Chapter 5: Oscillators

References:
Muhammad H. Rashid, “Microelectronic circuits : Analysis and Design”, PWS Publishing
Thomas L. Floyd, “Electronic Devices”, Prentice Hall
Jacob Millman & Christos C. Halkias, “Integrated Electronics”, McGraw-Hill

Part 1: Sinusoidal Oscillator

General Principles of Oscillation

A sinusoidal oscillator is a circuit that produces sinewave output without any external signal
excitation. It is a special case of feedback amplifiers that is intentionally made unstable. The output
supplies the necessary excitation to its own input terminals and results in an oscillation. It is also
known as positive feedback (as opposed to negative feedback) where the feedback signal reinforces
the output signal in a regenerative fashion.

The basic principle of oscillator circuit can be explained using the block diagram in Figure 1.
Consider first that the feedback network is not yet connected to the amplifier input. A signal vi applied
to the amplifier input will produce a signal vo at the amplifier output. The output of the feedback
network will be vf =  vo =  A vi. At suitable values of A & f, vf is identically equals to the externally
applied input signal vi. If the external source were removed and the feedback signal vf is momentarily
connected to the amplifier input, the amplifier will continue to produce the same output signal vo,
since it has no means of distinguishing the source of the input signal applied to it. Of course, the
statement vf = vi means that the 2 waveforms have exactly the same instantaneous magnitude and
phase. This condition is equivalent to  A = 1. A closed loop is formed when the feedback signal is
connected to the amplifier input. The product  A is called the loop gain.

vi Amplifier vo = A v i
A = voltage gain of the
A = A(f) amplifier, |A| > 1
 = voltage gain of the
vf =  vo
feedback network, || < 1
Feedback Network Both A &  are dependent
 = (f) of frequency (complex
valued functions).

Figure 1: Block diagram of a feedback amplifier.

The feedback network usually contains reactive elements (capacitors and/or inductors). The
magnitude and phase of the transfer function  vary with frequency. The condition  A = 1 will
probably be fulfilled at only one particular frequency. This means the only periodic waveform which
will preserve its form is the sinusoid. This condition is known as the Barkhausen Criterion which
states that: “the frequency at which a sinusoidal oscillator will operate is the frequency at which the
total phase shift as a signal flows from the input terminals, through the amplifier and the feedback
network, and back again to the input, is precisely zero (or an integral multiple of 2)”. At the desired
oscillation frequency, the total phase shift ( A) is zero if the feedback network introduces a phase
shift of 180o and the amplifier introduces another phase shift of 180o (i.e. an inverting amplifier).
Alternatively, if a non-inverting amplifier which introduces no phase shift is used, the feedback
network must gives a phase shift of 360o.
In order to fulfil the condition  A = 1, another requirement for the oscillation to sustain is
the amplifier gain magnitude |A| must be equal to the inverse of the feedback transfer function
magnitude 1/||. In other words, the magnitude of ||, which varies with frequency, must be equal to
1/|A| at the desired oscillation frequency. This is an integral part of the Barkhausen Criterion (i.e. 
A = 1).
The frequency range of oscillation is determined by the feedback components, RC
components generate a waveform at audio frequencies, i.e. from several Hz to several hundred of
kHz. LC components generate a waveform at radio frequencies, i.e. from 100kHz to 100MHz,
crystals generate waveform over a wide range from 10kHz to 10MHz. Due to relatively slow
frequency response of op-amps, BJTs or FETs are used for frequencies above 1MHz.

Frequency stability
Frequency stability is the ability of an oscillator to oscillate at an exact frequency. The oscillation
frequency is a function of circuit components. It can change in response to temperature changes,
device displacement, or parasitic elements. The frequency stability is increased if the change of phase
shift w.r.t. frequency ( |d/d| ) at resonance is increased. This is because a slight change in  is
sufficient to correct any phase shift and restore the loop gain phase to zero or multiple-2 phase shift.
The larger the quality factor Q [Q = reactive power/average power (PX/PR) at resonance, average
power is the power dissipated] of a circuit, the better the frequency stability, because |d/d| near
resonance is greater. Crystal oscillators are far more stable than RC or LC oscillator, especially at
higher frequencies. The equivalent electrical circuit of a crystal has a very high Q value, leading to a
high value of |d/d|. The Q-values of LC oscillators > that of RC oscillators.

Amplitude stability
The gain of practical amplifiers can change in response to changes in temperature, age, and operating
point. Hence, |A| might drop below unity. If |A| < 1, then an oscillating circuit ceases oscillating.
In practice, an oscillator is designed with a value of |A| that is slightly higher than unity, say by 5%,
at the oscillating frequency. The greater of the value of |A| will cause the greater of the amount of
its waveform distortion.

