7.
Oscillators
Oscillators are circuits with feedback which operate so that they are either critically stable (on the
boundary between stability and instability) or unstable. Sinewave oscillators are in the first category,
squarewave generators fall into the second. Oscillators are amplifiers with feedback which
compensate the losses of energy in the system.
7.1. Linear systems with feedback
Consider a linear time invariant (LTI) system with feedback in the frequency domain, where the input
signal is
, the output signal
, the feedback signal
, the amplifier input signal , the
amplifier gain , and the feedback ratio .Such a LTI system is schematically drawn in Figure
7.1.
The transfer function of such a system is a complex number
which indicates the gain and phase shift of the signal.
In electronics the signals are often voltages
, however they can be currents or power. The signals
can be represented as
Making all the substitution we obtain
The transfer function then becomes
Figure 7.1. Schematic representation of a linear time
invariant system with feedback
E L E C 3 1 7 P a g e | 1
The transfer function of a linear electronic circuit can be written as a ratio of two polynomials with
real coefficients
and
Or in the zero-pole form
The poles and zeros of the transfer function are either real or complex conjugate pairs.
For stability the denominator must have no roots within the right hand side of the
complex plane (with positive real part). According to the Nyquist criterion will have no
such roots if the Nyquist plot (of as a function of ) does not encircle the critical point
. This is graphically shown in Figure 7.2.
In order to design an oscillator the stability criterion should be applied in such a way that the circuit
oscillates at the design frequency.
The poles of the transfer function (due to real
and
) are
i. real
ii. complex conjugate
In case (i) a small disturbance at the input will cause the output to continuously increase. The system
is unstable but produces no oscillations, see Figure 7.3.
Figure 7.2. Graphical representation of the Nyquist criterion.
E L E C 3 1 7 P a g e | 2
In case (ii) the output will oscillate with an increasing amplitude, see Figure 7.4.
If there is a root of type (ii) without damping or gain, then we have containing
and if it can be rearranged so that there is no right hand side root (with positive real part),
then the response to an initial disturbance will be a continuous sinusoidal oscillation at the circuit's
output. In other words the circuit is an oscillator.
For this case may be written as
In the electronic formulation and
. It is clear that the frequency of oscillation will be
given by the value of . At , or i.e. the Nyquist plot passes
through the critical point at . In general both and are complex and may be written
Remembering the assumption that the feedback signal is inverted by the summator at the input of our
LTI, or ) we can write
Figure 7.3. Exponential increase of an unstable system without oscillation.
Figure 7.4. Exponential increase of an unstable system with oscillation.
E L E C 3 1 7 P a g e | 3
at the oscillation frequency. This may be separated into what is called the Barkhausen Criteria:
with an integer
This should be true only at
, the frequency of the oscillation) in other words the feedback
should be frequency selective.
This criterion is very useful for designing oscillators.
For both common emitter (CE) amplifier and common base (CB) amplifier so that
. For the CE amplifier
, therefore
etc. to make
. For
the CB amplifier
and, hence, we need
etc.
Figure 7.5. CE and CE amplifier circuit.
E L E C 3 1 7 P a g e | 4
7.2. RC oscillators (Wien oscillator)
The feedback network for the Wien oscillator is shown in Figure 7.6.
For this circuit the feedback is given by
where
. Thus
When
we obtain
In other words i.e. there is an attenuation of and zero phase shift.
Since for oscillation we require , this circuit may be used with
an amplifier having and zero phase shift , e.g. an operational amplifier as shown in
Figure 7.7.
Figure 7.6. The feedback network of
a Wien oscillator.
E L E C 3 1 7 P a g e | 5
The gain of the amplifier is given by
At
Since only when
, i.e.
will be the frequency of
oscillation. The capacitors (or resistors) may be ganged to allow adjustment of frequency.
Note that . For high frequency of oscillation will be small: e.g. if MHz and
, then . Clearly even at 1 MHz, becomes comparable with the stray capacitance,
for higher frequency operation must be reduced and this will increase the power dissipation. Hence
the Wien oscillator is unattractive above MHz. At higher frequencies, lower loss feedback
networks are needed such as circuits or crystal resonators.
