CH 10

CHAPTER 10
Introduction 803
10.1 The General Feedback Structure 804
10.8 Summary of the Feedback Analysis
Method 863
10.9 Determining the Loop Gain 863
10.2 Some Properties of Negative
Feedback 809 10.10 The Stabili ty Problem 868
10.3 The Four BasIc Feedback
Topologies 814
10.4 The Feedback Voltage Amplifier
(Series-Shunt) 823
10.5 The Feedback Transconductance
Amplifier (Series-Series) 834
10.6 The Feedback Transresistance
Amplifier (Shunt-Shunt) 846
10.7 The Feedback Current Amplifier
(Shunt-Series) 855
10.11 Effect of Feedback on the Amplifier
Poles 870
10.12 Stability Study Using 80de Plots 879
10.13 Frequency Compensation 884
Summary 890
Problems 890
IN THIS CHAPTER YOU WILL LEARN
1. The general structure of the negati ve-feedback amplifier and the basic
principle t hat underl ies it s operation.
2. The advant ages of negat ive feedback. how these come about, and at
what cost.
3. The appropriate feedback t opology to employ with each of the four am-
plifier types: vol t age, current , t ransconductance, and transresistance
amplifiers.
4. An intui t ive and insrght ful approach for the analysis of practical feedback-
amplifier circuits.
S. Why and how negat ive-feedback amplifiers can become unstable (i.e. ,
oscillat e) and how t o design the ci rcuit to ensure stable performance.
Introduction
Most physical systems incorporate orne form of feedback. It is imeresting to note, though,
that the theory of negative feedback has been developed by electroni cs engineers. In hi s
search for methods for the design of amplifier with stable gai n for use in telephone
repeaters, Harold Bl ack, an electronics engi neer with the Western Electri c Company,
invented the feedback amplifier in 1928. Si nce then the technique has been so widel y used
that it is almost impossible to think of electroni c circuits wi thout some form of feedback,
ei ther impl icit or expl icit. Furthermore, the concept of feedback and its associated theory
are currentl y used in areas other than engineering, such as in the modeling of bi ological
systems.
Feedback can be ei ther negative (degenerative) or positive (regenerative). In amplifier
design, negati ve feedback is applied to effect one or more of the following properties:
1. Desensitize the gain: that is, make the value of the gain less sensiti ve to variations
in the values of circuit components, such as might be caused by changes in temper
ature.
2. Reduce nonlmear distortion: that is, make the output proporti onal to the mput (in
other words, make the gain constant , independent of signal level).
3. Reduce the effect of noise: that is, minimi ze the contribution to the output of
unwanted electri c signals generated, either by the circuit components themselves, or by
extraneous interfe rence.
803
804 Chapter 10 Feedback
4 Control the inpllt and Olltpllt resistances that is, raise or lower the Input and
. . . fidb
resistances by the selection of an appropnate ee ack topology.
5. Extend the bandwidth of the amplifier.
All of the deSirable properti es above are obtained at the expense of a reduction In gam. It wi ll
be shown that the gam-reductIOn factor, call ed the amount of feedback, is the factor b)
which the CirCUIt IS desensIti zed, by which the Input resistance of a voltage amplifier is
increased, by which the bandWidth IS extended: and so on. In short, the basic idea a/nega.
tive feedback is to trade off gam for other deSirable propertres. Thl chapter is devoted t
the study of negati ve-feedback amplifiers: their analysis, design, and characteristics. 0
Under certain conditions. the negative feedback In an ampli fier can become positive and of
such a magnitude as to cause oscill ation. In fact, In Chapter 17 we will study the use of posltile
feedback in the design of OSCillators and bistable circutts. Here, In this chapter. however. we
are Interested in the design of stable amplifiers. We shall therefore study the stability prob-
lem of negati ve-feedback amplifiers and thetr potential for oscillation.
It should not be Implied. however, that posi tive feedback always leads to instability. In
fact, positi ve feedback is qUIte useful in a number of nonregenerative applicatIOns. such as
the design of acti ve fi lters, which are studied in Chapter 16.
Before we begin our study of negative feedback. we wish to remind the reader that we
have already encountered negati ve feedback in a number of applications. Almost all opamp
circuits (Chapter 2) empl oy negati ve feedback. Another popular application of negative
feedback is the use of the emitter resistance RE to stabilize the bias poi nt of bipolar transis
tors and to Increase the input resistance, bandwidth, and linearity of a BJT amplifier. In
additi on, the source follower and the emitter follower both employ a large amount of negative
feedback. The questi on then arises about the need for a formal study of negatt ve feedback. As
will be appreciated by the end of this chapter, the fOlmal study of feedback provides an
Invaluable tool for the analysis and design of electronic circuits. Also, the inSight gained b)
thinking In terms of feedback can be extremely profi table.
10.1 The General Feedback Structure
Figure 10.1 shows the baSIC structure of a feedback ampltfier Rather than showing voltages
and currents. Fi g. 10. 1 IS a signal-flow diagram, where each of the quant ities x can repre
sent either a voltage or a current signal. The apen-loop ampltfier has a gain A; thus its output
x, is related to the mput x by
(10.11
Source
A
.. t
Load -0-
\.
E
/l
Figure 101 Gene I I r h . d Ih quan
lili
';'
.. ra s ructure 0 t e feedback amplifi er. Thi s is a signal-now diagram. an e
represent either voltage or current signals.
The output x, is fed to the load as well as to a feedback netwo k h' h d
T ., r ,w IC pro uces a sample of
the output. hiS sample x
f
IS related to x, by the feedback facto r /3,
(10.2)
The feedback signal XI is subtracted from the source signal r whl'ch ' th . h
. I . - J ' is e mput to t e com-
plete feedback amphfier, to produce the Signal x" which is the input to the basic amplifier.
x, = x'-''f (10.3)
Here we note that it is this subtraction that makes the feedback negative. In essence, nega-
ttve feedback reduces that appears at the input of the basic amplifier.
Imphclt In the descnptton above is that the source, the load, and the feedback network
do not load the basic amplifier. That is, the gain A does not depend on any of these three net-
works. In practice this will not be the .case, and we shall have to find a method for casting a
real CIrcuit mtothe Ideal structure depicted In Fig. 10. 1. Fi gure 10.1 also implies that the for-
ward transtrusslOn occurs enttrely through the basic amplifier and the reverse transmission
occurs entirely through the feedback network.
The gain of the feedback amplifier can be obtained by combining Eqs. ( 10. 1) through
(I OJ):
A
(10.4 )
I+A/3
The quantity A/3 is called the loop gain, a name that follows from Fig. 10. 1. For the feed-
back to be negative, the loop gain A/3must be positive; that is, the feedback signal xfshould
have the same sign as x" thus resulting in a smaller difference signal x,. Equation (10.4) indi-
cates that for positive A/3the gain with feedback Af will be smaller than the open-l oop gain
A by a factor equal to I + A/3, which is called the amount of feedback.
If, as is the case m many circuits, the loop gain A/3is large, A/3 I, then from Eq. (IDA)
it follows that
I
A
f
=7J (10.5)
which IS a very mterestmg result: The gam of the feedback amplifier is almost entirel)' deter-
mined by the feedback network. Since the feedback network usually consists of passive com-
ponent which usually can be chosen to be as accurate as one wishes. the advantage of
negattve feedback In obtaining accurate. predictable, and stable gain should be apparent. In
other words, the overall gain will have very Itttle dependence on the gain of the basic ampli-
fier, A, a desirable property because the gain A is usuall y a function of many manufacturing
and application parameters, some of which might have wide tolerances. We have seen a dra-
matic illustration of all of these effects in op-amp ci rcuits in Chapter 2. where the closed-
loop gain (which is another name for the gain-with-feedback) is almost entirely detemlined
by the feedback elements.
Equations (10.1) through (10.3) can be combined to obtain the following expression for
the feedback signal Xi
.'1 =
A(J .
I + A(J "
( 10.6)
lin earlier chapters, we used the subscn pt "sig" for quantities associated with the signal source (e.g . I'
and R",). We dId thai to avoid confusion with Ihe subscnpl "s." which is usuall y used WIth FETs 10 de-
note quantitI es associated with the source termInal orthe transistor. At thi s point. however, it is expected
that readers have become sufficiently familiar wi th Ihe subJecllhat the possibility of confusion is min-
Imal. Therefore. we will revert to using the si mpler subscript s for SIgnal-source quantities.
o
o
o
o
Thus for AfJ:!> I we see that'r= X .' whIch implies that the Signal" at the Input ofthe bas,c
rtier is reduced to almost zero Thus if a large amount of negatIve feedback is employed,
signal x
J
becomes an almost Identical replica of the input Signal x. An Outcome
of thIS property is the tracking of the two Input termInals of an amp. The difference
between x and x/, which is x.' IS someumes referred to as .the er.ror Accordingly. the
input differencing circuit is often also called a c.ompanson CirCUIt . (It IS al so known asa
mixer.l An expresSIOn for Xi can be eaSIly detellllIned as
I
x, = I + A fJ
x
, (10.1)
from which we can venfy that for AfJ :!> I. x becomes very small . Observe that negati,e
feedback reduces the SIgnal that appears at the input tellllInals of the basic amplifier b) the
amount of feedback. (I + A/lJ. As will be seen later. it is this reduction of Input signal that
results In the increased lInearity of the feedback amplIfier
The nomnverting op-amp confi guration shown In Fi g. 10.2(a) pro, ides a direct ImplementatIOn of the
feedback loop of Fig. 10. 1
R
\ +
R
-
-
R
-
-
(a)
R.
0
+
\
V +
-
-
R
-
-
R.
-
-
rb)
Figure 10.2 (a) A nonlnvertlng op-amp CITCUII for Exampte to. t
with Its equivalent circuit.
\
R,
--
I'
AI
R,
-
-
h mp replaced
(b) The ClfCUIt In (a) with t e op-a
, . . . . f CE amplifier
We have In fact already seen exampl es of Ihl s: adding a reSlSlance R 111 the emItter a a for
(
'. . '. Ie . because
or a reSIstance R In Ihe SOurce of a CS amplifi er) increases the hneanty of these amp IIlers
the same InpUI SIgnal as before. v. and v are now smallar (by the amount of feedback).
(al Assume that the op amp has infimte InpUI resistance and zero output reSIstance. Find an expression for
the feedback factor fJ
(bl Find the condition under which the closed-loop gain A
J
is almost entirely determlOed by the feedback
network.
(c) I f the open-loop gain A = 10' V N. find R ,I R, to obtain a closedloop gain A f of 10 V N.
(d) What is the amount of feedback in decibels?
(e) If V, = I V. find Vo. Vf' and V,.
(f) If A decreases by 20%, what is the corresponding decrease in A
f
?
Solution
(al To be able to see more clearly the dlfect correspondence between the circuit in Fig. 10.2(a) and the
block diagram In Fig. 10. 1, we replace the op amp with its equivalentcircuit model. as shown in Fig.
10.2(b). Since the op amp is assumed to have infinite input resistance and zero output resi stance. its
model is simply an ideal voltage-controlled voltage source of gain A. From Fig. 10.2(b) we observe that
the feedback network, consisting of the voltage di vider (R
I
, R, ). is connected directly to the OUlput and
feeds a SIgnal V
f
to the inverting input termi nal of the op amp. lt is important at this point to note that the
zero output resi stance of the op amp causes the output voltage to be A V, irrespective of the values of R I
and R, and of R
L
. That is what we meant by the statement that in the block diagram of Fig. 10. 1. the
feedback network and the load are assumed not to load the basic amplifier. Now we can easily determine
the feedback factor fJ from
p=!i = RI
Vo RI +R,
Let ' s next examine how V
f
IS subtracted from V, at the tnput side. The subtraction is effectIvely per-
formed by the differential aCllon of the op amp; by its very nature. a differential -tnput amphfier takes the
dIfference between the signals at us two input terminal s. Observe also that because the input resistance of
the op amp IS assumed to be infinite, no current fl ows tn R,. Thus the value of R, ha, no bearing on I ', ;
or the source "does not load" the amplifier input. SimIlarly. because of the zero tnput current of the op
amp. V
f
will depend only on the rallo RIR, and not on the absolute value, of RI and R,
(b) The closed-loop gain Af is gIven by
To make Af nearly Independent of A, we must ensure that the loop gain AfJi' much larger than unity.
Since under such a condillon.
the condItIon can be stated as
AP :!> 1
A( RI ):!> I
R, + R.
I
Ai 7J =
R, + R,
RI
R,
- 1+-=
RI
(c) For A = \0
4
VN and Af = 10 VN. we see that A:!> A
J
thus we can seleci RI
I
P = - = 0. 1
If
and R, to obtain

Example 10.1 continued
Thu,.
whJeh Yield,
R, . R, =9
-
A more exact yalue for the ratio R1! R I can be obtamed from
which results in
and.
(d) The amount of feedback"
which is 60 dB.
(e) For I ' = I V.
10 =
A
I+AfJ
10'
1+ 10'fJ
fJ = 0.0999
R,
9.0 I - -=
R,
I+AfJ=:i-
A,
10'
10 = 1000
J', = A J J' = lO x I = 10 V
J J = fJl o= 0.0999 x 10 = 0.999 V
V = ~ = 10 = 0.001 V
, A 10'
Note that if we had used the approximate value of fJ = 0. 1, we would have obtamed J f = I V and
J',=OV,
(f) If A decreases by 20"k, thus becoming
A = 0.8 x 10' V N
the value of A! becomes
0.8 x 10'
AI = - - - - ' ~ ~ ~ - - = 9.9975 V' V
I + 0.8 x 10' x 0.0999
that is, It decreases by O.025"k, which IS lower than the percentage change In A by approximately a factor
(I + A fJ).
-
10.2 Some Properties of Negative Feedback 809
10.1 Repeat Example 10. 1,(c)to(f) for A = 100 VN.
Ans. (c) 10.11; (d) 20 dB; (e) IOV,0.9V,0.1 V; (f) 2.44%
10.2 RepeatExampleIO.I,(c)to(f) forA
f
= 10' VN. For(e)use V, = 0.01 V.
Ans. (c) 1110. 1; (d) 20 dB; (e) 10 V, 0.009 V, 0.001 V; (f) 2.44%
10.2 Some Properties of Negative Feedback
The properties of negative feedback were mentioned in the Introduction. In the following,
we shall consider some of these properties in more detail.
10.2.1 Gain Desensitivity
The effect of negative feedback on desensitizing the closed-loop gain was demonstrated in
Example 10. 1, where we saw that a 20% reduction in the gain of the basic amplifier gave
rise to only a 0.025% reduction in the gain of the closed-loop amplifier. This sensitivity-
reduction property can be analytically established as follows.
Assume that fJ is constant. Taking differentials of both sides of Eg. (10.4) results in
d4 = dA
I (I+AfJ)2
( 10.8)
DlVldmg Eg. (10.8) by Eg. (lOA) yields
dAr _ I dA
Af (I +AfJ) A
(\ 0.9)
which says that the percentage change m AI (due to variations in some circuit parameter) IS
smaller than the percentage change in A by a factor equal to the amount of feedback. For thiS
reason, the amount of feedback, I + AfJ, is also known as the desensitivity factor
o
10.3
An amplifier with a nom mal gam ~ = 1000 V'V exhibits a gam change of 10% as the operating
temperature changes from 25' C to 75C If it IS reqUired to constram Ihe change to 0.1 % by ap-
plying negative feedback, what is the largest closed-loop gam pOSSible? If three o;lhese feedback
amplifiers are placed in cascade, what o,erall gam and gam stabllt!} are acllle,ed
Ans. 10 VN; 1000 VN, With a maxmlUm ,anability of 0.3% over Ihe speCified temperature
range.
810 Chapterl0 Feedback
10.4
o
o
10.2.2 Bandwidth Extension
Consider an amplifier whose hi gh-frequency response is characteri zed by a single pole
gain at mid and high frequencies can be expressed as
AM
A (s) = -I -+-s"' l :-cv-'H (10.10)
where Alf denotes the midband gain and cvH is the upper 3-dB frequency. Application of neg.
ative feedback. with a frequency-mdependent factor P. around this amplIfier results in a
closed-loop gam AJs) given by
A(s)
A!.s) = 1+ PAis)
Substituting for A (s) from Eq. (10. I 0) results. after a little manipulatlon. in
A!.s) = (10.11)
Thus the feedback amplifier will have a midband gam of A All ( I + A MfJ) and an upper J.
dB frequency m
HI
given by
(10.12)
It follows that the upper 3-dB frequency IS increased by a factor equal to the amount of
feedback.
Similarly. it can be shown that if the open-loop gam is charactenzed by a dominant low
frequency pole giving ri se to a lower 3-dB frequency cv,. then the feedback amplifier will
have a lower 3-dB frequency m'l"
(IO.IJ)
Note that the amplifier bandwidth is increased by the same factor by which its midband
gain is decreased. mailltallling the gain bandwidth product at a constant value. This p01n1 is
further illustrated by the Bode Plot in Fig. 10.3.
Finally. note that the action of negative feedback in extending the amplifier bandwidth
should not be surprising: Negative feedback works to minimi ze the change in gam magnl'
tude. includmg its change with frequency.
Consider the non inverting op-amp CIrCUIt of Example 10. 1. Let the open-loop gam A have a lowfre
quency value of 10' and a un,form 6-dB/octave rolloffat high frequencies With a 3-dB frequency
of 100 Hz. Fmd the low-frequency gain and the upper 3-dB frequency of a closed-loop amplifier
With R, = I kG and R, = 9 kG.
Ans. 9.99 VN; 100.1 kHz
----
10.2 Some Propert,es of Negat,ve Feedback 811
Gam
(dB)
I
__________ _____ --;;-____ .!. )_
, t ' , ,
, I .20 dBldecade