Phase-Shift Oscillator
In a practical oscillator, the feedback network can be permanently connected in closed loop
with the amplifier. The loop gain must be slightly greater than unity. When the power supply is turned
on, noise signal will be amplified and the unstable loop will cause the oscillation signal to build up
until the amplifier output swings beyond the linear region (into the cutoff region or the saturation
region, for example). Hence, the amplitude of the oscillation is limited by the onset of nonlinearity.
 The Phase-Shift Oscillator circuit in Figure 2 is particularly suited to the frequency range
from several hertz to several hundred of kilohertz. The three cascaded arrangements of a capacitor C
and a resistor R form a high-pass filter. The phase-magnitude frequency response can be obtained
using a swept-frequency measurement or a theoretical analysis of the circuit network. The input
resistance of the inverting amplifier is equal to R1. The equivalent circuit is as shown in Figure below.

R1= R R2
-JX -JX -JX
Vf Vf
+ Vo Vo
R R
R
C C C

Rth
R R
Vth

Vo R
VTH 
R  jx
R( jx)
RTH   jx 
R  jx
 jRx  x 2  jRx ( R  jx  R)  jx(2 R  jx)
   jx 
R  jx R  jx R  jx

-JX Rth
Vf
V1
R Vth
R

Rth1
Vth1

Vo R 2
V R ( R  RTH )
VTH 1  TH 
R  RTH R  RTH
  jRx(2 R  jx)
R || RTH R  jx
RTh 1  
1  jx(2 R  jx)
R
R  jx
 jx(2 R  jx)  jRx(2 R  jx)
RTh1  2  2
R  jRx  x  j 2 Rx R  x 2  3 jRx
2

-JX Rth 1
Vf
V1
Vth 1
R

Vo R 2 Vo R 2 Vo R 2
VTH 1    2
R( R  jx)  RTH ( R  jx) R( R  jx)  jx(2 R  jx) R  jRx  2 jRx  x 2
 jRx(2 R  jx)
RTH 1  2
R  x 2  3 jRx

Vo R 2 .R
VTH 1 R R( R  jx)  jx(2 R  jx)
Vf  =
R  jx  RTH 1  jRx(2 R  jx)
R  jx  2
R  x 2  3 jRx

Vo R 2 .R

( R  jx)(R 2  x 2  3 jRx)  jRx(2 R  jx)

Vo R 2 .R

R 3  Rx2  3 jR 2 x  jR 2 x  jx 3  3Rx2  2 R 2 x  Rx2

Vo R 2 .R Vo R 2 .R
 
R 3  5Rx2  6 jR 2 x  jx 3 R 3  5Rx2  j ( x 3  6 R 2 x)
Vf R 2 .R
 
Vo R 3  5Rx 2  j ( x 3  6 R 2 x)

x3  6R 2 x
Phase Shift,   tan 1
R 3  5Rx2

This is an inverting amplifier with –Ve gain.

∴ To make Aβ = 1, the real value, the imaginary part of β must be zero. (or θ = 1080)
 x 3  6R 2 x  0
x 2  6R 2
xR 6
1
 R 6
wo C
Frequency of oscillatio n
1
 fo 
2RC 6
R3
∴ For Aβ = 1, |  | ∎ x2 = 6R2
R 3  5Rx2
R3 1
 3 
R  30 R 3
29
| A | 29
R
 2  29.
R1

Wien-Bridge Oscillator
The Wein-Bridge oscillator consists of a non-inverting amplifier and a bandpass filter feedback
network. It is a bridge because R1, (R2+R3), (R-jX) and (-jX//R) are located at the arms of a bridge,
respectively, X = 1/C.
D1
Vo
D2

R2
R Z1
+12V R2 + R3
R1
2 _ 7 R3
C
6 Vf
3 + Vo
4 - +
-12V

Vf R C
R1 C R Z2

C R

Figure: Wein-brigde oscillator

 Vf Z2  jX // R  jRX /( R  jX )
   
Vo Z 2  Z 1  jX // R  ( R  jX )  jRX /( R  jX )  ( R  jX )
 jXR  jRX XR
  
 jXR  R  jX 
2
 jRX  R  2 jRX  X
2 2
3 XR  j R 2  X 2  
In order for  A = 1, the imaginary part of  must be zero (or  = 0), i.e. R2 = X2. This gives:
1  1
X R  fo  o 
oC 2 2RC

Substitute this into the expression for , we get  = 1/3. Hence, an oscillation can be set up if the gain
of the non-inverting amplifier is slightly larger than 3.

