5

Oscillators and phase locked loops

The generation of a stable sinusoidal signal is a crucial function in most RF systems.
A transmitter will amplify and suitably modulate such a signal in order to produce its
required output. In the case of a receiver system, such a signal is fed into the mixer
circuits for the purposes of frequency conversion and demodulation. A circuit that gen-
erates a repetitive waveform is known as an oscillator. Such circuits usually consist of
an amplier with positive feedback that causes any input, however small, to grow until
limited by the non-linearities of the circuit. The feedback will need to be frequency
selective in order to control the rate of waveform repetition. This frequency selec-
tion is often achieved using combinations of capacitors and inductors, but can also
be achieved with resistor and capacitor combinations. In the present chapter, how-
ever, we will concentrate on feedback circuits based on capacitor/inductor combina-
tions. We consider a variety of oscillator circuits that are suitable for RF purposes and
investigate the conditions under which oscillation occurs. In addition, we consider
the issue of oscillator noise since this can often pose a severe limitation upon system
performance.
A particularly important class of oscillator is that for which the frequency can be
controlled by a d.c. voltage. Such an oscillator is an important element in what is known
as a phase locked loop. In such a system, there is a feedback loop that compares the
oscillator output with a reference signal and generates a control voltage based upon
their phase difference. When the system settles down, the oscillator is locked onto the
reference signal. Phase locked loops form a generic class of system that can been used
for purposes such as frequency control and demodulation. Here we consider the basic
principles of phase locked loops and investigate some of their applications.
5.1 Feedback
A general amplier (H) with feedback (G) is illustrated in Figure 5.1. For this system,
the relationship between input and output voltages is given by
v
o
=
H(j)
1 G(j)H(j)
v
i
. (5.1)
108
109 5.2 The Colpitts oscillator
H
G

v
o
v
i

+
Figure 5.1 General feedback system.
Negative feedback (the positive sign in Equation 5.1) provides a means of tailoring
the amplier frequency response and controlling both gain and linearity. When there is
positive feedback, however, the system can provide a means of generating oscillations.
For oscillation, we require a system that produces a signal without input (except for an
initial excitation). Consequently, for oscillations to occur at frequency
0
,
|G(j
0
)||H(j
0
)| = 1 (5.2)
and
arg{G(j
0
)H(j
0
)} = 0. (5.3)
This is the Barkhausen criterion and it ensures that any small component at frequency

0
will grow until limited by the non-linearities of the system.
5.2 The Colpitts oscillator
In a Colpitts oscillator, the feedback occurs via a series inductance -network. An
example, basedona common-emitter BJTamplier, is showninthe circuit of Figure 5.2.
By neglecting all but the current source of the BJT (r

= r
o
= and C

= C

= 0),
we obtain the simplied small signal model of Figure 5.3. Current balance at the BJT
collector will imply
sC
2
v

+ g
m
v

+
_
1
R
+sC
1
_
v
o
= 0 (5.4)
and, at the base,
sC
2
v

=
v
o
v

sL
. (5.5)
FromEquation 5.5 we obtain v
o
= v

(1 +s
2
C
2
L) and then, eliminating v
o
fromEqua-
tion 5.4,
sC
2
+ g
m
+
_
1
R
+sC
1
_
(1 +s
2
C
2
L) = 0. (5.6)
110 Oscillators and phase locked loops
C
BL
R
L
C
1
C
2
V
CC
RFC
bias
Figure 5.2 Colpitts oscillator.
c b
m
C
2
v
R
C
1
v v
o
g
Figure 5.3 Simple model of Colpitts oscillator.
Noting that s = j, and separating the real and imaginary parts of Equation 5.6, we
obtain
=
_
C
1
+C
2
LC
1
C
2
and
C
2
C
1
= g
m
R, (5.7)
where is the frequency of oscillation, C
2
/C
1
is the feedback ratio and g
m
R is the volt-
age gain of the amplier. Apractical design will normally set the transistor gain slightly
higher than the feedback ratio (C
2
/C
1
< g
m
R) to take account of component variations.
In this case, the oscillations will grow until the non-linearities in the device cause suf-
cient loss of gain for the Barkhausen criterion to be satised. This last point is important
as it means that the steady state operation of an oscillator is essentially non-linear. If v

is large, the transistor will make excursions into regions where it is switched off (we usu-
ally set the bias so that this is the case). Consequently, the transistor collector current I
t
will consist of periodic pulses with the peaks occurring where v

is maximum. The bias

current I
bias
will be the average of the transistor current pulses (I
bias
= (1/T)
_
T
0
I
t
dt ,
where T = 2/) and, by Fourier techniques, the fundamental component of the cur-
rent I
fund
is given by I
fund
= (2/T)
_
T
0
I
t
cos(t ) dt . The major contribution to I
fund
will
arise around the peak of v

and, as a consequence, I
fund
2I
bias
. If v

= V
osc
cos(t )
111 5.2 The Colpitts oscillator
D
G
S
R
1
R
S
R
2
C
2
L
C
BL
R
D
V
DD
C
BL
C
BP
C
1
Figure 5.4 FET Colpitts with bias circuits.
C
2
C
1
L
Figure 5.5 General Colpitts oscillator.
(feedback only occurs at the fundamental frequency) there will be an effective large sig-
nal transconductance of G
m
= 2I
bias
/V
osc
(see Lee). This is a general relationship, but
for a BJT it reduces to G
m
= (2V
T
/V
osc
)g
m
and for an FET to G
m
= [(V
GS

V
t
)/V
osc
]g
m
, where V
GS
is the d.c. component of the gatesource voltage. In designing
an oscillator, we will need to ensure that the Barkhaussen criterion can be satised
somewhere between the extremes of the large and small signal transconductances.
An FETversion of the Colpitts oscillator is shown in Figure 5.4. If we replace the FET
amplier by a generic unit, we obtain the generic Colpitts oscillator shown in Figure 5.5.
112 Oscillators and phase locked loops
C
3
C
1
L
C
C
2
L
1
L
2
Figure 5.6 Clapp and Hartley feedback circuits.
V
DD
bias
Figure 5.7 FET differential oscillator.
(Note that the feedback requires the amplier to have a phase shift of 180

, which
is the case for common-source and common-emitter ampliers.) The Colpitts circuit
employs a series inductance -network feedback, but there are alternative feedback
circuits that give rise to the Clapp and Hartley oscillators (see Figure 5.6). Whilst
oscillators based on a single-ended amplier input are common, it is also possible to
base an oscillator on a differential amplier. Figure 5.7 shows a design that is suitable
for CMOS implementation. Positive feedback is achieved by using both differential
input and output, the output of one side feeding the input of the other.
In practical applications, the oscillator will need to act as a source of RF signals.
Consequently, this will mean an additional load on the circuit. The signal is normally
taken from the output of the amplier on which the oscillator is based (the collector in
113 5.2 The Colpitts oscillator
Figure 5.2 and the drain in Figure 5.4). Unfortunately, the additional load can have an
adverse affect upon oscillator performance and so it is advisable to make this loading
as light as possible (a high impedance load). To this end, the additional load is normally
connected through a high input impedance buffer amplier (emitter or source follower
circuits are often used).
Example Design a JFET Colpitts oscillator for operation at 30 MHz. Assume a supply
voltage of 6 V and FET parameters V
t
= 3 V, K = 10
3
A/V
2
, C
GS
= 4 pF, C
GD
=
1.6 pF, C
DS
= 0.1 pF and r
d
= 30 k.
The design will be that shown in Figure 5.4. Since the circuit employs a JFET, we
can use the option of self bias and remove resistor R
1
. We will take R
2
= 100 ksince
this is well below the d.c. input resistance of a typical JFET and hence will suitably
ground the gate. The small signal model of a JFET is essentially the same as that for a
BJT, except that r