, ,
, 20 log ( I t I Ifll I , '0 dB/dc'.ldc
, ,
, ,
, ,
, ,
---....,;
20 log (..1,,,1
,
(
, ,
, , ,
,
I ,
, ,
, , I
log (I t A"IlI, Ilog ll t I ifill '
.' ' . . '
, , I
,
, , I
,
I
,
, , ,
,
I
,
, , ,
, I
ft fH fHI f(log scale)
(" I t I ifill
fl'
-
I + '\"Il I
Figure 10.3 Applicallon of neg alive feedback reduces Ihe midband gam. mcreasesfll' and reduces/" all by
Ihe same factor. (I +A.,/J), whICh is equal to Ihe amounl of feedback.
10,2,3 Interference Reduction
Negati ve feedback can be employed to reduce the mterference in an amplifier or. more pre-
cisely. to increase the ratio of signal to interference. However, as we shall now explain. Ihls
mterference-reducti on process is possible only under certam conditions. Consider Ihe situa-
tion illustrated in Fig. l OA. Fi gure 10.4(a) shows an amplifier with gain A" an input signal
V,. and interference, V,. It IS assumed that for ome reason this amplifier suffers from
IOlerference and that the interference can be assumed to be introduced at the input of the
amplifier. The signal-to-i nterference ratio for thi s amplifier is
S/1 = V,I v.,
(10.14)
Consider next the ci rcuit in Fi g. 10.4(b). Here we assume that it is possible to build another
amplifier stage with gain A, that does not suffer from the interference problem. If this is the
case, then we may precede our ori ginal amplifier A, by the clean amplifier A, and apply
negative feedback around the overall cascade of such an amounl as to keep the overall gam
constant. The output voltage of the circuit in Fig. 10.4(b) can be found by superposition:
A,A, V A,
v;, = V. 1 +A ,A, p+ " I +A,A,p
Thus the stgnal -to-interference ratio al the output becomes
S V
- ,
I V.
,
(10.15)
(10.16)
which is A, times higher than in the ongi nal case. .' .
We emphasize once more Ihal the improvemenl 10 s,gnal-to-.mterference ratIO by Ihe
applicati on of feedback is possible onl y if one can precede the IOlerference-prone stage
v,
+
-
+ A,
- -
v,
tI
AI
(a)
-(
{3
(b)
I
AI
+
I
Figure 10.4 Illustrating the application of negative feedback to improve the signal-la-interference ratio In
amplifiers.
by a (relallvely) Interference-free stage. This SltuallOn, however, IS not uncommon In
practice The best example is found in the output power-amplifier stage of an audio
amplifier Such a stage usually suffers from a problem known as power-suppl) hum
The problem arises because of the large currents that this stage draws from the power
supply and the difficulty of providing adequate power-supply filtering inexpensively
The power-output stage I, reqUired to provide large power gain but lillie or no voltage
gain. We may therefore precede the power-output stage by a ,mall-signal amplifier that
provides large voltage gain, and apply a large amount of negative feedback, thus restor
ing the voltage gain to IIs original value. Since the small-signal amplifier can be fed
from another, less heft y (and hence beller regulated) power ,upply, it will not suffer
from the hum problem. The hum at the output will then be reduced by the amount of the
voltage gain of this added preamplifier.
10.5 Consider a power-output stage \\lIh voltage gain A I = I, an Input signal I = I V, and a hum I; of
I V Assume that thiS power stage IS preceded by a smail-signal stage with gam ~ , = 100 VN and
that overall feedback with fJ = I is applied. I f I and ," remam unchanged, find the signal and mter-
ference voltages at the output and hence the Improvement in Sf!
Ans, = I V; = 0.0 1 V; 100140 dB)
10.2 Some Properties of Negative Feedback 813
Vo (V)
- 0.08 - 0.06 - 0.04 -0.02
4
3
I
- 3
- 4
(a) (b)
0.02 0. 08
, VI
Figure 10.5 Illustrating the application ofnegauve feedback to reduce the nonlinear distortion In ampli-
fiers. Curve (a) shows the amplifier transfer characteristi c (vo versus 1'/) without feedback. Curve (b)
shows the characteristic (vo versus v
s
) with negative feedback (P = 0.01) appli ed.
10.2.4 Reduction in Nonlinear Distortion
Curve (a) in Fig. 10.5 shows the transfer characteristic Vo versus v, of an amplifier. As indi-
cated, the characteristic is piecewise linear, with the voltage gain changing from 1000 to 100
and then to O. This nonlinear transfer characteristic will result in this amplifier generallng a
large amount of nonlinear distortion.
The amplifier transfer characteristic can be consi derably linearized (i.e .. made les non-
linear) through the application of negative feedback. That this is possible should not be too
surpOSlng, since we have already seen that negative feedback reduces the dependence of the
overall closed-loop amplifier gain on the open-loop gain of the basic amplifier. Thus large
changes in open-loop gain (1000 to 100 in this case) give rise to much smaller corresponding
changes In the closed-loop gain.
To illustrate, let us apply negative feedback with fJ = 0.0 I to the amplifier whose open-
loop voltage transfer characteristic is depicted in Fig. 10.5. The resulting transfer characteris-
tic of the closed-loop amplifier, Vo versus v
s
' is shown in Fig. 10.5 as curve (b). Here the
slope of the steepest segment is given by
1000
A/I = 1+IOOO xO. OI =
90.9
and the slope of the next segment IS given by
100
A f2 = -I +--:-:10""0c::.
x
--::
0
--:: .
0
:71 = 5
0
Thus the order-of-magnitude change in slope has been considerably reduced. The price paid, of
course, is a reduction in voltage gain. Thus if the overall gain has to be. restored, a
preamplifier should be added. This preamplifier shoul d not present a severe nonhnear-d" tor-
lion problem, since it will be dealing wi th smaller signals.
Finally. It should be noted that negative feedback can do nothing at all about am lifi
saturation. since m saturatIOn the gam IS very small (almost zero) and hence th P let
. e amOunt of
feedback is almost UnIty.
10.3 The Four Basic Feedback Topologies
Based on the quantity to be amplified (voltage or current) and on the desi red form of
or current). amplifiers can beclassified into four categories. These
discussed m Chapter I. [n the follOWIng. we shall review thiS amplifier classificat'
. . .
POInt out the feedback topology appropnate In each case.
10.3.1 Voltage Amplifiers
Voltage amplIfiers are Intended to amplify an input voltage Signal and provide an out
voltage signal. The voltage amplifier is essentially a voltage-controlled voltage
The input resistance is reqUired to be high. and the output resistance is required to be I
0'.
Since the signal source is essentially a voltage source. It is convenient to represent It in
terms of a Thevenin equivalent circuit. [n a voltage amplifier. the output quantity ofinter
est is the output voltage. [t follows that the feedback network should sample the output
voltage. Just as a voltmeter measures a voltage. Also. because of the Thevenin representa
tlOn of the source. the feedback signal XI should be a voltage that can be mixed with the
source voltage in series.
The most suitable feedback topology for the voltage amplifier is the voltage-mixing.
voltage-sampling one shown in Fig. 10.6. Because of the series connectIOn at the mput and
the parallel or shunt connectIOn at the output. this feedback topology is also known :II
series-shunt feedback. As will be shown. this topology not only stabilizes the voitagegain
but also results In a higher Input resl tance (IntUitively. a result of the series connection at
the input) and a lower output resistance (intuitively. a re ult of the parallel connection at
the output). which are deSIrable properties for a voltage amplifier.
R ,

+
- voltage R,
I '
I amplifier

J
I v,
I I Feedback
1 ,
-)
network
Figure 10.6 Block d I I 5
. h lagram a a feedback voltage amplifier Here Ihe appropriate feedback tOpO 01) I
senes- -5 unt.
10.3 The Four Basic Feedback Topologies 815
The mcreased input resistance because VI subtracts from V,. resulting in a smaller
signal V, at the mput of the basIc amplifier. The lower V,. in tum. causes the input current to
be smaller. With the result that the resistance seen by V, will be larger. We shall derive a for-
mula for the input resistance of the feedback voltage amplifier in the next section.
The decreased outpm resistance results because the feedback works to keep V. as con-
stant as If the current drawn from the amplifier output changes by Ma. the
change 6V
a
'? Va wIll be lower than It would have been if feedback were not present. Thus
the output resistance 6V/ Ma Will be lower than that of the open-loop amplifier. In the fol-
lowing section we shall derive an expression for the output resistance of the feedback volt-
age amplifier.
Three examples of series-shunt feedback amplifiers are shown in Fi g. 10.7. The ampli -
fier in Fig. 10.7(a) is the familiar noninverting op-amp confi gurati on. The feedback net-
work. composed of the voltage divider (R,. R,). develops a voltage V that is applied to the
negative input terminal of the op amp. The ubtraction of VI from V: is achieved by utiliz-
ing the differencing action of the op-amp differential mput. For the feedback to be negative.
VI must be of the same polarity as V, . thus resulting in a smal ler signal at the input of the
basic amplifier. To ascertain that this IS the case. we foll ow the signal around the loop. as
follows: As V, increases, Va increases and the voltage divi der causes VI to increase. Thus
the change in If is of the same polarity as the change in v,. and the feedback is negati ve.
The second feedback voltage amplifier. shown m Fig. 10.7(b). utili zes two MOSFET
amplifier stages in cascade. The output vol tage v" is sampled by the feedback network com-
posed of the voltage divider (R, R
2
). and the feedback signal VI is fed to the source termi -
nal of Q" The subtraction is implemented by applying V, to the gate of Q, and VI to its
source. with the result that the signal at this amplifier input V, = Vg, = V, - VI' To ascertain
that the feedback is negative. let V, increase. The drain voltage of Q, will decrease. and
since this is applied to the gate of Q2' its drain voltage Va will increase. The feedback net-
work will then cause VI to increase. which is the same change in polarity initially assumed
for V,. Thus the feedback is indeed negative.
+ +
Q,
V
t T
I
R
z
Rz
\'
V,
I R,
I R,
-
- - -
- - - -
- -
<al
(b)
Figure 10.7 Examples of a feedback voltagc amplifier. All these CIrcuits employ series-shunt feedback.
Note that the de bias circuits are only partially shown.
I "
-
-
-
-
10.6
\ '00
R,
Q
\
The third example of series shunt feedback, shown In fi g. 10.7(c), utilizes a CG transis
torQ with a fraction I ( of the output voltage I. fed back to the gate through a voltage
dl\ Ider (R,. R) Observe that the subtractIon of 1 I from J IS effected by applying r to the
source. thus the Input J to the CG amplifier IS obtained as J J A' U ' I h . '. _ . _ . J J :, sua . owever,"e
the polarity 01 the feedback If J', increases, J
d
(which IS J ,) will mcrease and
I "III correspondingly Increase. Thus J f and J , change in the same direction, verifymg
that the feedback IS negatIve.
For the CIrcuit In Fig. ID.7(c) let (R R) R r h . ,+ - D USing small-Signal analysIS, find e'pres Ions
lor t e open-loop gain A .. J J th I' db k r fJ . .
A = J V F ' " e ee ac lactor = J ,; 1'0' and the closed loop gam
I 0' or A fJ <> I, find an approXimate expression for A Neglect r
Ans. A = gmRo, fJ= R, (R +R
l
), I 0
A = gmRo ( R, \
I l+gRR ' (R R' 1+-'
mOl + l) R I
10.3 2 Current Amplifiers
The Input signal In a curr t I t- .
. en amp I ler " essentially a current, and thus the "gnal source II
most conveniently represented b . N .
Y Its orton equivalent. The output quantity of interesl
ll
current. hence the feedba k I .
c should sample the output current, Just as a curre
nl
metedr meah,ures a current. The feedback SIgnal should be in curren I form so that it may be
m/xe In S lint With the .
rent a I'fi' source current. Thus the feedback topology most suitable for a cur
mp I ler IS the current- . . . . d' Fig
108(a) B mIXing, current-sampling topology. Illustrate In .
. . ecause of the parallel ( h ton
or S unt) connection at the Input, and the series con nee I
I, R,
Q,
d
I,
+
-
-
Basic
current
amplifier
RF
>
1
(h)
10.3 The Four BasIc Feedback Topologies 817
E
I"
1
RL
R,."
-
-
Figure 10.8 (al Block diagram ofa feedback current amplifier. Here. the appropriate feedback lopology
is the shunt- senes. (b) Exampl e ofa feedback current amplifier.
at the output, thiS feedback topology IS also known as shunt-series feedback. As will be
, hawn, thi s topology not only stabilizes the current gam but also results in a lower input
reSIStance, and a hi gher output resistance, both desirable properties for a current amplifier.
The decrease In Input resistance results because the feedback current II subtracts from
the mput current I " and thu a lower current enters the basic current amplifier. This in turn
results in a lower voltage at the amplifier input, that is, across the current source I, . It fol-
lows that the mput resistance of the feedback current amplifier will be lower than that of the
open-loop amplifier. We shall deri ve an expression for R'f m Section 10.5.
The mcrease in output resistance is simply a result of the action of negative feedback in
keepmg the value of 10 as constant as possible. Thus if the voltage across RL is changed. the
resulting change in 10 will be lower than It would have been without the feedback. which
Implies that the output resistance is increased. An expression for Rol will be derived in
Section 10.5.
An example of a feedback current amplifier IS shown In Fig. 10.8Cb) II uI'I'
.llzesaCG
slage Q, followed by a CS slage Q2' The output currenl I. IS fed 10 a load reSislance R
sample of I" IS obtamed by placmg a small resistance R If 10 senes With R Th I A
"d' I . R h L evoltage
developed across R'I IS Ie via a arge reSlSlance F 10 I e source node of Q Th r
, I' e leedbac,
current If Ihal flows through RF IS subtracled from I , al Ihe source node resull
, 109 In the
mput currenl I, = I, -If' For the feedbad. to be negallve, If must have Ihe same 01 .
. h h" h . P amyas
I,. To ascertam I al I IS IS t e case, we assume an Increase 10 I , and follow Ihe change
around the loop: An mcrease 10 I , causes I, to mcrease and Ihe drain voltage of Q .
. S' h' I ' I' d h f h h I Will
mcrease. mce I IS vo tage IS app Ie 10 I e gate 0 I e p-c annel deVice Q ils I'ncr .
". ease will
cause 1
0
, Ihe dram curren I of Q" to decrease. Thus, Ihe voltage across R wI'11 d
. _. . . ' . \{ ecrease,
which Will cause If to mcrease. ThiS IS the same polanty assumed for the initial chan e In
I, venfying that the feedback IS mdeed negative g
For the feedback current amphfier shown m Fig 10.8(b), find expresSion, for the open-loop gain
A =1 I . the feedback factor fJ = (/f/l,). and the closed-loop gain Af =1,11, For stmphclty, neglect
the Early effect m Q, and Q2'
- -
- -
,
IA
I R
IR
-
-
I

y r a
I. t
-
- -
-
F Iguro 10.9 AnatySis oflhe feedback currenl ampilfier of Fig. IO.8(b) to oblaln ., ii I / 1 and fJ .. 1/ 10 ,
Solution
Figure 10.9 shows the Circuit prepa d f II ' ' '
th d
' S' . re or sma -Signal analySiS. Some olthe analy'" IS al,o indicated on
e "gram. mce, as md,cated ,
1= 'g RI
u m'1 LJ I
10.3 The Four BaSIC Feedback Topologies 819
the open-loop gain A is given by
10
A = 1= -gm,RD
,
To obtain fJ, we observe that 10 IS fed to a current divider formed by Rand R S' . .\1 F mce current mlxmg
results m a reduced mput reSistance the voltage at the source node of Q w'll be ltd R ' . . '. . 2 I CDse Qzero,an F in
effect appears m parallel With R AI, enabltng us to obtain fJ as
fJ=!L=- Rlf
10 RF+R'I
where the negali:e sign is aresult of the reference directions used for 10 and I, Note, however. that the
loop gam AfJ Will be POSItIve, as should always be the case in a negative feedback amphfier. We can
now comb me A and fJ to obtain A, as
10
A, = - = -
I ,
10.7 For the feedback current ampltfier of Fig. 10.8(b). analyzed m Example 10.2, find an approxImate
expression for the closed-loop current gain under the conditIon that the loop gam IS large. Also,
state the conditIOn precisely.
Ans. Af = -(I + RF ) , gm,RD'J>( 1 + RF )
R'I R\{
10.3.3 Transconductance Amplifiers
In transconductance amplifiers the Input signal IS a voltage and the output signal is a current.
lt follows Ihat the appropriate feedback topology is the voltage-mixing, current-sampling
topology, illustrated in Fig. 10. lO(a). The presence of the series connection at both the input
and the output gives this feedback topology the alternative name series-series feedback.
As In the case of the feedback voltage amplifier, the series connection at the input results
In an increased input resistance. The sampling of the output current results in an increased
output resistance. Thus the series-series feedback topology provides the transconductance
amplifier with the destrable properties of increased input and output resistances.
Examples of feedback transconductance-amplifiers are shown in Fig. 10.10 (b) and (c).
The cirCUIt m Fig. 10. IOCb) utilizes a differential amplifier A I followed by a CS stage Q2
The output current 10 is fed to RL and to a series resistance RF, which develops a feedback
VOltage V,. The latter is applied to the positive input terminal of the differential ampitfier
A I' The subtraction of V, from V, is performed by the differencing actIon of the
differential-amphfier input. At this pomt we must check that I', and I', have the same
polarity: A posHive change in V will result m a negative change at the gate of Q" which m
turn causes 10 to increase. The I ~ c r e a s e in 10 results in a positive change in I,. which is the
same polarity assumed for the change in V" verifymg that the feedback is negative.
The transconductance amplpifier in Fig.IO. IO(c) utilizes a CS amplifier Q, in cascade
with another CS amplifier, Q2 The oUlput current I , is fed to RL and to a senes reSIStance
R
\
Basic
tr.lJlsconduclance
\ .. amphtier
I
( b )
Feedback
network
(a)
-
-
E
1 ,
(c)
I
I
-
-
Figure 10.10 (a) Block diagram of a feedback transconductance amphficr. Here. the appropriate feedbacl
topology is seriessenes. (b) Example of a feedback transconductance ampll tier (c) Another example.
RF that develops a feedback voltage VI The latter IS fed to the source of Q" thus utilizing
the tnput of Q, to implement the subtraction. f , = I f I The reader is urged to venf)
that VI has the same polarity as I and thus that the feed bad. tS negative.
10.8 For the mcut! 10 Fig. IO.IOCb), let the dtfferential ampltfier A I have an tnfintte mput resistance.
Use small-stgnal analysIS to obtam expressIOns for the open-loop gam ., .. 10 I'" the feedback
factor fJ = V rio' and the closed-loop gam A I I I' If the loop gam tS much greater than umty,
lid
' J 0
10 an approximate expreSSIOn for A Neglect r
I 02'
Ans. A = A Igm" fJ = R, ' A = A Igml . A = I R .
I I +A g R' I ,
t m7 f
10.3 The Four BaSIC Feedback Topologies 821
10.3.4 Transresistance Amplifiers
In transresistance amplifiers the input signal, s current and the output signal is voltage.
It foll ows that the appropriate feedback topology is of the current-mi xing, voltage-
sampling type, hown in Fig. 10. I I(a). The presence of the parallel (or shunt) connection at
I, + R,
Q,

I, t I,
)
RF
-
-
(c)
Basic

Feedback
network
-
-
Q,
-
-
-
-
(b)
V"
I,
)
t 1
I, R,
-
-
-
-
(d)
- I fi Here the appropriate fel!dback
Figure 10.11 Ca) Block diagram of a feedback lransreSls,ance amp I ler. .;" hfiers.
topology IS shunt -shunt. (b), (c), and (d) Examples of feedback transreSistance P
10.9
both the input and the output makes thi s feedback topology also known as shunt-shuOI
feedback. . .
The shunt connectton at the input causes the Input resIstance to be reduced. The shuntcQ
nection at the output stabilizes the output voltage and thus causes the output resislance 10 he
reduced. Thus, the hunl- hunl topology eqUIps the amplifier wilh Ihe desu.
able attributes of a low input and a low output re istance.. . .
Three examples of feedback transreslstance amplifIers are shown In FIg. 1O.II(b), (e)
and (d). The circuit in Fig. 10. ll(b) utilizes an op amp with a feedback resislance RF Iha;
senses Vo and provides a feedback current If that is from 1, at the inpul node. To
see that the feedback is negative, let 1, Increase. The Input current 1, wIll Increase, causing
the voltage of Ihe negative inpul terminal to rise. In response, the . output voltage wiU
decrease, causing an increase in If' Thus If and 1, have the same polarIty, and Ihe feedback
is negative. .
The circuit In Fig. 10.1 1 (c) ulilizes a CG stage QI cascaded wIth a CS slage Ql' A feed.
back resIstor RF senses Vo and feeds a current If to the Input node, where Ihe subtraction
from 1, takes place. The reader is urged to show that If and 1, have the same polanry and thus
the feedback is negative.
Finally, the BJT feedback transresistance amplifier In Fig. 10.11 (d) utilizes aCE siage QI
cascaded with an emitter follower Ql' A feedback resistor RF senses Vo and feeds back a cur
rent If to the input node, where it is subtracted from I ,. The reader is urged 10 show Ihal the
feedback is indeed negative.
For the CIrCUIt In Fig. 10. II (b), let the op amp have an open-loop gaIn A, a dl fferential inpul reo
slstance R,d' and a zero output resistance. Analyze the cirCUIt from first pnnclples (i.e., donol
use the feedback analysis approach) to determine A
f
" 1'0' I, Under what conditions does
Af=-R
F
?
-- ---0 __ . __
-.. - - . . "--- . - -
10.3.5 A Concluding Remark
Throughout thIS section we introduced examples of the four dlfferenl types of feedbacl
amplifier. However, in order to use the feedback analysis approach. we had to make a
ety of approximations. For instance. in Example 10.2. to find f3 we had to assume Iha
6
1 e
. . . I . Exercise 10. we
Input resIstance of the closed-loop amplIfier was very low. A so. In
load Ihe ba<1C
assumed that (R
I
+ R
1
);!> RD Ihat IS that the feedback networl.. does nOI b
. " snodou l
amplIfier. The need to make such approximatIOns In a seemIngly ad hoc manner I . the
somewhat disconcerting to the reader. There is. however, very good newS: StartIng II"fi .
'. . . f f dback amp
next sectton we Will present a systematIc approach for the analYSIS 0 ee d hO(
that takes into account the various loading effects and thus obviates the need for a

approximat ions.
104 The Feedback Voltage Amplifier (Senes-Shunt) 823
10.4 The Feedback Voltage Amplifier (Series-Shunt)
10.4.1 The Ideal Case
As mentioned before, series-shunt is the appropriate feedback topology for a voltage amp Ii-
fier.The Ideal structure of the series--shunt feedback amplifier is shown in Fig. I 0.12(a). It
consists of a unila/eral open-loop amplifier (the A circuit) and an ideal voltage-sampling,
voltage-mixing feedback network (the f3 circuit). The A circuit has an input resistance R" an
open-ci rCUIt voltage gain A, and an output resistance Ro' It is assum.ed that the source and
load resistances have been absorbed InSIde the A CIrCUI t (more on thIS POInt later). Further-
more, note that the f3 circuit does no/load the A cirCUIt; that is, connecting the f3 cirCUit does
not change the value of A (defined as A " V'; V, ).
A circui t
r \------------,.,..-,
I Ro ---t1_1 __
v. + ,
-
S'
R,
S
+
V,
-
S'
+
V R , ,
AV, I
-
l
'------- --
. .J
1-------------1
I I
L-_
I
I
I
I
I
I
I
+
v.
-
I
I
I
I
I
I
I
I I
L _____________ ..J
(a)
o
Ii
o

