Ch1 Feedback and Stability

Download as pdf or txt
Download as pdf or txt
You are on page 1of 88

UEEA1333

Analogue Electronics
Topic 1
Feedback & Stability

Jan 2014
1

Reference
Donald A. Neamen, (2007),
Microelectronics Circuit Analysis and
Design, 3rd Edition, McGraw-Hill. ISBN10:007-125443-9. Chapter 12.
Sedra, Smith, (2004) 5th Ed.
Microelectronic Circuits, Oxford University
Press. ISBN 0-19-514252-7. Chapter 8.

Decibel
Pi

Ap

Pi = input power [W]

Po

Po = output power [W]

Po
Power gain in terms of ratio: AP =
Pi

Power gain in terms of dB:

AP ( dB )

[unitless]

Po
= 10 log10
Pi

[dB]

Decibel
Vi

AV

Vi = input voltage [V]

Vo

Voltage gain in terms of ratio:

Vo = output voltage [V]

Vo
AV =
Vi

Voltage gain in terms of dB: AV ( dB )

[unitless]

Vo
= 20 log10
Vi

[dB]

*This also gives the power gain if the input and output resistances are
the same
4

Example: Input voltage to a system


is 5 mV and output voltage is 0.5
V. What is the voltage gain of the
system in ratio and dB?

5 10 1
AV =
= 100
3
5 10
5 10 1
AV ( dB ) = 20 log10
= 40 dB
3
5 10
If Vo > Vi, we get positive value dB
=> gain
If Vo = Vi, we get 0 dB
If Vo < Vi, we get negative value dB
=> attenuation

Voltage
gain Av
32
16

decibel
value
30 dB
24 dB

18 dB

4
2

12 dB
6 dB

1
0.707
0.5
0.25
0.125
0.0625
0.03125

0 dB
- 3dB
- 6 dB
- 12 dB
- 18 dB
- 24 dB
- 30 dB

Frequency response & Bode plot


The voltage gain may be a function of frequency in
most systems i.e. it varies with frequency
Plot of voltage gain (dB) versus frequency on semilog
graph paper is used to show the frequency response
of a system

frequency

linear

log

log 1 = 0

log 4 = 0.602

log 7 = 0.845

log 2 = 0.301

log 5 = 0.699

log 8 = 0.903

log 3 = 0.477

log 6 = 0.778

log 9 = 0.954

Bandwidth
In general, an amplifiers gain versus
frequency graph resembles the figure below:

The gain at fL and fH is 3 dB less than the


maximum midband gain.
The bandwidth of the amplifier (in Hz) is
defined as fBW = f H f L
Frequency response is obtained from the
transfer function of the system (a mathematical
expression relating the output to input). The
magnitude and phase of the transfer function is
a function of frequency. Plot of these two
graphs consist the frequency response of the
system.
9

Introduction to feedback
Feedback can either be positive or negative
In positive feedback, a portion of the output
signal is added to the input signal
Example use in oscillators

In negative feedback, a portion of output


signal is subtracted from the input signal
Example use in amplifier systems

10

Configuration of Negative
Feedback Amplifier

Si = input signal (voltage or current)


So = output signal
Sfb = feedback signal
S = error signal

Forward path
Feedback path
Closed loop
11

Without feedback, the system gain = A


With negative feedback, the system gain becomes:

A
A
Af =
=
1 + A 1 + T

A = open loop gain or amplification factor


Af = closed loop gain or transfer function

= feedback transfer function


T = loop gain or loop transmission

Derivation:

S o = AS = A(Si S fb ) = ASi AS fb = ASi AS o


S o + AS o = ASi
So
A
A
Af =
=
=
Si 1 + A 1 + T

where T = A

12

If T >> 1,

A
A 1
Af =

=
1 + A A

So, Af is independent of A and depends on only


The feedback network usually consists of passive
components which can be chosen with high accuracy

A
1
Si Si
S o =

1 + A

S = S i S o S i

Si = 0

With a large loop gain, error signal decreases


to almost zero

13

Example: Assume that the open-loop gain of a


system is A = 105 and the closed-loop gain is
Af = 50. Determine .