Another issue with Wien oscillators is that for the circuit to oscillate we require the gain to be
precisely equal to 3. If the gain is less than 3 then even if an oscillation can be initiated it will die
away. If the gain is greater than 3 then the amplitude of oscillation will, for a linear circuit, increase
without limit.
Figure 7.7. Wien oscillator circuit.
E L E C 3 1 7 P a g e | 6
7.2.1. Stability of the oscillation amplitude
In practice, for all oscillators the gain is designed to be greater than the value required by the
Barkhausen criterion for small signal amplitude. Then, the inherent noise of the circuit is sufficient to
cause the build-up of an oscillation. A mechanism is designed into the circuit that reduces the gain
with increase of the signal amplitude. For some amplitude of oscillation the condition is
exactly satisfied. This principle is schematically drawn in Figure 7.8.
The gain control may be obtained in various ways based either upon nonlinear effects or an automatic
gain control.
The Wien operational amplifier oscillator, above, will build up an oscillation provided that
.
The oscillation will increase in amplitude until the output stage of the operational amplifier reaches
saturation. This reduces the effective gain and stabilises the amplitude of oscillation at the expense of
distortion of the waveform, see Figure 7.9.
The greater is the excess of
over the greater will be the distortion.
Lower distortion could be obtained by using a better controlled gain reduction mechanism. An
example of this is shown in Figure 7.10.
Figure 7.8. Graphical representation
of the oscillation condition.
Figure 7.9. Distortion stabilisation.
E L E C 3 1 7 P a g e | 7
For small amplitude of
the diodes
do not conduct, therefore
. For
large amplitude of
the diodes conduct and
. For some value of
,
and distortion can be quite small.
Another popular solution is to make
a resistor with high positive temperature coefficient of
resistance or
with high negative temperature coefficient of resistance. As the amplitude of
oscillation increases the temperature of the resistor will increase and, in either of the above cases,
cause the gain to fall to . This can lead to a very low distortion output.
7.3. Partitioning of the circuit
The operational amplifier used above may be regarded as ``ideal'' in this application (infinite input
impedance, zero output impedance and no reactive effects). More generally, amplifiers are not ideal
and the Barkhausen criteria are more difficult to apply. We might have the circuit given in Figure
7.11.
Now the loading of the amplifier by network,
and loading of the network by the amplifier,
, make it harder to find the oscillation frequency and the condition for oscillation.
Figure 7.10. Wien oscillator circuit including gain reduction
mechanism.
Figure 7.11. General circuit.
E L E C 3 1 7 P a g e | 8
It may be useful to partition the circuit differently as shown in Figure 7.12.
We will often prefer to draw the amplifier as shown in Figure 7.13.
Where sometimes
The feedback network will appear as a number of components in parallel with
.
We will see later that there are alternative methods of designing oscillators that avoid the problem of
finding a circuit partition that allows application of the Barkhausen criterion.
7.4. LC oscillators
RC phase shift networks are too dissipative to be usable at high frequencies.
The principles of an LC feedback are given in Figure 7.14.
Figure 7.12. Repartitioned circuit.
Figure 7.13. Amplifier.
E L E C 3 1 7 P a g e | 9
In Figure 7.14
And is a transconductive amplifier
, so that
Clearly only at the resonance, where
. At
and
for stable oscillations we must have i.e.
or
, thus
.
Some other alternatives are shown in Figure 7.15 and Figure 7.16.
At resonance
therefore we need
For the circuit in Figure 1.16,
so that . The Oscillation frequency is
. At resonance (Note that transformer action reduces the resistance value between the amplifier
output and ground.)
thus we require
.
There are many other possibilities to use CE, CB, and CC amplifiers with parallel or series resistance.
FETs and MOSTs are also used, as well as multiple device configurations.
Figure 7.14. LC Feedback circuit.
Figure 7.15. LC Feedback circuit.
E L E C 3 1 7 P a g e | 10
7.4.1. Common base amplifier
is often provided by a CB amplifier. Before analysing an actual oscillator circuit it is worthwhile
revising the CB configuration. The transistor equivalent circuit is shown in Figure 7.17.