(b)
J
0'
R .
{3 circuit
. )d I clure' (b) eqUivalent CIrcuit.
Figure 10.12 The series-shunt feedback amphfier' (a I ea sITU .
o
The CIrcuit of Fig 10. 12(a) exactly foll ows the ideal feedback model of Fig. 10.1 Th
fore the clo ed-loop vollage is gi ven by .
V A
A r '" i = -I ...:..+:"'-A""'jJ (10.11
The eqUIvalent cirCUli model of the series-shunt feedback amplIfier is shown in Fi
10. I 2(b). Observe that AI the open-circUlt voltage gain of the feedback amplifier R .Ig.
_ . ",IS115
Input resistance. and RoJ IS 115 output resIstance. ExpreSS IOns for R'I and Rol can be denve<l
as follows.
For R'I ' refer to the input loop of the circUlt In Fig. 10. 12( a). The series ml xmg
I f from I and reduces J' by a factor equal to the amount of feedback (Eq. 10.7).
I .
J' = ----,
, I + AfJ
Thus the input current I, becomes
I ' I
I, = R =
( I +
(10.18)
Since I, the current drawn from I . the Input resistance R'I can be as
and using Eq. (1018) IS found to be
R'I = (I + AfJ)R,
(10.19)
ThUS. as expected. the feedback results In an Increase In the amplifier "PUI
resIstance by a factor equal to the amount of feedback. ( I + A fJ). a highly demable propen)
for a voltage amplIfier
It should be clear from the above derivation that the Increased Input resistance IS a resuh
only of the senes miXing and is independent of the type of samplIng. the transconduc
tance amplIfier. whIch is the other amplIfier type In whIch senes mIxing is employed. will
also exhJbit an increased input resIstance even though the feedback network samples ItSOUt
put current (series samplIng).
To determine the output resistance Rol of the feedback amplIfier In Fig. 10.12(a). we set
V, = 0 and apply a test voltage J', between the output terminals. as shown m Fig. 10.1J.1f
the current drawn from J is I . the output resistance R"I IS
An equauon for the output loop YIelds
From the Input loop we see that
Now VI = fJVo= fJV, thus.
J ,
R =-
01 - 1
,
J' A J
I =
r RII
( 10.10)
(10.111
104 The Feedback Voltage Amplifier (Series-Shunt) 825
o-<S_---o--
+
v ,
-
S'
R A V , ,
o t
I
I
I
I
I
o
0'
Figure 10.13 Determmmg the oUlput restSlanee of the feedback amphfier of Fig. 1O.l2(a) R", V I [ .
1'= -fJV , ,
whIch when subsututed In Eq. (10.21) YIelds
1 = V,( I+ AfJ)
, R
o
Substituting this value of I, Into Eq. ( 10.20) provtdes the following expression for R
ol

R _ Ro
ol- I+ AfJ
(10.22)
Thus. as expected. the shunt sampling (or voltage sampling) at the output resulls in a
decrease in the amphfier output resistance by a factor equal to the amount of negative feed-
back. (I + AfJ). a highly desirable propert y for a vollage amphfier.
Although perhaps not entirely obvious. the reduction of the output resistance i a result
only of the method of sampling the output and does not depend on the method of mixing.
Thus. the transistance amplifier. which tS the other amphfier type in which shunt (or volt-
age) sampling tS employed. will also exhibit a reduced output resistance.
10.4.2 The Practical Case
In a practical series-shunt feedback amplifier. the feedback network will not be an ideal voltage-
controlled voltage source. Rather. the feedback network is usually resistive and hence will
load the basic ampl ifier and thus affect the values of A. R,. and R,. In addition. the source and
load resistances wi ll affect these three parameters. Thus the problem we have is as foUows:
Given a senes-shunt feedback amplifier represented by the block diagram of Fig. 10.14(a). find
the A circuit and the fJ circuit.
Our problem essenti al ly involves representing the amplifier of Fig. 1O.14(a) by the ideal
structure of Fi g. I 0. 12(a). As a first step toward that end we observe that the source and load
reSIStances should be lumped with the basic amplifier This. together with representing the
two-port feedback network in terms of its " parameters (see Appendix C). is illustrated in
Fig. 10. 14(b) The choice of h parameters tS based on the fact that this is the only parameter
set that represents the feedback by a series network at port I and a parallel network
at port 2. Such a representation is obviously conventent in view of the series connection at
the input and the parallel conneclton at the output.
Examinauon of the circuit in Fig. 10. 14(b) reveals that rhe current source ""I, represents
the forward transmission of the feedback network. Since the feedback network is usually
o
J' + ,
J
V
I
'I
R,
R
R,
--
(
(
,
I
I\",
I
1
I
I

I,
h"
Basic
amplifier
Feedback
network
(a)

amph!ier
h" V, h" / ,
Feedback network
(b)
+
h"
-
\ If CUI
/

R
" I I
I
hll I
I

"'" 1
L ________ ___________ _

----------------,
I I
+
___ -
I
I
0 __ 1-1__
I
I
t I ( _

(c)
/:. \.Ir It
J
R
\
Figure Dl.:rJvatum of the.: A clrCUll and P circuit for the SI;!r1CS shunt feedback amplifier. (a)
diagram of a practical shunt feedback amplifier. (b) The CirCUit 10 (a) With the feedback net\\
represented by Its II parameters. (e) The circuit 10 (b) with h neglected
104 The Feedback Voltage Amplifier (Series-Shunt) 827
passive, Its forward transnusslOn can be neglected In comparison to the much larger forward
transmission of the basic amplifier. We w!lltherefore assume that 11i"I'db", <li 1""lb,,,,
and thus omit the cont rolled source h" / , altogether. ""woe, .mphr."
Compare the circuit of Fig. 10. 14(b) (after eliminating the current source hI) with the
ideal circuit of Fi g. 10. I 2(a). We see that by including " " and "" with the ba: ic'amplifier,
we obtain the circuit shown in Fi g. 1O.14(c), which is very similar to the ideal circuit. Now.
If the basic amplifier is unil ateral (or almo t unilateral)-that is it does not contain internal
feedback- then the circuit of Fi g. 10.14(c) is equivalent to the ideal circuit. It follows then
that the A circuit is obtained by augmenting the basic amplifier at the input with the source
resistance R, and the resistance h" of the feedback network. and at the output with the load
resistance RL and the conductance h" of the feedback network.
We concl ude that the loading effect of the feedback network on the basic amplifier IS
represented by the components h" and "". From the definitions of the h parameters in
Appendix C we see that h" is the resistance looking into port I of the feedback network with
port 2 short-circuited. Since port 2 of the feedback network is connected In shunt With the
output port of the ampli fier. short-clfcUltlng port 2 destroys the feedback. Similarly. h" IS
the conductance look.ing into port 2 of the feedback network with port I open-circuited.
Since port I of the feedback network is connected in series with the amplifier input. open-
C1fcuiting port I destroys the feedback.
These observauons suggest a Simple rule for finding the loading effects of the feedback
network on the basic amplifier: The loading effect is found by looking into the appropriate
port of the feedback network while the other port is open-circuited or short-circuited so a to
destroy the feedback. If the connecti on IS a shunt one, we short-circuit the port; if it is a
series one. we open-circuit it. In Sections 10.5. 10.6, and 10.7 it will be seen that this simple
rule applies also to the other three feedback topologies. J
We next consider the determination of /3. From Fig. I 0.14( c). we see that /3 is equal to " "
of the feedback network.
/3 =",,=
- " '2 1\:0
(10.23)
Thus to measure /3. one applies a voltage to port 2 of the feedback network and measures the
voltage that appears at port I whil e the latter port is open-circuited. This result is Intuitively
appealing because the object of the feedback network is to sample the output voltage (/" = J)
and provide a voltage signal (V, = VI) that is mixed in series with the input source. The series
connecti on at the input suggests that (as in the case of finding the loading effects of the feed-
back network) /3 should be found with port I open-circuited.
10.4.3 Summary
A summary of the rules for finding the A cirCUit and /3 for a given senes .. shunt feedback
amplifier of the form In Fig. 10 14(a) is given In Fig. 10.15. As for uSing the feedback for-
mulas in Eqs. (10 19) and (10.22) to determine the Input and output resistances. It IS Impor-
tant to note that
1. R, and Ro are the Input and output resistances. respectively. of the A circuit In
Fig.1015(a).
2. Rand R are the input and output resistance. respectively, of the feedback ampli-
if uJ
fier, IIlCllidillg R, and RL (see Fig. 10. 14a).
) A simple rule to remember' I f the connection is .\'hunt, short It; If senes, sever it.
I' ,
Tle \ c r
" ,
R,
+
-
R,
\\ hen R 1 ohtamcJ I, lnl
Feedback
network
,nJ the 'lin \ donned \
(b, f3 IS ohtamed trom

\
I
\
I
\
Ba"c
amplifier
Feedback
network
R12 RL \ .
amI R l' ohtaincd from

I
, I
Feedback
network
+
I
R
R
Figure 10.15 Summary oflhe rules for findmg the A Clrcu,t and pfor the senes shunt case ofF'g. 1O.I4{al
3. The actual mput and output resistances of the feedback amplifier usually exclude R,
and R
L
. These are denoted Roo and Rom in Fig. 10. 14(a) and can be easily detemllnedas
(10.24)
(10.25)
Fi gure 10.16(a) shows an op amp connected m the noninverltng confi gurat ion. The or amp has an open
loop gamp, a differential inpul resIStance R,d' and an outpul resistance r , Recall that m our anaIYS,,::
op-amp CtrCUllS m Chapter 2, we neglecled the effects of R d (assumed II to be Infi nlle) and 0 1 I (a"um
illo be l era). Here we wish 10 use the feedback melhod analyze the circUIt taking both R,d and r. mto
account. Find expressions for A, fl, the closed-loop gain Vol V, , the input resIStance Roo (sec Fig.
and the OUlpUI resistance Roo,' Al so tind numeri cal values, given f.I = 10', Rid 100 kn, /'" = I U1. RI =-
kn, R, = I ill, R, = I Mn, and R, 10 kn.
104 The Feedback Voltage Amplifier (Senes-Shunt) 829
R,
+
v,
+
-
-
R,
-
-
R
+
1', R,d
\. ,
R,
RI I=(R, R,)
I 0

-
-
Figure 10.16 CircUlt, for Example 10J.
-
-
R,
-
( a)
r,
J.L \.' , RL
-
- -
-
(b)
R,
-
(c)
-
-
I' o
r
"
R'12 =
(R, ;-R,l
+
\
-
-
\
R.
E 'ample 10.3 continued
Solution
We observe that the feedback network conSIsts of R, and RI This network samples the output voltage V
and provides a voltage signal (across R
I
) that IS mIxed 10 series with the IOpUt source V. . .
The A circuit can be easily obtained followlOg the rules of FIg. 10.15. and IS shown 10 FIg. 10.16(b).
Observe that the loading effect of the feedback network at the input SIde IS obtalOed by shon-circuiting
pon 2 of the feedback network (because it is connected in shunt) and looking IOtO pon I. with the result
that R II = R I II R,. The loadi ng effect of the feedback network at the output S I ~ e IS found by open-cir-
cuiting pon 1 (because it IS connected 10 senes) and 100klOg IOto pon ~ . wIth the result that
R" = R, + R I . Forthe resulting A CIfCUlt in Fig. I 0.16(b). we can wnte by IOspectlon:
~ ~ RL II( R, +R,) R,d
A - - 11
=V; - [RLII(RI+R,)]+roR,d+R,+(R,IIR,)
For the values given. we find that A = 6000 VN
The circuit for determining fJis shown 10 Fig. 1O.16(c)_ from whIch we obtalO
V, R, '
fJ
10-' VN
="=R+R=
o I ,
The vohage gain with feedback can now be obtalOed as
Va A = 6000 = 857 VN
Af=V;=I+AfJ 7
The input resistance Rd determined by the feedback equations is the reSIstance seen by the external source
(see Fig. I 0.16a). and is given by
R'f = R,( I +AfJ)
where R, is the input resistance of the A circuit in Fig. 1O.16(b):
R, = R, + R,d + (R, II R,)
For the values given. R = III kn. resultlOg 10
R 'f = III x 7 = 777 kn
This. however. is not the resistance asked for What is required IS Roo. IOdicated 10 Fig. I 0. 16(a). To obtalll
Roo we subtract R, from Rif:
R," = R'f- R
For the values given. Roo = 739 ill. The resistance R
o
, given by the feedback equatIOns IS the output reSIS-
tance of the feedback ampitfier. including the load resistance R,. as Indicated In Fig. 10.16(a). R" 1\
given by
Ro
I+AfJ
where R. IS the output resistance of the A circuit R. can be obtained by inspectIon of Fig.10.16(b) as
For the values given. R. = 667 n . and
667
7
= 95.3 n
10.4 The Feedback Voltage Amplifier (Series-Shunt) 831
The resistance asked for. Roo,. is the output reSIstance of the feedback ampitfier excluding R,. From Fig.
10.16(a) we see that
Thus
As another example of a series-shunt feedback amplifier. conSIder the clfcuit shown in Fig. 10.7(b) which
IS repeated in Fig. I 0. 17(a). It is reqUIred to analyze thIS amplifier to obtain Its voltage gain Va J' ... input
resistance RIO' and output resistance R
o
,,' Find numerical values for the case gml = gm2 = 4 mAN.
RDI = RDZ = 10 kn. R, = I kn. and R,= 9 kn. For si mplicity. neglect ro of each of Q, and Q,.
Solution
We identify the feedback network as the vohage divider (R, R
z
). Its loading effect at the input is
obtained by short circuiting its pon 2 (because it is connected in shunt with the output). Then. looking
into its pon 1. we see R,ll R
z
. The loading effect at the output is obtained by open-circuiting port I of the
feedback network (because it is connected in series with the input). Then. looking into pon 2. we see R,
in series with R, . The A circutt will therefore be a shown in Fig. 10.17(h). The gain A is determined as
the product of the gain of Q, and the galO of Q, as follows:
A, = ;0 = -gm,[ R
D2
11 (R, + R,)]
dl
A = ~ = A,A,= gmI
R
Dlg",,[RD2 11 (R
,
+ R,)]
V, - l+gml(R,IIR,)
For the numerical values given.
A - 4 x IO x 4[101I(1 +9)] = 173.9 VN
- 1+4(1119)
The value of fJ is determined from the fJ CI1"CUl! 10 Fig. 10.17(c).
fJ='!J.= R,
Va R, + R,
For the numerical values given.
I
fJ= =0.1
1+9
The closed-loop gain V/ V, can now be found as
= 9.5 VN
A 173.9
-
I +AfJ - 1+ 173.9xO. 1
V
a - A -
-- -
V f

\ +
-
-
-
-
-
\
0
I
Q,
+
\
R
R,
- - - -
Q
R,
R,
-
-
(a)
-
-
Q
-0, -
-
-
R,
-
-
-
-
Ib)
\
Q,
-
-
R
-

R
R
I
"
R,
R

+
-
-
\
R,
-
- - -
I c)
Figure 10.17 Ia) Senesshunt feedback amplifier for Example 10.4. (b) The A Clrcu,t; (c) The pmcu,!.
The input resistance IS obvIOusly Infinite because of the infintte Input reSIStance of the MOSFET The
output resistance Ro", IS obtained as follows,
R"
I+AfJ
where R" is the output resIStance of Ihe A Clfeuit. From Fig. 10.17(b),
= 10 1110 = 5 kQ
104 The Feedback Voltage Amplifier (SeTles-Shunt) 833
The amount of feedback is
I+AfJ= 1+(173.9xO.I)= 18.39
Thus.
R = 5000 _
0 ", 18.39 -
272 Q
which IS relati vely low given that the open-loop amplifier has R.= 5000 Q.
EXERCISES
10.10
10.11
10.12
If the op amp of Example 10.3 has a untform 6-dS'octave high-frequency rolloff '\lth / " . =
I kHz. find the 3-dB frequency of the closed-loop gain I:.
Ans. 7 kHz
The CITCUlt shown in Fig. E I O. II conSists of a differential stage followed by an emitter follo\\ er.
with series shunt feedback supplied by the resistors R, and R,. Assuming that the dc component of
V is zero. and that fJ of the BJTs is very high, find the dc operating current of each of the three
transistors and show that the dc vol tage at the output is approximately zero. Then find the values of
A, fJ, A f " v" I 11,. R", and Roo<, Assume that the transistors have fJ = 100.
Ans. 85.7V'V;0.1 VY;8.96VN; 191 kQ. 19.1 Q.
+ 10.7 V
IOkfl R,=9kll
-
Q,
R, =
I Ul
\. + 5 mA
-
I mA
-
-
-
-
Figure El0.11
For the seri es-shunt amph fier III Fig. 10. 7( c I. f3. A / . R", and R"o,
Ans. t = gm[ R
D
II (R, + R,l ] . f3 = R, (R, + R, l,
A}= A ( I+A/J) , Rm= (1 g", )(I+A/J).
Ro", = [R
D
II (R, + R, )] ( I +t/J)
\ ,
-
-
Neglect "0 of Q.
10.5 The Feedback Transconductance Amplifier
(Series-Series)
As mentioned m SectIOn 10.3. the senes--senes feedback topology stabilizes VV and
is therefore best SUIted for transconductance ampli fiers. Figure 10. 18(a) shows the Ideal
structure for the series-series feedback amplifier It consIsts of a unilateral open-loop
ampltfier (the A cIrcuIt) and an Ideal feedback network The A cIrcuIt has an toput resls.
tance R . a short-circuIt transconductance .4" I., I" and an output resistance R, The
fJ CIrCUIt samples the short-cIrcuIt output current I, and proVIdes a feedback voltage V
that is subtracted from I . m the senes mput loop. ote that the fJ cIrcuIt presents zero
resIstance to the output loop. and thus does not load the amplt fier output. Also, the feed
back signal If = fJI, is an Ideal voltage source, thus the fJ circUIt does not load the
ampltfier mput. Also observe that whIle A is a transconductance, fJ is a trans resistance.
and thus the loop gam AfJ IS, as expected, a dImensIOnless quantity Finally, note thaI
the source and the load resistances have been absorbed inSIde the A cIrcuit (more on thIs
later).
+
S r - - - -

I
1\
R, A I' Ra I
I '
,
I
I
>
1
S' I
--'
0'
-
1-------------1
_ ---,,--J
S __ ---!.,
+
v,
5'
I
I
I
i
+
f3 l o
-
I I
I
I
I
L _____________ --.J -.;.
(a)
0 '
(b)
Figure 10.18 The senes -senes feedback amplifier (a) Ideal struclure; (b) eqUivalenl WCUI\.
10.5 The Feedback Transconductance Amplifier (Series-Series) 835
Since the structure of Fig. I 0.18(a) follows the Ideal feedback structure of Fi g. 10. 1, we
can obtam the closed-loop gam AI as
I
A = _ A
I - V, - I + AfJ
(10.26)
The feedback amplifier can be represented by the equi valent circUIt in Fig.
10.18(b). Note that AI IS the short-cIrcuIt transconductance. Because of the series mi xing,
the input resIstance WIth feedback, R'I' will be larger than the input resistance of the A cir-
cuit, R" by a factor equal to the amount of feedback,
R'I = R,( I + A/lJ ( 10.27)
Recall that the derivation we employed m the previous section to obtai n R of the senes-
shunt feedback amplifier dId not depend on the method of sampling. Thus /applies equall y
well to the series series amplifier we are considenng here.
ext we consIder the output resistance RoJ of the feedback transconductance amplifier.
From the eqUIvalent CIrCUIt m Fig. I 0.18(b) we observe that RoJ is the resistance seen by
breakmg the output loop (say at 00') and setting V, to zero. Thus to find the output res is-
tance RoJ of the series-series feedback amplifier of Fig. I 0. 18(a) we reduce V, to zero and
break the output circuit to apply a test current I" as shown in Fig. 10. 19:
( 10.28)
In this case, v. = - = -/ll. = -/ll ,. Thus for the cIrcuit m Fi g. 10. 19 we obtain
v, = (/x- AV,) R. = (f,+A/lI,)R.
Hence
R.
I
= (I + AfJ)Ro
(10.29)
That IS, in this case the negative feedback mcreases the output resistance. This should have
been expected, si nce the negative feedback tries to make 10 constant in spite of changes in the
output voltage, which means increased output resistance. This result also confirms our earlier
observation: The relationship between RoJ and R. is a function onl y of the method of sanlpling.
r---<5 __ ..!-_
__
+
AI ,
\, +
0 '
S'
------
-------'
Fi gure 10,19 Detennining the output resistance R"f of the series-senes feedback amplifier.
o
o
.
While voltage (shunt) samplIng reduces the output
resistance,
current (series) sampling
.
lficreaSes It
We conclude that the seriessenes feedback topology increases both the input and the
output resistance, a highly desirable outcome for a transconductance amplIfier
10.5.2 The Practical Case
Figure 10.20(a) shows a block diagram for a practical series -series feedback amplifier. To
be able to apply the feedback equations to thiS amplifier, we have to represent it by the Ideal
structure of Fig. I 0.18(a). Our objective therefore is to devise a Simple method for finding A
and fJ. Observe the definition of the amplIfier Input resistance R" and output resistance R
It is important to note that these are different from R'f and R 'f' which are determmed by the
feedback equatIons. as will become clear shortly
The series-senes amplifier of Fig. 10.20{a) IS redrawn in Fig. 10.20{b) With R andR , ,
shown closer to the baSiC amplifier, and the two-port feedback network represented by
= parameters (Appendix C). This parameter set has been chosen because it IS the only one
that provides a representation of the feedback network with a senes cirCUit at the input and a
senes circuit at the output. This is obviously convenient In view of the series connections at
input and output. The mput and output resistances with feedback, R'f and R"i' are indicated
on the diagram.
As we have done In the case of the series-shunt amplIfier, we shall assume that the
forward transmi SSion through the feedback network IS negligible m comparison to thai
through the basic amplifier, and thus we can dispense with the voltage source =,,/, In Fig.
10.20(b). DOing thi S, and redrawing the Circuit to include =" and = with the basic ampli-
fier. results in the cirCuit In Fig. 10.20(c). Now if the basic amplifier IS unilateral (or
almost Unilateral ), then the CIrcuit In Fig. 10.20(c) is eqUivalent to the ideal circuil of
Fig. 10. 18(a).
It follows that the A circuit is composed of the basic amplifier augmented at the Input
with Rand = and augmented at the output with R, and =:,. Since =" and =: are the imped
ances looking Into ports I and 2, respectively. of the feedback network with the other port
open-cirCUited. we see that finding the loading effects of the feedback network on the balic
amplifier follows the rule formulated in Section 10.4 That IS, we look Into one pan of the
feedback network while the other port IS open-circuited or short-Circuited so as to demo)"
the feedback (open if series and short If shunt) .
From Fig. 10.20(c) we see that fJ is equal to 2" of the feedback network,
(1030)
fJ
-. - VI
- -12 = -
/ ,
-
/. -0
This result is intuitively appealing. Recall that In thIS case the feedback network samples the
output current [/ = I,] and provides a voltage [I f = I ,1 that is mixed in series with Ihe
Input source Again, the series connection at the input suggests that fJ IS measured With
port I open.
10.5.3 Summary
For future reference we present in Fig. 10.21 a summary of the rules for f1ndiOg A andP
for a given senesseries feedback amplifier of the type shown in Fig. 10.20{a). Note thai
R, is the IOpUt resIStance of the A circuit, and its output reslqance is R", which can be
v, +
-
10.5 The Feedback Transconductance Amplifier (Series-Series)
R
1
,
"
<
Ba!) ic
amplifier
RIO
lJ R OUI
CD
Feedback
CD network
(a)
___ ---' R,....-_ ---::---1-- RL,,-_.::<=:/ "=--<>-_
BaSIC
amplifier
R.,
I.
)
II I,
R.
:> ZII
z"
'"
Feedback network
(b)
A Circuit
----------
I
t
I
R
II I ,