A
Af =
1 + A
5

10
50 =
1 + (105 )
= 0.01999

14

Example: Assume that the open-loop gain of a


system is A = -105 and the closed-loop gain is
Af = -50. Determine .

A
Af =
1 + A
10
50 =
1 + (105 )
= 0.01999
5

For negative feedback, A should be positive

15

Advantages of Negative Feedback


1. Gain Desensitivity variations in gain is
reduced.
2. Bandwidth Extension larger than that of
basic amplifier.
3. Noise Reduction may increase S-N ratio.
4. Reduction of Nonlinear Distortion
5. Control of Impedance Levels input and
output impedances can be increased or
decreased.

16

Disadvantages of Negative
Feedback
1. Reduced Gain reduced compared to that of
the basic amplifier.
2. Stability problem possibility that feedback
circuit will become unstable and oscillate at
high frequencies.

17

Gain Desensitivity
In most practical amplifiers, the openloop gain A is dependent on temperature
and the operating conditions of active
devices (transistors).
Negative feedback reduces the variation
in the closed-loop gain Af due to variation
in the open-loop gain A
18

If is constant,
d A

=
dA dA 1 + A

dA f
dA f

dA

dv
u

u dy v du
y= ,
= dx 2 dx
v dx
v
u = A, v = 1 + A

(
1 + A) A
1
=
=
2
2
(1 + A)
(1 + A)

dA f

dA
A
1 dA
=

=
2
Af
(1 + A) 1 + A 1 + A A

Percent change in Af is less than the corresponding


percent change in A by the factor (1+A)
19

Sensitivity of Af to A is defined as

Af
A

A f A f

1
=
=
A A
1 + A

For A >> 1, the sensitivity of Af to A becomes very


small, so, a significant change in A will cause only a
small change in Af
But it can be shown that Af changes directly with

Af

= 1

Thus, in designing a feedback amplifier, special care


should be taken to ensure that the variation in the
20
feedback factor is minimized

Example: The open-loop gain of a system is A


= 105, the closed-loop gain is Af = 50 and =
0.01999. Assume that the change in A is 10%.
Determine the percent change in Af .
dAf
1 dA
1
=
=
(0.1)
5
A f 1 + A A 1 + 0.01999(10 )
dAf
= 5 10 5 0.005%
Af

(Approximate answer)

A
1.1105
Af =
=
= 50.00227
5
1 + A 1 + 0.01999(1.110 )
dA f 50.00227 50
=
= 0.000045 0.0045%
Af
50
(Accurate answer)

21

Bandwidth extension
Let us consider an amplifier whose open-loop
gain A is dependent on frequency.

A0
A0
A(s ) =
=
1 + s H 1 + s (2f H )
where
A0 is low-frequency open-loop gain
(Note: low freqeuncy gain is taken at f = 0)
fH is open-loop 3-dB break (cutoff) frequency

s = j2f

22

A0
A0
A(s ) =
=
1 + s (2f H ) 1 + jf f H
A=

A0
1+ ( f fH )

The curve is plotted below.


The red line is an approximate curve which is often
used instead of the actual curve.
approximation
actual

Gain |A|

Ao
f  0, |A| = Ao
f = fH, |A| = Ao/2

f > fH,
|A| decreases at
20 dB/decade

20 log 1/2= -3 dB

The bandwidth
equals to fH

fH

f (Hz)

23

Assume is independent of frequency, using negative feedback


will result in change of the gain and bandwidth:

A0
low-frequency closed-loop gain A0 f =
1 + A0
closed-loop 3-dB break frequency f Hf = f H (1 + A0 )
It can be seen that bandwidth is extended by using feedback.
Note that the gain-bandwidth product GBW remains constant.

