 of the Bradley detector has a
The maximum phase
1,000-cps 40 per cent amplitude modulation superposed.
change is 25 deg. Evaluate the a-f output as a fraction of that produced by a

400-cps f d
14-27.

75 kc f-m signal.
that the ratio of the rms ordinate of

Show

OCD

OABD

in Fig. 14-49 to the

the improvement ratio s/S fa/ft14-28. Reconcile Eqs. (14-98) with the statements concerning variation of the
noise output with the bandwidth.
14-29. An a-m transmitter having an output power of 100 watts has a transmission-channel bandwidth of 10 kc and a signal/noise ratio of 15 db. Suppose
that it is planned to replace this communication link by an f-m transmitter having
a transmission-channel bandwidth of 120 kc. If it is desired that the signal/noise
ratio be 32 db, what is the required transmitter power?
14-30. A single-sideband suppressed carrier a-m signal is received on an f-m

rms ordinate

receiver.

of

is

Discuss the character of the receiver output.

CHAPTER

15

INFORMATION THEORY
15-1. Introduction.
This chapter will discuss several of the very
important results of "information theory," or the synonymous "communication theory." The discussion is necessarily restricted in scope and
will confine itself principally to certain topics

Many

that relate to transmission

mathematical
have been sacrificed in the interests of simplicity.
Of particular importance is the realization that it is possible to define
and measure the quantity of information that is contained in a message.
Moreover, it is possible to measure -the capacity of a communication
channel to transmit information. Since the definitions and measures are
applicable to all modulation systems, it is therefore possible to compare
the relative efficiencies of different modulation systems.
15-2. Discrete Systems.
Before introducing numbers or formulas,
certain fundamental considerations are important.
Consider initially
systems.

of the finer points of the theory, as well as

rigor,

Fig. 15-1.

signal

the term "message."

wave that

is

sampled at n points

in

In the discrete case, a message

to be a sequence of choices of possible symbols.

the period T.

may

The

be considered

possible symbols

the symbol alphabet, sl specified number of such letters of the

alphabet making up a message. For example, in ordinary speech or writIn telegraphy, the
ing, the alphabet may be the letters or the words.
alphabet will be the dot and dash. In music, the alphabet will be the
successive amplitudes of the signal wave.

make up

429

RADIO ELECTRONICS

430

[Chap. 15

Refer to Fig. 15-1, which shows a general signal of time duration T.

supposed that the signal is sampled at n points which are uniformly
spaced in time. A signal such as this may be sent from a transmitter to a
It is

receiver as the sequence of the sampled values.

For the signal to carry
information, there must be prior agreement between the transmitter and
receiver as to what meaning should be attached to the sequence of values

that have been received.

conditions are such that

what

signal

When
it is

has been done and when noise

this

known from

was transmitted, then

the received signal exactly

the information sent

all

is

received.

In this case the number of different sampling levels L makes up the symbol
alphabet, and a message may consist of the sequence of n sampling values.

The maximum

possible total

number

of different messages

which are

the different permutations of sampling values are not

distinguishable in a transmission channel, then the maximum rate of
transmission of information in the channel is greatly reduced. This
possible

is

problem

will receive

L".

If

now

Consider

some attention below.

the term "information in a message."

definition of

As stated by Goldman, the

a message is denned as

this expression is given in statistical terms.

amount

of

information received in

probability at the receiver of the

after the message is received

event
.ii

Information received

log

(15-1)

t-t-

probability at the receiver of the

event before the message is received

In the noiseless case the receiver

certain that the message received

is

correct, so that the probability of the event after the

is

Thus

unity.

Information

message

is

is

received

for the noiseless case

received

\
/ probability at the receiver of the
\event before the message is received/

'

Suppose now that the message alphabet consists of s symbols or letters

and that the message contains n symbols (not necessarily different) from
the message alphabet. Suppose also that there is associated the probability pi with each symbol s; of the total alphabet of s symbols.
The
average information

these probabilities.

Thus

in each

message

will evidently

be a function

of

in functional notation

= H(p h p

2,

,p n )

Shannon chose the following general and reasonable

be possessed by H
2

(15-3)

properties that

must

1.

2.

=
3.

H shall

be continuous in the Pi
the probabilities p, for each symbol of the set s are equal, then
1/s, and
shall increase monotonically with s.
If a selection of symbols is broken down into two successive selec.

If

INFORMATION THEORY

Sec. 15-2]

431

then
of the original selection of symbols should be equal to the
weighted sum of the H's of the two successive selections.
Subject to these properties, he deduced the following expression for the
average information in the message of n symbols:
tions,

= n

Since

^ p,

1,

may

Eq. (15-4)

pi log pi

(15-4)

=i

be written as

= n
= n

log

n log n2p<
~Lnpi log n

log

Swpi log npi

log n

"Znpi log p,

Snp< log

p;

or

H
Now

(15-5)

symbols (one for which the occurrence of

symbols is controlled by probability), the various messages become
In general, of course,
equally probable, for messages of sufficient length.
fewer
than
different
symbols, and these
n
messages of n symbols contain
frequencies.
In
fact, the number of
will generally occur with different
occurrences of the ith symbol will be
for

an ergodic system

of

n,

(15-6)

npi

Because of this repetition of symbols, the average information will be

reduced. The reduction of information that is due to the repetition of
the ith symbol may be written as

Hi =

m log n<

p,-n

log p{n

(15-7)

[The basis for this expression follows as a consequence of Eq. (15-8) now
being developed.] Thus, in the light of Eq. (15-7), Eq. (15-5) shows
that the average information contained in a message of n different symbols
is

H=
since the reduction of information

been accounted
letters, all

for.

letters

But n"

log

(15-8)

due to the repetition

gives the

being different.

If

number

symbols has
messages of n

of

of possible

M denotes this number

of possible

messages, then

M
and the

total information in the

message

(15-9)
is,

from Eq.

(15-8),

log

(15-10)

Generally, in information theory, the base 2 of logarithms

and the unit

of information is called the binit,

which

is

is

chosen,

a contraction of

RADIO ELECTRONICS

432

Thus Eq.

the term binary unity of information.

[Chap. 15

(15-4)

may

be written

as

n^pi log

binits

Pi

and similarly Eq.

(15-10)

H
Consider

which

now

(15-11)

is

Iog 2

binits

the expression for the average information per symbol,

is

binits/symbol

p, log 2 p,

(15-12)

This expression for information is similar to that for the entropy of a

system which is characterized by states of probabilities pi,
pn
Accordingly, the terminology
in the statistical sense of Boltzmann.
.

^ Pi log pi

entropy of the

set of probabilities pi

(15-13)

has become standard in information theory. The relation to the entropy

concept of statistical mechanics is somewhat indefinite, but the use
That is, the order-disorder
of the entropy concept is a valid one.
applicable
the
process of communicating
to
notion of entropy is directly
information.

As an extreme example, consider the case when one symbol has a

probability of unity, the probability of

all

other symbols being zero.

is zero, as is expected, since there is

According to Eq. (15-13), h

Consider now the case when

uncertainty.

equally probable, and such that

^p =
{

1.

all

no

the possible symbols are

In this case h attains

its

maximum

value.

This condition

of

maximum

entropy

is

since the greatest uncertainty exists in the selection, since

reasonable,
all

possible

symbols are equally probable.

The most significant early contribution to
15-3. The Hartley Law.
the theory of information was that of Hartley, 3 who developed a quantitative measure of the

amount

of information in a message, and,

based on

a measure of the capacity of a communication channel in terms of its

bandwidth. Subsequent work has extended these results to include the
it,

influence of noise.

Suppose that s, the number of letters in the symbol alphabet, is fixed

by the source or by other factors. It is desirable to determine n, the
number of letters in the message, which can be sent in a period of time
T equal to the message duration. That is, it is desired to find the rate
It is anticiat which symbols can be transmitted through the channel.

INFORMATION THEORY

Sec. 15-3]

433

pated, in the light of simple network considerations, that the transmission

depend on the bandwidth

rate will

of the

network.

It is desired to obtain

the relationship that exists.

Refer again to Fig. 15-1.

If

the signal

is

regarded as a single cycle of a

may be used to

repetitive wave, the graphical methods of Fourier analysis

ascertain the amplitudes of the harmonic

components which constitute the

such graphical schedule methods of analysis exist
(the Runge schedule would be particularly appropriate for a general
signal wave.)
For the case of n samples, n amplitudes may be determined. If the d-c component, if any, is omitted (since this can readily
be determined with the aid of a planimeter or by one of several methods of
numerical integration), the schedule analysis will yield the amplitude of
n/2 sine terms and n/2 cosine terms. Thus the amplitudes of the fundamental and all harmonics up to the (n/2)th are known. The period of the
highest harmonic is 2T/n.
The foregoing may be referred to as a
sampling theorem in the time domain, and it permits the specification of
continuous signals by discrete sets of values.
In the more general case, the waveform need not be periodic. Consider
a signal, in the form of a continuous function, which has passed through a
transmission system having a finite bandwidth.
For signals of this type,
Shannon 4 gives a sampling theorem which states

A number

signal.

If

a function

/(f)

of

contains no frequencies higher than

completely determined by giving

its

cycles per second,

it is

ordinates at a series of points spaced 1/2B

seconds apart, the series extending throughout the time domain.

The

proof of this theorem follows directly from considerations of the

Fourier transform and

spectrum

of f(t),

is

now

given.

If F(cc)

denotes the frequency

then
(15-14)

If

is assumed to be zero outside the band, the limits of the integral

be changed to the interval 2vB to +2x5. Thus

F(w)

may

(15-15)

Now let

n/2B, where n

is

any positive or negative

integer.

Then
(15-16)

The left-hand side of this expression is the value of f(t) at the nth sampling
The integral on the right represents the nth coefficient of the
complex Fourier-series coefficient obtained by expanding the function

point.

RADIO ELECTRONICS
series in the range from 2ttB

434

[Chap. 15

a Fourier
to +2irB.
Thus the
values of the sample f(n/2B) determine the coefficients of a Fourierseries expansion of F(u).
Consequently they completely determine F(u)
itself in the specified range.
Moreover, since F(u) is the frequency
F(<a) in

spectrum of f(t), the

coefficients or

samples therefore determine the func-

tion /(f).

Suppose that it is desired to transmit a signal which contains no harmonic components higher than the (n/2)th. Since the period of the
highest harmonic is 2T/n, the bandwidth of the transmitting channel
must be at least n/2 T to include all harmonics. Thus the bandwidth must
be at least

B = $r

15 -!7)

from which
n

now

Consider

= 2BT

(15-18)

the problem of transmitting a message that consists of

off) pulses contained in an

a series of

n equally spaced binary (on or

interval of

sec.

In so far as the transmitting channel

is

concerned,

problem is precisely the same as that considered above, and the

bandwidth of the transmitting channel is given by Eq. (15-17). Moreover, since the number of possible messages with a symbol alphabet of 2
= 2", then, by Eq. (15-10),
symbols, with n letters in the message, is
the total information in the message is
this

H
When

log

Thus the amount

n log 2

is

= n

binits

(15-19)

of information that the transmitting

will carry in

time

H
This

log 2"

referred to the base 2 system of logarithms, this yields

H
width

channel of band-

is

= 2BT

binits

(15-20,

the Hartley law.