If
--
B a.\Il'
amplitier
I
I
I l __________ ___ _
1----- ------ --, (I,
I
I Zl2 / a
'--___ 1 __ -
I
I
I
L _____________
(e) /3 mcult
Figure 10.20 Derivatton of the A circuit and the p cirCUli for series --senes feedback amplifiers. (0) A series
senes feedback amplifier. (b) The circuit of (a) with the fecdbad.. network represented by it s: parameters.
(e) A redrawi ng of the circuit 10 (b) \\-nh =21 neglected.
837
a
I
R
The -\ circuit is
R.
-
R,
where R is obtamed from
Feedback
I
network
(2)
and the "am 4 IS detincd A
I b) f3 IS obtained from
\
f3 '= ....L
1
t
,
\ "
o

I
\'
Basic
amplifier
(
I.
and R is obtained from
( I
Feedback
network
Feedback , \
t
network
,
"
Figure 10.21 Findmg Ihe A mcull and pfor Ihe senes-series feedback amplifier
L_-R"
1
-
Rn
determined by breaking the output loop and lookIng between Yand Y' whil e V IS selle
zero. R. and Ro can be used In Eqs. (10.27) and (10.29) to determine Rif and R,./ (see
10.20b). The Input and output resistances of the feedback amplIft er can then be found)
subtracting R, from R,/ and RL from R
o
/ '
(!OJ!)
(10.32)
As a first example of a feedback transconductance amplifier. consider the cirCUIt shown In Fig. 10.22('1,
ThiS IS the same circuit we presented in Fig. 10.1 Orb) and was the subject of Exercise 10.8, Here. for gen-
erality we not only assume that A I has finite Input and output resistances but include a source resIStance
R, . The objective is to analyze this circun to determine its closed-loop gain AI '= 1,/ 1' the input res{S-
- - . seen
tance of the feedback amphfier R", and the output resistance R
ou
, ' The latter tS the reststance
between the two terminals of R
J
looking back into the output loop,
10.5 The Feedback Transconductance Ampl ifier ( Series-Series) 8 39
R,
G
[)
-
v,
r-;
R
-
RL
-
R, R,
G::I R,
1
-)
R,
- -
- -
(a) (b)
R.
R,

+
-
r--- .... -
I ,
+
v,
-
-
-
-
-
-
-
S, "':'"
-
-
-
-
- -
-
R,
(c)
+
-
-
(d )
Figure 10.22 Ci rcuits for Example 10.5,
Solution
First we identify the basic amplitier and the feedback circuit. basic
tial amplifier A I cascaded with the CS PMOS tranststor Q2' e oulpu curren "
resistance R
F
. The latter IS the feedback network (Fig. IO.22b). It develops a voltage Vf that IS mixed m
series with the input loop.
The second step is to ascertam that the feedback is negative. We have already done this 10 Section
10.3.
Next, we detenmne an approximate yalue for Ai = 10 , V, under the assumptIOn that the loop gam
AfJ is much greater than unity. This value, found before any analYSIS IS undertaken, will help us deter
mme at the end whether our analysis IS correct: If the loop gain IS found to be much greater than umly,
then the final A f should be close to the value initially determined. From the circuit of Fig. 10.22(a),
and thus for large A fJ.
fJ = RF
I I
Af = - = -
fJ RF
Next. we determine the A Circuit. Since the feedback network (Fig. 1O.22b) I S connected 10 seri., with
both the mput and output loops, we include a resi stance RF in each of these loops (which is equivalent to
saying we include. at the input. the input reSIStance of the feedback circutt wtth pan 2 open and, at the
output, the input resistance of the feedback circuit with pon I open). Doing this, includmg R, and RL m
the A circui t. and replacing A I and Q, with their smallsignal models. results in the A cirCUit shown m
Fig. 10.22(c). Analysis of this circuit is straightforward:
(\ 0.33)
(10.34)
l' ,
I - J' 0_
o - -gm2 gJ1 r + R + R
02 L F
( 10.35)
Combming these three equatIOns results 10
( 10.36)
Usually R,d'l> (R, + R
F
), 1'o,:l> (R
L
+ R
F
). resultmg in the approximate expression for A:
A = A Igm'
( 10.37)
The input resistance R, can be found by mspectlOn as
(10.38)
The output resistance Ro is found by setting V, = 0, and breaking the output loop at any location. sa)
between D, and D ~ Thus.
Ro= r
u2
+R
L
+R
F
Finally. fJ can be found from Fig. 10.22(d) as
fJ=
V
f
R( - -
-
'.
( 10.39)
10.S The Feedback Transconductance Amplifier (Senes-Senes) 841
The loop gam A fJ is thus
AfJ = (A g , R)( R'd
I m. F R + R R
IJ _I +
(10040)
(IOAI)
With numerical values, one can now obtain the value of AfJ and determine whether it is indeed much
greater than unity. We next determine the closed loop gain
A _ A
r 1+ AfJ
Substttuting for A from Eq. (10.37) and for AfJ from Eq. (10.41). we have
which is the value we found at the outset.
The series mixing raises the mput resi stance with feedback.
R'f = R,( I + A/lJ
Substituting for R, from Eq. (10.38) and for AfJ from the full expression in Eq. (10.40). we obtam
R,j = (R,+ R,d+RF)(1 +A/lJ
R'f = R, + R,d+ RF + A Igm, RFR",
To obtam Roo. we subtract R, from R,j (see Fig. 10.22a):
(10.42)
which IS an mtultlvely appeahng result: The series mixing at the input raises the mput re,,,tance R", by a
factor equal to the approximate value of (I + A/lJ . .
To obtam R we note that the series connectton at the output ratses the output reSistance. thus.
of'
ROf = R, ( I + A/lJ
-
-
(1' ,+ RL + R,H I + A/lJ
o.
-
-
Substitutmg for AfJ from Eq. (10.40) and making the approximatIon Rid ~ (R, + RF ), we wnte
Rol = "02 + RL + RF + A Igm2
R
F" 02
To obtain Roo>' which is the resistance seen by RL m the circuit of Fig. 10.22(a), we subtract RL from
Rol '
Roo> = "02( I + A Igm2RF)
which is an intuitively appealing result: The senes connection at the output raises the output resistance of
Q2 ("02) by a factor equal to the amount of feedback. . ..
Finally, we note that we have deliberately solved this problem m great detail to Illustrate the beauty
01013 Forthe circuit analyzed in Example 10.5, select a value for RF that will result in AI = 5 mAN.
Now, for A, = 200 VN, gm2 = 2 mAN, R,d = 100 kO, " 0' = 20 kO , and assuming that
R, <a; Rid and RL <a; r02 ' find the value of AI realized and the mput and output resistances of the
feedback transconductance amplifier. Iffor some reason g m2 drops m value by 50%, what is the
corresponding percentage change mAl?
Ans. 2000; 4.94 mAN; 8 I MO; 1.62 MO, -1.25 %
c____ __ ~ r . -
Because negative feedback extends the amplifier bandwidth, it is commonly used in the design of broad
band amplifiers. One such amplifier is the MC 1553. Pan of the cirCUit of the MC 1553 is shown in Fig.
I 0.23(a). The circuit shown (called a feedback triple) is composed of three gain stages with series-senes
feedback provided by the network composed of REI' RI' and R".
Observe that the feedback network samples the emitter current 10 of Q3 ' and thus 10 is the output
quantity of the feedback amplifier. However, practically speaking, 10 IS rather difficult to utiltze Thus
usually the collector current of Q3' 1" IS taken as the output. ThiS current is of course almost equal to 10 ,
1, = alv ' Thus, as a transconductance amplifier with I, as the output current, the output reSIStanCe of
mterest is that labeled Roo> m Fig. 10.23(a). In some applications, 1, IS passed through a load resistance,
such as Rc3 , and the voltage Vo is taken as the output. Assume that the bias CITCUlt, which IS not shown;
establishes I CI = 0.6 rnA, 1 C2 = I rnA, and 1 C3 = 4 rnA. Also assume that for all three tranSiStors,
h
l
,.= 100 and "0 = 00.
'To avoid possible confusIOn of the BJT current gain fJ and the feedback factor fJ, we sometimes use hi' to denote the trtUl
slstor fJ.
10.5 The Feedback Transconductance Amplifier (Series-SeTtes) 843
(al Antlclpatmg that the loop gain will be large, find an approximate expression and value for the closed-
loop gam AI = 10/ V, and hence for 1/ V,. Also find V.I V,.
(b) Use feedback analysis to find A, fJ, AI' V.lV" R
m
, and Roo>, For the calculation of Roo>, assume
that ,.o of Q3 IS 25 kO .
Solution
(a) When AfJ ~ I,
I
= -
fJ
where the feedback factor fJ can be found from the feedback network. The feedback network is high-
li ghted in Fi g. 10.23(a), and the determination of the value of fJ is illustrated in Fig. 10.23(b), from which
we fi nd
+
\
+
V,
-
-
Re i = 9 kn
-
-
(b)
-
-
100 x 100 = I 1.9 0
100 + 640 + 100
1
R,
-
-
)
)
R
e
) = 600 n
i,=:::::;-<J \
) . 1
Rp = 640 n
~ - - - - - - ~ - - - - ~
- 100 n
R,,=loon
-
- -
-
<al
I"
- -
- -
Figure 10.23 CirCUits for Example 10.6.
-
- -
-
ReI
-
-
Q
-
\ -
- - - - -
R,
- -
-
-
- - -
Ie) ( d)
Figure 10.23 continued
Thus,
I
II. 9
= 84 mAN
Vo _ - f( R{ l
V I = 84 x 0.6 = -50A V IV
(fib)dEmPIOYing}he 10adlOg rules gIven in Fig. 10.21. we obtatn the ,.j cirCUli shown in Fig. IO.l3(c). To
lo A" 10 I j we first determine the aain of th t- . . . eo e Irs! stage. ThIS can be wnllen by Inspection as
'.1 _ -al(RC'I 1'.,)
. .. I '. r"+ IR,, II (RF+R"ll
SInce Q, IS bIased at 0.6 rnA I' - 41 7 n T' .:
100.40 = 2 5 kn S .' .' - . ranslStor Q IS bIased as I mA, thus 1', = hI, 'g, 1 =
. ubstltutlng these salues together with a = 099 R = 9 kn R = 100 n R =
640 n, and R, = 100 n. results In ' ." , '. ' I
J, I
-
- 14.92 VN
J ,
Next, we determine the gam of the second st h' . . .. .
V . age, w leh can be wntten by IOspectlon (noting that I " = I ,, )al
'2
I'd = gm, {Rc' II (hf<-+ I )[r,.,+ (Rnll (R,.+RI.I ))I}
10.5 The Feedback Transconductance Amplifier (Series-Series) 845
Subslttutlng gm2 = 40 mAN, Rc, = 5 kn, h" = 100, r ,) = 25/ 4 = 6.25 ",
,... d R 100"" I ' " " R,., = 100 n, RF = 640
'L, an EI = >L, resu ts lo
~ - ,
= -131.2 VN
V<\
Finall y, for the third stage we can write by inspection
10 _ I e)
-
~ 2
V
bl
= 6.25+(10011
74
0) = 10.6 mAN
ComblOlOg the gains of the three stages results 10
10
A" V = - 14.92 x- 131.2 x 10.6x 10-)
,
= 20.7 A/ V
The closed-loop gain Ar can now be found from
A= ~ _ A
'- ' ;- I+AP
20.7 8
= I +20.7x 11.9 = 3.7 mAN
which we note is very close to the approximate value found in (a) above.
The voltage gain IS found from
Vo - 1, RC) - l oRC)
- - = 1/ =-AJRo
~ ~ s
= - 83.7 X 10 ') x 600 = -50.2 VN
whIch is al so very close to the approximate value found in (al above.
The lOput resi stance of the feedback amplifier is given by
R" = R" = R,( I + AjJ)
where R, is the lOput resistance of the A circuit. The value of R, can be found from the circuit lo Fig.
10.23(c) as follows:
R
j
= (h". + 1)[1',1 + (REI II (RF+ Rnl)]
= 13.65kn
R" = 13 .65( I + 20.7 x 11 .9) = 3.38 Mn
To determine the output resistance RO"" which IS the resistance looking into the collector of Q" we face
a dilemma. The feedback does not sample I, and thus we cannot employ the feedback formulas dIrectly.'
Nevenheless, we present a somewhat Indirect solution to this problem below. Here we note parentheti-
cally that had Q, been a MOSFET, this problem would not have existed, since I" = I,.
SInce the feedback senses the emitter current 1
0
, the output resistance gIven by the feedback analysis
will be the resistance seen In the emllter cirCUIt, say between Y and Y ~
RoJ = Ro( I + AjJ)
'ThIS Imponant POlOt was first brought to the authors' attention by Gordon Robens (see Robens and Sedra. 1992).
where Ro can be detennined from the A Clfcuit in Fig. 10.23(c) by breaking the circuit between Yand Y'
The resi,tance looking between these two nodes can be found to be
RC2
Ro =[Rdl (RF+REI)]+r.-) +, I
1t' +
which, for the values given, Ro = 143.911. The output resistance RO[ of the feedback amplifier can now
be found as
Raj = Ro(l +AfJ) = 143.9(1 +20.7 x 11 .9) = 35.6 kl1
We can now use the value of RoJ to obtam an approximate value for Roo, To do thiS, we assume that the
effect of the feedback is to place a resistance R'J (35.6 k(1) in the emitter of Q" and find the output resis-
tance from the equivalent Clfcuit shown in Fig. 10.23(d). This is the output resistance of a BJT with a
resIstance Raj m its emitter and a resistance Rn m Its base. The formula we have for thIS (Eq. 7.50) does
not unfortunately account for a reSistance m the base The formula, however, can he modified (see Prob-
lem 10.48) to obtain
= 25 + [35.611 (0.625 + 5)]
= 2.19 Ml1
Thus Roo, is increased (from raJ) but not by (I + AfJ)
[
I + ]
. 25 + 5
010.14 For the feedback triple m Fig. I 0.23(a), analyzed m Example 10.6, modIfy the value of RF to ob-
tain a closed-loop transconductance 10' 1', of approximately 100 mNV Assume that the loop
gam remains large. What IS the new value of R F? For thi s value, what IS the approximate value
of the voltage gai n If the output voltage IS taken at the coll ector of QJ ?
Ans, 800 n ; - 60 VN
10.6 The Feedback Transresistance Amplifier
(Shunt-Shunt)
10,6.1 The Ideal Case
As menti oned In Secti on 10.3, the shunt- shunt feedback topology stabilizes Va' I , and IS thus
best sUited for transreslstance amp Ii fi ers. Figure I 0.24( a) shows the ideal structure for the
shunt- shunt feedback amplifier. It consists of a unil ateral open-loop ampl ifi er (the A CltCUIt)
an .,deal feedback network. The A circuit has an input resistance R" an open-circUli
ansreslstance A = Vol I" and an output resistance Ro' The fJ circuit samples the open-clfCUlI
10 .6 The Feedback Transresist ance Ampli fier (Shunt-Shunt ) 847
_ ___ / ...1 CtTCUIl
I
I )I R
0
+---.,-- -----:0
+
S

0 '
,------- ------1
I
I I
I I
I I
L-___ I__ >---'_ -----.:
I I
L _______ ______
{3 circuit
I
R",
o

S
R,f \ ..
S' :r---=--- -
o
(b)
Figure 10.24 (a) Ideal structure for the shunt-shunt feedback ampli fier. (b) EquIvalent CIrcuit of the
amplifier 10 (a).
output voltage Va and provides a feedback current I, that IS subITacted from the signal-source
current I, at the input nodes. ote that the fJ circuit presents an infmite impedance to the
amplifier output and thus does not load the amplifier output. Also, the feedback signal IJ = fJVo
is provided as an ideal current source, and thus the fJ circuit does not load the amplifier input.
Also observe that while A is a ITansreslstance, fJ is a transconductance and thus the loop gain
AfJ is, as expected, a dimensionless quantity. Finally, note that the source and load resistances
have been absorbed inside the A circui t (more on thi s later).
ince the structure of Fig. 10.24(a) foll ows the ideal feedback structure of Fi g. 10.1, we
can obtai n the dosed-loop gai n A J as
A
I +AfJ
(1 0.43 )
The feedback trans reSIStance amplIfier can be represented by the equivalent circuit in Fi g.
10.24(b). ote that A
J
is the open-circuit trans resistance. To obtain the input resistance R,r'
refer to the ,"put side of the bl ock dtagram m Fig. 10.24(a). The shunt connectIOn at the
mput causes the feedback current to subtract from I , resulting in a reduced current I , mto
the A circuit ,
I , = I, I,
o
o
Subslitullng If = /31 '0 = /3A I and rearrangmg, results in
J = I ,
, I + A/3
which indicates that the shunt mixmg reduces the Input current by the amount of ~ db
f d
I' fE 0 7 eeack
ThiS IS, 0 course, a Irect app IcatlOn 0 q. ( I . ), where m the case of shunt .'
h
. . . h ~ db k miXing
x, = I f and x, = I ,. T e mput resistance Wit lee ac , R'I' can now be obtained from '
V, V
R = - - '
if - I , - ( I + A/3J/ ,
Substituting for I ',> I, = R" which IS the input resistance of the A circuit, results in
R
R - '
'f - I +A/3
(10.44)
Thus, as expected, the shunt connectIOn at the input lowers the input resistance by a faclor
equal to the amount of feedback. The lowered mput resistance IS a welcome result for Ihe
transreSIStance amplifier; the lower the mput resistance, the easier it is for the signal current
source that feeds the amplifier Input
Turning our attention next to the output reSistance, we can follow an approach Identical 10
that used In the case of the senes-shunt amplifier (Section 10.4) to show that the shunt connec
tion at the output lowers the output resistance by a factor equal to the amount offeedback ,
R _ Ro
of - I + A/3
( 10.45)
This also is a welcome result for the transreslstance amplifier as It makes Its voltage-output
CirCUit more Ideal; the output voltage will change less as we draw current from the amplifier
output. Finally, note that the shun I f eedback connection, whether at Ihe input or at the Ollt
put, always reduces the corresponding resistance.
10.6_2 The Practical Case
Figure 10.25 shows a block dtagram for a practical shunt .. shunt feedback amplifier To be
able to apply the feedback equations to thts amplifier, we have to represent It by the Ideal
structure of Ftg. 10.24(a). Our obJecttve therefore is to devise a stmple method for finding
the A circutt and /3. Building on the instght we have gamed from our study of the series-
shunt and senessenes topologies, we present the method for the shunt shunt case, without
I,
Figure 10.25
,
Basic
amplifier
Feedback
network
Btock di agram for a t I h
prac lea s unt -- shunt feedback amplifier.
R
R, JI
10.6 The Feedback Transresistance Amplifier (Shunt-Shunt) 849
(a) The: A circuit I ~
I,
R,
R,
where R, is obtained from
I
RII
Feedback
network
and the gam A IS defined
\
A - -
I,
(b) f3" oblalned from
+v
I 00)
If
\ v 0
f3
B a ~ l c
amplifier
and R I ~ obtained from
Feedback
0 network
Feedback
network
+
- v"
Figure 10.26 Findmg the A cirCUIt and fJ for the shunt shunt feedback amplifier m Fig. to.25.
+
\ '