Here
and
are neglected, more accurate calculations including their effect can be performed
using a computer.
The hybrid equivalent circuit may be redrawn for CB use as shown in Figure 7.18.
Here
CB gain. Miller's theorem has been applied to
. is often large and with little
phase shift. Using pessimistic numbers
, we estimate
, a
negative resistance. It is in parallel with a much smaller positive resistance and has the effect of
increasing this resistance slightly.
has been drawn as two current sources in series. If a connection is made at the dashed line the
circuit's function will be unchanged because no current will flow through it.
Figure 7.16. LC Feedback circuit.
Figure 7.17. CB transistor equivalent circuit.
Figure 7.18. CB transistor hybrid equivalent circuit.
E L E C 3 1 7 P a g e | 11
The following replacement can be made, see Figure 7.19.
Hence the common base equivalent circuit may be redrawn as in Figure 7.20.
Note the absence of a feedback capacitance, which makes the CB amplifier attractive for high
frequency operation. In practice there will be some capacitance, but this will be
.
Since the circuit is designed from the common emitter hybrid equivalent circuit it should not be
considered reliable above
.
7.4.2. Useful circuit transformations
Induction-Resistance
At one frequency the two circuits in Figure 7.21 can have the same impedance if the L's and R's are
chosen correctly. These circuits can be distinguished only if the frequency is changed.
The impedance of the series circuit may be written as
Figure 7.19. Replacement.
Figure 7.20. CB equivalent circuit.
Figure 7.21. Induction-
resistance equivalence.
E L E C 3 1 7 P a g e | 12
With
. It is now possible to define
With
Of particular interest is the high quality factor inductor for which
( will be
usually sufficient) when
And
As an example, we compare the circuits given in Figure 7.22.
The transformation, of course, works in the opposite direction with
. The transformation
is normally deemed to be acceptable both at, and near, the frequency
of the calculation.
Resistance-Capacitance
We can compare the two circuits given in Figure 7.23.
They can have the same impedance at some frequency if the R's and C's are chosen correctly. The
impedance of the series circuit may be written as
Figure 7.22. Induction-resistance
equivalence, example.
E L E C 3 1 7 P a g e | 13
Where
It is now possible to define
With
If the capacitor's dissipation factor,
, is low, say, then
and
And
As an example, we compare the circuits given in Figure 7.24.
Figure 7.23. Capacitance-
resistance equivalence.
Figure 7.24. Capacitance-resistance
equivalence, example.
E L E C 3 1 7 P a g e | 14
These transformations can be useful for:
1. designing oscillators
2. designing tuned amplifiers
3. designing power matching circuits.
7.4.3. Colpitts oscillator
The circuit diagram for Colpitts oscillator may be drawn, see Figure 7.25.
Capacitors marked are AC short circuit.
, and
are chosen to establish correct quiescent
condition. Note that the transistor operates in CB mode. If the oscillator is to operate at high
frequency (>100 kHz), then care must be taken with layout and choice of components to ensure that
stray impedances are not significant.
The AC equivalent circuit has been drawn in Figure 7.26.
For application of the Barkhausen criteria it is convenient to repartition with
as part of the
filter circuit. Analysis is simplified if
(more correctly
), write
and
. These changes are drawn in Figure 7.27.
Figure 7.25. Circuit diagram for a Collpitts oscillator.
Figure 7.26. AC equivalent circuit for a Collpitts oscillator.
E L E C 3 1 7 P a g e | 15
Provided that
then
.
Approximations are introduces in order to simplify analytical solutions. In practice, a computer may
be used to find exact solutions (to accuracy of data).
The circuit including all the simplification is shown in Figure 7.28.
Where
. If the inductor has a series resistance
, then its
equivalent parallel resistance will be
which also contributes to
.
Since has zero phase shift, we require to have zero phase shift too:
Clearly, the phase shift is zero when , i.e. when
, which will be
the frequency of oscillation. At
. For we have
and when
. A suitable design will satisfy both equations and
have
.
Generally, there are many combinations of components that will satisfy these equations. Design is
therefore by trial and error until a set of values that are technically satisfactory has been found.