can be neglected. Consequently,

=
_
C
1
+C
2
LC
1
C
2
and
C
2
C
1
= g
m
R
D
(5.8)
for oscillations to occur. To reduce the effect of transistor capacitance, C
1
and C
2
should have values much greater than the FET output capacitance (0.1 pF) and input
capacitance (4 pF), respectively. We choose bias conditions such that the d.c. component
of gatesource voltage V
GS
is 2V. Suchclose proximityto V
t
will ensure that operation
is pushed well into the non-linear regime. From the saturation region characteristic
equations, this will require a drain current of 1 mA and hence a source resistor R
S
of
2 k. The value of transconductance is calculated from g
m
= 2K(V
GS
V
t
) and, for
the above bias conditions, will have the value 2 10
3
S. Resistor R
D
is given the
value 1 k to satisfy the usual design rule that the quiescent V
DS
be approximately
V
DD
/2. For oscillations to start, we will require that C
2
/C
1
< g
m
R
D
which implies
that C
2
/C
1
< 2. To guard against component variations, it is advisable to make this
constraint well satised and so we take C
1
= C
2
. Furthermore, to satisfy our original
constraints on C
1
and C
2
, we choose C
1
= C
2
= 50 pF. For 30 MHz oscillations, the
conditions in Equation 5.8 will imply a value of 1.125 H for inductor L. This value,
however, will need to be slightly reduced due to the effect of the parasitic reactances
within the transistor. As the oscillations grow, the large signal transconductance G
m
will
become appropriate and we will have C
2
/C
1
= G
m
R
D
when steady state is achieved.
Since G
m
= [(V
GS
V
t
)/V
osc
]g
m
at this point, the amplitude of oscillations at the gate
will be around 2 V. This value, however, will tend to be an overestimate since the
drain current will deviate from the ideal pulse behaviour that is assumed by the theory
(Figure 5.8 illustrates the drain current behaviour for this oscillator).
Another variant of the Colpitts oscillator, shown in Figure 5.9, is useful when the
amplier has less than unity voltage gain. (Note that the phase shift of the feedback
is zero, consistent with the phase shifts of source and emitter follower ampliers.)
114 Oscillators and phase locked loops
Time
D
r
a
i
n

c
u
r
r
e
n
t
Figure 5.8 The drain current for a Colpitts oscillator.
L
C
2
C
1
Figure 5.9 Alternative Colpitts oscillator.
Consider an oscillator based on the n-channel JFET source follower of Figure 5.10.
Capacitors C
1
and C
2
of the feedback circuit need to be chosen so that the total loop
gain is greater than 1 (choosing C
1
= C
2
is usually sufcient). As the amplitude of
oscillation rises, the positive swings will eventually be clipped by the action of the
diode (at a level of about 0.7 V for a silicon diode). This results in a d.c. voltage that
pushes the transistor gatesource voltage towards V
t
. As a consequence, there will be
a reduction in gain that continues until equilibrium is reached (i.e., the Barkhausen
criterion is satised).
We have noted that C
1
and C
2
need to be chosen to give adequate loop gain, but
other considerations can affect their choice. Firstly, their values should not be so small
115 5.2 The Colpitts oscillator
output
input
C
BL
RFC
R
G
R
D
V
DD
C
BP
Figure 5.10 Source follower amplier.
g
m
v

L C
2
v

v
C
1
R
i
2
i
1
i
3
Figure 5.11 Simplied Colpitts oscillator model.
that the internal capacitance of the transistor becomes a signicant factor and hence
introduces a susceptibility to device variations. Secondly, they should not be so large
as to prevent oscillation. The idealised model of Figure 5.11 (it ignores the diode and
internal parasitics of the FET) helps to explain the last point (note that R represents
the resistance of the inductor). The currents in the circuit are related to the voltages
through
i
1
= v

jC
2
(5.9)
i
2
=
v +v

R +jL
, (5.10)
i
3
= vjC
1
. (5.11)
116 Oscillators and phase locked loops
We must have i
3
= i
1
+ g
m
v

, from which
v

(jC
2
+ g
m
) = vjC
1
(5.12)
and i
1
+i
2
= 0, from which
v

jC
2
+
v +v

R +jL
= 0. (5.13)
If we eliminate v between Equations (5.12) and (5.13), we obtain
jC
2
+ g
m
+jC
1
[jC
2
(R +jL) +1] = 0 (5.14)
and, on taking imaginary and real parts,
C
2

3
C
1
C
2
L +C
1
= 0 (5.15)
and
g
m

2
C
1
C
2
R = 0. (5.16)
Equation 5.15 implies that the oscillations will occur at frequency =

(C
1
+C
2
)/C
1
C
2
L and, from Equation 5.16, it is clear that too large a value for C
1
C
2
will prevent the oscillator from nding equilibrium. We need to have C
1
C
2
g
m
/
2
R
for oscillation to occur.
Figure 5.12 shows an alternative source follower Colpitts oscillator in which a source
resistor is used to produce gain compression. As the level of oscillation rises, the d.c.
component of the source current will also rise and force V
GS
towards V
t
. As a conse-
quence, the gain will reduce until a point is reached where the Barkhausen criterion is
satised.
RF output
L
C
2
V
DD
C
1
Figure 5.12 An FET Colpitts oscillator that allows output to be taken from the drain.
117 5.3 Stability and phase noise in oscillators
Note that, since the drain resistor has no bypass, an RF voltage will develop at the
drain and output can be taken from this point with very little effect upon the quality
of oscillation. Because of its pulse nature, the drain current will be rich in harmonics
and, if the drain resistor is replaced by a suitably tuned circuit, it is possible to extract
power at a harmonic frequency.
5.3 Stability and phase noise in oscillators
Consider the positive feedback system of Figure 5.13. The amplier (assumed to be
ideal) has voltage gain A and the feedback is provided by a series combination of
inductance, capacitance and resistance. The relationship between source and output
voltages (assuming A 1) will be given by
v
o
=
v
i
A
1
R
i
A
R
i
+R+j
_
L
1
C
_
. (5.17)
We can rearrange this into the form
v
o
=
v
i
A
1
1
1+jQ
_

_
AR
i
R
i
+R
, (5.18)
where
o
= 1/

LC is the resonant frequency of the feedback loop and

Q =

o
L
R
i
+ R
(5.19)
is its quality factor. What is clear from Equation 5.18 is that a disturbance in the phase
transfer characteristics of the amplier (i.e., in the characteristics of A) will require an
adjustment in oscillation frequency if the Barkhausen criterion is to remain satised.
R
i
A
R
L C
v
i
v
o
Figure 5.13 Oscillator with series LCR feedback.
118 Oscillators and phase locked loops

A
m
p
l
i
t
u
d
e

0
D
Figure 5.14 Phase noise.
It will be noted, however, that the requisite adjustment in frequency will reduce as the
value of Q increases. Consequently, high Q circuits are clearly the key to frequency
stable oscillators.
Close to the oscillation frequency
o
,
v
o

v
i
A
o
2jQ(
o
)
, (5.20)
where A = (R
i
+ R)/R
i
for the Barkhausen criterion to be satised. It is clear from
this that the quality of oscillations will be strongly affected by disturbances containing
frequencies close to resonance. In particular, noise in the oscillator amplier can cause
a spread of output frequencies around that which is desired (see Figure 5.14). The
amplier noise can be expressed as a voltage source v
ni
at the input of the amplier
with mean square value satisfying v
2
ni
= kT R
i
F for a bandwidth of 1 Hz (F is the
amplier noise factor). The corresponding output noise voltage v
no
will satisfy
v
2
no
=
A
2