R,
R
.-
den vat IOn, In Fig. 10.26. As In prevIous cases, the method of Fig. 10. 26 assumes that the
basIc amplifier IS unilateral (or almost so) and that the feedforward transmi sSion through the
feedback network is negligibly small .
As indicated In Fig. 10.26, the A CIrcuit is obtained by Induding R, across the Input tenni-
nals of the amplifier and RL across its output temllnals. The loading effect of the feedback net-
work on the amplifier input is represented by the resistance R" and its loading effect at the
output IS represented by the resistance Rn. The value of R" is obtained by looking into port I
of the feedback network while port 2 is shorted (because it is connected in shunt ). Similarly,
R" is found by looking into port 2 while port I is shorted (because it is connected in shunt ).
Finally, observe that si nce the feedback network enses 1'0' it is fed by a voltage I ',, ; and since
It delivers a current I[ that is mixed in shunt at the input, its port I is short-circUited and /3 IS
found as I}' 1
0
, where I[ is the current that fl ows through the short circuit.
The open- loop resistances Rand Rare detennined from the A cirCUit and are used in
, "
Eqs. (10.44) and ( 10.45) to deten11lne Ru and Rot . Finall y, the resistances Roo and Rom that
characteme the feedback amplifier are obtained from Roo and Ro[ by reference to Fi g. 10.25
as fa ll ows:
Roo = I
( 10.46)
Rout
( 10.47)
Figure I 0.27(a) shows a feedback transresistance amplifier. It IS formed by connectlOg a resistance R .
I
Ofi th . ,In
the negative-feedback path of a voltage amp I ler WI gam fl. an IOput resistance R,d' and an output
resistance roo The amphfier fl can be Implemented With an op amp. a simple differential amphfier. a sin-
gle-ended inverting amplifier. or. in the limit. a single-transistor CE or CS amphfier. The laller case will
be considered in Exercise 10. 15. Of course. the higher the gain fl. the more ideal the characteri stics of the
feedback transresistance amplifier will be. simply because of the concomitantlOcrease 10 loop gam.
I,
R,
>--- -
-
-
I,

R
R,
-
-
-
-
-
-
R
(a)
-
-
(b)
- -
- -
(C)
\
R,
-
-
Vo
+
\
- -
- -
Figure 10.27 (a) A feedback IransreSi stance amplifier; (b) the !3CJrcuit; (e) the A ClfCUIt.
t
\
-
-
10.6 The Feedback Transresistance Amplifier (Shunt-Shunt) 851
(a) If the loop gam IS large. find an approximate expression for the c10s d I 0 0
fth
C db k I' fi e - oop open-clfcull transresls-
tance 0 e lee ac amp 1 ter.
(b) Find the A Clfcuit and expressIOns for A. R,. and R
o
'
(c) Find expressions for the loop gam. AI R R Rand R
. Ii' In' 0/' out'
(d) Find the values of R, . Ro. A. /3. AI' R'I' Rm. R
ol
' and RoO! for the case f.1 = 10' VN R
ro = 100 Q, RF = 10 kQ, and R, = RL = I kQ. ,,/ =
(e) If instead of a current source I having a source resistance R - I ,,-, the a I' fi . f d f . l. .f - mp 1 leT IS e rom a
voltage V, havmg a source reSistance R, = I kQ. find an expression for and the value of the
voltage gam Vol V, .
Solution
(a) If the loop gain A/3 is large.
Vo I
AI = - = -
I, /3
where /3 can be found from the /3 circuit in Fig. 10.27(b) as
Thus.
I{ I
/3=-=--
Vo R,
Vo
- =-RF
I,
(10.48)
Note that in thi s case the voltage at the input node (lhe inverting input terminal of fl) will be very close to
ground and thus very little. if any. current flows into the inputtenninal of the amphfier. Nearly all of I ,
will flow through R
F
resulting in Vo = 0 -I,RF = -I,RF' This should be reminiscent of the inverting
op-amp configuration studied in Section 2.2.
(b) Since the feedback network conSists of R,. the loading effect at the amphfier input and output will
simply be R
F
. This is IOdicated 10 the A cirCUit shown in Fig. 10.27(c). The open-loop transresistance A
can be obtained as follows:
where
R, = R'd IlRFIIR,
(R
F
II R
L
)
Vo = -flV,dro +(RFIIRL)
CombinlOg Eqs. (10.49) and (10.51) gives
A = = -f.1
R
(RFII R
L
)
I, 'ro+(RFIIRL)
(10.49)
(10.50)
(10.51)
(10.52)
The open-loop output resistance can be obtamed by inspecuon of the A circuit with I, set to O. We see that
V,d=O.and
Ro = roll RFII RL
(c) The loop gam A/3can be obtained by combining Eqs. (10.48) and (10.52),
(
R, ) (RFIIRL)
A/3= fl R" ro+( RFIIR
L
)
( 10.53)
852 Chapter 10 Feedbac k
Observe that although both A and jJ are negative. AjJ is positive, a comforting fact confirming that the
feedback is negative. Also note that AjJ is dimensIOnless, as it must always be.
The closed-loop gam A J can now be found as
f'o A
AJ = I, = 1+ AjJ
Thus
(10.54)
Note that the condition of A/l p I which results In A
J
= -RF corresponds to
(10.55)
The input resistance with feedback, R'J' is obtained by diViding R, by (I + AfJ> with the result
R _ R,
'J - I + AjJ
or
I Iij} Ii!...
-=-+ =-+
R'J R, R, R, RF
(R
F
II R
L
)
1'0 + (R
F
II R
L
)
Substituting for R, from Eq. (10.50) and replacing ,II(R
F
II R
L
) [1' 0 + (R
F
II R
L
)] by ,II', where fl ' IS
lower than but usually close to the value of ,II, results in .
R,, = R,d RF II R. II (R
F
,II' )
The two terms containing R F can be combined,
R,r= R II R,dll lR
F
I (,II'+ I)] (10.56)
Since R'J = R II R". we see that
Usually R,d IS large and thus
R = R, = '!.!:
In p' + I p'
(10.57)
from which we observe that for large amplifier gain ,II, the Input resistance Will be low.
The output resi stance with feedback RoJ can be found by dividing Ro by (I + A fJ>
R _ Ro
oJ- I +AjJ
Thus,
10.6 The Feedback Transresistance AmplIfier (Shunt-Shunt) 853
Substi tuting for Ro from Eq. (10.53),
I I I I R, I
= -+-+-+,11--
RoJ RL RF ro RFro
I I I ( R,
= - + - + - I + ,11-
RL RF 1'0 R
Thus,
Si nce, moreover,
we obtam for RoUi
from whi ch we see that for large ,II, the output resistance will be considerably reduced.
(d) For the numeri cal values given:
R, = R'd IlRFII R,
= ~ 1110 III = 0.91 kQ
Ro = ro II RFII R,
= 0. 1 1110 III = 90 Q
(R
F
II Rd
A = - ,IIR' l'o+(R
F
IIR
L
)
_ - 10' x 0 91 x (1
0
III ) = -8198 kQ
- . 0.1+(10111)
jJ= -i = -
AjJ = 819.8
I +AjJ = 820.8
I
- = -0.1 mAN
10
= A = _ 8198 = -9.99 kQ
AI I + AjJ 820.8
which is very close to the ideal value of - RF = - 10 kQ.
R, _ 910 _ 1.11 Q
R'J = I +A/l - 820.8-
I
I
=I.IIQ
R;o
- -
-
I I
I I
-
---
R'J
R,
1.11 1000
which is very low, a highl y desirable property. We also have
R
., - Ro = 90 = 0.1\ 0
0, - I +AP 820.8
I
Ro", = I I
- -
= _ ~ I ' - - - , - - = O. \I 0
I I
0. 1\ 1000
which as well is very low, another hi ghly desirable property.
(el If the amplifier is fed with a voltage source V, having a resistance R, = I kO , the output voltage can
be found from
Thus,
~ = ~ = _9.99 kO = -9.99 VN
V, R, I kO
10.15 For the transresistance ampli fie r m Fig. E10.15, replace the MOSFET with its equlvalentClrcult
model and use feedback analysIs to show the followmg:
t 1 (ideal)
_Q
R(lU'
R,
R,
- -
- - -
-
(a) For large loop gam (which cannot be achi eved here) A
, f
(b) A = -( R, II R,}gm("o II R,)
Figure E10.15
" V.l I , = - R,.
f I + (R,II Rf }gm("o II RrJI R,
R
(c) R = (
m [I +gm( roIlR,) ]
10.7 The Feedback Current Ampli f ier (Shunt-Series) 855
_ II R,
(d) Ro", -"0 1+ gm(R, II R,}
(e}Forgm= 5 rnAIV' ''o = 20 kO , R, = 10 kO , andR, = I kO , find A, p , Ap,A, R" R
o
'
R.
/
, RIOI Ro/' and R out-
Ans. (e) -30.3 kO; - 0.1 rnAIV; 3.03; -7.52 kO (compare to the Ideal value of - 10 kO );
9090; 6.67 kO; 226 0 ; 291 0 ; 1.66 kO; 1.66 kQ
---- -- - -- - -
---- -.-.--
10,6.3 An Important Note
The feedback analysis method is predicated on the assumption tbat all (or most) of the feed-
forward transmission occurs in the basic ampli fier and all (or most) of tbe feedback trans-
mIssion occurs in the feedback network. The ClfClllt considered in Exercise 10.15 above is
SImple and can be analyzed directly (I.e., without mvoking the feedback approach) to deter-
mine A,. In this way we can check the validity of our assumptions. This point is illustrated
to Problem 10.58, where we find that for the ci rcuit m Fig. E I 0.15, all of the feedback trans-
miSSIOn occurs in the feedback circuit. Also, as long as gm is much greater than 1/ R" the
assumption that most of the feed forward transmi ssion occurs in the basic amplifiers is valid,
and thus the feedback analysis is reasonably accurate.
10.7 The Feedback Current Amplifier (Shunt-Series)
As mentIOned m Section 10.3, the shunt- serIes feedback topology is best suited for current
amplifiers: The shunt connection at the mput reduces the input resistance, making it easier to
feed the amplifier with a current signal; the sampling of output current stabilizes 10 , whi ch
is the output signal in a current amplifier, and the series connection at the output increases
the output resistance, making the output current value less susceptible to changes in load
resIstance.
Figure 10.28(a) shows the ideal structure for the shunt- series feedback amplifier. It con-
sists of a unilateral open-loop amplifier (the A circuit) and an ideal feedback network. The A
circuit has an input resistance R" a short-circuit current gain A ,, 1/ Ii ' and an output resi s-
tance Ro. The p ci rcuit sampl es the short-circuit output current 10 and provides a feedback
current If that is subtracted from the signal-source current I, at the input node. Note that the
f3 circuit present s a zero resistance to the output loop and thus does not load the amphfier
output. Al so, the feedback signal If = PI
o
is proVIded as an ideal current source, and thus
the P ci rcuit does not load the amplifier input. Al so observe that both A and p are current
gains and AP is a dimensionless quantity. Finally, note that the source and load resistances
have been absorbed inside the A circuit (more on this later).
Smce the structure of Fig. 10.28(a} follows the Ideal feedback structure of Fi g. 10. 1, we
can obtain the closed-loop current gam A, as
A
I+AP
(10.59)
The feedback current ampli fier can be represented by the equi valent circuit in Fig. 10.28(b).
o
o
I,
S'
S
S'
------------ -
I I, )
t 1
J
I _
R, AI, R"
4 ClfCUlt
/'
I
I
I
, -----------,
I
I
I

L-____ I __ -'
I
I
I
I
L- _________ __
f3 Ctrcult
I I
_ _

R
f
Al l. , R ol
0'
(b)
I.
0
0'
Figure 10.28 (a) Ideal struclure for Ihe shunl-series feedback amphfier (b) Equlvalenl CIrculi ofthe
amplifier In (3).
Note that A, is the short-circuit current gain. The Input resistance R" IS found by dividing
R by (I + A/ll, "hlch IS a result of the shunt connection at the input. Thus,
R _ R,
'f- I+AjJ
(10.60)
The output resistance RoJ IS the resistance obtained by setting I = 0, breaking the shon
cirCUit output loop, at say 00' , and measuring the resistance between the two tenmnals Ihus
created. Since the series feedback connection always raises resistance, we can obtain R,( by
multiplYing Ro by (I + A/ll,
( 10.61)
10,7.2 The Practical Case
. I fi To be
Figure 10.29 shows a block diagram for a practtcal shunt senes feedback amp I ler
able to apply the feedback equattons to this amplifier, we have to represent it by the Ideal
structure of Fig. 10.28(a). Our objective therefore is to deVise a Simple method for findmg
the A and jJ CIrCUitS. Building on the insight we have gained from the study of the thfee
other topologies, we present the method for the shunt senes case without derivation, In Fig
10.30. As In prevIOus cases, the method of Fig. 10.30 assumes that the basic
lateral (or almost so) and that the feed forward transmission in the feedback network IS n g
gibly small
10.7 The Feedback Current Amplifier (Shunt-Series) 857
I,
t
R,
RiJ
Ron
Basic
amplifier
Feedback
network
10
E
Y Y'
L
I"
RoJ
RoUi
Figure 10.29 Block dIagram for a pracllcal shunl senes feedback amphfier.
(a) The A CIrculi is
I,
t
R,
R,
where R!I is obtained from
CD
R
Feedback
network
and Ihe gam A IS defined as
I
A - 2
I,
( b) f3 IS oblatned from
f3
'L
1 > 0
Basic
amplifier R"
and R" IS obtained from
Feedback
network

CD
Feedback
network
RL
y Y'
I"
Figure 10.30 Finding the A circuit and f3 for the current-mixmg current-sampling {shunt series} feedback
amplilier of Fig. 10.29.
As mdicated In Fig. 10.30. the A circuit IS obtained by Including R, across the tnput termi-
nals of the ampltfier and RL In senes with its output loop. The loading effect of the feedback
network on the ampitfier input IS represented by the resistance R II_ and ItS loading effect at
the amplifier output is represented by resistance Rn. The value of RII is obtained by look-
Ing into port I of the feedback network whtle its port 2 IS open-cirCUited (because It IS con-
nected in series). The value of R 22 IS obtained by looking into port 2 of the feedback
R
network while Its port I IS short-Clfcuited (because It IS connected In shunt ), Finally, observe
that since the feedback network senses 10 , It IS fed by a current l a' and Since It deliv
current I, that IS mixed tn shunt at the input, its port I is short-circUIted and f3 is f o u : : ~
I I where I IS the current that flows through the short cIrculI.
, " I '.
The open-loop resistances R, and Ro are deterrmned from t ~ e A cIrcuIt as indicated.
Observe that Ra IS found by breaking the output loop at say } Y and measunng the resis.
tance between Yand Y' Resistances R, and Ro are then used in Eqs. ( 10.60) and (10,61),
respecttvely. to determtne R'f and ROf Finally, the resIstances Roo and RoUi that character.
ized the feedback ampltfier are obtained from R'i and Raf by reference to Fig, 10,29, as
follows:
( 10.62)
( 10,63)
Figure 10.31 shows a feedback current amplifier formed by cascading an inverting voltage amplifier
Jl with a MOSFET Q. The output current la is the drain current of Q. The feedback network, con
sisting of resistors R, and R2 senses an exactly equal current, namely, the source current of Q. and
provIdes a feedback current signal that is mIxed wIth I at the input node Note that the bIas arrange-
ment is not shown.
-
-
'f
f
R R
-
-
/l
Q
-
I
R,
I
+
R,
R,
(a)
Figure 10.31 CirCUit for Example 10.8.
-
-
I,

\ o
I, 0
I , +
-
-
I ,
~
R,
R,
R,
-
-
(b)
---+
-
/l
- -
R,
-
-
(e)
R,
-
R id
R,
10.7 The Feedback Current Amplifier (Shunt-Selles) 859
t I"
-
-
R,
+
V,
-
- -
-
-
R
I I t -'
R,
I R.
/lV,
R.
-
R,
The amplIfier Ji can be implemented 10 a \'anety 01'\\ a)"s, includmg by means of an op amp, a dif-
ferential amplifier. or a smgle-ended m\erting amplIfier. The simplest approach IS to Implement 11
\\ Ith a CS MOSFET amplifier Howe\er. 10 such a case the loop gam \\ "I be very limited. Assume
that the amplIfier Ji has an input resistance R,d' an open-circuit voltage gain Ji, and an output
resIstance rol .
(a) If the loop gam IS large. find an approximate e"presslOn for the closed-loop gam At "loll,.
(b) Find the A circuit and derive e"presslOns for R . and R,
(c) Give e"presslOns for A p. A,. R". R '" . Ro,' and Roul '
(d) Fmd numerical values for A. p, AP. A,. R,. R.I' R,o' Ro' R"" and R OUI lor the following case'
Ji = 1000 YV. R = 00. R d = 00,1"0 1 = I kn. R = 10 kn. R, = 90 kn. and ror Q' g. 5mA
Yand I" = 20 kn
Solution
(a) When the loop gam AP <>1. Aj
Thus.
I I p. To determine P refer to Fig. 10.31 (b).
If
P
- - - - -
I ,
RI
RI + R,
I ( R,)
Af - 7J = \ I + II
( 1064)
( 10.65)
To see what happens m thIS case more dearly. refer to Fig. Ill., I (c). Here we have a"umed the loop gam
to be large. so that I, 0 and thus I {= I , Also nole that because I , ().' , \\" I be dose to zero.
Thus. we can easily determme the \oltage at the source of Q as I, R, - I R, The current through RI
Will then be I, R , I R I The source current of Q \\ III be '(/, + I ,R , I R I ) \\ hllh means that the output cur
rent 10 w1I1 be
lo=/ l I+t )
which confirms the expre"ion for A} obtamed above (Eq. 10.65).
(b) To obtam the A circuit we load the input side of the baSiC amplifier \\ Ith Rand RII The latter
in this case is simply R, + R, (because port 2 of the feedback network IS opened). We also load the
output of the basic amplifier with R ,. which In this case is R I II R, (because port I of the feedback
network is shorted). The resultmg 4 CIrCUit IS shown m Fig . IO.3i(d). where we have replaced the
amplifier Ji with its equivalent CIrcuit. AnalySiS of the A CirCUit IS stralghtlilf\\ ard and proceeds as
follows:
R = R II R,d II (R I + R,)
,- = I R
, ,
I
, . I 1',2
= 'Ji
, '1 / gm+(R11IR,II ..
0
') I"u +(R
I
IIR
2
)
Combining Eqs. (10.67) and (IO.6R) results in.4
I"
A_i, = Ji l
1'", + ( R I II R, )
(10.661
(10.67)
( I 0.68)
(10.69)
10.7 The Feedback Current Amplifier (Shunt-Series) 861
For the case I ' gm"" (R, II R, lIr
o
') '
R
A - Ji '
RI IIR,lIro'
Whi ch reduces to
A = - Ji R, (10.70)
R I II R,
Notmg that Ro is the output resistance of Q, which has a resl'stance (R II R) . ' t I d I , 10 I S source ea . we
can wnte
(c) The loop gam IS obtamed by combmmg Eqs. (10.64) and (10.69).
The mput reSistance R'f IS found as
r 0'1 R I
"u,+(RIII R,) RI+R,
R'f = R,. (I + A/ll
-'-- = .!. + :iP
R'f R, R,
(10.71)
(10.72)
(10.73)
We can substitute for AP from the full expression 10 Eq. (10.72). For the approximate case. we use AP
from Eq. ( 10.73).
That IS,
I
- =
R,
R'f = R, II...;
Ji
Subslltuting for R, from Eq. (10.66). we wnte
R,
R'f = R,II R",II (RI + R,) II Ii'
Since by defiOilion.
we can easily find Rm as
(10.74)
Usually the third component on the nght-hand side is the smallest; thus,
R,
R = -=
00 11
For the output resistance, we have
Rol = Ro( I +AjJ) = APRo
(10.75)
Subst,tuting for Ro for Eq. (10.71) and for A P from the approximate express ton tn Eq. ( 10.73), we
have
R,
R
ol
= I1R R (gm
r
o2) R,
, + ,
-
Finally, we note that
Cd) For the numerical values given,
Since I / gm=
R, = = 100kQ
0.2 kQ <t ( 10 1190 II 20) ,
R
= - 1000 100 = _ II II x 10' A' A
101190
P= _ R, =
R, + R2
10
10 + 90= - 0.1 NA
AP= 1111
A = _1111 X 10' = - 999 NA
I 1+1111
which is very close to the Ideal value of
AI = -(I = -( I = - 10 N A
R, 10
R = = 90 kQ = 90 Q
00 11 1000
R. = gmru2 (R,1I R
2
)
= 5 x 20( 101190) = 900 kQ
Ro", = (I +A/3}R. = 1112 x 900 = 1000 Mil
( 10.76)
(10.77)
-
10.8 Summary of the Feedback AnalysIs Method 863
EXERCISES
For the amplifier tn Example 10.8, find the values of A Rand R h th I f' 10
. I . f In' oul W en eva ue 0 J1 I S
limes ower, that IS when 11 = 100.
10.16
Ans. -9.9 1 NA;900 Q; 100 MQ
10.17 If in the circuit tn Fig. 10.31(a), R2 IS short-clfcuited, find the ,deal value of A For the case
R, = Rid = give expressIOns for R RAp A Rand R I
I' 0' f ' In ' out
Ans. AI= I NA,R = R, ' R = 1' 2 A = - "g R . p- I' A R '( 1 R )
, '0 0' " m ,. - -, I = I1g
m
, + I1g
m
, ;
Roo = I/ l1gm' Ro", = l1(gmro2) R,.
10.8 Summary of the Feedback Analysis Method
Table 10.1 provides a summary of the rules and relationships employed in the anal ysis and
deSign of the four types of feedback amplifier. In addition to the wealth of information tn
Table 10.1, we offer the following important analysis tips.
1. Always begin the analysis by determtning an approximate value for the closed-loop
gatn AI' assuming that the loop gain AP is large and thus
This value should serve as a check on the final value you find for AI' How close the
actual A I IS to the approximate value Will depend on how large A P is compared to
umty.
2. The shunt connecllon at mput or output always results in reducing the correspond-
ing resistance (input or output). The series connection at input or output always
results 10 mcreasmg the correspondmg resistance (input or output).
3. In utilizmg negative feedback to improve the properties of an amplifier under
deSign, the starting pomt 10 the deSign is the selection of the feedback topology
appropriate for the application at hand. Then the required amount of negative feed-
back (I + AjJ) can be ascertained utilizmg the fact that it is this quantity that deter-
mmes the magnitude of Improvement in the vari ous amplifier parameters. Also, the
feedback factor P can be determmed from
10.9 Determining the Loop Gai n
We have already seen that the loop gam A{J IS a very important quantity that characterizes a
feedback loop. Furt hermore, 10 the foll owing secti ons it will be shown that A{J determines
whether the ampli fier is stable (as opposed to oscillatory). In this section. we shall
descri be an alternative approach to the determinati on of loop gain.
o
o
C\
o
-
o
a.
{?
.-
--
-
a.
E