24

Bandwidth extension
Gain

frequency
Bode plot of magnitude response of an amplifier

25

Without feedback:
Low-frequency gain
Bandwidth BW
Gain-bandwidth product GBW
With feedback:
Low-frequency gain

= A0
= fH
= A0 fH

A0
A0 f =
1 + A0

Bandwidth BW
fHf = fH(1 + A0)
Gain-bandwidth product GBW
=A0f fHf=A0fH
26

Example: A feedback amplifier has an openloop low-frequency gain of A0 = 104, an openloop bandwidth of fH= 100 Hz and a closedloop low frequency gain of A0f = 50.
Determine the bandwidth of the closed-loop
system.
4

A0
10
A0 f =
50 =
1 + A0 = 200
1 + A0
1 + A0
f Hf = f H (1 + A0 ) = 100(200) = 20 kHz
27

Noise reduction
Negative feedback may reduce the noise level
in amplifiers, thus increasing signal-to-noise
ratio (SNR = signal power/noise power)
Feedback can help reduce the noise generated
in an amplifier, but it cannot reduce the effect
when the noise is part of the input signal
Thus, SNR can be increased only if a noisy
stage is preceded by a relatively noise-free
stage e.g. in an audio amplifier
28

Hum noise
preamplifier

Output stage

In audio amplifier, the output stage suffers from


power-supply hum problem (a type of noise)
caused by difficulty in providing stable dc voltage
due to large currents that this stage uses.
A preamplifier stage with large voltage gain can be
added before the output stage (this preamplifier
should not have hum problem). Thus, SNR can be
increased.
29

Example:

vo = A1 A2 v + A2 vn = A1 A2 (vi vo ) + A2 vn
(1 + A1 A2 )vo = A1 A2 vi + A2 vn
A1 A2
A2
vo =
vi +
vn 100vi + 0.1vn
1 + A1 A2
1 + A1 A2
noise
signal
S o 100vi
Si
=
= 1000
N o 0.1vn
Ni

30

Reduction of Nonlinear Distortion


Ideally, to reproduce a signal without distortion, an
amplifier should have a transfer characteristic (plot of
So against Si ) which is linear:
So = AvSi
where Av is independent of Si

So

Si

31

Practical amplifier contains nonlinear devices such as


transistors, causing the transfer characteristic to be not linear.
Thus, if the input signal is a sinusoidal waveform, the output
voltage will not be sinusoidal, i.e. the output signal will be
distorted
input

output

32

However, closed loop gain Af is almost independent of open loop


gain. Thus negative feedback can linearize the transfer
characteristic of the amplifier, reducing non-linear distortion in
the output signal

Open loop transfer characteristic

Closed loop transfer characteristic

33

Basic Feedback Configurations

Four amplifier types


Voltage amplifier
Input: voltage
Output: voltage

Current amplifier
Input: current
Output: current

35

Transconductance
amplifier
Input: voltage
Output: current

Transresistance
amplifier
Input: current
Output: voltage

36

Four Basic Feedback Configurations


Series-shunt
(Voltage amplifier)

Shunt-series
(Current amplifier)

37

Series-series
(transconductance amplifier)

Shunt-shunt
(transresistance amplifier)

38

Summary of Feedback Amplifier

39

Ideal Series-Shunt Feedback

Input resistance of
feedback circuit is infinite

40

Vo = AV
v
Vfb = Vo = vVo
Assuming RS is negligible,
V = Vi Vfb
Vo
Av
Avf = =
= Closed-loop voltage gain.
Vi 1+ v Av

41

With feedback, input resistance of the system is :


Rif = Ri (1 + v Av )
This is larger than Ri, the input resistance without feedback.
Large input resistance is desirable for a voltage amplifier to reduce
loading effect on input signal source.

Derivation:

Vi = V + V fb = V + vVo = V + v ( AvV )
V =

Vi
1 + v Av

V
Vi
Ii =
=
Ri Ri (1 + v Av )
Vi
Rif = = Ri (1 + v Av )
Ii

42

With feedback, output resistance of the system is :


Rof = Ro / (1 + v Av )
This is smaller than Ro, the output resistance without feedback.
Small output resistance is desirable for a voltage amplifier to reduce
loading effect on output signal when load is connected.
Derivation:
Set the input signal voltage source equal to zero (a short circuit), and
apply test voltage to the output terminals.