2
o
kT R
i
F
4Q
2
(
o
)
2
(5.21)
which indicates that the output noise density will fall away as (
o
)
2
. Very close to
resonance, however, the above behaviour will be moderated by gain compression and
far fromresonance there will be a oor that is set by the amplier noise itself. Due to the
effect of gain compression, the main contribution to noise will arise from uctuations
in phase and, as a consequence, oscillator noise is often referred to as phase noise. It is
clear from the above considerations that phase noise can be reduced by using high Q
components in the oscillator circuit.
The phase noise performance of an oscillator is normally described in terms of the
relative phase noise density, dened (in terms of dB/Hz) by
L = 10 log
_
v
2
no
v
2
sig
_
, (5.22)
119 5.3 Stability and phase noise in oscillators
IF
Frequency
A
m
p
l
i
t
u
d
e
RF1
RF2 LO
Figure 5.15 Illustration of reciprocal mixing.
where v
sig
is the noise free signal level. Phase noise imposes an important limitation
of oscillator performance and oscillator specication will normally include values of
relative phase noise at various frequency offsets from the intended frequency. In partic-
ular, for a receiver that down converts the input RF signal to an intermediate frequency
(IF), the phase noise performance of the receiver local oscillator (LO) can be important
because of the possibility of reciprocal mixing. Although the desired signal will mix
with the intended LO frequency to produce the IF frequency, there is the possibility
that energy fromthe local oscillator frequency skirts could mix with strong out-of-band
signals to also produce the IF frequency. This is known as reciprocal mixing and will
lead to interference that could be unacceptable in some applications. The concept is
illustrated in Figure 5.15 where RF1 represents the desired signal and RF2 the strong
out of band signal.
It is clear that we require high Q resonant circuits for good oscillator performance.
Extremely high Q resonators can be constructed out of quartz crystals and these are
used extensively in RF circuits. Such resonators are electromechanical in nature and use
the piezoelectric effect to translate high quality mechanical vibrations into electrical
oscillations. (Other devices withhigh Qare coaxial andceramic resonators.) Figure 5.16
shows a typical circuit model for a quartz crystal (valid near the fundamental frequency
of resonance). The inductance is extremely large (hundreds of henries) and the shunt
capacitance C
2
is typically tens of picofarads (the series capacitance C
1
is very much
less). From the circuit model, we obtain the following expression for the impedance of
the crystal
Z =
(1
2
LC
1
) +jRC
1
j(C
1
+C
2
)
_
1

2
LC
1
C
2
C
1
+C
2
_

2
RC
1
C
2
. (5.23)
120 Oscillators and phase locked loops
crystal C
2
C
1
L
R
series
resonance
parallel
resonance
Z

Figure 5.16 Crystal resonator model.

C
2
C
1
R
Figure 5.17 Crystal Colpitts oscillator.
It is clear that the device will exhibit both series and parallel resonance (note that
R can be neglected due to the very high Q). The frequencies of these resonances will,
however, be very close. Figure 5.17 shows a typical example of a crystal controlled
Colpitts oscillator (bias components and d.c. supply not shown).
5.4 Voltage controlled oscillators
An oscillator with a voltage controlled frequency is often required in applications such
as phase locked loops. A varicap (variable capacitance) diode (sometimes known as a
varactor) can be used to achieve this. Diodes can be manufactured such that the reverse
bias junction capacitance changes quite dramatically with voltage
C(V
bias
) =
C
0
_
1
V
bias
V
diff
_1
2
, (5.24)
121 5.5 Negative resistance approach to oscillators
C
BP
V
DD
C
BP
output
tuning
bias
RFC
RFC
Figure 5.18 A Colpitts VCO.
cathode anode
n n n
Figure 5.19 Gunn diode.
where C
0
is the zero bias capacitance and V
diff
has a value of about 0.6 V for a silicon
diode. Figure 5.18 shows a Colpitts voltage controlled oscillator (VCO) that is based
on such devices (note the addition of a source resistor to provide gain compression).
A variable frequency crystal oscillator (VXO) can be constructed by replacing the
inductance with a quartz crystal, but the achievable frequency variation is often very
small.
5.5 Negative resistance approach to oscillators
When a tuned circuit is excited by a pulse, it will ring at the resonant frequency. The
circuit resistance will, however, cause a rapid damping of these oscillations. This can be
overcome by introducing a device that has negative resistance in order to cancel out the
circuit resistance. An example of such a device is the Gunn diode shown in Figure 5.19.
A Gunn diode exhibits higher energy states for which the current carriers have lower
mobility and this will cause negative resistance under suitable bias conditions (see
Figure 5.20). The negative resistance can be used to cancel out the damping resistance
of a tuned circuit and hence create an oscillator (see Figure 5.20 and Collin).
We can also generate negative resistance using an FET(or a BJT) and this can provide
an alternative way of analysing oscillators. The circuit shown in Figure 5.21 is capable
122 Oscillators and phase locked loops
Gunn
diode
C R
L
V
time
low mobility states
high mobility states
I
V
Figure 5.20 Oscillator based on the Gunn diode.
m
Z
i
v
i
C
1
R
L
V
DD
C
1
v

g
C
2
R
L
i
i
RFC
C
2
i
i
v
i
v

Figure 5.21 Negative resistance circuit based on an FET.

of generating negative resistance (gate bias not shown). This can be analysed through
the model that is also shown in the Figure 5.21 and from which
v
i
=
i
i
jC
1
+
_
i
i
+ g
m
i
i
jC
1
_
1
jC
2
. (5.25)
The input impedance Z
i
will be given by
Z
i
=
v
i
i
i
=
1
jC
1
+
_
1 +
g
m
jC
1
_
1
jC
2
(5.26)
which has a negative real part
R
i
=
g
m
C
1
C
2