The information capacity

of binits per

of the channel, C, is defined as the

number

second carried by the channel, so that

C =
15-4. Hartley's

= 2B

Law and Continuous

binits/sec

Signals.

limitations exist in the Hartley law, which will

now

(15-21)

Certain fundamental
be discussed. In this

connection, refer to Table 1-1, which gives the binary code for a 4-pulse

code in pulse-code modulation.

In this tabulation there

is

seen to be an

Sec. 15-4]

INFORMATION THEORY

435

alphabet of 2 symbols, and each message consists of 4 letters. The total

number of possible messages is 2 4 = 16, as shown. Note that each

message specifies a sample point. Clearly, the 4-pulse coded pulse-code

modulation permits a total of 16 quantizing levels, the total information
in each message being 4 binits.
Suppose now that the alphabet of 2 symbols be retained, but now
consider that each message is to consist of 7 letters.
In this case the
total number of possible messages is 2 7 = 128, and each message contains
an average information of 7 binits. That the amount of information
should increase with the number of letters per message is made evident
by a simple example. Suppose that a given wave is to be sampled. If a
If a 7-pulse
4-pulse code is used, 16 amplitudes may be represented.
Clearly, in the latter
code is used, 128 amplitudes may be represented.
case, the available amplitude levels are specified within much narrower
limits than in the former case, whence the uncertainty is less, or the
information is greater.
In the limit for a signal that can be measured with an infinite degree
of precision, an infinitely large amount of information is contained in each
message. The transmission could therefore take place at an infinite rate
through a channel, if an infinite number of inputs can be correctly recognized at the output. Observe that the output signal does not have to be
identical with the input signal in this case, for if the channel alters the
signal in some known fashion, a precise knowledge of this change will

Of course, if the signal is

changed in some unknown or random fashion, the number of input signals
which may be recognized becomes finite, when the channel capacity
This means that random fluctuations or noise in a
also becomes finite.
permit the recognition of the original signal.

transmitting channel will limit a channel to finite capacity.

The

foregoing discussion can be extended to the case of a continuous

signal.

Refer again to Fig. 15-1, but consider the signal to be a

length of continuous signal.

There is an

infinite

finite

number of sample points,

Moreover, there is an infinite

It would appear, therefore,
that a continuous signal of finite length is represented by an alphabet
containing an infinite number of symbols, there being an infinite number of
symbols in each message. This would lead to the conclusion that a continuous signal, even of finite length, contains an infinite amount of information. Two reasons exist in the practical case why the foregoing is not
true.
First, all actual signals possess spectra of finite bandwidth, and,
according to the sampling theorem in the time domain, only the values
of the signal at the sampling points can be chosen independently.
This
reduces the continuous signal to the equivalent of a finite sequence of
symbols. Second, as noted above, random fluctuations or noise will limit
the value of each of which

number

is its

ordinate.

of possible values of the ordinate.

RADIO ELECTRONICS

436

[Chap. 15

a channel to finite capacity, since the number of distinguishable amplitude

reduced to a finite value.

levels is

Two important conclusions can be drawn at this point. First, it

appears that continuous signals would be expected to have the same order
of information as discrete signals.
Second, information theory provides
no

significant information for continuous signals without noise.

15-5.

finite

The Modified Hartley Law.

As noted

capacity of the transmitting channel

in the foregoing section,

imposed by the presence

is

of random fluctuations or noise.

This finite capacity of the channel may
be expressed in terms of the relative magnitudes of the signals and of the
fluctuations that are imposed upon them during transmission. 6 If the
fluctuations are of the character of random noise, then on the average a
change in signal amplitude can be detected only if it is at least equal to
the rms value En of the noise.
In all practical types of detectors, the
extraction of information from a mixture of signal and noise requires that
some threshold level of S/N ratio be exceeded. Even above this threshold the reliability of information increases with S/N ratio.
If Es is the
rms value of the signal, then for large S/N ratios (actually the ratio of
peak signal to peak noise need only be greater than 2) the number of
allowable letters in the symbol alphabet corresponding to the reliable
recognition of the signal amplitude is (Es
En)/En or 1 + E S /EN
Moreover, the sampling of the amplitudes may occur at the rate of 2B per

second, since the signal

is

confined to the bandwidth B.

In the light of these considerations, the information

contained in a
message containing n symbols chosen from an alphabet of s symbols is

H
which

may

is

log

log 2

log s"

+ y~ )

binits

(15-22)

be written

H
This

= 2BT

the modified Hartley law.

C =

= 25

log 2

lo S2

binits

The corresponding channel

(l

fr;)

binits/sec

(15-23)

capacity

is

(15-24)

This expression specifies the maximum channel capacity of a system for

a specified S/N ratio and bandwidth B while at the same time obtaining
a vanishingly small percentage of errors.

Equation (15-24) indicates that neither the bandwidth nor the sigThat is, a given

nal/noise ratio inherently limits the channel capacity.

INFORMATION THEORY

Sec. 15-6]

437

message can be transmitted through a given channel of fixed bandwidth

a certain time T with a specified signal/noise ratio. The same message
can be transmitted through a different channel, if the factors are so
chosen that the value of the right-hand side of Eq. (15-20) under the new
conditions equals that under the original conditions.
It must not be
inferred, however, that the factors in the modified Hartley law may be
varied at will. The type of modulation that is used will impose certain
fundamental limitations on one or another of the parameters in the
equation.
In fact, only coded systems permit any flexibility in the
magnitudes of the transmission factors.
Attention is called to the very important fact that the modified Hartley
law expresses the optimum information capacity of a given transmitting
channel.
It says nothing about the actual capacity that may be realized
by any particular type of modulation. Nor, in fact, does it say anything
about the degree to which the parameters may be varied in any particular
modulation system. The discussion in the next section will examine
certain particular systems of modulation in the light of the foregoing.
A point of considerable practical importance that is contained in Eq.
(15-24) is that bandwidth can be "traded" for signal/noise ratio in the
It is only with coded systems that one is able to achieve
ideal system.
the full exchange of bandwidth for signal/noise ratio.
15-6. Evaluation of A-M, F-M, and Various Pulse Systems.
It is
desired to examine several important systems of modulation in the light
of the modified Hartley Law.
This evaluation may be accomplished
relatively directly, if it is first noted that in a communication system one
may refer to two transmission channels, the primary information channel
and the transmission channel. The primary information channel refers
to the information channel before modulation or after detection and in an
a-m or f-m system for voice communication would denote the audio channel.
The transmission channel refers to the carrier channel.
For the ideal case, we may write
in

2BT

log

( l

fi) = 2B'T

log

( 1

+ ff

where the unprimed quantities refer to the primary channel and the
primed quantities refer to the transmission channel. This expression
states that the information contained in the messages in a given time T
is the same in both channels.
It follows from this expression that

(15-25)

Here B'/B

is

the bandwidth-expansion factor.

This expression shows

RADIO ELECTRONICS

438

[Chap. 15

that the signal/noise ratio can be increased, at least ideally,

by

increasing

the bandwidth.
a. A-M Systems.
Consider that a 5-kc signal bandwidth is to be
transmitted in a conventional double-sideband system. The transmissionchannel bandwidth must be equal to twice the primary-information-

channel

bandwidth.

In

so

far

as

fluctuation-noise

interference

is

concerned, a greater transmission-channel bandwidth does not improve the

transmission, nor does

it

permit the use of a lower signal/noise

ratio.

Conversely, a transmission-channel bandwidth narrower than twice the

primary information channel will result in the loss of part of the information, and this loss cannot be compensated by increasing the signal/noise
ratio.
This means that bandwidth expansion cannot be employed in an
a-m system to improve the signal/noise ratio.
In the case of single-sideband suppressed-carrier transmission an appreciable

improvement over double-sideband transmission

single-sideband

transmission,

is

With
bandwidth is
As a result of

possible.

the transmission-channel

equal to the primary-information-channel bandwidth.

this factor of 2 in bandwidth, the improvement in signal/noise ratio over

Now also, since the power in the sidethe double-sideband case is 3 db.
bands, for 100 per cent modulation, is one-half that contained in the
carrier,
If

then only one-third of the total transmitted power is intelligence.

If the carrier and
is suppressed, there is a gain of 4.77 db.

the carrier

one sideband are suppressed, there is a gain of 6 db. Hence the singlesideband system possesses an inherent advantage of 9 db over the doublesideband system. The practical problem in comparing the actual advantage of single sideband over double sideband is complicated by the fact
that single-sideband power amplification is accomplished in class B amplifiers, whereas double sideband is ordinarily produced in class C stages.
In so far as trading bandwidth for signal/noise ratio in single-sideband
is concerned, the same limitations exist as for the doublesideband system, and no effective exchange is possible.
In the case of an f-m system, a certain amount of
b. F-M Systems.
transmission

trading between bandwidth and signal/noise ratio is possible. The

amount of signal/noise improvement that is possible was examined in
detail in Sec. 14-20.
According to Eq. (14-103), which is valid
provided that the peak noise amplitude is less than one-half the peak
signal amplitude, the signal/noise ratio varies directly with the frequency
Moreover, the total signal/noise ratio in the receiver output
deviation.
is dictated by the audio bandwidth, which would be set nearly equal to
the bandwidth of the primary information channel. As illustrated in

some

Fig. 14-49, the signal/noise ratio

sion.

of

is

proportional to the bandwidth expan-

The improvement can be made somewhat

preemphasis

circuits, since

higher through the use

the preemphasis of the higher audio fre-

INFORMATION THEORY

Sec. 15-6]

439

Howquencies makes the noise relatively lower at these frequencies.

improvement is not nearly as large as that per-

ever, the signal/noise

mitted by the Hartley law.

The term pulse modulation as distinct from pulsec. Pulse Systems.
code modulation will be used to designate unquantized pulse modulation.
The pulse-modulation systems to be considered are pulse-amplitude

modulation (PAM), pulse-duration modulation (PDM), and pulse-posi-

(PPM).
bandwidth required

tion modulation

for pulse-modulation systems depends

on the pulse repetition rate, the pulse waveshape, and the pulse width.
Under normal operation the r-f bandwidth required is given approximately by

The

over-all

B =

(15-26)

where t is the pulse width in microseconds and B is in megacycles. Consequently pulse-modulation systems require relatively large bandwidths
in their operation.
is the most efficient form of pulse modufrom the standpoint of bandwidth B. However, pulse-amplitude
modulation is subject to the same transmission-link stabilities and
That is, the
signal/noise characteristics as any other a-m systems.
bandwidth employed will govern the effect of the noise, broader bandwidth being accompanied by an increased noise without an improved
Hence pulse-amplitude modulation is not able to improve the
signal.
signal/noise ratio in exchange for bandwidth.
Pulse-duration modulation and pulse-position modulation, both being
These do much to
of constant amplitude, permit the use of limiters.
improve the signal/noise ratio of the system by reducing the effect of
impulse noise, just as in frequency modulation. However, noise does
remain, as discussed in Sec. 1-7, this noise appearing on the leading and
Impulse noise may cause a displacement
trailing edges of the pulses.
of the pulse edge from the position corresponding to the modulating

Pulse-amplitude modulation

lation

may also affect the pulse amplitude, shape, or slope of the

In pulse-duration modulation the effect ultimately appears
In pulse-position modulation the
as a change in the effective duration.
effect will appear as a change in position.
If the bandwidth in an uncoded pulse-modulation system is doubled,
signal.