V + Vfb = V + vVx = 0
or

V = vVx

Vx Av ( vVx ) Vx (1 + v Av )
Vx AV
v
=
=
Ix =
Ro
Ro
Ro
Vx
Ro
Rof = =
smaller the Ro
I x (1 + v Av )

43

To determine the output resistance:

Voltage
source is
shorted

Apply test
voltage at
output

44

Equivalent Circuit for


Series-Shunt Feedback Circuit

Avf =

Av
1 + v Av

Rif = Ri (1 + v Av )

Rof =

Ro
1 + v Av

45

Example: A series-shunt feedback amplifier


has an open-loop gain of Av = 105 and a closedloop gain of Avf = 50. Assume the input and
output resistance of the basic amplifier are Ri =
10 k and R0 = 20 k respectively.
5

Av
10
Avf =
1 + v Av =
= 2000
1 + v Av
50
Rif = Ri (1 + v Av ) = 10( 2000) = 20 M

20
Rof =
=
= 10
1 + v Av 2000
Ro

46

Ideal Shunt-Series Feedback

47

I o = Ai I
I fb = I o = i I o
Assuming ideal signal source, RS =
I i = I + I fb
Io
Ai
Aif = =
Closed-loop current gain.
I i 1 + i Ai

48

With feedback, input resistance of the system is :


Rif = Ri / (1 + i Ai )
Small input resistance is desirable for a current amplifier to reduce
loading effect on input signal source

Derivation:

I i = I + I fb = I + i I o = I + i ( Ai I )
I =

Ii
1 + i Ai

I i Ri
Vi = I Ri =
(1 + i Ai )
Vi
Ri
Rif = =
I i (1 + i Ai )

49

With feedback, output resistance of the system is :


Rof = Ro (1 + i Ai )
Large output resistance is desirable for a current amplifier to reduce loading
effect on output signal when load is connected

Derivation:
Set input signal current equal to zero (an open circuit), and apply test
current to the output terminals.

I + I fb = I + i I x = 0
I = i I x

Vx = Ro (I x Ai I ) = Ro [I x Ai ( i I x )] = I x (1 + i Ai )Ro
Vx
Rof =
= (1 + i Ai )Ro
Ix
50

To determine the output resistance


Open
circuit to
remove
current
source

apply test
current to the
output

51

Equivalent Circuit for


Shunt-Series Feedback Circuit

Aif =

Ai
1 + i Ai

Rif =

Ri
1 + i Ai

Rof = (1 + i Ai ) Ro

52

Ideal Series-Series Feedback


Circuit

53

Equivalent Circuit for


Series-Series Feedback Circuit

Agf =

Ag
1 + g Ag

Rif = Ri (1 + g Ag )
Rof = (1 + g Ag ) Ro

54

Ideal Shunt-Shunt Feedback Circuit

55

Equivalent Circuit for


Shunt-Shunt Feedback Circuit

Azf =

Az
1 + z Az

Rif =

Ri
1 + z Az

Rof =

Ro
1 + z Az
56

Series-Shunt Feedback Circuit


Example: Non-inverting amplifier

57

In the ideal feedback circuit, the amplification factor Av is very


large. The transfer function is
Vo
Av
1
Avf = =

for v Av >> 1
Vi 1 + v Av v
For the ideal noninverting op-amp amplifier,
Vo R2
Avf = = 1 +
Vi R1

Therefore, the feedback transfer function v is


1
v =
R2
1 +
R1

With finite amplifier gain, it can be shown that Avf , Rif and Rof
has the same form as in ideal series-shunt configuration
58

Example: Consider the non-inverting op-amp


with parameters Ri = 50 k, R1 = 10 k, R2 =
90 k and Av = 104. Determine the input
resistance of the circuit.