.><
u
'"
IJ
"0
'"
'"

:J
o

'"
.c

o
-
a.
-
.c

c
o
-
-
'"
a:
-o

'" E
E
:J
'"
o

Q/
-..Q
r:.
"
-
a:
- ,
o -

C 0
-5 1-

o
...JZ
"
r' ..
0
"" '"
- -
--
. . .
"-
'"
.; ..
-o
c.
-o
C
-
-
0
-
- -
""-
. .."
'"
+
-
Ei.
...,
+
-
-
'"
=
-
C
'j
,
"
,...
-
--
..:: - ' 1-"
""- .: 1-
."
-
..< ' .
'..J o;.Lj
o 0
.<. -
-00

" 0 _f-
-
-
-
-
'" -.
0
-
-
"
'" '" - OG r' N
. . .
- 0 0
-
- - -
Ei.
...,
+
-
-. .
'"
""-
- ."
'"
+
-
c
o
t:
o
Z
_ "'. 1-
-
-'
-
-
,

-
"
.,
0
M
0
-
-
-
-
- -.
-
-
-
'" -
C
-
-
""-
""
+
-
-
'"
Ei.
"" +
-
-.
'"
=
- c
'j
,
U

f-
0
r,
0
-
-1-
-1-

-
-
-
-
,

c

,
,
u
C
'j
/ .
-
r' .
C
-
-
" -
-
..,

-
r,
"
.
- 0
-
0
- - -
""-
<
...,
'"
+
-
""-
- ."
'"
+
-
c
o
t:
o
7
-
-
-
-
'"
r,
0
-
10.9 Determlnln9 the Loop Gain 865
10.9.1 An Alternative Approach for Finding AfJ
First, conSider again the general feedback amphfier shown In Fig. 10.1. Let the external
source x, be set to zero. Open the feedback loop by breaking the connectIOn of x to the feed-
back network and apply a test signal x" We see that the signal at the output of feedback
network is'j= /3x,; that at the input of the baSiC amplifier is x. = -/1x,; and the signal at the
output of the amplifier, where the loop was broken. will be x" = -A/1x,. It follows that the
loop gain A/3 is given by the negative of the ratio of the rellimed signal to the applied test
SIgnal; that IS, A/3 = -xolx,. [t should also be obvious that thIS applies regardless of
where the loop is broken.
However, in breaking the feedback loop of a practical amplifier circuit, we must ensure that
the conditions that existed prior to breakmg the loop do not change. This is achieved by tenni-
nating the loop where 11 is opened with an Impedance equal to that seen before the loop was
broken. To be specific, consider the conceptual feedback loop shown in Fig. 10.32(a). If we
break the loop at XX' , and apply a test voltage I to the terminals thus created to the left of
XX' , the termmals at the rIght of XX' should be loaded with an impedance Z, as shown In
Fig. 10.32(b) The impedance Z, IS equal to that previously seen looking to the left of XX'
The loop gain A/3 is then determmed from
A/3 =
/ ' ,
- -
/' ,
(10.78)
Finally, it should be noted that in some cases it may be convenient to determine A/3 by
applying a test current I, and finding the returned current Signal I,. In this case. A/3 = -I, I, .
An alternative equivalent method for determining A/3 (see Rosenstark, 1986) that is usu
ally convenient to employ especially in SPICE simulations is as follows: As before. the loop
is broken at a convenient point. Then the open-cirCUIt \oltage transfer function T IS deter-
mIned as indicated In Fig. 10.32(c), and the ,honcIrculI current transfer function T IS
determined as shown in Fig. 10.32(d) . These two transfer functions are then combmed to
obtain the loop gam A/3,
( 10.79)
This method is particularly useful when it is not eas) to determine the termination Imped-
ance l, .
To ill ustrate the process of determining loop gain, we consider the feedback loop shown
In Fig. 10.33(a) ThIS feedback loop represents both the invel1lng and the nOlllnvening op-amp
conliguratlons. Usmg a simple eqUlvalent-circLllt model for the op amp. we obtain the circuit
of Fig. 10.33(b). Examination of this circui t reveals that a convenient place to break the loop
is at the input temllnals of the op amp. The loop. broken in this manner. IS shown in Fig.
10.B(c) wi th a test signal /', applied to the nght-hand-side terminals and aresistance R, ter
minatmg the left -hand-side terminals. The returned \oltage / is found by mspectlon "'
/' = ,
pi {RIll [R,+ R,I ( R",+ R)]} [R
,
(R ,, +R)]
I { R ,. [ R,+ R I I ( R ," + R)]} + 1' 0 [R I (R ,J + R) 1 + R,
( 10.80)
This equati on can be used directl y to hndthe loop gain L ;1/3= -I, I,; I , I,
Since the loop gam L is generally a function of frequency. IllS usual to call,t loop trans-
mission and to denote it by l.( s ) or L(j UJ) .
o
o
\
I
I
, /

r
(e)
\
I

t I
,
(d )
T
,
,
' .I
Figure 10.32 A conceplual feedback loop IS broken al XX' and a test voltage I', is applied. The I m ~ '
ance Z, IS equal to that prevIously seen lookmg to Ihe left of XX'. The loop gam A/J ~ - 1:11',. wheref."
the returned voltage. As an alternative, AP can be detennmed by linding the open-CircUit transfer funcllon
1". , as in (c), and the short-cIrcUit transfer function T . as in (d), and combtnlOg them as mdicated.
Finally, we note that the value of the loop gatn determined ustng the method discussed
here may differ somewhat from the value deterrruned by the approach studied in the prell'
ous seClions. The difference stems from the approximations made 10 the feedback anal)'11
method utilized in the prevtous secttons. However, as the reader will find by soIvmg the
end-of-chapter problems, the dIfference IS usually limlled to a few percent.
10.9.2 Equivalence of Circuits from a Feedback-Loop point of
View
F
. . d denl of the
rom the study of CIfCUll theory we know that the poles of a Clfcull are m epen
I
. . ' appropnal
e
externa excItatIon. In fact the poles, or the natural modes (whIch IS a more
10.9 Determlntng the Loop Gain 867
R2
-
-
(c)
Figure 10.33 (a) A feedback loop that represents bolh the mvenmg and the nonmvenmg opamp config-
urations; (b) equivalent Clfcuit; (c) detennlnatlOn of the loop gam.
name), can be determtned by settmg the external excitation to zero. It follows that the poles
of a feedback amplifier depend onl y on the feedback loop. This wiU be confmned in a later
section, where we show that the characteristic equation (whose roots are the pole) is com-
pletely determined by the loop gai n. Thus, a given feedback loop may be used to generate a
number of ci rcuits having the same poles but different transmission zeros. The closed-loop
gam and the transmi ssIOn zeros depend on how and where the input signal is injected into
the loop.
As an example, return to the feedback loop of Fig. 1O.33(a). This loop can be used to
generate the nontnverting op-amp cirCUIt by feedmg the input voltage signal to the terminal
of R that IS connected to ground; that IS. we lift thi s terminal off ground and connect it to r',.
The same feedback loop can be used to generate the inverting op-amp circuit by feeding the
input voltage SIgnal to the terminal of RI that IS connected to ground.
Recognnion of the fact that two or more circuits are equivalent from a feedback-loop
point of view IS very useful because (as will be shown in Section 10. 10) stability is a func-
tion of the loop. Thus one needs to perform the stability analysis only once for a gIven loop.
In Chapter 16 we shall employ the concept of loop equivalence In the synthesIs of aclive
filters.
10.18
10.19
Find the loop gam A fJ for the feedback amplIfier m Fig. 10.17 (Example 1004). Set V, = 0, break
the loop at the gate of Q2' apply a voltage V, to the gate of Q2' and d e t e ~ m e the returned voltage
V, at the dram of Q" Evaluate the expression for AfJ for the values given m Example lOA and
compare to the value obtamed in Example lOA. Neglect 1' 01 and 1'02'
AfJ
= gm2RD2 RIRD; ; 16.67 (compared to 17.39 obtamed 10 Example 10.4)
Ans. ( I )
R
D2
+R
2
+ Rill R, +g
gm2 ml
Fmd the loop gain A fJ forthe feedback amplifier m Fig. E I 0.15 (Exercise 10.15) Set I , = 0, break
the loop at the gate of Q, apply a voltage V, to the gate of Q. and determme the voltage V, that
appears across R, . Find the value of AfJ using the component values given m ExerCise 10.15. and
compare to the value given in the answer to Exercise 10.15.
Ans. AfJ =
gml'oR, ; 3.22 (compared to 3.03 obtamed in Exercise 10.15)
l'o+R
f
+R,
o
o
o
10.10 The Stability Problem
10.10.1 Transfer Function of the Feedback Amplifier
in a feedback amplifier such as that represented by the general structure of Fig. 10.1, the open
loop gain A is generally a function of frequency, and It should therefore be more accurately
called the open-loop transfer function, A(s). Also, we have been assuming for the most part
that the feedback network is resistive and hence that the feedback factor fJ is constant, but this
need not be always the case. We shall therefore assume that in the general ca e the feedback
transfer function is fJ(s). It follows that the closed-loop transfer function A/s) is given by
A(s)
A/s) = I +A(s)j3(s)
(10.81)
To focus attenti on on the points central to our discussion in thiS ection, we shall assume
that the amplifier is direct coupled wi th constant dc gain Ao and with poles and zeros occur
ring in the high-frequency band. Also. for the time being let us assume that at low frequen
cies fJ(s) reduces to a constant value. Thus at low frequencies the loop gain A(s)P(s)
becomes a constant. whi ch should be a positive number; otherwise the feedback would not
be negative. The questIOn then is: What happens at higher frequencies?
For physical frequencies s = jO). Eq. ( 10.81) becomes
AIJO) - AUO) ( .
- I +AUO)fJUO) .
Thus the loop gam A(jO)fJ(jO) is a complex number that can be represented by Its mag
nl
'
tude and phase.
LU 0) '" AU O)j3u 0))
= IA U O)j3u 0)1 eJ(lUl
( 10.83)
10.10 The Stability Problem 869
It is the manner m which the loop gain vanes with frequency that determmes the stability or
instabilIty of the feedback amplIfier. To appreciate this fact, consider the frequency at which
the phase angle ( 0) becomes 180. At this frequency, 0)"", the loop gain A!}O))fJ(jO)) will
be a real number WIth a negative sign. Thus at this frequency the feedback will become pos-
llive. If at 0) = 0)"" the magnItude of the loop gai n is less than unity, then from Eq. (10.82)
we see that the closed-loop gain A, UO) wi ll be greater than the open-loop gam ..I I} 0)), since
the denominalOr of Eq. (10.82) wi ll be smaller than unity. Nevertheless, the feedback ampli-
fier will be stable.
On the other hand, if at the frequency 0),., the magnitude of the loop gain is equal to UnIty,
it follows from Eq. ( 10.82) that Aj UO) will be mfinIte. This means that the amplifier will have
an output for zero input; this is by definition an oscillator. To visualize how this feedback loop
may oscillate, conSider the general loop of Fig. 10. 1 with the external input x set to lero. Any
dISturbance in the circuit, such as the closure of the power-supply switch, will generate a Signal
x,(t) at the input to the amplifier. Such a noise signal usually contains a wide range of frequen-
Cies, and we shall now concentrate on the component with frequency 0) = 0)"", that is. the sig-
nal ,,, sin (O)' J1(, t) . This input signal will result in a feedback Signal given by
Stnce Xj is further multiplied by - I in the summer block at the tnput, we see that the feed-
back causes the Signal X, at the amplifier Input to be sustained. That is. from this point on.
there will be sinusoidal signals at the amplifier input and output of frequency 0)", . Thus the
ampli fier is said to oscillate at the frequency 0),,0-
The questIon now is: What happens if at 0)" 0 the magnItude of the loop gatn IS greater
than unity? We shall answer thi s question, not in general. but for the restricted yet very
Important cia s of circuits in which we are interested here. The answer, which is not obvious
from Eq. (10.82), is that the circuIt Will oscillate. and the OSCillations will grow 10 amplitude
until some nonlinearity (which is always present in some form) reduces the magnItude of the
loop gam to exactly unity, at which point sustained osci llations will be obtained. This mech-
anism for starting oscillations by using positive feedback WIth a loop gain greater than unity.
and then using a nonlinearity to reduce the loop gain to UnIty at the desired amplitude. will
be exploi ted In the design of sinusoidal oscillators in Chapter 17. Our objective here is Just
the opposi te: Now that we know how oscillatIons could occur in a negative-feedback amplI-
fier, we wish to find methods to prevent their occurrence.
10.10.2 The Nyquist Plot
The Nyquist plot IS a formalized approach for testIng for stability based on the discus;ion
above. It is simply a polar plot of loop gai n with frequency used as a parameter. Figure
10.34 shows such a plot. Note that the radial distance IS IAj31 and the angle is the phase
angle rp. The solid-line pl ot IS for pOSItive frequencies . Since the loop gain and for that
matter any gain function of a physical networkhas a magnItude that is an even function
of frequency and a phase that is an odd functIon of frequency. the AjJplOl for negatl\e fre-
quencies (shown m Fig. 10.34 as a broken line) can be drawn as a mirror image through the
Re aXIs.
The NyqUISt plot mtersects the negatIve real axis at the frequency w",," Thus, if th\S
IntersectIon occurs to the left of the point ( I, 0), we know that the magnItude of loop gaIn
at this frequency is greater than unity and the amplifier Will be unstable. On the other
hand, If the Intersection occurs to the right of the POInt (-I. 0) the amplIfIer Will be stable.
It follows that if the Nyquist plot enCIrcles the point ( 1,0) then the ampl lfl cr Will bc
1m
W negatIve and
. .
Increasmg m
__ - - ~ magmtude
~ ......
/ "-
/ "
/ '\
r \
I \ W o
\ , ~ ~
-,
~ - I , 0
--'
Re
W = Wl80
W posItive
and increasing
Figure 10.34 The Nyqui st plot of an unstable amplifier.
unstable. lt should be mentioned, however, that this statement is a simplified version of
the Nyquist criterion; nevertheless, it applies to all the circuits in which we are inter
ested. For the full theory behind the Nyquist method and for details of it s application,
consult Haykin (1970).
10.20 Consider a feedback ampltfier for whtch the open-loop transfer function A(s) tS gtven by
A( S)= ( 10 )3
I +sI IO'
Let the feedback factor /3 be a constant mdependent of frequency. Find the frequency wIS' at
whtch the phase shift is 180. Then, show that the feedback amplifier WIll be stable if the feedback
factor /3 is less than a cntical value /3" and unstable if /3 ~ /3", and find the value of /3".
Ans. (0180 = Jj x 10' radls; /3" = 0.008
10.11 Effect of Feedback on the Amplifier poles
The amplifier frequency response and stability are determined directly by its poles. There
fore we shall investigate the effect of feedback on the poles of the amplifier.'
6 For a brief review of poles and zeros and related concepts, refer to Appendix F.
10.11 Effect of Feedback on the Amplifier Poles 871
10.11 .1 Stabil ity and Pole Location
We shall begin by considering the relationship between stability and pole location. For an
amplifier or any other system to be stable, its poles should lie m the left half of the s plane. A
pair of complex-conjugate poles on thej(O axis gives ri se to sustained sinusoidal oscillations.
Poles in the right hal f of the s plane give rise to growing oscillations.
To verify the statement above, consider an amplifier with a pole pair at s = 0"0 j (O,. If
this amplifier is subjected to a disturbance, such as that caused by closure of the power-
supply switch, its transient response wi ll contai n tenns of the fonn
qat +) W"t -j ru"t 0"0'
v(t) = e [e + e 1 = 2e cos(w,1) (10.84 >
'"
This is a sinusoidal signal with an envelope e O. Now if the poles are in the left half of the
s plane, then 0"0 will be negative and the oscillatIons will decay exponentially toward zero. as
shown m Fig. 1O.35(a), indicating that the system is stable. If. on the other hand. the poles are m
JW
x
x
.
JW
Figure 10.35
s plane
s plane
X
x
s plane
a
a
a
-
(a>
-
(b)
- --
(c)
Relati onshIp between pole location and transient response.
Time
Time
Time
--
o
o
the nght half-plane. then 0;, will be poSitive. and the oscillations will grow exponentially (until
some nonlinearity Iimus their growth). asshown 10 Fig. 1O.35(b). Fmally. If the poles are on the
jwaxlS. then 0'0 will be lero and the oscillations will be sustamed: as shown in Fig. IO.3S(c).
Although the discussion above is in terms of complex-conjugate poles, it can be shown
that the existence of any right-half-plane poles results 10 mstabillty
10.11.2 Poles of the Feedback Amplifier
From the closed-loop transfer function 10 Eq (10.81). we see that the poles of the feedback
amplifier are the zeros of I + A(.I)P(s). That is, the feedback amphfoer poles are obtained by
sohi ng the equation
I + = 0 (10.85)
which is call ed the characteristic equation of the feedback loop. It should therefore be
apparent that apply 109 feedback to an amplifier changes Us poles.
In the followmg, we shall conSider how feedback affects the amplifier poles. For this
purpose we shall assume that the open-loop amplifier has real poles and no finite zeros (i.e.
all the zeros are at .\ = This Will Simplify the analysis and enable us to focus our attention
on the fundamental concepts involved We shall also assume that the feedback factor pis
lOde pendent of frequency
10.11.3 Amplifier with a Single-Pole Response
Consider forst the case of an amplifier whose open-loop transfer function is characterized by
a single pole:
04 0
04(5) = - =---
1+5 Wp
The closed-loop transfer functi on is given by
( \086)
04(5)= Ao/( I +AoP) (10.87)
, I + S / Wp ( I + A ofJJ
Thus the feedback moves the pole along the negati ve real axis to a frequency WPI'
w
p
! = W
p
( I + (10.88)
Thi s process is Illustrated 10 Fig. 10 36(a). Fi gure IO.16(b) shows Bode plots for IA; and
Note that whil e at low frequencies the diffe rence between the twO plolS is
20 log( I the two curves coincide at high frequencies. One can show thatthos indeed
is the case by apprOXimating Eq . ( 10.87) for frequencies w w,.< I + AullJ:
Aowp
A,(s)= -A(s)
s
(10.891
PhYSicall y speaking, at such hi gh frequencies the loop gam is much small er than umt) and the
feedback is ineffecti ve.
Fi gure 10 36(b) cl earl y Illustrates the fact that appl ymg negati ve feedback to an ampli
fier result s in extendong It s bandWidth at the expense of a reducti on In gam. Since the pole 01
the closed-loop amplifier never enters the ri ght half of the .\ plane, the single-pole amplifier
IS stable for any value of p. Thus thos amplifier is said to he unconditionally Itable. 1111\
10.11 Effect of Feedback on the Ampl ifi er Poles 873
JW dB
20 log (I + Aom
s plane
IAol ----
( o