2
. (5.27)
123 5.7 Analysis of a phase locked loop
phase
comparator
LFP
VCO
V
D
V
C
V
V
V
R
reference
Figure 5.22 General phase locked loop.
By connecting the above impedance in parallel with an inductor, a Colpitts oscillator
is formed. The magnitude of R
i
will need to be larger than the intrinsic resistance of
the inductor in order for oscillation to occur and this will lead to the same condition as
was derived in Section 5.2.
5.6 Phase locked loops
A phase locked loop (PLL) is a feedback system in which the feedback is based on
phase difference alone. These systems have a large variety of applications including
frequency control and demodulation. Figure 5.22 shows a typical PLL architecture.
The PLL compares the output of a voltage controlled oscillator (VCO) with a reference
signal andproduces a control voltage that is proportional tothe phase difference between
them. This voltage then adjusts the VCOsuch that it moves closer to the reference signal
in terms of phase. When the system settles down, the VCO is basically locked onto
the reference signal. The low-pass lter (LPF) helps remove unwanted high frequency
components that are present at the phase comparator output and the amplier ensures
an adequate level of control voltage. In essence, a PLL produces a less noisy version
of the reference signal, but slightly out of phase (the phase difference can be reduced
by increasing the amplier gain). The low-pass lter characteristics will be dictated by
the application and, in the case of demodulation, will need to exhibit a bandwidth that
is at least that of the baseband signal.
5.7 Analysis of a phase locked loop
If the reference signal v
R
has the form V
R
cos[
0
t +
R
(t )] and the VCO signal v
V
has
the form V
V
cos[
0
t +
V
(t )], the output of the phase detector v
D
will be
v
D
(t ) = k
D
[
R
(t )
V
(t )], (5.28)
where k
D
depends on the nature of the phase detector. We assume that the phase detector
is linear and that
0
is the free running frequency of the VCO (the frequency for which
v
C
= 0). After passage throughthe lter (transfer function H) andthe amplier (gain A),
ACTIVE FILTERS AND OSCILLATORS
300 Chapter 5
1,
Figure 5.46. Parasitic oscillation example.
capacitance of the transistor and the meter
capacitance resonated with the meter in-
ductance in a classic Hartley oscillator cir-
cuit, with feedback provided by collector-
emitter capacitance. Adding a small base
resistor suppressed the oscillation by re-
ducing the high-frequency common-base
gain. This is one trick that often helps.
5.19 Quartz-crystal oscillators
RC oscillators can easily attain stabilities
approaching O.l%, with initial predictabil-
ity of 5% to 10%. That's good enough for
many applications, such as the multiplexed
display in a pocket calculator, in which a
multidigit numerical display is driven by
lighting one digit after another in rapid
succession (a lkHz rate is typical). Only
one digit is lit at any time, but your eye
sees the whole display. In such an appli-
cation the precise rate is quite irrelevant
- you just want something in the ballpark.
As stable sources of frequency, LC oscil-
lators can do a bit better, with stabilities
of O.OlO/o over reasonable periods of time.
That's good enough for oscillators in radio-
frequency receivers and television sets.
For real stability there's no substitute
for a crystal oscillator. This uses a piece
of quartz (same chemical as glass, silicon
dioxide) that is cut and polished to vibrate
at a certain frequency. Quartz is piezo-
electric (a strain generates a voltage, and
vice versa), so acoustic waves in the crys-
tal can be driven by an applied electric
field and in turn can generate a voltage
at the surface of the crystal. By plating
some contacts on the surface, you wind
up with an honest circuit element that can
be modeled by an RLC circuit, pretuned
to some frequency. In fact, its equiva-
lent circuit contains two capacitors, giving
a pair of closely spaced (within 1%) se-
ries and parallel resonant frequencies (Fig.
5.47). The effect is to produce a rapidly
changing reactance with frequency (Fig.
5.48). The quartz crystal's high & (typ-
ically around 10,000) and good stability
make it a natural for oscillator control,
as well as for high-performance filters (see
Section 13.12). As with LC oscillators, the
crystal's equivalent circuit provides posi-
tive feedback and gain at the resonant fre-
quency, leading to sustained oscillations.
I
Figure 5.47
Figure 5.49 shows some crystal oscilla-
tor circuits. In A the classic Pierce oscilla-
tor is shown, using the versatile FET (see
Chapter 3). The Colpitts oscillator, with
a crystal instead of an LC, is shown in B.
An npn bipolar transistor with the crystal
OSCILLATORS
5.19 Quartz-crystal oscillators 301
as feedback element is used in C. The re-
maining circuits generate logic-level out-
puts using digital logic functions (D and
El.
capacitive I
B
Figure 5.48
The last diagram uses the convenient
MC 120601 1206 1 series of crystal oscillator
circuits from Motorola. These chips are in-
tended for crystals in the range lOOkHz to
20MHz and are designed to give excellent
frequency stability by carefully limiting the
amplitude of oscillation via internal ampli-
tude discrimination and limiting circuitry.
They provide sine-wave and square-wave
outputs (both "TTL" and "ECL" logic lev-
els).
An even more convenient alternative,
if you're willing to accept a square wave
output only, and if utmost stability isn't
needed, is the use of complete crystal os-
cillator modules, usually provided as DIP
IC-sized metal packages. They come in
lots of standard frequencies (e.g., 1, 2, 4,
5, 6 , 8, 10, 16, and 20MHz), as well as
weird frequencies commonly used in mi-
croprocessor systems (e.g., 14.3 18 18MHz,
used for video boards). These "crystal
clock modules" typically provide accura-
cies (over temperature, power supply volt-
age, and time) of only 0.01% (lOOppm),
but you get it cheap ($2 to $ 9 , and you
don't have to wire up any circuitry. Fur-
thermore, they are guaranteed to oscillate,
which isn't by any means assured when
you wire your own oscillator: Crystal os-
cillator circuits depend on electrical prop-
erties of the crystal (such as series versus
parallel mode, effective series resistance,
and mount capacitance) that aren't always
well specified. All too often you may find
that your home-built crystal oscillator os-
cillates, but at a frequency unrelated to
that stamped on the crystal! Our own ex-
perience with discrete crystal oscillator cir-
cuits has been, well, checkered.
Quartz crystals are available from about
lOkHz to about lOMHz, with overtone-
mode crystals going to about 250MHz.
Although crystals have to be ordered for
a given frequency, most of the commonly
used frequencies are available off the shelf.
Frequencies such as 1 OOkHz, 1 .OMHz,
2.0MHz, 4.0MHz7 S.OMHz, and 1O.OMHz
are always easy to get. A 3.579545MHz
crystal (available for less than a dollar) is
used in TV color-burst oscillators. Digital
wristwatches use 32.768kHz (divide by 215
to get lHz), and other powers of 2 are
also common. A crystal oscillator can
be adjusted slightly by varying a series
or parallel capacitor, as shown in Figure
5.49D. Given the low cost of crystals
(typically about 2 to 5 dollars), it is
worth considering a crystal oscillator in
any application where you would have to
strain the capabilities of RC relaxation
oscillators.
If you need a stable frequency with a
very small amount of electrical tunability,
you can use a varactor to "pull" the
frequency of a quartz-crystal oscillator.
The resulting circuit is called a "VCXO"
ACTIVE FILTERS AND OSCILLATORS
302 Chapter 5
2.5rnH
1 OOOpF
output
-
A. Pierce oscillator 0. Colpitts oscillator C
CMOS inverter
Figure 5.49. Various crystal oscillators.
v output
(voltage-controlled crystal oscillator), and
combines the good-to-excellent stability of
crystal oscillators with the tunability of LC