As an example we will design a Colpitts oscillator for 20 MHz operation.
Assuming availability of a transistor with high
, we might begin with
mA, so that
must satisfy
, choose
, so that
.
It is possible to arrange for
to be dominated by
Figure 7.27. Repartitioning equivalents.
Figure 7.28. Collpits oscillator circuit diagram.
E L E C 3 1 7 P a g e | 16
Then
To ensure the oscillations we should make e.g. 3. Then
, however,
(because
) therefore
. is then
, and
for
we must have
.
Note that
. This is so low a value that
could be made quite small,
e.g. 1-5 without affecting the above calculations significantly.
7.4.4. Common base Hartley oscillator
The circuit for the common base Hartley oscillator is shown in Figure 7.29.
The oscillation frequency
and
are chosen to establish oscillation. From the
design point of view, there is the problem of discovering the degree of coupling (mutual inductance)
between the two sets of windings: might be approximated as two independent inductors when is
small, or as an autotransformer, when .
Figure 7.29. Circuit diagram for a Hartley oscillator.
E L E C 3 1 7 P a g e | 17
7.4.5. Clapp (Gouriet) oscillator
The circuit for a Clapp (Gouriet) oscillator is shown in Figure 7.30.
A disadvantage of the Colpitts oscillator is the fact that
and
must establish both the frequency of
oscillation and the condition for oscillation. If a third capacitor is added in series with , then
considerably more flexibility is available.
and
can be chosen to be much larger than in a
Colpitts design and the smaller capacitor
can be chosen, with , to set the frequency of oscillation.
Writing
we have
. If
, then
.
7.5. The negative resistance design method
Consider the circuit shown in Figure 7.31.
The resistance
, where
is positive, and
is negative. The current is defined by
Choose , a unit impulse (a small noise impulse in practice). Then and
with
and
.
Figure 7.30. Circuit diagram for a Clapp (Gouriet) oscillator.
Figure 7.31. Negative resistance.
E L E C 3 1 7 P a g e | 18
If
then and , the current, , in that case is schematically depicted in Figure
7.32.
If
then and , is a perpetual sinewave - an oscillator, see also Figure 7.33.
If
then and , is an ever increasing sinewave, see Figure 7.34, until
nonlinearities prevents further growth.
Figure 7.32. when and .
Figure 7.33. when and .
Figure 7.34. when and .
E L E C 3 1 7 P a g e | 19
7.5.1. Negative impedance circuit
Consider the AC circuit and its equivalent (Not including unimportant components) as shown in
Figure 7.35.
Colpitts, Clapp, Pierce and other circuits have this general form. Hartley circuit can be analysed
similarly with the capacitors replaced by inductors. We choose
to be very large in the Clapp
circuit, but this condition is not normally a problem even with Colpitts and Pierce circuits. This
assumption is introduced only to simplify the analysis. Writing
and assuming
the equivalent circuit may be drawn, see Figure 7.36.
What is the impedance
Clearly
and
, so that
Or
Thus
The impegance
can be found from
or
The total impedance is given by
Figure 7.35. An AC circuit (left) and its equivalent (right).
Figure 7.36. Redrawn equivalent circuit.
E L E C 3 1 7 P a g e | 20
7.5.2. Negative impedance including
If we include
we obtain the circuit as given in Figure 7.37.
, but what is
? Please note that
includes
Then
Or
Hence
i.e. it is the capacitor
in series with
. Writing this as an admittance gives
This comprises of two components in parallel, the first is a capacitor
and the
second is a negative resistance
as shown in Figure 7.38.
If we want to be able to neglect
we require
i.e.
Figure 7.37. Redrawn equivalent circuit
including
.
Figure 7.38.
E L E C 3 1 7 P a g e | 21
Or
Thus
can be chosen, so that this condition is satisfied. Then, the total
impedance is a negative resistance in parallel with capacitors (which is the usual approximation) as
shown schematically in Figure 7.39.
Using this equivalence the Clapp oscillator may be drawn as shown in Figure 7.40.
If
is the resistance of , see also section 1.4.2,
can also include the
effects of
etc.}), then the circuit can support an oscillation at
) indefinitely.
can be chosen to maintain oscillations.