Noise

pulse edges.

In pulsethe transmission symbol can be located twice as accurately.

duration modulation and pulse-position modulation, since the uncertainty

time of the pulse, and

which improve directly with the bandwidth,
proportional to the bandwidth, and hence pro-

of the position of the pulse is a function of the rise

also the

decay time, both

the signal/noise ratio

is

of

RADIO ELECTRONICS

440

[Chap. 15

portional to the bandwidth expansion.

While this means that both

pulse-duration modulation and pulse-position modulation trade bandwidth for an improvement in signal/noise ratio, it is not nearly so large

an increase as is possible, according to the Hartley law.

In pulse-code modulation or other possible coded systems, full advantage can be made of added bandwidth.
Suppose, for example, that the
transmission-channel bandwidth is doubled. As a result, if the number
of letters in the symbol alphabet is unchanged, a message containing
twice as many symbols, or two messages of the original symbol content,
can be transmitted in a given time. But if each message has 1 + E'S /E'N
possible signal amplitudes, the combination of the two messages has
(1

E's /E'ff) 2 possible values.

This states, in

effect,

that

where the exponent is the bandwidth-expansion factor. By comparison with Eq. (15-25), this is the form governing the signal/noise ratio
improvement that is possible in the ideal system, as given by the Hartley
law.
15-7. A Geometrical Approach to the Modified Hartley Law. A
rather interesting geometrical approach to the modified Hartley law was

developed by Shannon. 6 As discussed in Sec. 15-3, a signal of duration T

and bandwidth B can be specified exactly by a set of 2BT numbers which
are uniformly distributed in time, at the instants 1/25 sec apart.
If one

now

considers a multidimensional space of

specification of

2BT

2BT

dimensions, then the

coordinates defines a point in this 2BT-dimensional

Moreover, the "distance" from the origin to the point is the

square root of the sum of the squares of the coordinates of the point. If
the coordinates are potentials, the sum of the squares is a quantity that
space.

is

proportional to the power in the signal.

power

less

than

radius that

is

P s must lie inside

proportional to

\/

Hence

all

signals of total

a 2-BT-dimensional "sphere" having a

s.

Now

consider the situation after the signal has been transmitted

through a transmission channel, being thereby affected by noise. As a
result, the signal point is displaced in signal space.
But since the noise

random, the displacement

probable in any
displacement being proportional to -\/PnThe situation is then somewhat like that illustrated graphically in Fig.
15-2 for the very simple case of a two-dimensional space.
Since the total
power of the perturbed signal is P s
all
such
signals
are confined
Pn,
to the sphere of radius y/Ps
Also,
the
number
of
signals t hat
Pncan be reliably distinguished is the number of spheres of radius \ZPn
is

direction, the

amount

of the signal point is equally

of the

INFORMATION THEORY

Sec. 15-7]

that can be included in the sphere of radius

y/Ps

441

Pn-

This number

given by the ratio of the volume of the large sphere to the volume of the

is

Fig. 15-2. Cross section of signal space showing the transmitted, random-noise,
received powers.

small sphere.

But

since the

volume

of

an w-dimensional sphere

and

of radius

r is 6

n/2

V =

the ratio of the two volumes, and hence the ratio of distinguishable
signals, is

M
Hence by Eq.

= (VPs

Pk)

2BT

(15-27)

"2BT

(15-10) the total information in the message

= BT

log

= BT

log

Ps

is

+ Pn

or

log

(15-28)

This form of the modified Hartley law differs somewhat from that
given in Eq. (15-23) and appears to be the more general form. Note,
however, that since Eq. (15-23) is restricted to large S/N ratios, the

two expressions are

practically the same.

REFERENCES
1.

2.

Goldman, S., "Information Theory," Prentice-Hall, Inc., New York, 1953.

Shannon, C. E., "The Mathematical Theory of Communication," University of Illinois Press, Urbana, 111., 1949.

RADIO ELECTRONICS

442

[Chap. 15

Hartley, R. V. L., Bell System Tech. J., 7, 535 (1928).

Shannon, C. E., Proc. IRE, 37, 10 (1949).
Tuller, W. G., Proc. IRE, 37, 468 (1949).
Sommerville, D. M. Y., "An Introduction to the Geometry of N-dimensions," p. 135, E. P. Dutton, New York, 1929.
A good general discussion is given by:
Leifer, M., and W. F. Schreiber, "Communication Theory," in L. Marton,
ed., "Advances in Electronics," vol. Ill, Academic Press, Inc., New York,

3.

4.
5.
6.

7.

1951.

PROBLEMS
15-1.

when

How much

information

is

contained in the message that a pair of dice,

rolled, resulted in a six?

It
15-2. A language is drawn from an alphabet of four symbols, A, B, C, D.
encoded into a secondary channel having an alphabet of two symbols 1 and
according to the code
is

A->11

B->10

C->01

D->00

What is the
a. Assume that the symbol probabilities are each equal to 0.25.
language-transmission capacity of the secondary channel, for unit duration of
the symbols 1 and 0?
The symbol probabilities are now, respectively, 0.3, 0.3, 0.2, 0.2. What
the language-transmission capacity of the secondary channel under these
conditions?
15-3. The elements of a pulse-coding transmission system are illustrated in
the accompanying figure. Specify the missing data called for on this figure,
6.

is

assuming a noiseless system.

Signal

Network
Bandwidth

Encoder

Quantizer

Sampler

B
S=
n-

S=32

S=2

n-

n-

15-4. A 5 by 5 matrix of lights is available, any bulb of which can be lighted in

50 millisec. If a specification of the lighted bulb is transmitted by means of a
sequence of on-off pulses:
a. What is the required bandwidth of the transmission channel?
b. What is the channel capacity in binits per second?

APPENDIX A
NOTES ON GENERAL NETWORK ANALYSIS
A-l. Reference Conditions and Notation. Mention is made at several
points in the text of the reference positive polarity and reference current
It is important to examine the significance of these terms and
direction.

network analysis.
amounts to the finding of expressions
network
problem
The solution of a
Somevarious
points in the circuit.
currents
at
for the potentials and
required
also.
For
the
a-c
case
the
linkages
are
times charges and flux
potentials
are
conpolarities
of
the
currents
and
the
directions of the
Algebraic notation must always be employed.
tinually reversing.
their relationship to general

All measurable electrical quantities (current, potential, etc.) are called

Their algebraic representations (e, i, etc.) are called

Physical entities change their direction or polarity,
algebraic quantities.
while their algebraic representation changes their sense (algebraic sign).
physical entities.

The

direction of current,

when the symbol which

represents

it is

posi-

tive, is called its reference direction.

Similarly, the polarity of a potential,

when the symbol which

it is

polarity.

represents

In this latter case, the

and

positive, is called its reference

signs not only are used to

designate the potential but also are used in algebraic interpretations.

The two
dition

is

The general term reference coneither

reference
direction or a reference polarity.
imply
a
used to
uses should not be confused.

Thus, in general terms, the reference condition of a physical entity corresponds with the positive sense of the algebraic quantity which represents
it.

A-2. Notation for Sinors. It is customary in the discussion of amplifier

response to assume that the input signal is sinusoidal and of the general

form
ei

= \/2 Ex

sin (wt

6)

Ordinarily the calculations are not carried out in terms of the trigono-

metric functions, but use

work

analysis.

The use

is

of

made of complex-number theory in a-c netcomplex-number theory stems from the fact

* For more detail, refer to W. LePage and S. Seely, "General Network Analysis,"
McGraw-Hill Book Company, Inc., New York, 1952.

443

RADIO ELECTRONICS

444

may

that the sinusoid

d=
where

[App.

be expressed in exponential form by the relation

y/2 Ex

sin (at

= Im

X)

y/2

1 e'^'

+0 ^

(A-l)

Im

denotes that the imaginary part of the Euler* expansion of this

chosen.
Now, it is found in subsequent analysis that the
essential information relating to the sinusoid is contained in the quantity

quantity

Erf'* 1 ,

is

which does not include the time. This permits the analysis to be
complex-number theory and the associated alge-

carried out in terms of

braic manipulations of this discipline, and, as a final step, the results may
be correlated with, and written in trigonometric form involving the time.

Many of the results obtained by the use of complex numbers can be

obtained through geometric means alone, by plotting the complex numbers as directed lines in a plane.

vectors in two dimensions.

vectors

can

arise.

quantities having
quantities.

As

The

directed lines

may

be regarded as

However, confusion with three-dimensional

Therefore,

in

electrical-engineering

two dimensions (complex

applications,

quantities) are called phasor

discussed, therefore, phasor quantities can be used to

represent symbolically sinusoidal functions of time.

This

is

not the only

For example, the impedance function R + jwL

of a simple series circuit is a complex number but represents quite a
different quantity from the complex number which symbolically represents the sinusoid.
Those complex quantities (phasor quantities) which
use of them, however.

symbolically represent sinusoidal functions are called sinors.

sinor is a special type of phasor.

Thus a

In setting up the correspondence between a sinor and a sinusoid, the

form of writing the sinusoid is chosen arbitrarily. The correspondence between a sinor and sinusoid is written
sine

Eie' 61 symbolically represents

The

multiplier

is

introduced to

make Ei

V2 E

sin (ut

6i)

(A-2)

the effective value.

Emphasis is placed on the fact that either the sine or the cosine function
could be used on the right-hand side of relation (A-2). However, in this
text the foregoing symbolism always means the sine function.
Without
adherence to such a rule, the use of sinors and complex notation could
lead to confusion.

Since a sinor

a complex quantity which symbolically represents a

the previous definition of a reference
condition would indicate that a sinor does not have a reference condition,
since it is not a varying quantity.
It is a directed line in a plane, but this
is

sinusoidally varying quantity,

direction is not an indication of a reference condition.

It is related,
however, to the reference condition in an indirect way, to be explained.
*

The Euler expansion

of

an exponential has the form Ee ia

E(cos a

+ j sin a).

NOTES ON GENERAL NETWORK ANALYSIS

Sec. A-3]

Figure A-l illustrates

by a

sinusoidal wave,

how

a potential or current

and then

in turn

by a

sinor.

sinusoidal wave, in terms of the physical conditions

is

445

represented,

The meaning
shown on the

first

of the
circuit

JL

Fig. A-l. Interpretation of sinor notation in terms of a sinusoidal wave.

diagram,

is

referring

them

The

discussed in Sec. A-l.

sinors

E and

I are interpreted

by

to the sinusoidal waves; thus

symbolizes e

I symbolizes

=
=

y/2

sin

y/2 I sin

(cot
(cot

+
+

a)

(A-3)

/3)

This gives meaning to the reference conditions, in terms of the instantaneous variables which are symbolized by the sinors.
A slightly different practice is shown in Fig. A-2. The sinor symbol
used on the circuit diagram, and the intermediate sinusoidal wave is omitted. This system is convenient, as it is more direct than
In the event of any
that illustrated in Fig. A-l.
question of interpretation, the thought sequence
indicated in Fig. A-l is implied by Fig. A-2.
Double subscripts may be used on sinor symis

imply the reference condition for the

In fact, it may
it symbolizes.
be noted that identical systems of notation are

bols, to

variable which

applicable to instantaneous variables

The only

difference is that,

and

sinors.

when applied

to

Fig. A-2.