oscillators. The best approach is proba-
bly to buy a commercial VCXO, rather
than attempt to design your own. Typ-
ically they produce maximum deviations
of f lOppm to f IOOppm from center fre-
quency, though wide-deviation units (up to
f 1000ppm) are also available.
Without great care you can obtain fre-
quency stabilities of a few parts per mil-
1 OMR
1 -
oscillator) with somewhat better perfor-
mance. Both TCXOs and uncompensated
oscillators are available as complete mod-
ules from many manufacturers, e.g., Bliley,
CTS Knights, Motorola, Reeves Hoffman,
Statek, and Vectron. They come in various
sizes, ranging down to DIP packages and
TO-5 standard transistor cans. TCXOs
deliver stabilities of lppm over the range
0C to 50C (inexpensive) down to 0. lppm
over the same range (expensive).
- +5v
lion over normal temperature ranges with
crystal oscillators. By using temperature-
Temperature-stabilized oscillators
com
p
ensation schemes yo; can- make a For the utmost in stability, you may need a
TCXO (temperature-compensated crystal crystal oscillator in a constant-temperature
Tri r 1
5
11 16
1 OOk
1 ) 10; 4 )
32,768Hz
-7
sine
1 ?OPF
g outputs 9 1
- -
- MC12060 (100kHz-2MHz) -
-L
-
- MC12061 (2MHz-20MHzl
D E
+ slne 4
2 -
0
-sine
3 - d} re out
10 TTL out
SELF-EXPLANATORY CIRCUITS
5.20 Circuit ideas 303
oven. A crystal with a zero temperature
coefficient at some elevated temperature
(80C to 90C) is used, with the thermo-
stat set to maintain that temperature. Such
oscillators are available as small modules
for inclusion into an instrument or as
complete frequency standards ready for
rack mounting. The 108 1 1 from Hewlett-
Packard is typical of high-performance
modular oscillators, delivering 1 OMHz
with stabilities of a few parts in 10'' over
periods of seconds to hours.
When thermal instabilities have been
reduced to this level, the dominant ef-
fects become crystal "aging" (the frequency
tends to decrease continuously with time),
power-supply variations, and environmen-
tal influences such as shock and vibration
(the latter are the most serious problems
in quartz wristwatch design). To give an
idea of the aging problem, the oscillator
mentioned previously has a specified aging
rate at delivery of 5 parts in 10'' per day,
maximum. Aging effects are due in part
to the gradual relief of strains, and they
tend to settle down after a few months,
particularly in a well-manufactured crystal.
Our specimen of the 108 1 1 oscillator ages
about 1 part in 1011 per day.
Atomic frequency standards are used
where the stability of ovenized-crystal
standards is insufficient. These use a mi-
crowave absorption line in a rubidium gas
cell, or atomic transitions in an atomic ce-
sium beam, as the reference to which a
quartz crystal is stabilized. Accuracy and
stability of a few parts in 1012 can be ob-
tained. Cesium-beam standards are the of-
ficial timekeepers in this country, with tim-
ing transmissions from the National
Bureau of Standards and the Naval Ob-
servatory. Atomic hydrogen masers
have been suggested as the ultimate in
stable clocks, with claimed stabilities ap-
proaching a few parts in loi4. Recent
research in stable clocks has centered on
techniques using "cooled ions" to achieve
even better stability. Many physicists be-
lieve that ultimate stabilities of parts in
1018 may be possible.
SELF-EXPLANATORY CIRCUITS
5.20 Circuit ideas
Figure 5.51 presents a variety of circuit
ideas, mostly taken from manufacturers'
data sheets and applications literature.
ADDITIONAL EXERCISES
1. Design a 6-pole high-pass Bessel filter
with cutoff frequency 1 kHz.
2. Design a 60Hz twin-T notch filter with
op-amp input and output buffers.
3. Design a sawtooth-wave oscillator, to
deliver 1 kHz, by replacing the charging re-
sistor in the 555 oscillator circuit with a
transistor current source. Be sure to pro-
vide enough current-source compliance.
What value should RB (Fig. 5.33) have?
4. Make a triangle-wave oscillator with
a 555. Use a pair of current sources I.
(sourcing) and 210 (sinking). Use the 555's
output to switch the 210 current sink on
and off appropriately. The following figure
shows one possibility.
0 - 555 - -
U-Lr
out put
fl
A
2 + T -
Figure 5.50
C H A P T E R
11
Oscillators
Oscillators are autonomous dc-to-ac converters. They are used as the frequency-
determining elements of transmitters and receivers and as master clocks in
computers, frequency synthesizers, wristwatches, etc. Their function is to
divide time into regular intervals. The invention of mechanical oscillators
(clocks) made it possible to divide time into intervals much smaller than the
Earths rotation period and much more regular than a human pulse rate.
Electronic oscillators are analogs of mechanical clocks.
11.1 Negative feedback (relaxation) oscillators
The earliest clocks used a verge and foliot mechanism which resembled
a torsional pendulum but was not a pendulum at all. These clocks operated as
follows: torque derived from a weight or a wound spring was applied to a
pivoted mass. The mass accelerated according to Torque = I d
2
/dt
2
(the angular
version of F=ma). When reached a threshold,
0
, the mechanism reversed
the torque, causing the mass to accelerate in the opposite direction. When it
reached
0
the torque reversed again, and so on. The period was a function of
the moment of inertia of the mass, the magnitude of the torque, and the
threshold setting. These clocks employed negative feedback; when the con-
trolled variable had gone too far in either direction, the action was reversed.
Most home heating systems are negative feedback oscillators; the temperature
cycles between the turn on and turn off points of the thermostat. Negative
feedback electronic oscillators are called relaxation oscillators. Most of
these circuits operate by charging a capacitor until its voltage reaches an
upper threshold and then discharging it until the voltage reaches a lower
threshold voltage. In Figure 11.1(a), when the voltage on the capacitor builds
up to about 85 V, the neon bulb fires. The capacitor then discharges quickly
through the ionized gas (relaxes) until the voltage decays to about 40 V. The
bulb then extinguishes and the cycle begins anew.
120
The circuit of Figure 11.1(b) alternately charges the capacitor, C, until its
voltage reaches 2.5 V, and then discharges it until the voltage has fallen to
2.5 V. (V
1
(t) decays alternately toward +5 or 5 volts. When it reaches zero
volts, the left-hand op-amp abruptly saturates in the opposite direction, kicking
V
1
(t) to the voltage it had been approaching. The voltage then begins to decay in
the opposite direction, and so forth.) Voltage-to-frequency converters are usu-
ally relaxation oscillators in which the control voltage determines the slope, and
hence the oscillation period, of a fixed-amplitude sawtooth wave. Relaxation
oscillators typically contain waveforms that are ramps or exponential decays. In
the verge and foliot clock, the angle (t) consists of a sequence of parabolic arcs.
Note that relaxation oscillators are nonlinear circuits which switch alternately
between a charge mode and a discharge mode. Positive feedback oscillators, the
main subject of this chapter, are nominally linear circuits. They generate sine
waves.
11.2 Positive feedback oscillators
Clock makers improved frequency stability dramatically by using a true pen-
dulum, a moving mass with a restoring force supplied by a hair spring or
gravity.
1
As first observed by Galileo, a pendulum has its own natural fre-
quency, independent of amplitude. It moves sinusoidally in simple harmonic
motion. A pendulum clock uses positive feedback to push the pendulum in the
direction of its motion, just as one pushes a swing to restore energy lost to
friction.
Figure 11.1. Relaxation
(negative feedback) oscillators.
V
1
(t)
5
+5
1/2 TL082
1/2 TL082
R
(b)
C
V
out
5 V
5
0
f =1/(2 R C In2)
(a)
5 V
5 V
2k
2k