In practice the design will set
for small amplitude of oscillation (this causes net
resistance to be negative for small amplitudes). An oscillation then builds up until some other
mechanism, such as saturation, reduces the value of
and causes the equality
to
be achieved.
7.6. Frequency stability
In LC oscillators the values of and can change with temperature and age, and hence the frequency
of oscillation will also change. It is common practice to use combinations of positive and negative
temperature coefficients to minimise frequency variations, but there are limitations to what can be
achieved in this way. An alternative is to use quartz crystals in place of LC resonant circuits, see
Figure 7.41.
Figure 7.39.
Figure 7.40. Equivalent Clapp oscillator.
E L E C 3 1 7 P a g e | 22
By virtue of the piezoelectric effect, a thin slice of quartz crystal with opposite surfaces metallised and
bonded to electrical connections behaves like a high resonant circuit. The orientation of the crystal
planes can be chosen to minimise the temperature coefficient to a level that is orders of magnitude
better than is achievable with LC circuits.
Components can be manufactured for specific frequency of operation with very high precision.
Crystals are 'oven' temperature controlled for the highest quality of performance.
The electrical equivalent circuit of the crystal is given in Figure 7.42.
Typical values for a 1 MHz crystal are
and .
Crystals exhibit both serial resonance and parallel resonance, which is also known as antiresonance,
see Figure 7.43.
Figure 7.41. Quartz crystal.
Figure 7.42. Electrical
equivalent circuit.
E L E C 3 1 7 P a g e | 23
The equivalent circuit for the antiresonance is shown in Figure 7.44.
At the antiresonance we have
Doing series to parallel transformation we obtain
The circuit impedance is real at the resonance
or
I.e.
Figure 7.43. Serial and anti-resonance.
Figure 7.44. Equivalent circuit for anti-
resonance.
E L E C 3 1 7 P a g e | 24
whence
with
since
if
and
then
i.e.
and
are quite close in practice.
If the crystal is in parallel with an external capacitor
then the antiresonant frequency becomes
In practice for antiresonant operation of crystals, manufacturers specify the value of
that will make
equal to the design frequency.
may be varied to adjust the value of
. Similarly, a series
capacitor may be used to vary
.
7.6.1. Overtones
All mechanical structures exhibit multiple resonances (e.g. the string of a musical instrument).
Crystals behave in the same way. A more accurate equivalent circuit is therefore given in Figure 7.45.
Overtone operation is possible at the 3rd, 5th, etc. overtones in series mode only. Since it is difficult
to manufacture very thin crystals for high fundamental frequencies, they are operated at the 3rd
overtone above 20 MHz, and at the 5th overtone above 50 MHz.
Figure 7.45. Equivalent circuit of
harmonics.
E L E C 3 1 7 P a g e | 25
7.6.2. Crystal oscillator
There are many possible configurations, e.g. and
of the Clapp oscillator may be replaced by a
quartz crystal. The circuit is shown in Figure 7.46.
using results of section 7.5.
, and
are resistance, inductance and capacitance of the crystal respectively (the effect of
parallel capacitance is included in
). See also Figure 7.47.
is the equivalent of
in series with
since
. Note the use of a slight extension of the series to parallel transformation, see Figure
7.48.
Figure 7.46. Circuit diagram for a Clapp (Gouriet) oscillator
including a quartz oscillator.
Figure 7.47.
E L E C 3 1 7 P a g e | 26
Clearly, the circuit will oscillate if
. The frequency of oscillation will be
governed by the crystal if
The same circuit with emitter grounded is known as a Pierce oscillator, see Figure 7.49.
7.7. Schmitt trigger (relaxation) oscillator
A Schmitt trigger has a transfer characteristic as shown in Figure 7.50.
Figure 7.48. Series to parallel transformation.
Figure 7.49. Pierce oscillator.
Figure 7.50. Schmitt trigger transfer
characteristics.
E L E C 3 1 7 P a g e | 27
may be an analogue signal but, ideally,
has two values. Transitions of output level should be
instantaneous but 'stray' impedances limit transition times. An essential feature of the transfer
characteristic is that it displays hysteresis. For
there are two possible output levels
- the output value is determined by the previous history.