The use

of sinor

symbols on a circuit diagram to imply the reference conditions shown in

Fig. A-l.

with reference conditions

through the intermediate variable quantity, as in Fig. A-l.
A-3. Loop Analysis. The general complete solution of a network by
the loop method of analysis may be most easily presented in terms of a
The circuit of Fig. A-3 serves this purpose. All
specific example.
sources are potential sources, and all have the same frequency. The
impedances are drawn as boxes for simplicity. Each box is assumed to
sinors, the connection

is indirect,

RADIO ELECTRONICS

446

[App.

be no more complicated than a series combination of resistance, inductance,

By

and capacitance.
an application of the Kirchhoff potential law around each

of the

indicated loops, the equations that result are

+ Z d )l, - ZJ + = E a
+ Z d + Z )I - Z l = E,
- ZJ + (Z + Z.)I, = -E,

(Z.

-Z dh +

(Z 6

The

by the application

O
Fig. A-3.

is

(A-4)

solutions of Eqs. (A-4) are most easily written in terms of determi-

nants,

it

Cramer's

of

For writing these

rule.

solutions,

to
A typical

network labeled

for loop analysis.

convenient to define certain symbols to represent determinants.

These are
Z

-Z d

Zd

Z d

A =

Z;,

Zd

-Z
=

Zb

fl

Ejj
Z

tt

Zd

Zd

A =
3

-\-

z.

Zc

Ej

Ze

Z<J

Zb

tn terms of these quantities, the three

IT 2

Ze
E

Ze

unknown
*

-T"

(A-5)

Z d

-Z

Ai

Ea

Zd

Zd

-Z

Ze

-E^
Za

Zc

-Zd

E
Ai

Ze

Efl

-Efl
currents are

(A-6)

by the application

of Cramer's rule.
should be realized that, even though the example above is for three
loops, the general form of the equations will be similar for a circuit of

It

N loops.

For the network with

N loops, the general equations will be of

NOTES ON GENERAL NETWORK ANALYSIS

Sec. A-4]

447

the form

piJi

ptflll

+ P12I2 +
+ P22I2 +

0i\T 2 l2

+ pillar = Ei
+ Qwlti = E

'

^^

QNlflN

EiV

In these equations, the quantities Ei, E 2

denote the algebraic sums
the sources on the various loop peripheries or contours, with a potential rise in the loop-current direction taken as positive.
,

of all

have the dimensions of impedance. When p = q, they

impedances on a loop contour. When p ^ q, gPQ is +
the impedance in the branch common to loops p and q, according as

The

factors g pq

are the
or

sum

of all

same or opposite directions in the common

impedance. The g coefficients are given the name copedance, a contraction for the phrase, coefficient of impedance.
As an illustration, the analysis of Fig. A-3 may be written in the
generalized notation of Eqs. (A-7)
In this case
these loop currents are in the

+ &12I2 + P13I3 = Ei
+ 022I2 + P23I3 = E
+ P32I2 + P33l3 = E3

(?nli

P2J1

(A-8)

P3lll

where

Qu =
e 22
033

The
must

Z<,

= Zb
= Z

+Z
+ Zd +
+Z

P31

=
=

9l3

023

032

921

Z.

P12

= Z,j
=
=

Ea
Ei
E2 = E
E3 = E#

(A-9)

fi

resulting expressions are identical with those in Eq. (A-4), as they

be.

Special attention

is

required

when magnetic

^r-

coupling exists, as the formulas for the coped-

ances of the elements

off

the diagonal of the

In this connection,
proof, it is noted
that the sign to be'Jassociated with the term
jo>M representing the effect of the mutual coudeterminants are affected.
refer to Fig. A-4.

pling

is

positive

if

Without

the currents enter or leave

the dotted terminals and

is

negative

if

FlG A-4. Example

t

one

.-

of

mag-

netlc C0U P lin &-

current enters and the second current leaves the dotted terminal.

For

the case illustrated

P2i

A-4. Junction Analysis.

ei2

The

= -Z +jo>M

(A-10)

presentation of the junction analysis

same scheme as used for the loop analysis. All sources have
the same frequency, and all are current sources.
The admittances are
follows the

RADIO ELECTRONICS

448

[App.

drawn as boxes

for simplicity.
Each box is assumed to be no more
complicated than a parallel combination of resistance, inductance, and

capacitance.

The rules for writing the equilibrium equations, which are based on the
Kirchhoff current law, are applied to the network of Fig. A-5. The equa-

Dafum j unci ion

network labeled for junction
;

Fig. A-5.

A typical

tions are the following,

when the current

analysis.

at each junction

is

equated to

zero:

+ Yd )Ei - Y E, +
+ Y d + Y)E - Y E
- Y E + (Y + Y)E

(Y.

d Ei

(Y

The

3
3

solution of this set of simultaneous equations

rule.

= I
= Lj
= -Lj
is

(A-ll)

given by Cramer's

Write the determinants

Y

A =

-Yd

+ Yd

-Y,

Y + Yd +
b

-Y.

Y,

Y +

-Y.
I

-h
Y + Yd
A =
-Yd
2

-Y d
+ Yd + Y

A =
3

Yd

-Y.

Y +

-Y.

Y.

(A-12)

-Y.

Is

-I,

Y + Yd

Y +Y
c

Yd
Y + Yd + Y
6

I
e

-Y.
In terms of these determinants, the

1^

-I

unknown

potentials are given

by the

ratios

Ea

by the

application of Cramer's rule.

(A-13)

NOTES ON GENERAL NETWORK ANALYSIS

449

This type of solution can be written for a general network of

datum junction, by the set of equations

N junc-

Sec. A-5]

tions plus the

+ dl2E2 +
+ 022E2 +

<*llEl
<*2lEl

Ja'iEi

The

itrsEi

'

'

+ dl2E2 =
+ ^2nEn =

I2

+ dif/fEff =

Ijy

Ii
/

1 A \

(A-14)

branch admittances in the followadmittances connected to the path

q, d ps is the negative of the admittance connecting
directly.
These coefficients are called comittances, as
a contraction of the phrase, coefficients of admittance. The currents
Ijv appearing on the right are junction current sources.
A junction current source is the algebraic sum of all current sources connected to the
junction.
A current having a reference direction toward the junction
is taken as positive in forming the algebraic sum.
A-5. Network Theorems. A number of network theorems are ound
to be of considerable assistance in solving the network problems involving
vacuum tubes. Several such theorems will be given, although the proofs
will not be given in all cases.
The theorems which will here be given are
the Helmholtz equivalent-source theorems (TheVenin and Norton
theorems), the Millman theorem, and dual-circuit construction rules.
coefficients d represent the various

p q,
junction.
If p 9*
junctions p and q
ing way: If

dP4 is the

sum

of all

Helmholtz Equivalent-source Theorems.

a.

A-6 represent a general network

with
independent loops,
+1

junctions,

and

N loop potential sources,

.AT

all

M junction current sources,

One branch
network is shown

Let the outside terminals

isolated.
.

/.....

be closed by a fictitious potential

loop potential

sources

of the same frequency.

of this general

Let the rectangle of Fig.

Or

junction current
sources

a.

A:6

Circuit for illustrating the

equivalent-source theorems

L
Helmholtz
( T hevenin and Norton theorems).
,

may be the potential

drop across a passive element between these two points. A general
analysis of such an active network leads to the conclusion that any
two-terminal linear network with any number of current [and potential
sources of the same frequency can be viewed as a source.
This
source can, in turn, be represented by either of the simple circuits of
Fig. A-7.
When the potential-source form is used, the network is said
to be represented by a Helmholtz-TheVenin potential-source equivalent.
The current-source form is called the Helmholtz-Norton current-source
source, which

equivalent.

The

series

impedance Z

t,

or the shunt admittance Y, appearing in

Fig. A-7, are the inverse of each other,

i.e.,

Z = 1/Yn
(

Also,

is

the

RADIO ELECTRONICS

450

impedance

[App.

network as measured at the terminals aa when all

and current sources are reduced to zero.
The potential source E is the potential E which appears across the
of the

internal potential

terminals aa

when

current source I

these terminals are an open circuit.

the current in the short circuit which

is

Similarly, the
is

placed across

the terminals aa.

r
i
a.

Fig. A-7.

a.

The Helmholtz

equivalent-source representations.

The parameters E<, I, Z Y can be found by any method of circuit

They may also be found experimentally for physical systems.
t,

solution.
b.

Millman Theorem.

Certain simple combinations of potential-source

equivalents are of use because they offer simplification in the solutions

of

more extensive networks

sider Fig. A-8a,

which

is

in

which the combinations occur.

Con-

typical of the situation to be considered

E3

first.

z1

z2

z3

z3
(b)

YJ+ Y2 +Y3
Fig. A-8. Combination of potential sources.

The
E.

is accomplished by finding the potential

each potential source and its series impedance be

solution of the network

To do

this, let

replaced by a current-source equivalent, to yield the form shown in Fig.

A-86.
In this diagram the sources have been separated from their individual admittances to give the grouped arrangement shown.

ultimately to the circuit of Fig. A-8c.

is given directly by the relation

E =

The

This leads

solution of this circuit for

E]Yi + E2Y2 + E3Y3

Yi + Y 2 + Y 3

(A-15)

NOTES ON GENERAL NETWORK ANALYSIS

Sec. A-5]

451

Duality and Dual-circuit Construction.

Throughout the developnetwork analysis a parallelism of statements is necessary, once

for potentials and once for currents.
There are potential sources and
current sources; a Kirchhoff potential law pertaining to a loop, and a
Kirchhoff current law pertaining to a junction, etc. These ideas are part
of a larger pattern which exists in network analysis.
The simultaneous existence of two similar systems of analysis is given
the name duality and is founded on the interchange of independent and
dependent variables in the equations expressing the behavior of the circuit
c.

ment

of

WW\

JU1SLSL'-

-t-

Fig. A-9. Basic series and parallel circuits.

Duality does not imply equivalence; it means only that the

mathematical representations of the circuits are similar in form.
Consider the basic series and parallel circuits illustrated in Fig. A-9.
If the series combination is excited by a potential, and the parallel comelements.

bination by a current, the appropriate relations for the respective circuits

For the series circuit,
follow.
e

For the

eR

eL

ec

ic

ih

Ri

di
dt

(A-16)

e dt

(A-17)

dt

parallel circuit,

ia

= Ge

de

^+

dt

Table A-l contains the dually related quantities,

works illustrated in Fig. A-9 are dually related.

Clearly, the

two net-

TABLE A-l
DUAL RELATIONSHIPS
Loop concept

Loop

interior

Circuit exterior

Potential source in an external

branch
Potential source in series in a

Junction concept

Junction other than datum junction

junction
Current source with one terminal connected to the
datum junction
Current source shunting a common branch

Datum

common branch
Loop current
Branch impedance

Junction potential to the datum junction

Branch admittance

RADIO ELECTRONICS

452

[App.

It is of interest to compare Eqs. (A-4), which apply to Fig. A-3, with

Eqs. (A-ll), which apply to Fig. A-5. Since there is a systematic interchange of Z with Y and of I with E in these expressions, the two networks
are dually related.
Among the important dual relationships implied

herein are those contained in Table A-l.