+
+
90 V
90 V
NE-2
Neon bulb
V
0
t
1
The Salisbury Cathedral clock, when installed around 1386, used a verge and foliot
mechanism. Some 300 years later, after Christian Huygens invented the pendulum clock
based on Galileos observations of pendulum behavior, the Salisbury clock was converted
into a positive feedback pendulum clock, its present form.
121 Oscillators
Electronic versions of the pendulum clock are usually based on resonators
such as parallel or series LC circuits or electromechanical resonators such as
quartz crystals. They use positive feedback to maintain the oscillation. A
resonator with some initial energy (inductor current, capacitor charge, or
mechanical kinetic energy) will oscillate sinusoidally with an exponentially
decaying amplitude as shown in Figure 11.2. The decay is due to energy loss in
the load and in the internal loss of the finite-Q resonator. In Figure 11.2 the
resonator is a parallel LCR circuit.
To counteract the exponential decay, a circuit pumps current into the reso-
nator when its voltage is positive and/or pulls current out when its voltage is
negative. Figure 11.3 shows how a transistor and a dc supply can provide this
energy. In this example circuit, the transistor is shown in the emitter-follower
configuration simply because it is so easy to analyze; the emitter voltage tracks
the base voltage and the base draws negligible current. The single transistor
cannot supply negative current but we can set it up with a dc bias as a class-A
amplifier so that current values less than the bias current are equivalent to
negative current.
All that remains to complete this oscillator circuit is to provide the transistor
drive, i.e., the base-to-emitter voltage. We want to increase the transistors
conduction when the output voltage (emitter voltage) increases, and we see
that the emitter voltage has the correct polarity to be the drive signal. Since an
emitter followers voltage gain is slightly less than unity, the base needs a drive
V(t)
V
t
Figure 11.2. Damped oscillation
in a parallel LCR circuit.
Base-to-emitter
drive voltage
R
L
V
dc
Figure 11.3. Transistor and dc
supply replace energy lost to
damping.
122 Radio-frequency electronics: Circuits and applications
signal with slightly more amplitude than the sine wave on the emitter.
Figure 11.4 shows three methods to provide this drive signal. The Armstrong
oscillator adds a small secondary winding to the inductor. The voltage induced
in the secondary adds to the emitter voltage. The Hartley oscillator accom-
plishes the same thing by connecting the emitter to a tap slightly below the top
of the inductor. This is just an autotransformer version of the Armstrong
oscillator if the magnetic flux links all the turns of the inductor. But the top
and bottom portions of the inductor do not really have to be magnetically
coupled at all; most of the current in the inductor(s) is from energy stored in
the high-Qresonant circuit. This current is common to the two inductors so they
essentially forma voltage divider. (Note, though, that the ratio of voltages on the
top and bottom portions of the inductor ranges from the turns ratio, when they
are fully coupled, to the square of the turns ratio, when they have no coupling.)
If we consider the totally decoupled Hartley oscillator no mutual inductance
and then replace the inductors by capacitors of equal (but opposite) reactance
and replace the capacitor by an inductor, we get the Colpitts oscillator. Note that
each oscillator in Figure 11.4 is an amplifier with a positive feedback loop. No
power supply or biasing circuitry is shown in these figures; they simply indicate
the ac signal paths.
Using the Hartley circuit as an example, Figure 11.5 shows a practical circuit.
It includes the standard biasing arrangement to set the transistors operating
point. (A resistor voltage divider determines the base voltage and an emitter
resistor then determines the emitter current, since V
be
will be very close to
0.7 V.) A blocking capacitor allows the base to be dc biased with respect to the
emitter. Abypass capacitor puts the bottomof the resonant circuit at RF ground.
In practice, one usually finds oscillators in grounded emitter circuits, as
shown in Figure 11.6. The amplitude of the base drive signal must be much
smaller than the sine wave on the resonant circuit. Moreover, the polarity of the
base drive signal must be inverted with respect to the sine wave on the collector.
You can inspect these circuits to see that they do satisfy these conditions. But,
on closer inspection, you can note the circuits are identical to the circuits of
Figure 11.4, except that the ground point has been moved from the collector to
(a) Armstrong (b) Hartley (c) Colpitts
R
L R
L
C
2
C
1
R
L
Figure 11.4. Feedback loop
details define (a) Armstrong;
(b) Hartley; (c) Colpitts
oscillators.
R
L
V
dc
Figure 11.5. Hartley oscillator
circuit including bias circuitry.
123 Oscillators
the emitter. In these oscillators, an amplifier is enclosed in a positive feedback
loop. But, because there is no input signal to have a terminal in common with
the output signal, oscillators, unlike amplifiers, do not have common-emitter,
common-collector, and common-base versions.
The Colpitts oscillator, needing no tap or secondary winding on the inductor,
is the most commonly used circuit. Sometimes the transistors parasitic
collector-to-emitter capacitance is, by itself, the top capacitor, C
1
, so this
capacitor may appear to be missing in a circuit diagram. A practical design
example for the Colpitts circuit of Figure 11.6(c) is presented later in this
chapter.
11.2.1 Unintentional oscillators
In RF work it is common for a casually designed amplifier to break into
oscillation. One way this happens is shown in Figure 11.7. The circuit is a
basic common-emitter amplifier with parallel resonant circuits on the input and
output (as bandpass filters and/or to cancel the input and output capacitances of
the transistor). When the transistors parasitic base-to-collector capacitance is
included, the circuit has the topology of the decoupled Hartley oscillator. If the
feedback is sufficient, it will oscillate. The frequency will be somewhat lower
than that of the input and output circuits so that they look inductive as shown in
the center figure. This circuit known as a TPTG oscillator, form Tuned-Plate
Tuned-Grid, in the days of the vacuum tube.
Figure 11.6. Grounded-emitter
oscillator circuits.
(a) Armstrong
C
2
C
1
(c) Colpitts (b) Hartley
Figure 11.7. Tuned amplifier as
an oscillator
C
bc
C
bc
C
bc
Hartley oscillator Tuned amplifier
124 Radio-frequency electronics: Circuits and applications
With luck, the loop gain of any amplifier will be less than unity at any
frequency for which the total loop phase shift is 360 and an amplifier will be
stable. If not, it can be neutralized to avoid oscillation. Two methods of
neutralization are shown in Figure 11.8.
In Figure 11.8(a), a secondary winding is added to provide an out-of-phase
voltage which is capacitor-coupled to the base to cancel the in-phase voltage
coupled through C
bc
. In Figure 11.8(b), an inductor from collector to base
resonates C
bc
to effectively remove it (a dc blocking capacitor would be placed
in series with this inductor). In grounded-base transistor amplifiers and
grounded-grid vacuum tube amplifiers the input circuit is shielded from the
output circuit. These are stable without neutralization (but provide less power
gain than their common-emitter and common-cathode-counterparts).
11.2.2 Series resonant oscillators
The oscillators discussed above were all derived from the parallel resonant
circuit shown in Figure 11.2. We could just as well have started with a series
LCR circuit. Like the open parallel circuit, a shorted series LCR circuit executes
an exponentially damped oscillation unless we can replenish the dissipated
energy. In this case we need to put a voltage source in the loop which will be
positive when the current is positive and negative when the current is negative,
as shown in Figure 11.9.
C
bc
Figure 11.8. Amplifier
neutralization.
Figure 11.9. Series-mode
oscillator operation.
V(t)
R
L
R
L
125 Oscillators
While a bare transistor with base-to-emitter voltage drive makes a good
current source for a parallel-mode oscillator, a low-impedance voltage source
is needed for a series-mode oscillator. In the series-mode oscillator shown in
Figure 11.10, an op-amp with feedback is such a voltage source.
Since no phase inversion is provided by the tank circuit, the amplifier is
connected to be noninverting. An emitter-follower has a low output impedance
and can be used in a series-mode oscillator (see Problem 11.4). When the series
LC circuit is replaced by a multisection RC network, the resulting oscillator is
commonly known as a phase-shift oscillator (even though every feedback
oscillator oscillates at the frequency at which the overall loop phase shift is
360). An RC phase-shift oscillator circuit is shown in Figure 11.11. Op-amp
voltages followers make the circuit easy to analyze.
For the three cascaded RC units, the transfer function is given by V
2
(t)/V
1
(t) =
1/(RC+1)
3
. The inverting amplifier at the left provides a voltage gain of 16/2
= 8, so V
1
(t)/V
2
(t) = 8. Combining these two equations yields a cubic
equation with three roots: RC = 3j,