Schmitt trigger circuits are available as gates in all logic families, or may be constructed using
operational amplifiers or discrete components, e.g. 74 LS 14 - hex inerter, or 74 LS 132 - quad two
input nand. Figure 7.51 shows the schematic representation of a Schmitt trigger.
The typical transfer characteristic is given in Figure 7.52.
7.7.1. Schmitt trigger based on an operational amplifier
If an operational amplifier is used with a positive feedback (see Figure 7.53) it will become a Schmitt
trigger.
Figure 7.51. Schematic Schmitt trigger.
Figure 7.52. Typical transfer characteristic.
Figure 7.53. Operational amplifier with
positive feedback.
E L E C 3 1 7 P a g e | 28
The open and closed loop characteristics of this circuit are shown in Figure 7.54.
and
are type dependent. Typical values are:
and
.
Given
and
, what are
?
Let us consider the circuit operating at a threshold condition.
When
and
, see Figure 7.55.
Clearly
Or
e.g.
, then
.
When
and
, see Figure 7.56.
Again
Figure 7.54. The open and closed loop characteristics.
Figure 7.55.
Figure 7.56.
E L E C 3 1 7 P a g e | 29
Or
e.g.
, then
.
7.7.2. Discrete component Schmitt trigger
This circuit is based on emitter coupled pair (differential amplifier), see the circuit given in Figure
7.57.
and
are switching speed-up capacitors. The transistors are in the opposite states: either
is on
and
is off, or
- off and
- on.
7.7.3. Use of Schmitt trigger in oscillators
An oscillator can be constructed using an inverting Schmitt trigger, see Figure 7.58.
Figure 7.57. Discrete component Schmitt trigger.
Figure 7.58. Inverting Operational
amplifier.
E L E C 3 1 7 P a g e | 30
The characteristics are shown in Figure 7.59.
determine the threshold voltages
A combination of the inverting Schmitt trigger and a integrating circuit makes an oscillator, such a
circuit is shown in Figure 7.60.
After each transition of
, the capacitor will charge exponentially toward
or
with time
constant , see Figure 7.61.
Knowing and the initial and final conditions, we can write down an equation for the voltage across
the capacitor without analysis
Figure 7.59. Transfer characteristics.
Figure 7.60. Inverting Schmitt trigger in
combination with a RC integrating circuit.
Figure 7.61. The charging of C.
E L E C 3 1 7 P a g e | 31
The output waveform is shown in Figure 7.62.
At
we obtain
and at
Hence
can be found using
Or
With some algebra we obtain
In the same way
At
and at
noting that for
Or
we obtain
The period of oscillations is then given by
Figure 7.62. The output waveform.
E L E C 3 1 7 P a g e | 32
If
the analysis can be simplified. In this case the threshold voltages
and is
charged and discharged, in effect, by constant currents
Now,
with
. The voltage rate can be calculated using the definition
of capacitance
Thus
Therefore
And
The period of oscillations is
An identical result can be obtained from the previously derived equation for the period assuming that
the threshold voltages
and using a formula .
Many ICs use the Schmitt oscillator principle of operation, e.g. 555 timer and 566 voltage controlled
oscillator.
The oscillators of the 56x range of ICs (phase locked loops) use constant current sources to charge
and discharge the timing capacitor, see Figure 7.63.
If
then S connects to
.
If
then S connects to
Figure 7.63. 56x range IC oscillator.
E L E C 3 1 7 P a g e | 33
The signal diagram is given in Figure 7.64.
is proportional to an externally applied voltage
therefore
we have a voltage
controlled oscillator (VCO).
7.8. Questions
Q7.1. Show that is maximum at
for the Wien feedback circuit.
Q7.2. Express:
as a resistor in parallel with a capacitor.
Figure 7.64. The signal diagram.
Figure Q7.3.
E L E C 3 1 7 P a g e | 34
Q7.3. Consider the following circuit
1. Determine the circuit resonance frequency.
2. Determine the voltage gain of the CB amplifier A.
3. Calculate the open loop gain A and can this circuit sustain an oscillation?
Figure Q7.8.