In the case of

method

series or parallel circuits, the

of going

from one

circuit to its dual is illustrated in

It is convenient to

Fig. A-9.

be

able to construct a circuit which

is

the dual of another in the general

case.

dual

and can be

exists,

found, for any circuit, provided that

can be made

it

flat

and includes

transformers in such a

way

that

they can be replaced by equivalent

circuits.

The
solved
Fig. A-10.
analysis.

work

essential
is

problem

to

be

the following, and for con-

A typical circuit setup for loop

venience a typical three-loop netis to be considered.

The netrepresented by the equations

work

specified

is

>llll

P2J1
>3lll

The symbols S e

+
+
+

P12I2
P22I2
P32I2

+
+
+

P13I3
P23I3

P33I3

= Si S
= S e2
= S

e;

(A-18)

e3

are used to represent potential sources, and each I repre-

sents a loop current.

It is

now

required to find the network which

is

represented by the equations

<*iiEi
<*2lEi

<J3iEi

in

which each

source.

The

Each

is

is

+
+
+

d"i 2

E2 +

+
E +

(J22E2

d 32

=
=
=

<Ji3E 3
0"23E 3

S,-i

Si2

(A-19)

Si2
S3

a junction response potential and each S,

to be the same as the corresponding p.

details of the construction follow:

three junctions, in addition to the

datum

is

a current

The new network must have

junction.

A dot is then placed

and the network is enclosed by a continuous line. Each

internal dot becomes a nondatum junction, and the outside line is the
datum junction. The junctions are numbered to correspond with the
loops in which they appear. Junctions 1 and 2 are to be connected by a
branch which is the dual of the branch common to loops 1 and 2. When
in each loop,

this

branch

is

drawn,

it

in essence crosses the

branch to which

it is

dually

Sec. A-5]

NOTES ON GENERAL NETWORK ANALYSIS

related.

This procedure

datum junction.
In setting up the dual

is

453

carried out for each pair of junctions, includ-

ing the

circuit, it is

necessary to choose reference condiTwo considerations

tions in accordance with a consistent set of rules.

are involved, (1) the reference direction of the current sources,

Pig. A-ll.

Two

and

(2)

the

Da-tum
which are duals of each other.

circuits

reference polarity at the junctions.

The

rules for a consistent representa-

tion are:
1.

its

If

dual

2.

If

a potential source in a loop is a rise in the clockwise direction,

is a current source directed toward the dually related junction.
a current has a clockwise direction in a loop, its dual is a positive

potential at the dually related junction.

Based on the above, Fig. A-10, appropriately marked, and

given in Fig. A-ll.

its

dual are

APPENDIX B
PLATE CHARACTERISTICS OF VACUUM TUBES
average:

plate:

characteristic

TYP : 5U4 G
E f = 5.0 V OLTS A.C.

y
7

3
<

J 400
_>
2

u
I-

>

<

zoo

20

40

80

60

PLATE VOLTS
Fig. B-l.

5U4-G

454

D.C.

diode.

100

120

92C-6I09

455

456

457

458

459

460

PLATE CHARACTERISTICS

461

40

30

20

3
-

10

15

10

0-C PLATE VOLTAGE

Fig. B-8.

6H6

M VOLTS

diode.

20

25

462

463

saaadwmih

ni

iNauano 3J.vid

464

6 .

465

466

<0

-r

S3b3dHvmm

c\

ni

cm

iN3ayno aivid

467

-i

"

"

468

469

RADIO ELECTRONICS

470

[App.

200

/
A-

150

j..

too

it
*

50

X*

10

20

30

0-C PLATE VOLTAGE

Fig. B-17.

6X5

IN

40

VOLTS

diode.

50

APPENDIX C
CHARACTERISTICS OF TRANSMITTING TUBES

471

472

RADIO ELECTRONICS

[App.

 O! N
-h
''
I

++

IN
,

~H

CO 5
t>. CM

<D
w >o
N (O
O o S
Q M
oo o
o
ft

ID r<
lO

t~-

1-H

*
OO

1-t

1-H

<>
o

++
van
WON
-

Q
Z
i

* O N O N
In 2 h
o
O
I

+++

CM 00 D IO
tO CO
CO
"O to

OS to
O

o ~- CM
oo

^a*

'

>

<

'

i-4

i-t

<

aa

'

>

'

>

h-5

fa

O
w

eg

o o

i
i

++++

8 83
888

>

W
CO
CO
W

IXC
^

<

'

"

++

>

r- co
n

ih

+
++

i-H

i-4
i-l

I-H

<

H
fa
Q O
X CO
Q O
H
W O
Cm
- Z
P
fa

t^-

CJO
i-t

<

S CO

<

'

->f<

t>-

'

PC

<
1

+
IN ^ to O

V iO
S
o e
o
h
o cn

+++
+++
+
* O OO cm 2; cm oo o CO
us
r
i
IQ N 0 M
CO CM tO
v a
Sn
00 CO
oS h w co
eo wa rCO
rH CM
-h
O
O CM
o

'"'
+
++++++++
+++
<D O O CO N iQ
C4 2 a s

''<

9
+ If +
+
p O N
55 S a h ^ io
I5 h o
~h
CM W

H
CO

N
N
f
OO
IH O
H
i

<E>

MS OO
*0
os o>
H

CO CO rH o

h c5
O CO

oo

_<
_
?
M
C4 O) 0 W
n
a
*
(D Q
N
(O
tO CO

55
CM ^H
CM CM
o

"'

++
++++
if CO <0
H
OO M (p CO
os
cm
"85CO CO N to CO OOo CO o
~h o
O
++++
N f CO
OO CM tO
CO Ui
CO

CO
CO CO
o
O
S3 5 a

o

o o
-

+
I

+++

~H o> iO OS iO

<

""Jl
-^f u?
CO U)

CS

OC

o>

CO iO 5

12

a N O (D
H
N
CO 0>
~+

O co
o

rl

O ^
oooooooooo

++
+++
++

i-H

'

?????????????
'

<

CX)

"J-

i-h

I
I

<

CM OO
~H CO

++++++++

-*<-<
co "5
o
CO H O
388
S3
CM CO
o o

++++++++++
0
OJU'll'tDNHNMMCSCNIP'-l
i

eg cm

l
i

<

os
i-i

<q J
cl

<

<

>C

to

WOitOOOOO
HCOHHHO)pN(C)N>HpO

????
++++
IO H O) NN^flOQCOf'-i
^"
N
?3 <5 s CM CM
CM CM CO
OOQOO00Q

++++++++++++
sg
<e o h v oo o to IO CO OO CO CO CO O CO
h
Oi
co ^ x) >4 N
o
Sifl

CO CM
CM





++
M N
tp O0 CO CM CO OS to
N OO C4
V t1*

S CO CO " rt co CM 5 e3 O
o

e
l
<

<
-

ie>

+
,

i-h

<
-

<

1-

1-H

>''""

+++++++++++

N CM
O) 0O
*
N O O IO h
*2 2
SCO
r- ^i
CM *0
HM

S8

++++++
o0 w n ^ *

3

""<

++
+++
CM
CM
s
O) lO 1" Q O
s * o CS
OS CO CO CO
CO
^
CO O CO if
-<
tO

rf<
**"
i-h

1-H

<

<

i-l

i-H

I
t

'<
i

+-

O
O)
CO CO
w
w

to
OQ
*

a> t oo a>
CO JO
CM US CO i-H

sea
sss s

???? + + + +
ST

CM
O

<

+++

^P

i-H

'

'*'<
-f

co

t<

ooci-ii

473

CO

K}

CM

INDEX
A

Amplifiers, class A, radio frequency (see

Tuned amplifiers)
resistance-capacitance-coupled (see

supply, 114

Abnormal glow discharge, 41

Amplification, voltage (see Amplifiers,
potential gain)
Amplification factor, 28, 30

Resistance-capacitance-coupled
amplifier)

measurement, 93

199

self-bias, 115,

single-stage, 83

numerical values for several tubes,

455 ff.
tetrode, 37
triode, 30
Amplifiers, cascode, 238
cathode-coupled, 132
cathode follower, 172 ff.
double, 185

ff.

tetrode, 34

transformer-coupled (see Transformer-coupled amplifiers)

tuned

(see

Tuned

voltage gain
class

below)
AB, definition, 101

push-pull

gain, 173

amplifiers)

(see potential gain,

(see

Push-pull amplifier)

class B, definition, 101

graphical analysis, 175

input capacitance, 174
output admittance, 175
class A, beam power tubes, 196
cascade, 119
potential gain, 121
definition, 101
degenerative (see Feedback)
distortion, 189
dynamic characteristic, 87
equivalent circuit for, 87

push-pull (see Push-pull amplifier)

tuned (see Tuned amplifiers, class

power)
class C, definition, 101
tuned (see Tuned amplifiers, class

power)
degeneration

(see

Feedback)

difference, 132

direct-coupled, 130 ff.

battery-coupled, 131

(See also Equivalent circuit)

grid-bias potential in, 83
(See also Self-bias in amplifiers)

grid-signal potential in, 83

inductance-capacitance-coupled, 134
input admittance, 108
inverse feedback (see Feedback)
load line, 85
load resistance for maximum power
output, 188
maximum power output, 188, 191
output transformer, 190
pentode, 196

phase relations, 86
plate dissipation, 193
plate efficiency of, 193
potential gain (see potential gain,
below)

cathode-coupled, 132
resistance-coupled, 132
series-balanced, 141
double-tuned (see Tuned amplifiers)
driver stage, 205, 206
dynamic characteristic, construction
from static, 86
equations, 95
push-pull, 203
feedback (see Feedback)
in feedback oscillator, 302
frequency characteristic, 127
frequency classification, 102
frequency distortion (see Frequency
distortion)

frequency response

(see

Frequency

distortion)
gain (see potential gain, below)
graphical analysis of output, 85

power output, 188

push-pull (see Push-pull amplifier)
quiescent operating point, 85

grid-bias potential, 84
(See also Self-bias in amplifiers)

475

B
C

RADIO ELECTRONICS

476

Amplifiers, grid-signal voltage, 84

grounded grid, 104

harmonic distortion

(see

Amplifiers, regenerative (see Feedback)

resistance-capacitance-coupled (see

Resistance-capacitance-coupled

Nonlinear

amplifier)

distortion)

inductance-capacitance coupled, 134

interelectrode capacitances, 107
intermediate frequency (see Tuned
amplifiers, class A)
internal impedance, effect of feedback
on, 160 ff.
inverse feedback (see Feedback)
inverted, 142
limiters, 401

linear, class

class

(see

Tuned

amplifiers,

power)

conditions for, 101, 246

load line, 41

maximal

flatness,

resistance-coupled (see direct-coupled,

above)

shunt feed, 189, 195

stagger-tuned, 228, 229
transformer-coupled (see Transformercoupled amplifiers)

tuned (see Tuned amplifiers)

Amplitude distortion (see Nonlinear
distortion)

Amplitude modulation
Angstrom unit, 19

Anode

modulated wave
amplifiers, class

narrow-band

(see

(see

Tuned

power)

Tuned

42

fall,

ignitor rod, 47

amplifiers)

mercury pool

noise (see Noise)