3
p
, and

3
p
. The first root corresponds
to an exponential decay of any initial charges on the capacitors while the two
imaginary roots indicate that the circuit will produce a steady sine-wave oscil-
lation whose frequency is given by RC

3
p
. In practice, the 16k resistor
would be increased to perhaps twice that value to ensure oscillation. Note that
Figure 11.11. An RC phase-shift
oscillator.
+

+ + +
R R R
V
2
(t)
V
1
(t)
2k
f = 0.27/(RC)
16k
C C C
R
1
R
2
R
L
+

A
V
=

1

+

R
2
/R
1
=

just over 1
Figure 11.10. An op-amp series-
mode oscillator.
126 Radio-frequency electronics: Circuits and applications
this circuit is a positive-feedback sine-wave oscillator even though it does not
contain a resonator. When the 16k resistor value is increased, the loop gain for
the original frequency becomes greater than unity, but for the new gain, there
will be a nearby complex frequency, j, for which the loop gain is unity. The
time dependence therefore becomes e
j(j)t
= e
jt
e
t
, showing that the oscil-
lation amplitude grows as e
t
. This circuit illustrates how any linear circuit with
feedback will produce sine-wave oscillations if there is a (complex) frequency
for which the overall loop gain is unity and the overall phase shift is 360. (Of
course must be positive, or the oscillation dies out exponentially.)
11.2.3 Negative-resistance oscillators
In the circuits described above, a transistor provides current to an RLC circuit
when the voltage on this circuit is positive, i.e., the transistor behaves as a
negative resistance. But the transistor is a three-terminal device and the third
terminal is provided with a drive signal derived fromthe LCR tank. Figure 11.12
shows how two transistors can be used to make a two-terminal negative
resistance that is simply paralleled with the LCR tank to make a linear sine
wave oscillator that has no feedback loop.
The two transistors form an emitter-coupled differential amplifier in which
the resistor to V
ee
acts as a constant current source, supplying a bias current, I
0
.
The input to the amplifier is the base voltage of the right-hand transistor. The
output is the collector current of the left-hand transistor. The ratio of input to
output is 4V
T
/I
0
, where V
T
is the thermal voltage, 26 mV. This ratio is just the
negative resistance, since the input and output are tied together. This negative-
resistance oscillator uses a parallel-resonant circuit, but a series-resonant ver-
sion is certainly possible as well.
Any circuit element or device that has a negative slope on at least some
portion of its IV curve can, in principle, be used as a negative resistance.
Tunnel diodes can be used to build oscillators up into the microwave frequency
range. At microwave frequencies, single-transistor negative-resistance oscilla-
tors are common. A plasma discharge exhibits negative resistance and provided
a pre-vacuum tube method to generate coherent sine waves. High-efficiency
R
V
ee
R
L
R
L
R
(a) (b)
V
cc
V
cc
Figure 11.12. A negative-
resistance oscillator.
127 Oscillators
Poulsen arc transmitters, circa World War I, provided low-frequency RF power
exceeding 100 kW.
11.3 Oscillator dynamics
These resonant oscillators are basically linear amplifiers with positive feedback.
At turn-on they can get started by virtue of their own noise if they run class A.
The tiny amount of noise power at the oscillation frequency will grow expo-
nentially into the full-power sine wave. Once running, the signal level is
ultimately limited by some nonlinearity. This could be a small-signal non-
linearity in the transistor characteristics. Otherwise, the finite voltage of the
dc power provides a severe large-signal nonlinearity, and the operation will shift
toward class-C conditions. The fact that amplitude cannot increase indefinitely
shows that some nonlinearity is operative in every real oscillator. Any non-
linearity causes the transistors low- frequency 1/ noise to mix with the RF
signal, producing more noise close to the carrier than would exist for linear
operation. An obvious way to mitigate large-signal nonlinearity is to detect the
oscillators output power and use the detector voltage in a negative feedback
arrangement to control the gain. This can maintain an amplitude considerably
lower than the power supply voltage. Alternatively, if the oscillator uses a
device (transistor or op-amp circuit) with a soft saturation characteristic, the
amplitude will reach a limit while the operation is still nearly linear. For
example, the amplifier in the oscillator of Figure 11.10 might have a small
cubic term, i.e., V
OUT
= AV
IN
BV
IN
3
, where B/A is very small (see Problem
11.5).
11.4 Frequency stability
Long-term (seconds to years) frequency fluctuations are due to component
aging and changes in ambient temperature and are called drift. Short-term
fluctuations, known as oscillator noise, are caused by the noise produced in
the active device, the finite loaded Q of the resonant circuit, and nonlinearity in
the operating cycle. The higher the Q, the faster the loop phase-shift changes
with frequency. Any disturbances (transistor fluctuations, power supply varia-
tions changing the transistors parasitic capacitances, etc.) that tend to change
the phase shift will cause the frequency to move slightly to reestablish the
overall 360 shift. The higher the resonator Q, the smaller the frequency shift.
Note that this is the loaded Q, so the most stable oscillators, besides having the
highest Q resonators, are loaded as lightly as possible. In LC oscillators, losses
in the inductor almost always determine the resonator Q. A shorted piece of
transmission line is sometimes used as a high-Q inductor. Chapter 24 treats
oscillator noise in detail.
128 Radio-frequency electronics: Circuits and applications
11.5 Colpitts oscillator theory
Let us look in some detail at the operation of the Colpitts oscillator. Figure 11.13
shows the Colpitts oscillator of Figure 11.6(c) redrawn as a small-signal
equivalent circuit (compare the figures). The still-to-be-biased transistor is
represented as a voltage-controlled current source. The resistor r
be
represents
the small-signal base-to-emitter resistance of the transistor.
The parallel combination of L and the load resistor, R, is denoted as Z, i.e., Z =
jLR/(jL+R) = jL
S
+R
S
, where L
S
and R
S
are the component values for the
equivalent series network. Likewise, it is convenient to denote r
be
1
as g. The
voltage V
be
, a phasor, is produced by the current I (a phasor) from the current
source. This is a linear circuit, so V
be
can be written as V
be
= I Z
T
, where Z
T
is a
function of . We will calculate this transfer impedance using standard circuit
analysis. Since the current I is proportional to V
be
, we can write an equation
expressing that, in going around the loop, the voltage V
be
exactly reproduces
itself :
g
m
V
be
Z
T
V
be
or
1
Z
T
g
m
: (11:1)
This equation will let us find the component values needed for the circuit to
oscillate at the desired frequency, i.e., the values that will make the loop gain
equal to unity and the phase shift equal 360.
We can arbitrarily select L, choosing an inductor whose Q is high at the
desired frequency. Equation (11.1), really two equations (real and imaginary
parts), will then provide values for C
1
and C
2
. To derive an expression for Z
T
,
we will assume that V
be
= 1 and work backward to find the corresponding value
of I. With this assumption, inspection of Figure 11.13 shows that the current I
1
is
given by
I
1
jC
2
g: (11:2)
Now the voltage V
c
is just the 1 volt assumed for V
be
plus I
1
Z, the voltage
developed across Z:
V
c
1 jC
2
gZ: (11:3)
V
c
l
1
C
2
C
1
Z
R
L
I = g
m
V
be
V
be
r
be =1/g
Figure 11.13. Colpitts oscillator
small-signal equivalent circuit.
129 Oscillators
Finally, the current I is just the sum of I
1
plus V
c
jC
1
, the current going into C
1
:
I jC
2
g 1 jC
2
gZ jC
1
: (11:4)
Since we had assumed that V
be
=1, we have Z
T
=1/I or
1
Z
T
jC
2
g 1 jC
2
gZ jC
1
: (11:5)
Using this, the condition for oscillation, Equation (11.1) becomes
g
m
jC
2
g 1 jC
2
gZ jC
1
0: (11:6)
The job now is to solve Equation (11.6) for C
1
and C
2
. If we assume that is
real i.e., that the oscillation neither grows nor decays, we find from the imag-
inary part of this equation, that
C
2
C
1
gR
S
C
1
L
S
C
1
C
2