C1
Wp( wp( I + Aom
Wp
WPf W (log scale)
(a) (b )
FIgure 10.36 Effect of feedback on (a) Ihe pole location and (b) Ihe frequency response of an amplifi er
having a smgle-pole. open-loop response.
resull, however, is hardly sorprising, since the phase lag associated with a single-pole response
can never be greater than 90. Thus the loop gain never achieves the 180 phase shift required
for the feedback to become positive.
r __ ___ _
10.21 An op amp having a single-pole rolloff at 100 Hz and a low-frequency gam of 10' is operated m
a feedback loop with P= 0.01 What is the factor by which feedback shifts the pole? To what fre-
quency? If PIS changed to a value that results m a closed-loop gam of + I , to what frequency does
the pole shift?
Ans. 1001,100.1 kHz; 10 MHz
10.11.4 Amplifier with Two-Pole Response
Consider next an amplifier whose open-loop transfer function is characterized by two real-
aXIs poles:
Ao
( 10.90)
A(s) = (I +s/wPI)(1 +s/ WP2)
In this case, the closed-loop poles are obtaoned from I + A(s)P= 0, which leads to
/ + s(W
pl
+ w
p
,) + (I + Ao/l>wpi WP2 = 0 (10.91)
Thus the closed-loop poles are gi ven by
S = w
pi
+ WP2) J( W
PI
+ Wp,)' - 4( I + Ao/l> WPI WP2 (10.92)
From Eq. ( 10.92) we see that as the loop gain AoP is increased from zero, the poles are
brought closer together. Then a value of loop gain IS reached at which the poles become
coincident. If the loop gain is further Increased, the poles become complex conjugate and
move along a vertical line. Fi gure 10.37 shows the locus of the poles for mcreasmg loop
gai n. Thi s pl ot is called a root-locus di agram, where "root" refers to the fact that the poles
are the roots of the characteri sti c equati on.
o
.
JW
t
;.
t
..
-Wn
WPI
WPI + WP2
2
s plane
0 a
Figure 10.37 Rool-iocus diagram for a feedba,l
amplifier whose open-loop transfer funchon has Ill'
real poles.
From the root-locus diagram of Fig. 10.37 we see that this feedback amplifier also is
unconditionally stable. Again, thts result should come as no urprise; the maximum phase
shift of A(s) in this case is 180 (90 per pole), but this value tS reached at 1iI= 00. Thus there
is no finite frequency at which the phase shift reaches 180.
Another observation to make on the root-locus diagram of Fig. 10.37 is that the open
loop amplifier might have a dominant pole, but this is not necessari ly the case for the closed
loop amplifier. The response of the closed-loop amplifier can, of course, always be ploued
once the poles have been found from Eq. (10.92). As is the case with second-orderresponses
generally, the closed-loop response can show a peak (see Chapter 16). To be more specific,
the characteristic equation of a second-order network can be written In the standard form
( 10.93)
where % is called the pole freq uency and Q is called pole Q factor . The poles are complex
if Q is greater than 0.5. A geometric interpretation for % and Q of a paIr of complex-conjugate
poles is given in Fig. 10.38, from which we note that % is the radtal di stance of the poles
from the origin and that Q indicates the distance of the poles from the jlil aXIs. Poles on the
jlilaxis have Q = 00.
By comparing Eqs. (10.91) and (10.93) we obtain the Q factor for the poles ofthefeed
back amplifier as
.
JW

I w.
I
I
I
I
I
I
*
WJL
2Q
, pl ane
o
a
Q = J( I + AofJJ COpl COP2
COPI + CO
P2
( 10.94)
Figure 10.38
conjugate poles.
. . of
Definilion of % and Q of a pair
- po
" -
I
0
-
0
'"
_- Q 0.707 (maximally flat response)
"

-

E
0 - 12 dBloctave
Q 0.5
z
Q 0.3
w (log scale)
Figure 10.39 Normalized gain of a two-pole feedback amplifier for vanous values of Q. NOle thai Q IS
determined by the loop gain according to Eq. (10.94).
From the study of second-order network responses in Chapter 16, it will be seen that the
response of the feedback amplifier under consideration shows no peaking for Q $ 0.707. The
boundary case corresponding to Q = 0.707 (poles at 45 angles) results in the maximally
flat response. Figure 10.39 shows a number of possible responses obtained for various val-
ues of Q (or, correspondingly, various values of AofJJ.
10,22 An amplifier with a low-frequency gam of 100 and poles at 10' radls and 10' radls IS mcorporated
10 a negative-feedback loop with feedback factor p. For what value of Pdo the poles of the closed-
loop amplifier COinCide? What is the corresponding Q of the resulung second-order system? For
what value of P IS a maXimally flat response achieved? What is the low-frequency closed-loop
gain in the maximally flat case?
Ans. 0.245; 0.5; 0.5; 1.96 VN
As an Illustration of some of the Ideas Just di scussed, we consider the positive-feedback circuit shown in
Fig. 10.40(a). Find the loop transmi ssion L(s) and the characteristic equation. Sketch a root-locus diagranl
for varying K, and find the value of K that results in a maximally flat response and the value ofK that
makes the Circuit oscillate. Assume that the amplifier has frequency-idependent gam, mfintte mput Imped-
ance, and zero output impedance.
Solution
To obtain the loop transmission, we short-circuit the signal source and break the loop at the ampli fier
input. We then apply a test voltage V, and fi nd the returned voltage V" as indicated in Fig. 10AO(b). The
+
\ .
-
-
-
-
A
Q
R
+};
(a )
K = I Sg6
{I 0 707
o
-
1
A
{I
R
c
\
JW
'C)
\
" j
{I ~
I
-
-
l 11ane
" 1
{I
t
\ .
-
-
Figure 10.40 CirCUli' and plot for hamplc 10.9.
Lt,)
I ,
-
I I
Kn,)
r---------,
I R e I
I
I I
R
I I
L _ ~ ___ ~ __ ~
\
-
-
(b)
(10.951
where Tel) is the transfer funeti f h' . F'
, on 0 t t.: two-port RC network .... hown m ... itlc the broken-line nnx iO Ig.
I040(b):
Tel ) -
r
I
s ( I f eR) (10.96)
.1 + .1( 3 I C R ) + ( I f C R )'
Thus.
L(s) = -s(KICR)
s' + s( 3/CR) + (I /CR)'
(10.97)
The charactenstlc equatIon IS
I + L(s ) = 0 (10.98)
that IS,
, 3 ( I )' K
s +s + -s = 0
CR CR CR
, 3 -K ( I )'
S +s CR + CR = 0
(10.99)
By comparing this equatIOn to the standard form of the second-order characteristic equallon (Eq .
10.93) we see that the pole frequency % is given by
and the Q factor IS
I
% = CR
I
Q = 3- K
(10. 100)
(10. 101 )
Thus for K = 0, the poles have Q = 1 and are therefore located on the negative real axis. As K is
increased, the poles are brought closer together and eventually coincide (Q = 0.5, K = I). Further increas-
109 K results 10 the poles becoming complex and conjugate. The root locus is then a circle because the
radial distance % remains constant (Eq. 10.100) mdependent of the value of K.
The maximally flat response IS obtamed when Q = 0.707, which results when K = 1.586. In this case
the poles are at 45 angles, as indicated in Fig. 10.40(c). The poles cross thejlOaxis into the right half of
the s plane at the value of K that results 10 Q = ~ , that is, K = 3. Thus for K > 3 this CIrCUit becomes unsta-
ble. This might appear to contradict our earlier conclusion that the feedback amplifier with a second-order
response IS unconditionally stable. Note, however, that the circuit in this example is quite difTerent from
the negallve-feedback amplifier that we have been studying. Here we have an amplifier with a positive
gam K and a feedback network whose transfer functIon T(s) is frequency dependent. This feedback IS III
fact positive. and the circuit will oscillate at the frequency for which the phase of T(jlO) is zero.
Example 10.9 illustrates the use of feedback (positi ve feedback in this case) to move the
poles of an RC network from their negative real-axis locations to complex-conjugate loca-
tlons. One can accomplish the same task usi ng negati ve feedback. as the root-locus diagram
of Fig. 10.37 demonstrates. The process of pole control is the essence of active-filter design.
as will be discussed in Chapter 16.
10.11.5 Amplifiers with Three or More Poles
Figure 10.41 shows the root-locus diagram for a feedback amplifier whose open-loop
response IS characteri zed by three poles. As indicated. increasing the loop gain from lero
moves the hi ghest-frequency pole outward while the twO other poles are brought closer
together. As A"fi is increased further, the twO poles become coincident and then become
complex and conjugate. A value of A,fiexlSts at which thiS pair of complex-conjugate poles
enters the ri ght half of the s pl ane, thus causmg the amplifier to become unstable.
10.23
.
JW
( ) (
s plane
o a
Figure 10.41 Root-locus diagram for
an amplifier with three poles. The arro\\ S
mdlcate the pole movement as AoP IS
Increased.
ThtS result is not enttrely unexpected. since an ampitfier with three poles has a phase
shift that reaches -270 as W approaches Thus there eXISts a I' lOtte frequency. at
which the loop gain has 180' phase shift.
From the root-locus diagram of Fig. 10.41. we observe that one can always maintain
amplifier stability by keeptng the loop gam AofJ small er than the value corresponding to the
poles entering the right half-plane. In terms of the Nyquist diagram. the cntical value of AoP
is that for which the diagram passes through the (- J. 0) point. Reducing Ao/l below this
value causes the Nyquist plot to shrink and thus tntersect the negative real axts to the nght of
the (-I. 0) pOtnt. tndicattng stable amplifier performance On the other hand. increasingA.,p
above the critical value causes the Nyquist plot to expand. thus enctrcling the (-1.0) point
and tndtcattng unstable performance
For a given open-loop gatn A" the conclusions above can be stated 10 terms of the feed
back factor fJ. That IS. there exists a lIIa.mnulII l'alue for fJ above which the feedback ampli
fier becomes unstable Alternatively. we can state that there exists a IIIl1lillltllll \'altle for the
closed-loop gatn AjU below whtch the amplifier becomes unstable To obtain lower values of
closed-loop gatn one needs therefore to alter the loop transfer function L(s). ThIS is the pro
cess known as !requellc\ compensalion. We shall study the theory and techniques of fre
quency compensation in Section 10. 13.
Before leaving thts sectton we point out that constructton of the root-locus diagram for
ampitfiers havtng three or more poles as well as fintte zeros tS an tnvolved process for which
a systematic procedure exists. However. such a procedure will not be presented here. and the
tnterested reader should consult Haykm (1970). Although the root-locus dtagram provides
the amplifier designer with conSiderable insight. other. 'tmpler techniques based on Bode
plots can be effecttvely employed. as will be explatned in Section 10. 12.
Consider a feedback amplifier for which the open-loop transfer functton 4(5) is gtven by
I
A( .I) ( 10 )3
1+5110'
10.12 Stability Study Using Bode Plots 879
Let the feedback factor pbe frequency tndependent. Find the closed-loop poles as functtons of P .
and show that the root locus is that of Fig. EI0.23. Also find the value of pat which the amplifier
becomes unstable. (Nole. This IS the same amplifier that was considered m Exercise 10.20.)
JW
60'
60
Ans. See Fig. EIO.23; Pm'". = 0.008
s plane (normalized
to 10' rad/s)
o
10.12 Stability Study Using Bode Plots
10.12.1 Gain and Phase Margins
Figure 10.E23
.- -
From Secttons 10. 10 and 10. II we know that whether a feedback amplifier is or is not sta-
ble can be determined by examimng tts loop gain A/l as a function of frequency. One of
the simplest and most effective means for doing this is through the use of a Bode plot for AfJ.
such as the one shown in Fig. 10.42. (Note that because the phase approaches -360. the net-
work examined is a fourth-order one.) The feedback amplifier whose loop gain is plotted in
Fig. 10.42 will be stable. since at the frequency of 180 phase shift. WIS.' the magnitude of the
loop gam tS less than unity (negative dB). The difference between the value of IA fJl at wlSO
and unity. called the gain margin. is usually expressed in decibels. Thegainmargin repre-
sents the amount by which the loop gain can be mcreased whtle stabthty tS mmntamed.
Feedback ampltfiers are usually deSigned to have sufficient gain margin to allow for the
inevttable changes in loop gain with temperature, time. and so on. .
Another way to investigate the stabiltty and to expres its degree is to examttle the Bode
plot at the frequency for which IA fJl = I. which is the POttlt at whtch the magnttude
crosses the O-dB line. If at this frequency the pha e angle is less (ttl magnttude) than 180
then the amphfier IS stable. This is the si tuation illustrated in Fig. 10.42. The .dtfference
between the phase angle at thi s frequency and 180 is termed the phase. margttl . On th:
. t de the phase lag tS m excess of 180 .
other hand. tf at the frequency of untty loop-gam mag"' u .
the ampitfier will be unstable.
10.24
IAPI dB
Gain margin
Wl80
0-----
w, I
: 1\1
I I
I I
I I
w (log scale)
w, I I
o - - ; ; ; , _ ~ : . c --tl-:- w:- ,,-o----w-(-Io-g-s-c-ale)
~
o
" -
00
"
'"
"
~
'"
-90
-180
.c -2700
0..
-360
I
I
t
Phase margin
Figure 10.42 Bode plot for the loop gam AfJ ,lIustratmg the defimttons of the gam and phase margins
Constder an op amp having a Single-pole. open-loop response with Ao = 10 and,t;,= 10 Hz. Let
the op amp be Ideal otherwtse (tnfintte tnput impedance, zero output impedance, etc.). If this am
plIfier tS connected 10 the nonlnverting configuration with a nomtnallow-frequency, closed-loop
gatn of 100, find the frequency at which IAPI = I. Also, find the phase margtn.
Ans. 10' Hz; 90
10.12.2 Effect of Phase Margin on Closed-loop Response
Feedback amplifiers are normally designed with a phase margin of at lea t 45. The amount
of phase margin has a profound effect on the shape of the closed-loop gain response. To
see thIS relationship consid f db k . . . .
, er a ee ac amplIfIer with a large low-frequency loop gam,
A,j3 jl> I. It foll ows that the closed-loop gain at low frequencies is approximately liP Deno
t

109 the frequency at whIch th " F' e magnttude of loop gam is unity by OJ we have (refer to Ig.
10.42) ,.
(10.102a)
where
() = 180 - phase margi n
(IO.IOlb)
At OJ, the closed-loop gain IS
Afj 0J, 1 = -;--A--,Ue.,.OJ -,-,I.!..,l---:
I +AUOJ,l/l
Substituting from Eq. (l0. 102a) gives
Afj 0J, )
Thus the magnitude of the gain at w, is
= (1/{J)e-)8
I + e -)8
I l /l
II+e')81
For a phase margin of 45, ()= 135; and we obtain
10.12 Stability Study USing Bode Plots 881
(10103)
(10.104)
(10105)
(10.106)
That is, the gain peaks by a factor of 1.3 above the low-frequency value of 1/ fl. This peaktng
increases as the phase margin is reduced, eventually reaching ~ when the phase margin is
zero. Zero phase margin, of course, implies that the amplifier can sustain oscillations [poles
on thejwaxis; Nyquist plot passing through (-1, 0)].
. -. -_. ----
EXERCISE
10.25
Find the closed-loop gatn at W, relative to the low-frequency gatn when the phase margtn IS 30.
60, and 90.
Ans. 1.93; 1,0.707
10.12.3 An Alternat ive Approach for Investigating Stability
Investigating stability by constructing Bode plots for the loop gain AP can be a tedious and
time-consuming process, especially if we have to investi gate the stability of a given amplifier
for a variety of feedback networks. An alternative approach. which is much simpler, is to
construct a Bode plot for the open-loop gain A(jW) only. Assuming for the lIme being that P
IS tndependent of frequency, we can plot 20 log( 11/3) as a horizontal straight line on the same
plane used for 20 10g1AI. The difference between the twO curses will be
2010g1AUOJl l- 2010g1 = 2010g1A/l1
(10.107)
which IS the loop gain (in dB). We may therefore study 5tabllit) by examIning the difference
between the two plots. If we wI&h to evaluate stabihty for a different feedback factor. we SIm-
ply draw another hori zontal straight line at the level 20 log( 1/ /l).
To illustrate, consider an amplifier whose open-loop transfer function is characterited by
three poles. For simplicity let the three poles be widely separated-say, at 0.1 MHz, 1 MHz,
and 10 MHz, as shown in Fig. 10,43. Note that because the poles are widely separated. the
10
10
10'
dB
100
90

20 log
80
X,
,....:-----,- (a)
I
85 dB (stable)
I 25 dB gain margin
70
20 log I
60 ----------+--
for lero margins I I /X, -40 dB/decade
(b)
40
30
20
10
o
10'
'"
20 log I
I I
50 dB (unstable) I I
10' 10'
I
I I I
I I I
I I I
I I I
I ./, .. I I
I ' I I
10'
I, I I
, I
10'
I
I
60 dB/decade
,
10 I( Hl )
10' 0 10' 10' I I 10'
I I
10' I( Hl)
- 45'
-90"
- 135'
-180'
- 270'
I I
108'
,---_1 1 ___
------------
--
I 72 pha,e
I
Figure 10.43 Stability analySis using Bode plot orlAl.
phase is approxtmately -45' at the first pole frequency, 135 at the second, and -225' at
the third. The frequency at which the phase of A<JaJ) tS - 180 Ites on the -40dB/decade
segment, as mdicated in Fig. 10.43
The open-loop gain of thi s amplifier can be expressed as
A =
10'
(I + jf 10 )( 1+ if 10' )( I +/f 10' )
( 10.108)
from whtch IAI
obtamed as
can be eastly determined for any frequency I (In Hz), and the phase can be
(10.109)
The magnitude and phase graphs shown in Fig. 10.43 are obtained using the methOd for
constructing Bode plots (Appendix F). These graphs provide approXImate values for
10.12 Stability Study Using Bode Plots 883
important amplifier parameters, with more exact values obtainable from Eqs. (10. 108) and
(10.109). For example, the frequency j,,,, at which the phase angle is 180 can be found from
Fig. 10.43 to be approximately 3.2 X 10' Hz. Using this value as a starting pOint, a more
exact value can be found by trial and error using Eq. (10. 109). The result isj,1l = 3.34 x 10'
Hz. At this frequency, Eq. (10.108) gives a gain magnitude of 58.2 dB, which is reasonably
close to the approximate value of 60 dB given by Fig. 10.43.
Consider next the straight line labeled (a) in Fig. 10.43. This line represents a feedback
factor for which 20 log( 1I,Bi = 85 dB, which corresponds to f3 = 5.623 x 10-' and a c1osed-
loop gain of 83.6 dB. Since the loop gain is the difference between the IAI curve and the 1If3
line, the pOint of intersection X, corresponds to the frequency at which IAf31 = I . Using the
graphs of Fig. 10.43, this frequency can be found to be approximately 5.6 x 10' Hz. A more
exact value of 4.936 x 10' can be obtained using the transfer-function equallon . At thIS
frequency the phase angle IS approximately -108. Thus the closed-loop amplifier, for
which 20 log( 1/,Bi = 85 dB, will be stable WIth a phase margin of 72. The gam margin can
be easily obtained from Fig. 10.43; it is 25 dB.
ext, suppose that we wish to use thi amplifier to obtain a closed-loop gain of 50-dB
nonunal value. Since Ao = 100 dB, we see that Aof3 I and 20 10g(AofJ) = 50 dB, resuillng
in 2010g(l/,Bi = 50 dB. To see whether this closed-loop amplifier is or is not stable, we
draw line (bl in Fig. 10.43 with a height of 50 dB. This line intersects the open-loop gain
curve at point X" where the corresponding phase is greater than 180. Thus the closed-loop
amplifier with 50-dB gain will be unstable.
In fact, it can easily be seen from Fig. 10.43 that the minimum value of 20 10g(l/,Bi that
can be used, with the resulting amplifier being stable, is 60 dB. In other words, the minimum
value of stable closed-loop gain obtained with this amplifier is approximately 60 dB. At thIS
value of gain, however, a manufactured version of this amplifier may still oscillate, since no
margin IS left to allow for possible changes in gain.
Since the 180-phase point always occurs on the -40-dB/decade segment of the Bode
plot for IAI, a rule of thumb to guarantee stability IS as follows: The closed-loop amplifier
will be stable if the 20 log( 1/,Bi line illtersects the 20 10giAI curve at a point 011 the - 20-dBI
decade segment. Followmg this rule ensures that a phase margin of at least 45 is obtained.
For the example of Fig. 10.43, the rule implies that the maximum value of f3is 10 .... , which
corresponds to a closed-loop gain of approxImately 80 dB.
The rule of thumb above can be generalized for the case in which f3 is a funcllon of
frequency. The general rule states that at the intersection of 20 log [I 'l flU aJ)l) alld
20 10giA U aJ)1 the difference of slopes (called the rate of closure) should 1I0t exceed
20 dB/decade.
-. -
10.26
Consider an op amp whose open-loop gam is idenucal to that of Fig. 10.43. Assume Ihat the op
amp IS Ideal otherwise. Let the op amp be connected as a dlfferentiator Use the rule of
above to show that for stable performance the d,fferenllator time con slant should be greater than
159 ms. [HII1I Recall that for a differentiator, the Bode plot for I lf3uwll has a slope of +20
dB/decade and intersects the O-dB hne al IIf, where TIS the dlfferenuator ume conslanL]