170
output impedance, effect of feedback
on, 167
output transformer, 190
output waveform, 86
parallel or shunt feed, 135, 189
paraphase (see Push-pull amplifier)
path of operation, 85, 248, 254, 255
phase characteristic, 127
phase inverter, 205 ff.
phase relations, 86

fall,

41
externally heated, 43
high-pressure, 45
initiation, grid control, 43

221

modulated, 336
for

Modulation)

42

fall,

Arc, anode

cathode

(see

(see

Mercury-arc

rectifier)

oscillations,

Arc back in mercury rectifiers, 49

Argon, in high-pressure diodes, 45
in phototubes, 20
in thyratrons, 47
Automatic amplitude control, oscillators
with, 301

Automatic gain control, 362

delayed, 363

plate-circuit efficiency (see Platecircuit efficiency)

plate-current components, 86
plate dissipation (see Plate dissipation)
plate-modulated, 331, 336
potential gain, db gain, 104
effect of feedback on, 150
effect of interelectrode capacitances,

Balanced modulator, 327

Bandwidth, effect of feedback on, 151
tuned amplifiers, 216, 218, 225
reduction in cascaded stages, 225
untuned amplifiers, 127
reduction in cascaded stages, 129
Barkhausen criterion for oscillators, 287
Beam power tubes, 38
Beat-frequency oscillator, 368

108
multistage amplifier, 121
resistance-capacitance coupling,
122 ff.
resistance coupling, 132
resistance load, 91
transformer coupling, 135
tuned, double-tuned, 220
single-tuned, direct coupling, 215
transformer coupling, 217

power, 187

ff

power supply, common, 114

push-pull (see Push-pull amplifier)
quiescent operating point, 85
radio frequency (see Tuned amplifiers)

supply, 115

Backfire, 49

Bel, 103

Bessel functions, 380, 473

Bias, grid, 101

impedance, effect in amplifiers, 156

Bleeder resistance, 75
Blocking, capacitor, 120
in oscillators, 293
Bombardment by positive ions, 22, 42
Bradley detector, 411
Breakdown, gas tubee, 41
mercury-arc tubes, 42
Breakdown potential, 41

INDEX
Bridge measurements of triode
cients, 93
Bridge rectifier, 62
By-pass capacitor, 115

coeffi-

supply, 115
Capacitance, interelectrode, highfrequency effects, 112
numerical values for several tubes,

455

ff.

pentodes, 111
tetrodes, 111
triodes, 109

Capacitor

filters (see Filter)

Carrier suppression, 6, 328

Carrier wave, disappearance in frequency

modulation, 383
suppression, 6, 328
Cathode, directly heated, 18
indirectly heated, 18
inward-radiating, 18

mercury

as,

42

oxide-coated, disintegration, 45
emission, 17
work function, 17
photoelectric, 18
power for heating, 114
thermionic (see Thermionic cathodes;
Thermionic emission)
thoriated tungsten, 17
tungsten, 17
Cathode efficiency, 17

Cathode

fall, arc, 42
glow discharge, 41
Cathode modulation, 341

Characteristic curves, tube, 455 ff

Child's law (see Space-charge-limited
current)

Choke input

filter (see Filter)

477

Condensed-mercury temperature and

vapor pressure, 44
Conductance, mutual, 31, 286
Conduction, gaseous, 40
(See also Arc)
Constant-current characteristic, 30, 33,
257
Constants, multielectrode tube, 37
triodes, 331
Conversion, general theory, 370
square-law, 369
transconductance, 369
expressions for, 372
Converter, pentagrid, 372
Coupling, amplifier, cathode, 132
inductance, 134
resistance, 132
resistance-capacitance, 119
transformer, 135
coefficient, critical, 221
transitional, 223
networks, design, 242
Critical grid-control curves, 46
Critical inductance, niters, 75
Crystal oscillators, 300, 301
Current density, 24
Current feedback, 156
Current-potential locus, push-pull
amplifier, 201
resistance load, 85
Current-source equivalent circuit, 89
Cutin, capacitor input filter, 69
inductance filter, 66
Cutout, capacitor input filters, 69
inductance filter, 67
L-section filters, 75
Cylinders, space-charge-limited current,
22, 24

Decibel gain, 103

Class A amplifiers (see Amplifiers)

Class B, C, amplifiers (see Amplifiers)
Class C modulated amplifier, 336

Decoupling filters in amplifiers, 116

Degenerative feedback, 151
Deionization, 50
Delay, age systems, 363

Clipping, sideband, in modulators, 324

Cloud, electron, 22
Coefficient, coupling, 220

Demodulation, 348
definition, 348

Coil,

of,

214

Cold-cathode triodes, 43
Collisions in arc tubes, 42
ionizing, 22
by positive ions, 22
Colpitts oscillator, 296
Communication system, elements,

distortion, 102
ff

(See also Detection)

Demodulator, definition, 348

(See also Demodulation; Detection;
Detector)
Detection, for amplitude-modulated

waves, 348
1 ff.

Composite characteristics, push-pull,

dynamic, 203
static, 200

ff

conversion, 368
definition,

diode

348

(see linear, below)

for frequency-modulated waves, 401

RADIO ELECTRONICS

478
Detection, linear, 351
diagonal clipping, 357

Distortion, in square-law detection, 350

Distributed capacitance, 138
Doubler, frequency, 278
potential, 63

distortion, 357
effect of load on,

361
354
input impedance, 356

Driver stage, 205, 206

Duality, 451

efficiency,

rectification characteristic,

360

Dushman
Dynamic

single-sideband, 367
square-law, 349
distortion in, 350

equation, 16
characteristics, class A, 87

construction, 87
general power-series representation, 95
nonlinear distortion, 102
parabolic, 95
push-pull, 203

with triode, 349

superregenerative, 364

suppressed carrier, 366

Detector, Bradley, 411
diode, 351
discriminator, 401

rectifier,

55

resistance load, 87

Dynatron,

372
linear, 351
ratio, 409

oscillator,

309

first,

Efficiency, cathode, 17

square-law, 349
(See also Detection)
Deviation, frequency, in frequency

detection, 354
plate-circuit (see Plate-circuit

modulation, 380
Deviation ratio in frequency modulation,

rectification, 57

380
Diode, characteristics, operating, 26, 49
cylindrical, 25
detector, 351
discriminator, 401
high-pressure gas, 45

glow, 41
Electron, secondary (see Secondary
emission)

23

ratings, 26, 49

as rectifiers (see Rectifiers)

static characteristics, 25,

55

Direct-coupled amplifiers (see Amplifiers)

Discharge, electrical, in gases, 41
Discriminator for frequency modulation,
401
Disintegration of cathodes, 45
Disintegration potential, 45
Dissipation (see Plate dissipation)
Distortion, amplitude (see Nonlinear
distortion)
amplifiers, 102

in class

causes, 102

permissible

maximum,

tuned-circuit transfer, 244

Electric arc (see Arc)
Electrical discharge in gases, 40
arc (see Arc)

breakdown, 41

hot-cathode gas, 44
plane-parallel,

efficiency)

191

for push-pull operation,

204

cross modulation, 342

delay, 102

feedback effect, 154

frequency (see Frequency distortion)
harmonic (see Nonlinear distortion)
in linear detectors, 357
in modulation, 324, 330
nonlinear (see Nonlinear distortion)
phase, 102

Electron-coupled oscillator, 298

Electron-current density, drift, 24
Electron emission, bombardment, 22
photoelectric, 18, 20

secondary (see Secondary emission)

thermionic, 16, 17, 114
Electron tubes, classification, 100
(See also Diode; Pentode; Tetrode;
Thyratrons; Triode)
Emission, electron (see Electron emission)
Entropy and information, 432
Equivalent circuit, linear class A, 87 ff.
analytical derivation, 87
applicability, 88
current source, 89
interelectrode capacitance, effect of,

108
pentode, 111
potential source, 87
push-pull, 199
tetrode, 110

Excitation potential, in amplifiers, a-c

d-c components, 84
multistage, 121
push-pull, 197

INDEX
Feedback, in amplifiers, 150 ff
effect of, on frequency and phase
distortion, 154
on input impedance, 169
on internal impedance, 160
on noise, 155
on nonlinear distortion, 154
on output impedance, 167
on stability of amplification, 153
gain, 150

stability,

170

criterion for oscillation, 172,

284

155

circuits,

compound, 159
current, 156
potential, 157

Feedback
Feedback
Feedback

factor, 150

284
complex, 286

oscillators,
ratio,

Frequencies, sideband, in frequency-

modulated wave, 380

Frequency, angular, instantaneous, 378
carrier, 321
mid-band, 123
multiplication, 278
threshold, 18
Frequency characteristic, 127,

Field emission, 21

Filament (see Thermionic cathodes)

Filament potential, 114
Filter, capacitor input, 68

direct-coupled, 130

inductance-capacitance-coupled
amplifier, 135
resistance-capacitance-coupled, 127
transformer-coupled, 139
tuned potential, 212 ff.
tuned power, 242 ff.

Frequency doubler, 278

Frequency-modulated oscillator, 387
Frequency modulation (see Modulation)
Frequency-modulation detection, 401
Frequency-modulation receivers, 400,

cutin, 69

401

cutout, 69
effect on inverse peak potential, 73
peak-tube current with, 69

72
choke or inductor input, 66
cutout, 67
ripple, 68
controlled rectifier, 76
decoupling, in amplifiers, 115
L-section, 73
bleeder resistance, 75
critical inductance, 75
cutout, 74
ripple, 73
swinging choke, 76
multiple L-section, 76
ripple, 77
pi section, 78
ripple, 78
Flash-back potential, 49
ripple,

Forward potential
Fourier

rating, rectifiers, 49

series, amplifier

pull,

output, push-

204

single-tube, 96

output, single-phase, full-wave,

65
half-wave, 65
Frequencies, half-power, 124, 126, 215
midfrequency range, 123
sideband, in amplitude-modulated
wave, 321
rectifier

139, 215,

221, 228, 231

with inverse feedback, 151

Frequency conversion, 368

Frequency converter tubes, 372
Frequency distortion, description, 102

negative, 151
positive, 151

and

479

Frequency response

(see

Frequency

distortion)

Frequency stability of
Frequency translation
Modulation)

oscillators,
(see

297

Detection;

Full-wave rectification, 58

Gain, potential

(see Amplifiers, potential

gain)

Gain-bandwidth product, tuned

amplifier, 227
untuned amplifier, 127
Gas amplification, mechanism, 21
phototube, 20

Gas diodes, 43
Gas phototubes, 20
Gaseous conduction, 41
Gated-beam tube, 416

Glow
Glow

discharge, 41
tubes, diodes, 42

triodes, 43
Graphical determination of Fourier
components, 96
Grid, control, definition, 28
cold-cathode triode, 43
grid-control curves, 29
in

vacuum

tubes, control of current,

29
heating of, 248, 262
nonuniform, 38

RADIO ELECTRONICS

480

Grid, screen, 34
suppressor, 37
in thyratron, 47
Grid-bias modulation, 336, 340
Grid-bias potential, self -bias, 115
Grid-control curves, thyratron, 46, 47
Grid current, causes, 29, 248
class
class