2
(11:7)
and, from the real part, that

2
C
1
C
2
R
S
g
2
L
S
C
1
1 g
m
: (11:8)
Solving Equations (11.7) and (11.8) simultaneously for C
2
and C
1
produces
C
2

g
m
L
S
2R
S
1

4R
S
1 gR
S
g
m
g
g
m
2
L
S

2
s
!
(11:9)
and
C
1

C
2

2
L
S
C
2
1 gR
S
: (11:10)
Normally C
2
will have a much larger value than C
1
and % 1=

LC
1
p
.
Moreover, the second term in the square root of Equation (11.9) is usually
much less than unity so C
2
g
m
L
S
/R
S
.
11.5.1 Colpitts oscillator design example
Let us design a practical grounded-emitter Colpitts oscillator. Suppose this
oscillator is to supply 1 mW at 5 MHz and that it will be powered by a 6 V dc
supply. Assuming full swing, the peak output sine wave voltage will be 6 V. The
output power is given by 0.001 W=(6 V)
2
/(2R
L
) so the value of the load
resistor, R
L
, will be 18 k ohms. Assuming class-A operation, the bias current
in the transistor is made equal to the peak current in the load: I = I
pk
= 6 V/
18 k = 0.33 mA. If we let the emitter biasing resistor be 1.5 k, the emitter bias
voltage will be 1500 0.33 mA= 0.5 V. Assuming the typical 0.7 V offset
between the base and emitter, the base voltage needs to be 1.2 V. A voltage
divider using a 40 k resistor and a 10 k resistor will produce 1.2 V from the 6 V
130 Radio-frequency electronics: Circuits and applications
supply. These bias components are shown in the schematic diagram of
Figure 11.14.
A 0.05 F bypass capacitor pins the base to ac ground and another bypass
capacitor ensures that the dc input is held at a firm RF ground. Note that the
1.5 k emitter bias resistor provides an unwanted signal path to ground. This path
could be eliminated by putting an inductor in series with the bias resistor as an
RF choke, but this is not really necessary; the1.5k resistor is in parallel with C
2
,
which will have such a low reactance that the resistor will divert almost no
current from it.
With the biasing out of the way, we nowdeal with the signal components. The
transconductance of the transistor is found by dividing the bias current, I
0
, by
26 mV, the so-called thermal voltage,
2
i.e., g
m
=0.33 mA/ 26 mV=0.013 mhos.
The small-signal base-to-emitter resistance, r, is given by r = V
thermal
/I
0
. For a
typical small-signal transistor, such as a 2N3904, is about 100, so r
in
=
1000.026 V/0.33 mA=8000 ohms.
Using Equation (11.9) and (11.10), the values of C
1
and C
2
are 102 pF and
0.023 F, respectively. These are the values for which the oscillator theoretically
will maintain a constant amplitude. In practice, we increase the feedback by
decreasing the value of C
2
to ensure oscillation. This produces a waveform that
grows exponentially until it reaches a limit imposed by circuit nonlinearity. The
frequency becomes complex, i.e., becomes j and the time dependence
therefore becomes e
j(j)t
= e
jt
e
t
. Suppose we want to be, say 10
5
, which
will cause oscillation to grow by a factor e every 10 sec. (Fast growth would be
important if, for example, the oscillator is to be rapidly pulsed on and off.) How
do we find the value of C
2
to produce the desired ? To avoid doing more
analysis, it is convenient to use a standard computer program such as Mathcad
to find the root(s) of Equation (11.6) for trial values of C
2
. In this example, if we
decrease C
2
to 0.020 F, we obtain the desired .
+5 V
18k
C
2
0.020 F
C
1
102 pF
2N3904
1.5k
40k
10k
0.05 F
0.05 F
10 Hy
0.01 F
Figure 11.14. Colpitts oscillator:
5 MHz, 1 mW.
2
The thermal voltage is given by V
thermal
= 0.026V = kT/e, where k is Boltzmans constant,
T is the absolute ambient temperature, and e is the charge of an electron.
131 Oscillators
Problems
Problem 11.1. Draw a schematic diagram (without component values) for a bipolar
transistor Colpitts oscillator with the collector at ground for both dc RF. Include the
biasing circuit. The oscillator is to run from a positive dc supply.
Problem 11.2. Design (without specifying component values) a single-transistor
series-mode oscillator based on the emitter follower circuit.
Problem 11.3. A simple computer simulation can illustrate how an oscillator builds up
to an amplitude determined by the nonlinearity of its active element. The program shown
below models the negative-resistance oscillator of Figure 11.12(a). The LC resonant
frequency is 1 Hz. This network is in parallel with a negative-resistance element whose
voltage vs. current relation is given by I = (1/R
n
)*(VV
3
), to model the circuit of
Figure 11.12. The small-signal (negative) resistance is just R
n
. The termV
3
makes the
resistance become less negative for large signals. The program integrates the second-
order differential equation for V(t) and plots the voltage versus time from an arbitrary
initial condition, V=1 volt.
Run this or an equivalent program. Change the value of the load resistor R. Find the
minimum value of R for sustained oscillation. Experiment with the values of R and R
n
.
You will find that when the loaded Q of the RLC circuit is high, the oscillation will be
sinusoidal even when the value of the negative resistor is only a fraction of R. When Q is
low (as it is for R=1), a low value of R
n
such as R
n
= 0.2 will produce a distinctly
distorted waveform.
QBasic simulation of negative-resistance oscillator of Figure 11.12a.
SCREEN 2
R=1:L=1/6.2832:C=LtheparallelRLCcircuit:1ohms,1/2pihenries,1/2pifarads
RN = .9 run program also with RN=.2 to see non-sinusoidal waveform
E = .01 negative resistance: I= (1/RN)*(V-EV^3)
V = 1: U = 0 initial conditions, V is voltage, U is dV/dt
DT = .005 step size in seconds
FOR I = 1 TO 3000
T = T + DT increment the time
VNEW = V + U * DT
U=U+(DT/C)*((1/RN)*(U-3*E*V*V*U)-V/L-U/R)
V=VNEW
PSET(40*T,100+5*V)plotthepoint
NEXTI
Problem 11.4. In the oscillator shown below, the voltage gain of the amplifier decreases
with amplitude. The voltage transfer function is V
out
=2 V
in
0.5V
in
3
. This characteristic
will limit the amplitude of the oscillation.
132 Radio-frequency electronics: Circuits and applications
R
1
R
2
V
out
V
out
=2 V
in
0.5V
in
3
V
in
V
in
1.5
1.5
1 1
Find the ratio R
2
/R
1
in order that the peak value of the sine wave V
in
will be one volt.
Hint: assume V
in =
sin(t). The amplifier output is then 2 sin(t) 0.5sin
3
(t). The
second term resembles the sine wave but is more peaked. The LC filter will pass the
fundamental Fourier component of this second term. Find this termand add it to 2sin(t).
Then calculate the ratio R
2
/R
1
so that the voltage divider output is sin(t).
133 Oscillators