1
1
v =
=
= 0.10
1 + R2 / R1 1 + 90 / 10
Rif = Ri (1 + v Av ) = 50[1 + (0.10)(10 )] 50 M
4

59

Stability

In negative feedback, Sf = So is subtracted from Si to


produce the error signal Se.
Open loop gain is a function of frequency (due to
transistor parameters and capacitances) i.e. A(s = j)
60

At some frequencies, the open loop gain may


change such that subtraction becomes addition;
then negative feedback become positive,
producing an unstable system and oscillation.
Af ( s ) =

A( s)

1+ A(s)
14243
negative feedback
uuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuur
s changes until A ( s ) changes sign
Af ( s ) =

A(s)
1 A( s)
14243
positive feedback
61

Af ( s ) =

So ( s )
Si ( s )

A(s)
1+ A( s) ( s)

Lets write this in phasor form:

Af ( j ) =
=
=

A ( j )
1 + A ( j ) ( j )
A ( j )
1 + A ( j ) ( j ) e
A ( j )
1 + T ( j )

j ( )

A ( j )
1 + T ( j ) ( )
magnitude

phase

62

The system can become unstable at radial


frequency at which phase angle () = 1800

A( j )
A( j )
A f ( j ) =
=
1 + T ( j ) 180 1 T ( j )
e j = cos ( ) + j sin ( )
= 1 + j 0 = 1

Criteria to determine stability:


If |T(j)| < 1 when phase angle () = 180,
the system is stable
If |T(j)| 1 when phase angle () = 180,
the system is unstable

63

Example: One-pole amplifier


The single-stage common-emitter
current amplifier is an example of
a one-pole amplifier. The smallsignal current gain can be written
in the form:
Io
Ai = =
Ii

Aio
f
1 + j
f1
f
tan
2
f1
f

f1

Aio

1 +

magnitude

phase

1
where f1 =
2R C1
is the 3 dB break frequency
64

Bode plot of the current gain magnitude.

65

Bode plot of the current gain phase is shown


below (note that does not affect the phase,
thus the phase of T and A are the same)

Amplifier is stable because phase angle does not reach 180


66

Example: Two-pole amplifier


If we connect two one-pole amplifiers in cascade (series) to
become a two-stage amplifier, we obtain a two-pole amplifier.

The curent gain can be written in the form :


Aio
Ai =

f
f
1 + j f 1 + j f
1
2

1 f

Aio
1 f
Ai =
tan + tan
2
2
f1
f2

f
f
1+ 1+
f1
f2
f1 is the upper 3-dB frequency of the first stage, and f2 is the
67
upper 3-dB frequency of the second stage.

Bode plot of the current gain magnitude,


assuming f1 << f2.

68

Bode plot of the current gain phase, assuming


f1 << f2.

Amplifier is stable because phase angle does not reach 180


for any finite frequency

69

Example: Three-pole amplifier


An op-amp is a three-stage amplifier
Since each stage has an equivalent input
resistance and capacitance, this circuit is an
example of a three-pole amplifier.

70

The overall gain can be expressed as


A=

Ao

f
f
f
1 + j 1 + j 1 + j
f1
f2
f3

where Ao is the low-frequency gain factor.

71

Assuming the poles are far apart, i.e., f1 << f2


<< f3. The Bode plot of the magnitude gain is

72

The Bode plot of the current gain phase is

Amplifier may become unstable since phase angle becomes


73
180 at some frequency

Example: Consider a three-pole feedback


amplifier with a loop gain given by
T( f ) =

100
f

1 + j 5
10

Determine the stability of the amplifier for =


0.20 and = 0.02

74

Write the loop gain as a phasor (in terms of the magnitude and phase):