A
C

amplifier,
amplifier,

oscillators,

248
258

292

thyratrons, 47
Grid driving power, class

amplifiers,

262
Grid glow tube, 43
Grid leak and capacitor-bias potential,
292
build-up, 292
effect on oscillator stability, 293
Grid-plate transconductance, 31, 37
Grid potential, bias, 115
(See also Self-bias)

246
Grounded-grid amplifier, 104
cutoff, 101,

Inductance, swinging choke, 76

Inductance-capacitance-coupled
amplifier, 134
Information theory, 429 ff.
discrete systems, 429
entropy, 432
Hartley law, 432
continuous signals, 434
effects of noise on, 436
system evaluation, 437
sampling theorem, 433
Input admittance, effect on, of feedback,
169
of lead inductance, 112
pentode, 111
tetrode, 110
transit-time effect, 113
triode, 108
Input capacitance of amplifiers, 109
effect on operation, 125
Input potential (see Excitation potential)
Input resistance, triodes, 110
negative, conditions for, 110
Instantaneous amplitude, 320
Instantaneous angular frequency, 320,

Harmonic

distortion (see Nonlinear

distortion)
Harmonic generation, in linear detector,

362
in modulation, 330

63

350

in triodes, 95
(see

Fourier series; Nonlinear

distortion)

Hartley law, 432

continuous signals and, 434
Hartley oscillator, 284
Heat-shielded cathodes, 9
Heising modulator, 329
Heptode, 38
Hexode, 38

High

field emission, 21
High-level modulation, 330
High-pressure arcs, 45

Ignitor rod, 47
Ignitron rectifier, 47
Image force, 16

Impedance matching,

Insulation stress in transformers, 57, 60,

Interelectrode capacitance, 107

in square-law detection,

Harmonics

378
Instantaneous phase, 320, 378

effect

ff.

on input admittance, 109

equivalent circuit including, 107

numerical values for several tubes,
455 ff
pentode, 111
reduction by screen grid, 34
tetrode, 111

Intermediate frequency, 12, 369

Inverse feedback, 151
Inverse-peak-potential rating, 27, 49
bridge circuit, 62
potential doubler, 63
single-phase, with capacitor input
filter, 73
full-wave circuit, 60
half-wave circuit, 57
Inverted amplifier, 142
Inward-radiating cathodes, 18
Ionization, description, 21
by electron bombardment, 21
in gas phototubes, 20
,

188, 192

by transformer, 190
Impedance-stabilized oscillator, 299
Inductance, cathode lead, 112
for filters, 66
critical, in L-section, 75
leakage, 138
primary, of transformer, 136

photoelectric, 22
in plasmas, 41

by

positive ions, 22
Ionization time, 50
Ions, 21
function in gaseous conduction, 42

INDEX
Ions, in glow-discharge cathode-fall space,

41

plasma, 41

in

Johnson noise, 145

Junction analysis, 447

481

Metals, image force, 16

secondary emission, 21
thermionic emission, 16
work function, 16
Millman network theorem, 450
Misch metal, 42
Mixer, 368
Mixing (see Conversion)

Modulated wave, components,

Kinetic energy, electrons and electron
velocity, 24

Konal metal, 17

filter, 73
Langmuir-Child's law, 24

L-section

(See also Space-charge-limited current)

Leakage reactance, 139

Limiter, amplitude, 401
Linear amplifier, class A, 95
class B, 246
Linear detector, 351

Load

line, nonlinear, in

push-pull

amplifier, 203

pentode amplifier

196
half-wave, 55

circuit,

rectifier, single-phase,

triode amplifier circuit, 85

resistance, choice in pentodes, 196

Load

class

amplifier, 85

maximum power

output, 188, 192

matching, 190
push-pull amplifier, 202, 203
Local oscillators, 368, 372
Loop analysis, 445
Low-level modulation, 325

Master

oscillator,

Maximal

298

221
Maximum power output, class A
amplifier, 188, 192
push-pull connection, 202, 203
from linear modulator, 334
undistorted, 192
Mercury, vapor pressure of, 44
Mercury-arc rectifier, arc back, 49
arc drop, 42, 49
cathode disintegration, 45
evaporation of mercury, 44
flash-back potential, as function of
temperature, 49
ratings, average current, 49
forward potential, 49
inverse potential, 49
surge current, 49
vapor pressure, 44
Merit, figure of, vacuum tube, 129
flatness,

amplitude-modulated, 321
frequency-modulated, 380
envelope, 378
Modulation, amplitude, 3, 320 ff
comparison with frequency
modulation, 320, 419
definition, 320
expression for, 321
frequency spectrum, 322
information-theory considerations,
438
linear, 330
sidebands, 4, 322
sinor representation, 322
square-law, 325
angular, 378
instantaneous, 378
carrier suppression, 6, 327
cathode, 341
cross, 342
definition, 320
feedback to improve linearity, 324
frequency, 5, 377 ff.
capture, 423
characteristics, 377
comparison with phase modulation,
386
definition, 378
deviation ratio, 380
expression for, 380
frequency deviation, 383
information-theory considerations,
438
narrow-band, 395
spectrum, 380
wide-band, 387
frequency translation by, 322
grid-bias, 336
advantages, 340
high-level, 330
index, 321
information theory, 437
low-level, 325
percentage, 321
phase, 385
comparison with frequency
modulation, 386
expression for, 386

482

RADIO ELECTRONICS

Modulation, phase, sinor representation,

396
plate, 331
advantages, 340
process of, 320
pulse, 7

information-theory considerations,
439
sidebands, 321
single-sideband, 6, 328
sinor representation, 322
square-law, 325

suppressor grid, 342

transformer, 335
types, 320

van der Bijl, 325

Modulation characteristics, 331
Modulation factor, 321, 380
Modulation frequency, 321
Modulator, amplitude, 321 ff.
balanced, 327
grid-modulated class C amplifier
336
Heising, 329
plate-modulated class C amplifier
331
van der Bijl, 325
frequency, 387
Armstrong method, 395
phasitron, 398
reactance tube, 387
stabilization, 393
phase, 385
single-sideband, 328
suppressed carrier, 328
(See also Modulation)
Multielectrode tubes, 34
beam power, 38
shield-grid thyratron, 47
(See also Pentode; Tetrode)
Multiple L-section filters, 76

as,

as,

Multiplication in high-vacuum tubes,

frequency, 278
potential, 63
Multistage amplifiers, 119
Mutual characteristics, 29
Mutual conductance, 31

ff.,

212

ff.

measurement, 94
numerical values of several tubes,
455 ff.

Negative feedback, 151

Negative input resistance of amplifiers,
110, 309
Negative plate resistance in tetrodes, 35,
309

Negative resistance oscillator, 309

Negative transconductance oscillator, 319
Network analysis, general, 443 ff.
Network theorem, Millman, 450
Norton, 450
Thevenin, 449
Neutralization, in amplifiers, 273
electron space charge, 42
Noise, 143 ff.
induced grid, 147
quantizing, 9
resistance, equivalent, 146
shot, 145
thermal agitation, 144
Noise figure, 148 ff.
networks in cascade, 148
Nonlinear circuit element, as modulator,
325
as rectifier, 53 ff.
Nonlinear distortion, description, 95 ff.
distortion factor, 97
effect of feedback on, 154
five-point schedule, 97
general dynamic curve, 95
parabolic dynamic curve, 95
push-pull amplifiers, 204
Nonlinearity and oscillator stability, 287
Normal current density, 24
Normal glow discharge, 41

Operating point, 85
Oscillation, in amplifiers, 150

ff.

Barkhausen criterion for, 287

Nyquist criterion for, 172
283 ff.
amplitude stability, 301
automatic amplitude control, 306
basic circuits, 296
bridged-T, 307
class A, 301
classification, 283
Colpitts, 296
criteria for oscillation, 387
crystal-controlled, 300
circuits for, 301
equivalent circuit, 301
stability, 301
dynatron, 309
electron-coupled, 298
feedback, 284
self -excitation in, 284
Franklin, 313
frequency-modulated, 387
frequency stability, 297

Oscillators,

grid bias in, 291

Hartley, 296

INDEX
Oscillators, intermittent operation,

293

372
locked, 412
master, 298
Meacham, 318
local, 368,

negative resistance, 309

negative transconductance, 310
nonlinearity and stabilization, 286

phase shift, 308

power, 293
quench, in superregenerative detector,
365
resistance-capacitance tuned, 302
resistance-stabilized, 299
self-starting, 292
stabilization, amplitude, impedance,
299
resistance, 299
stabilization factor, 297
transconductance, effect of, 287
transitron, 319
tuned-grid, 296
tuned-grid tuned-plate, 296
tuned-plate, 293
frequency of oscillation, 295
self-excitation, 294
twin-T, 307
Wien-bridge, 305
Output impedance, 167
effect of feedback, 167

Output potential

of amplifier (see

Amplifiers, potential gain)

Oxide-coated cathodes, 17
'

disintegration, 45

work function, 17

Parallel-feed amplifiers, 189, 195

Paraphase amplifiers, 205

Peak-inverse-potential rating (see

Inverse-peak-potential rating)

Pentode, 37
in class

A
B

amplifier, 196

amplitude-frequency
amplifier, 203

in class

distortion,

38
circuit, 111

equivalent
input admittance, 111
interelectrode capacitances, 111
load resistance, 196

modulation with, 342

plate characteristics, 37
potential gain with, 111

Phase distortion, 102

Phase inverter, 205
Phase modulation (see Modulation)
Phase relations, amplifier, 86, 127
Phase-shift characteristics of tuned
amplifier, 215
Phase-shift distortion, 102
Photocells (see Phototubes)
Photoelectric emission, 18
gas amplification, 20

Phototubes, vacuum, 20
gas-filled, 20
potential-current curve, 20
Pi-section filters, 78
Piezoelectric crystals in oscillators, 300
Plasma, description, 41
Plate characteristics, 455 ff.
Plate-circuit efficiency, amplifier, class

A, 193
class B, tuned, 251

modulated, grid, 339

334
tuned, 261
Plate current, average value, 250, 255,
258, 332
total differential, 30, 88
Plate dissipation, 194
in class C amplifier, 262
during modulation, 335, 339
permissible, 253
in tuned class B amplifier, 252
Plate efficiency, amplifier, class A, 193
series-fed, 195
shunt-fed, 195
tuned class B, 251
tuned class C, and operating angle,
262
Plate modulation, 331, 340
Plate-plate resistance, 200
class C,

plate,

Plate resistance, diode, 53

measurement, 94
multielectrode tubes, 35, 37
negative, in tetrodes, 35
numerical values of several tubes,

455

ff.

push-pull amplifiers, 200

tetrode, 37
triode, 31
variable nature, 31

Polar diagram of feedback amplifier, 172

Positive feedback, 151
Positive-ion bombardment, 21, 41
Positive-ion current, 21, 40
Positive ions (see Ions)

power, 196

remote cutoff, 38
in tuned amplifier, 212
Phase characteristic, 127

483

ff.