R 2  R3
A 1 3
R1

Similarly, the amplitude of oscillation can be reduced by means , where R2 + R3 is slightly

of modifying > 2R
the linear operating
1
so that
range using a pair of diodes. It also stabilises the oscillation frequency. A is slightly > 3.

LC Oscillators
A general form of oscillator circuit is as shown below. The amplifier can be a BJT, FET, or op-amp.
E.g., an npn BJT is used, Vf is fed into the base, the emitter is connected to ground through a coupling
capacitor and the collector provides Vo, coupling capacitors may be required for the base and the
collector to keep the BJT at desired biasing point. To simplify the analysis, we assume the input
impedance of the amplifier is arbitrarily large.

BJT, FET or OP-AMP Ro

Vo
+
A<0 V’o = -AVf
Vo -

Vf
jX3
Ztot = Ro + jX2//(jX1+jX3)
Pi-to-tee transformation,
ZB = (jX1)(jX2)/(jX1+jX2+jX3)
jX1 jX2 ZB is located as in the phase–
shift oscillator case.

Figure: LC oscillator

Consider the amplifier has an output resistance Ro, the feedback is redrawn as follows :
 Ro
jx3
Vf
Vo

jx1 jx2

After Wye-Delta Transformation the circuit becomes as follows:

Ro

ZA ZB
Vf Vo

ZC

Where the values of the elements are

𝑗𝑋1 ∗ 𝑗𝑋3 𝑗𝑋1 ∗ 𝑋3

𝑍𝐴 = =
𝑗𝑋1 + 𝑗𝑋2 + 𝑗𝑋3 𝑋1 + 𝑋2 + 𝑋3
𝑗𝑋2 ∗ 𝑗𝑋3 𝑗𝑋2 ∗ 𝑋3
𝑍𝐵 = =
𝑗𝑋1 + 𝑗𝑋2 + 𝑗𝑋3 𝑋1 + 𝑋2 + 𝑋3
𝑗𝑋1 ∗ 𝑗𝑋2 𝑗𝑋1 ∗ 𝑋2
𝑍𝐶 = =
𝑗𝑋1 + 𝑗𝑋2 + 𝑗𝑋3 𝑋1 + 𝑋2 + 𝑋3
𝑗𝑋1 ∗ 𝑋2
𝑉𝑓 𝑍𝐶 𝑋1 + 𝑋2 + 𝑋3
= =
𝑉𝑜 𝑅𝑜 + 𝑍𝐵 + 𝑍𝐶 𝑗𝑋2 ∗ 𝑋3 𝑗𝑋1 ∗ 𝑋2
𝑋1 + 𝑋2 + 𝑋3 𝑋1 + 𝑋2 + 𝑋3 + 𝑅𝑜
+

−𝑋1 ∗ 𝑋2
=
−𝑋2 (𝑋1 + 𝑋3 ) + 𝑗𝑅𝑜 (𝑋1 + 𝑋2 + 𝑋3 )

The inverting amplifier has a voltage gain A = - |A|. In order for A = 1, the imaginary part of  must
be zero or  = -180. This gives:

Vf X1 X
X 1  X 2  X 3   0 and    1
Vo X1  X 3 X2

The circuit will oscillate at the resonant frequency of the series combination of X1, X2 and X3. The
reactive elements X1 and X2 must be of the same type (so that  is a negative value). |A| must be
slightly larger than 1/|| to self-start up oscillation.
For Colpitt Oscillator
X1 and X2 are capacitors, So

1 1
X1   X2   X 3   o L3
 o C1  o C2
1 1
X1  X 2  X 3      o L3  0
 o C1  o C 2

1 1 𝐶1 + 𝐶2 1
𝜔𝑜2 𝐿3 = + = =
𝐶1 𝐶2 𝐶1 𝐶2 𝐶𝑇

1 C1C2
fo  where CT 
2 L3CT C1  C2

𝛽 = − 𝑋𝑋1 = − 𝐶𝐶2
2 1
So, the required amplifier gain at resonant frequency is A = -C1/C2.

Hartley Oscillator
For Hartley Oscillator, X1 and X2 are inductors and X3 is capacitor.

1
X 1   o L1 X 2   o L2 X3  
 o C3
1
X 1  X 2  X 3  o L1  o L2  0
o C3
1
 fo  where LT  L1  L2
2 LT C3
 = -L1/L2 and the required amplifier gain at resonant frequency is A = -L2/L1.

R1 Rf R1 Rf

+ Vo + Vo

L3 C3
C2
C1 L1 L2

Colpitt Oscillator Hartley Oscillator