T( f ) =

100
f

1
+
j

105

100
f
f
f

j
j
j
1
1
1
+
+
+

105
105
105

100
2
2
2

1 f
f
f
f

1 f
1 f
1+

1
+
1
+

tan
+
tan
+
tan
5
5
5
5
5
5

10
10
10
10
10

10

100
3

1+
3 tan 1 f

105
105

100
1 f
=

3
tan
5
3
2
10

1+

10

75

T( f ) =

100

f
3 tan ( 5 )
3
10
2
f
1+ 5
10
1

The frequency at which the phase becomes - 180 is found :

f
phase of T ( f ) = 3 tan ( 5 ) = 180 f = 1.73 105
10
The magnitude of the loop gain at this frequency :
1

For = 0.20
0.20(100)
= 2.5
| T ( f ) |=
8
For = 0.02

system is unstable

0.02(100)
| T ( f ) |=
= 0.25
8

system is stable
76

Phase and Gain Margins


The methods to determine stability of a feedback
system are:
Nyquist stability criterion using Nyquist diagram
Phase and gain margins from the Bode plots

We can determine whether a feedback amplifier is


stable or unstable by examining the loop gain as a
function of frequency from the Bode plots.
We can also use this technique to determine the
degree of stability of a feedback amplifier.
77

At the frequency at which the loop gain


magnitude is unity, if the magnitude of the
phase is less than 180, the system is stable.
The difference (magnitude) between the phase
angle at this frequency and 180 is called the
phase margin.
Phase margin indicates how much the loop
gain can increase and still maintain stability
A typical desired phase margin is in the range
of 45 to 60.
78

(0 dB)

79

A second term that describes the degree of


stability is the gain margin.
It is defined to be the difference between loop
gain |T(j)| at the frequency where the phase
is 180 and unity, expressed in dB
It also gives an indication of how much the
loop gain can increase and still maintain
stability.

80

Frequency Compensation
This topic will discuss a method for modifying
the loop gain of a feedback amplifier to make
the system stable.
The general technique of making a feedback
system stable is called frequency
compensation.
One basic method involves introducing a new
pole in the loop gain function, at a sufficiently
low frequency that |T(f)|=1 occurs when || <
1800
81

Consider the Bode plots of a three-pole loop


gain magnitude and phase given in Figure
below and shown by the solid lines.
In this case, when the magnitude of the loop
gain is unity, the phase is nearly 270 degrees
and the system is unstable.
If we introduce a new pole fPD at a very low
frequency, and if we assume that the original
three poles do not change, the new Bode plots
of the magnitude and phase will be as shown
by the dotted lines in the Figure.
82

83

In this situation, the magnitude of the loop


gain becomes unity when the phase is || <
1800, and the system is stable.
Since the pole is introduced at a low frequency
and since it dominates the frequency response,
it is called a dominant pole.
This can be done by adding a fourth stage with
an extremely large input capacitance.
In practice, the Miller capacitance effect of
transistor is used. The second stage has a
feedback capacitor connected between the
output and input. This capacitor CF is called a
84
compensation capacitor.

Given the gain A, the effective input Miller


capacitance is
CM = CF(1 + A)

85

Exercises
Draw the block diagram of a negative feedback
system and derive the formula for the closed-loop
gain.
Explain the advantages and disadvantages of negative
feedback.
Name the four basic feedback configuration and the
type of amplifier associated with each.
Derive the closed-loop gain, input resistance and
output resistance of an ideal series-shunt feedback
system. Repeat for ideal shunt-series feedback.
86

Write the formula for the closed-loop gain, input


resistance and output resistance of each of the four
basic feedback configuration.
Explain why a negative feedback system can become
unstable. Under what condition does the system
becomes unstable?
Sketch the Bode plot of the magnitude and phase
response for a one-pole amplifier. Repeat for twopole and three-pole amplifiers. Discuss the stability of
each type of amplifier.
Explain the term phase margin and gain margin with
the help of diagrams.
With the help of diagram, explain what is meant by
frequency compensation.
87

Appendix: Complex numbers


Complex form: z = x + jy
where

Phasor form: z = |z|

| z |= x 2 + y 2
1

= tan ( y / x)

z1 z2 = (1 + 2 )
z1 | z1 |
=
(1 2 )
z2 | z2 |
88

You might also like