Potential, breakdown (sparking), 41

disintegration, 45

RADIO ELECTRONICS

484

Potential distribution, 22, 41

glow discharge, 41
parallel-plane diode, space-charge
flow,

22

in plasma, 41

Potential gain, cascade amplifier, 121

decibels, 104

with feedback, 151

(See also Amplifier, potential gain)

Potential-source equivalent circuit, 87

Power, rectifier, input, 56, 59

maximum

(see

Maximum power

output)
output, 56, 59
Power amplifiers, 187 ff.
Power oscillator, 293
Power sensitivity, 103
Power series, 95
Power supply, 115
Predistorter in frequency modulation, 398
Preemphasis of high frequencies, 392
Pressure, mercury, as function of temperature, 44
Pulse modulation, 7, 439
Push-pull amplifier, class A, 197 ff.
circuit, 198
composite dynamic characteristic,
203
composite static characteristic
curves, 200
distortion, 204

equivalent circuit, 199

harmonics

in,

204

plate-plate resistance, 200

power output, 204

single-ended, 211
class B,

205

driver stages for, 205

single-tube paraphase, 206

two-tube paraphase, 206

point, 85
Quadrupler, potential, 64

Radio communication, amplitude

modulation, 4
frequency modulation, 5, 377, 400
modulation essential for, 3, 320
pulse modulation, 7
single sideband, 6,

328

suppressed carrier, 6, 328

Radio-frequency amplifiers (see
amplifiers)

Ratio detector, 409

Tuned

Reactance, primary, effect on amplifier

operation, 136
transformer, leakage, effect on
amplifier operation, 139
Reactance tube, 388
Receiver, automatic gain control, 362
bandwidth, 5
for amplitude-modulated waves, 239
for frequency-modulated waves, 5,
400
for pulse-modulated signals, 439
frequency-modulation, 13, 400
discriminator, 401
limiter, 401
selectivity, 212, 226
single-sideband, 367
superheterodyne, 12, 375
superregenerative, 365
tracking, 370
tuned radio frequency, 12
volume control, 362
Recombination, ions, 41
Rectification, definition, 54
Rectification characteristics, 360
Rectifier filters, 76
Rectifier meter, 62
Rectifiers, circuits, bridge, 62
full-wave single-phase, 58
gas tubes, 60
half-wave, single-phase, 54
potential doubling, 63
potential quadrupling, 64
Rectigon tube, 45
Reflected resistance, 190
Regeneration in amplifiers, 151
Regulation, potential, 115
Remote-cutoff tubes, 38
Resistance, bleeder, 75
grid, for high-vacuum tubes, 84
input, of tubes, 110
load (see Load resistance)
negative, 309
plate (see Plate resistance)
plate-plate, 200
reflected, 190
Resistance-capacitance-coupled
amplifier, 119 ff.
applicability, 119
cascaded stages, 129
feedback, 160
frequency characteristic, 127
gain of, high-frequency, 125
low-frequency, 123
mid-frequency, 123
gain-bandwidth product, 127
half-power frequency, 124, 126
universal amplification curve, 126

INDEX
Resistance-capacitance tuned oscillator,

302
Resistance-coupled amplifiers

(see

Sidebands, phase-modulation, 386

Signal/noise ratio, amplitude-modulated
system, 420

bandwidth trading, 438

Amplifiers, direct-coupled)
Resistance-stabilized oscillator, 299

capture effect in frequency modulation,

423
comparison of amplitude modulation
and frequency modulation, 423
frequency modulation, 421
Hartley law, 432
Signal potential (see Excitation potential)
Sinors, 443
Sources of electrons, 15
(See also Electron emission)
Space charge, cloud, 22

Resonance, double-tuned, 221

parallel, 214
transformer-coupled circuit, 218
universal resonance curve, 215
Ripple factor, capacitor filter, 72
definition,

485

58

full-wave circuit, 60
half-wave circuit, gas tube, 62
vacuum tube, 58
inductor filter, full-wave, 68
L-section filter, 73
multiple, 77
pi-section filter, 78

glow discharge, 41
limitation of current by, 22, 24
in

in plasma, 41

positive-ion neutralization, 42
(see Space-chargelimited current)
Space-charge-limited current, factors
influencing, 22
parallel planes, current between, 24

Space-charge flow

Sampling theorem
433

in

information theory,

Saturation, in phototubes, 20
in screen-grid tubes, 35
space-charge-limited current, 25
temperature-limited current, 25

tungsten filament, 25
Schedule method in Fourier analysis, 97
Screen grid in tubes, 34
Screen-grid tubes (see Tetrode)
Screen supply potential, 115
Second harmonic (see Nonlinear
distortion; Ripple factor)
Secondary emission, by ion bombardment, 21
suppression, in beam tubes, 39
in pentodes, 37
in tetrodes, 35
Self-bias in amplifiers, feedback resulting
from, 115
in push-pull amplifiers, 199
Semigraphical analysis, class B and class
C amplifiers, 256
Series, Fourier (see Fourier series)
power, 95
Taylor, 30, 88
Shield-grid thyratron, 47
Shunt-feed amplifiers, 189, 195

by feedback,

153
Stagger-tuned amplifier, 228
Starter probe, 42
Static characteristics, 455
Supercontrol tubes, 38

ff.

Superheterodyne, 12, 375

Superregenerative receiver, 365
Suppression of carrier wave, 6, 328
Suppressor-grid modulation, 342
Surge-current rating, 49
Swinging chokes for filters, 76

Tank

circuits for tuned power amplifier,

242
Taylor series for triode, 30, 88

(See also

Power

series)

arc cathodes, 18
glass envelope of tubes, 26
mercury pressure, 44

380

receivers (see Receiver)

separation of carrier from, 328
Sidebands, amplitude-modulation,
separation from carrier, 328
clipping in detectors, 357
frequency-modulation, 380

Stabilization of amplifiers

Temperature, anodes in diodes, 26

Side frequencies, detection (see

Detection)
for frequency modulation,

three-halves-power dependence, 24
Sparking potential, 41
Square-law modulation, 325
Stability of oscillators, amplitude, 299
frequency, 297
reactance tube, 387

4,

322

rating of diodes, 26, 49

thermionic cathodes, 16
Temperature-limited current, 25
Tetrode, 34
coefficients, 37

RADIO ELECTRONICS

486

Tetrode, disadvantages, 37
equivalent circuit, 111
gas, 47
input admittance, 110
interelectrode capacitances, 110

modulation with, 342

mutual conductance, 36
negative resistance, 35
plate characteristic, 35
screen grid, 34
secondary emission, 35
(See also Beam power tubes)
Thermionic cathodes, carbonization, 17

directly heated, 18
efficiency, 17

heat-shielded, 18
indirectly heated, 18
oxide-coated, 17, 45
temperature, 17
thoriated-tungsten, 17
Thermionic emission, 16

Transformer-coupled amplifiers, impedance matching, 190

push-pull amplifier, 200
tuned, in amplifiers, 216-218
Transformer insulation stress, full-wave
rectifier, 60
half-wave rectifier, 57, 73
Transformers as impedance-matching
devices, 190
Translation, frequency, by modulation,
322
Transmitters, amplitude-modulated, 11,
346
frequency-modulated, 13, 387, 395, 398
using phase-shift modulators, 395
using reactance-tube modulators,
387
stabilization of, 393
Triode, gas (see Thyratrons)
vacuum, 28 ff.
amplification factor (see Amplification factor)
characteristic curves, 455 ff.

constants of (A 6o, Eiv), 16

Dushman equation, 16
efficiency, 17
power for heating cathodes, 114
saturation, 16
work function, 16
(See also Work function)
Third harmonic (see Nonlinear distortion)
Thoriated-tungsten cathodes, 17
Three-halves-power law, 24
Threshold frequency, 18
Threshold wavelength, 19
Thyratrons, 45
,

47
18
(See also Oxide-coated cathodes)
critical grid-control curves, 46
grid glow tube, 43
ionization time, 50
mercury-vapor pressure, 44
positive control tube, 46
shield-grid, 47
Time deionization, 50
Time delay in amplifiers, 102
Time ionization, 50
Transconductance, grid-plate (see Mutual
conductance)
Transfer characteristic, 29
Transformer, interstage, 135
capacitances in, 136
output, push-pull, 200
r-f air core, 212
Transformer-coupled amplifiers, 135 ff.
gain, high-frequency, 138

coefficients, 31

constant-current characteristic, 30
grid current, 29
(See also Grid current)
grid heating, 262

input admittance, 109

interelectrode capacitances, 107

mutual characteristics, 29
mutual conductance (see Mutual
conductance)
plate characteristics, 29
plate resistance, 30
space charge, 28
square-law detection with, 349
square-law modulation with, 325

argon-filled,

cathode

for,

low-frequency, 137
mid-frequency, 136

Taylor

Tube

series for, 30,

88

constants, multielectrode tubes, 37

relation

among, 31

triodes, 31

Tube drop,

in arcs, 42

glow tubes, 42
Tubes, 455 ff.
in

figure of merit, 129

Tuned

amplifiers, class A,

212

ff.

bridged-T, 236
cascode, 238
coupling, 212
distortion, 212

double-tuned, 218

ff.

bandwidth, 216, 218, 225

cascade, bandwidth reduction, 226
gain, 220
gain-bandwidth product, 227

maximal

flatness,

221

INDEX
Tuned

amplifiers, class A, double-tuned,

transitional coupling, 223

single-tuned, 213 ff.

cascade, bandwidth reduction, 225
direct-coupled, 213

bandwidth, 216
gain, 215
gain-bandwidth product, 227
transformer-coupled, 216
bandwidth, 218
gain, 217
optimum coupling, 217
stagger-tuned, 228
comparison with double-tuned,
229
class B power, 246 ft.
analysis tabulation, 261
analytical treatment, 249
applications, 324
optimum conditions, 253
phase relations, 249
plate efficiency, 252
semigraphical analysis, 248, 256
class C power, 253 ff.
analysis tabulation, 261
approximate analytical solution, 268
cathode-modulated, 341
conditions for high efficiency, 261
design considerations, 264
frequency multiplication, 278
grid-bias-modulated, 336
grid current, 258
grounded-grid, 277
idealized, solution, 255
plate current, 258
plate dissipation, 262
plate efficiency, 261
plate modulation, 331
power considerations, 260
push-pull, 272
Q of load, 245
semigraphical analysis, 256
tubes, 274
neutralization, grid, 273
plate, 273
twin-T, 230 ff.

487

Tuned amplifiers, twin-T, bandwidth, 232

circuits,

233

gain, 235

input impedance, 232

Wien-bridge, 236

Tuned

circuits, class

B and

class

242
double-tuned, 218
amplifier,

half-power frequencies, 215, 221

in oscillators, 291
RC networks, 230, 236
single-tuned direct-coupled, 213
single-tuned transformer-coupled, 216
stagger-tuned, 228

Tuned-grid oscillator, 296

Tuned-grid tuned-plate oscillator, 296
Tuned-plate oscillator, 293
Tuned power amplifier (see Tuned amplifiers, class B power, class C power)
Tungar tubes, 45
Tungsten, filaments, 17

work function, 17

Universal resonance curve, 215

Van der Bijl modulation, 325

Variable-mu tube, 38
Volt-ampere characteristics, 455
Voltage, disintegration, 45
Voltage-doubling rectifier, 63
Voltage gain (see Potential gain)
Voltage-quadrupling rectifier, 64

ff.

Voltage-source equivalent circuit, 87

Wavelength, threshold, 19

Work

function, 16

definition, 16

Dushman equation involving, 16

oxide-coated cathodes, 17
photoelectric emission, 19
table of values, 17
thoriated tungsten, 17
tungsten, 17

ff.