10.13 Frequency Compensation
In thi s sectlon. we shall discuss methods for modifying the open-loop transfer function Als)
of an amplIfier havmg three or more poles so that the closed-loop amplIfier is stable f
Wa
given desired value of closed-loop gam.
dB
Y' l'
100
,
I
20 dB/decade
,
I ,
I
,
, A"
I ,
I
,
,
I ,
I 40 dB/decade ,
A'
,
I
60 ,
I '
I "
20 log 1/(3 40 dB Iy
,
l
40
I \ - 60 dB/decade
I
\
\
I
\
20 I
\
I \
I \
0
I \
.
10
'
10 10- 10' 10' 10' 10 10'
I (Hz)
t t t t
t\
\
If) Ii, fp, Jp Jp \
Figure 10.44 Frequency compensal,on for f3= 10 The response labeled ." IS oblamed by lfllroducmg
an add!llonal pole al/D The r response IS oblamed by moving Ihe onglnallow-frequency pole 10 I';
10.13.1 Theory
The simplest method of frequency compensatIOn conSISts of mtroducing a new pole in the
A(s) at a sufficiently low frequency. f
o
such that the modified open-loop gain.
A (s). Intersects the 20 log( 1/1.01) curve wllh a slope difference of 20 dB/decade As an
example. let it be reqUired to compensate the amplifier whose A(s} IS shown," Fig. 10.44
such that closed-loop amplifiers with .0 as high as 10 - (i.e .. closed-loop gains as 10_ :I>
approximately 40 dB) wIiI be stable. First. we draw a horizontal straight line at the 40-dB
level to represent 20 log( II .0). as shown in Fig. 10.44. We then locate pomt } on this line al
the frequency of the first pole. f
p
,' From Y we draw a hne with -20-dB/decade slope and
determine the point at which this hne intersects the dc gam Ime, pomt Y' This latter poml
gives the frequency fo of the new pole that has to be mtroduced in the open-loop transfer
function.
The compensated open-loop response A'(s) IS indicated m Fig. 10.44. It has four
poles: atfoJ pl'it'1' andfPJ' Thus IA1 begins to roll off with a slope of -20 dB/decade at!o'
Atf" the slope changes to -40 dB/decade. at!n it changes to -60 dB/decade. and so on.
Smce the 20 log( IIfJ} line intersects the 20 loglA1 curve at point Yon the -20-dB/decade
segment. the closed-loop amplifier with this {3value (or lower values) will be stable.
A serious disadvantage of this compensatIOn method is that at most frequencies the open-
loop gam has been drastically reduced. This means that at most frequencies the amount of
feedback available will be small. Since all the advantages of negative feedback are directly
proportional to the amount of feedback. the performance of the compensated amplifier will
be impaired.
Careful examination of Fig. 10.44 shows that the gain A'(s) is low because of the pole at
I.,. If we can somehow eliminate this pole. then-rather than locating point Y. drawmg I Y'.
and so on-we can Starl from point Z (at the frequency of the second pole) and draw the hne
ZZ' This would result in the open-loop curve A"(s). which shows considerably higher gain
than A'(s).
Although It IS not possible to ehminate the pole atfpl' it IS usually possible 10 shift that
pole fromf = h, to f = f 6. This makes the pole dominant and ehminates the need for intro-
ducing an additional lower-frequency pole. as will be explained next.
10.13.2 Implementation
We shall now address the question of implementing the frequency-campen at ion scheme
discussed above. The amplifier circuit normally consists of a number of cascaded gain
stages. with each stage responsible for one or more of the transfer-function poles. Through
manual andlor computer analysis of the circuit. one identifies which stage introduces each of
the Important polesh,Jn ' and so on. For the purpose of our discussion. assume that the first
pole f p, is introduced at the interface between the two cascaded differential stages shown m
Fig. 10.45(a}. In Fig. 10.45(b} we show a simple small-signal model of the circuit at thi s
tnterface. Current source I represents the output-signal current of the Q,-Q, stage. Resis-
tance R, and capacitance C, represent the total resistance and capacitance between the two
nodes Band B' . It follows that the polefp, is given by
I
(10.110)
/P, = 2nC R
, ,
Let us now connect the compensating capacitor C, between nodes Band B'. This will
result in the modified equivalent circuit shown in Fig. 10.45(c) from which we see that
the pole introduced will no longer be at fp,; rather. the pole can be at any desired lower
frequency!; :
r' - I
JD - C
2mC,+ c) R,
(IO.III)
We thus conclude that one can select an appropriate value for C( to shift the pole frequency
from!PI to the value!; determined by point l' in Fi g. 10.44.. .
At this juncture it should be pointed Ollt that adding the capacitor Cc Will usually result
in changes in the location of the other poles (those atf!"! andj.,). One nughttherefore need to
calculate the new locati on of r and perform a few iteratIons to arrive at the reqUired value
In
fo r C(""
I
I
I
I
I
I
B' t-------------+---,

-----
-------------'
t
B'
B
(bl
Rx
I
I
-'-- Cx
<a)
B'
R,
B
<e)
C,
I
I
Figure 10.45 <a) Two cascaded gam stages of a muillstage amplifi er. (b) Equi valent CIrculi for the inter
face bety,:een the two stages In (a). (e) Same circuit as In (b), but with a compensatmg capacitor C( added_
Note that the analySIs here applies equally well to MOS amplifiers,
A disadvantage of this implementatIOn method tS that the reqUired value of C
c
is usuall)
quite large. Thus If the amplifier to be compensated is an IC op amp, it will be difficult, and
probably Impossible, to include this compensating capacitor on the IC chip. (As pointed OUI
in Chapter 7 and in Appendix A, the maximum practical size of a monolithic capaCilor IS
about 100 pF.) An elegant solution to this problem is to connect the compensating capaCHor
in the feedback path of an amplifier stage. Because of the Miller effect, the compensall ng
capacitance will be multiplied by the stage gain, resulting in a much larger effective capaci'
tance. Furthermore, as explained later, another unexpected benefit accrues.
10.13.3 Miller Compensation and Pole Splitting
Figure 10.46(a) shows one gain stage 10 a multistage amplIfier. For Simplicity, the Slage IS
shown as a common-emitter amplifier, but 10 practice It can be a more elaborate WCUIl. In
the feedback path of thiS common-emitter stage we have placed a compensating capacitor C,
Figure 10.46(b) shows a Simplified equivalent circuit of the gain stage of Fig. 10.46131,
Here R, and C, represent the total resistance and total capacitance between node Band
ground. Similarly, R, and C, represent the total resistance and total capacitance between node
C and ground. Furthermore, it is assumed that C includes the Miller component due 10
. C I . I'f! stage
capacitance Jft and C, IOcludes the input capacitance of the succeed 109 amp J ter
Finally, I, represents the output signal current of the preceding stage.
c
--
B
c
+
C, R, V.
R, --
B
-
-
-
-
-
-
-
-
-
-
- -
-
-
-
<al (bl
Figure 10.46 (a) A gam stage In a multistage amplifier with a compensating capacitor connected in the
feedback path, and (b) an equivalent circu,t. ate that although a BJT IS shown, the anal ySIS applies equally
well to the MOSFET case,
In the absence of the compensating capacitor Cf' we can see from Fig. 10.46(b) that
there are two poles one at the input and one at the output. Let us assume that the e two
poles arefp, andf", of Fig. 10.44; thus,
I
.!PI = 2nC
I
R
I
(10.112)
With C, present, analysis of the circuit yields the transfer functi on
_ (sC,-gm)RIR,
T, - I + s[CIR
I
+ C,R, + C/gmRIR, + RI + R,) + s'[C, C, + C/C, + C, )R, R,
(10.113)
The zero usually at a much higher frequency than the dommant pole, and we shaJl neglect
ItS effect. The denominator polynomial D(s) can be written in the form
,
D(s) = (I + )(1 + = I +s( I, + (LII, ) + {LI'
(J)p I {J)P2 wp I P2 PI P2
(10.114)
where {LI' and {LI' are the new frequencies of the two poles. Normally one of the poles will
PI P2
be dominant; <!; {LI;,. Thus,
D(s) =
s S'
1+,, + //
{LIp I {LIp I (LIn
(10.115)
Equating the coeffi cients of s in the denominator of Eq. (10.113) and in Eq. (10. 115) results
In
I ,
(LIp, = CI RI+C, R, +Cj (gmRIR,+RI +R,)
whi ch can be approximated by
,
{LIp I
(10.116)
o
o
To obtain W;, we equate the coefficients of s' in the denominator of Eq. (10.113) and
Eq.(l0.IIS)anduseEq.(10.116): In
gmCr
,.,' -
U/P2 -
(10. 117)
From Eqs. (10.116) and (10.117), we see that as C, IS mcreased, W;I IS reduced and 0/ .
increased. This action is referred to as pole splitting. Note that the mcrease In m;, is
beneficial; It allows us to move pOint Z (see Fig. 10.44) further to the nght, thus resuhmg
m
!ugher compensated open-loop gam. Fmally note from Eq. (l0 I 16) that C, is multiplied by
the MIller-effect factor g.,R" thus resultmg 10 a much larger effective capacitance, g),C,
In other words, the reqUIred value of C, wIll be much smaller than that of C
c
in Fig. 10.45.
Consider an op amp whose open-loop transfer functIOn IS Identical to that shown In Fig. 10.43. We wish to
compensate this op amp so that the closed-loop amplifier with resisti,e feedback is stable for any gam
(i.e., for /3 up to unity). Assume that the op-amp circuit includes a stage such as that of Fig. 10.46 with C
= 100 pF. C, = 5 pF, and g. = 40 mAN, that the pole atl
Pl
is caused by the mput cirCUIt of that stage, and
that the pole atlP2 is introduced by the output circuit. Find the value of the compensating capacitor for
cases: either ifit is connected between the mput node B and ground. or 10 the feedback path of the tranSlS
tor.
Solution
First we deterrmne R. and R from
Thus,
Thus,
10'
2tr n
10'
R, = n
- tr
[f a compensatmg capacitor C, IS connected across the mput termmals of the transistor stage, then the
frequency of the first pole changes from / . to j"
. PI Do
f
' I
lJ =
2J!(C
I
+ Cc)R
I
The second pole remams unchanged at I-MHz The reqUired value for If, IS determmed by draw 109 a -20
dB/decade Ime from the I-MHz frequency pomt on the 20 log(l /3) 20 log I 0 dB line. This hne will
mtersect the 100-dB dc gam Ime at 10 HI Thus
,
rf,=IOHz= I
2J!(C
I
+ Cc)R
I
which results 10 Cc = I which IS quttelarge and certamly cannot be mcluded on Ihe Ie chip.
Next, If a compensatmg capacitor C, IS connected In the feedback path of the transistor then both
poles change locatIOn to the values given by Eqs. (10. 116) and (10.117): '
1; 1 = I I;, = gm
C
[
2n[C, C,+C[(C
I
+C,)] (10.1\8)
To determme where we should locate the first pole, we need to know the value of JP,. As an approxima-
tion, let us assume that C[ p C" whIch enables us to obtam
r' gm
}P, = 2n(C
I
+ C,) = 60.6 MHz
Thus it appears that this pole will move to a frequency higher thanl" (which is 10 MHz). Let us therefore
assume that the second pole wIll be at 1M' ThiS requires that the first pole be located at
Ipi = k1 = 10
7
Hz = 100 Hz
Ao 10'
Thus,
Ipl=100HZ=2 I

which results 10 C, = 78.5 pF. Although this value is mdeed much greater than C" we can determine the
location of the pole}p, from Eq. (10. 118), which yields If, = 57.2 MHz, that this pole has
mdeed been moved past/Pl .
We conclude that using Miller compensation not only results in a much smaller compensating capac-
Itor but, ow 109 to pole splittmg, also enables us to place the dommant pole a decade higher m frequency.
ThiS results 10 a wider bandwidth for the compensated op amp.
EXERCISE
10.27 A multi pole amplifier havmg a first pole at I MHz and an open-loop gam of I 00 dB is to be com-
pensated for closed-loop gams as low as 20 dB by the introduction of a new dominant pole. At
what frequency must the new pole be placed?
Ans. 100 Hz
10.28 For the amplifier descnbed m Exercise 10.27, rather than mtroducing a new dommam pole, we
can use additional capaCItance at the ClfCUIt node at which the first pole is formed to reduce the
frequency of the first pole. If the frequency of Ihe second pole IS 10 MHz and If It remams un-
changed while additIOnal capacitance is introduced as mentioned, find Ihe frequency to which the
first pole must be lowered so that the resultmg amplifier IS stable for closed-loop gams as low as
20 dB. By what factor must the capacitance at the controlling node be mcreased'
Ans. 1000 Hz; 1000
:j '

00

.1:1

.J
Summary
Negative feedback is employed to make the amplIfier
gain less sensitive to component variations; to control 10-
put and output impedances; to e,tend bandwidth; to
reduce nonlinear distortion; and to enhance signal-to-
Interference ratio.
The advantages abo\'e are obtaIned at the expense of a
reduction In gain and at the fISk of the amplIfier becom-
Ing unstable (that is, oscillating). The latter problem IS
sohed by careful design
For each of the four basic types of amplIfier, there is an
appropriate feedback topology. The four topologIes, to-
gether \\ Ith thelT analysis procedure and thelT effects on
Input and output Impedances, are summarized m Table
10. 1 In Section 10.8.
The key feedback parameters are the loop gaIn (Ath
which for negative feedback must be a positive dimen-
sionless number, and the amount of feedback (I + AP)
The latter dlTectly determInes gaIn reduction, gaIn de-
sensitivity, bandwidth exten lon, and changes In R
and R
SInce A and P are In general frequency dependent, the
poles of the feedback amplIfier are obtaIned by solvIng
the charactenstlc equation I + A(.nP(s) = O.
Computer Simulation Problems
Iii:. Problems IdentIfied by this icon are Intended to
demonstrate the value of usmg SPICE simulatIOn to ver-
Ify hand analysis and design, and to mvestlgate imponant
Issues such as allowable signal swing and amplIfier n _
I d' on
lneaT Istortion. Instructions to assist In setting
PS . up
pice and Multlslm Simulations for all the mdlcated
problems can be found 10 the corresponding files on the
diSC. Note that If a panicular parameter value is not spec-
Ified In the problem statement. you are to make a reason-
able assumption difficult problem: more difficult.
very challengmg andlor time-consuming' 0 d . '
bl
' eSlgn
pro em
For the feedback amplIfier to be stable Its p I
. 0 es must all
be In the left half of the s plane.
StabilIty IS guaranteed If at the frequency for whICh the
phase angle of AP IS 180' (I.e., lU, .. ),IA PI IS less th '
ty; the amount by which It IS less than unity ex an urn
d
. . ,. ' pressed In
eClbels, IS the gaIn margIn. Alternatively the am I'fi '
. pllerlS
stable If, at the !Tequency at which IAPI = I th , __ .
1
' 1 h ,ep,,,,,
ang e IS ess t an 180'; the difference is the phase margin.
The stabilIty of a feedback amplifier can be analyzed by
construcung a Bode plot for IAI and superimpoSing on ita
plot for I/IPI. StabilIty IS guaranteed if the
sect \\ Ith a difference In slope no greater than 6 dBOCt.lve.
To make a given amplIfier stable for a given feedback[".
tor P, the open-loop !Tequency response is SUItably modi.
fied by a process known as !Tequency compensation.
A popular method for !Tequency compensation mvolles
connecting a feedback capacitor across an in
the amplifier. This causes the pole formed at the mputofthe
amplifier stage to shift to a lower frequency and thus be
come dominant, whi le the pole formed at the output of!be
amplIfier stage is moved to a very high !Tequency and !bus
becomes unImportant. This process is known as pole
splittIng.
Section 10.1: The General Feedback Structure
10.1 A negatIVe-feedback amplIfier has a closed-loop gam
A( = 100 and an open-loop gaIn A = 10', What is the feed
back factor P? I f a manufacturing error results m a reduction
of A to 10 ', what closed-loop gaIn results? What is thep'"
centage change In A, correspondIng to this factor of 10 ,educ
tion in .1'J
10.2 Consider the op-amp clTcuit shown m Fig PI O.l,
where the op amp has mfimte mput resistance and zero Ollt-
put resistance but fiMe open loop gam .,
(a) Convince yourself that P= R, (R , +R,) 'v
(b) If R, = 10 k11, find R) that results m A) = 10 \
for the followmg three cases: (i) A = 1000 ViV; (II)
A = 100 VN; (Iii) A = 12 VN
(c) For each of the three cases m (b), find the percentage
change mAl that results when A decreases by 20%. Com-
ment on the results.
+A
-
+
+ v,
R,
-
-
- -
-
-
-
Figure Pl0.2
10.3 The noninvenmg buffer op-amp configuration shown
m Fig. PI 0.3 provides a direct implementation of the feed-
back loop of Fig. 10. \. Assummg that the op amp has mfi-
OIte input resistance and zero output resistance, what is fJ? If
A = 1000, what is the closed-loop voltage gain? What IS the
amount of feedback (in dB)? For V, = I V, find V, and v,. If
A decreases by 10%, what is the corresponding percentage
decrease in A?
R,

,
v, +
- -
- -
Figure Pl0.3
10.4 In a panlcular CITCUlt represented by the block dIagram
of Fig. 10.1, a signal of I V from the source results m a dif-
ference Signal of 10m V being provided to the amplItying ele-
ment A, and 10 V applied to the load, For this arrangement,
identify the values of A and pthat apply,
10.5 Find the loop gain and the amount of feedback of a
voltage amplifier for which A,and I 'P differ by (a) 1%, (b)
5%, (c) 10%, (d) 50%.
10.6 In a panlcular amplifier design, the P network con-
sists of a linear potentIOmeter for which pis 0.00 at one end,
1.00 at the other end, and 0.50 10 the middle. As the potenti-
ometer IS adjusted, find the three values of closed-loop gam
that result when the amplifier open-loop gaIn IS (a) I V'V,
(b) 10 VN, (c) 100 VN, (d) 10,000 V:V
10.7 A newly constructed feedback amplIfier undergoes a
performance test with the following results: With the feed-
back connection removed, a source signal of 5 111 V IS
Problems 891
required to provide a 10-V output to the load, with the feed-
back connected, a 10-V output requITes a 200-m V source
signal. For thiS amplifier, Identity values of A, p, AfJ, the
closed-loop gam, and the amount of feedback (in dB).
Section 10.2: Some Properties of Negative
Feedback
10,8 For the negative- feedback loop of Fig. 10.1, find the
loop gaIn AP for which the sensitivity of closed-loop gam to
open-loop gain [i .e., (dA/A, )/ (dA/ A)] is -40 dB. For what
value of APdoes the senSItivity become 1/2?
o 10.9 A deSigner IS consldenng two pOSSible designs of a
feedback amplifier. The ultimate goal is AI = 20 V IV. One
deSign employs an amplIfier for which A = 500 VN and
the other uses A = 250 V IV. Find P and the desensltlVlty
factor in both cases. If the A = 500 amplifier units have a
gain uncertamry of I 0%, what is the gain uncertainty for
the closed-loop amplifiers utiliZIng this amplIfier type? If
the same result IS to be achIeved With Ihe A = 250 amplI-
fier. what is the maximum allowable uncertainty in its gain?
o 10.10 A designer is reqUITed to achieve a closed-loop
gain of 25 + I % VN usmg a baSIC amplifier whose gain
variation is + I 0 %. What nominal value of A and P
(assumed constant ) are reqUITed?
o 10.11 A CITCUl t deSigner requITes a gain of 25 I % VN
using an amplIfier whose gam vanes by a factor of 10 over
temperature and time. What IS the lowest gaIn required? The
nom mal gain? The value of P?
o 10,12 A power amplifier employs an output stage whose
gam vanes from 2 to 12 for various reasons. What is the
gain ofan ideal (non varying) amplIfier connected to dnve It
so that an overall gain With feedback of 100 5% V N can
be achieved? What is the value of P to be used? What are
the reqUIrements If AI must be held wlthm 0.5 %? For
each of these situatIOns, what preamplIfier gain and feed-
back factor P are required If AI IS to be 10 VN (with the
two possible tolerances)?
o 10.13 It is required to deSign an amplIfier with a gain
of 100 that IS accurate to within I %. You have available
amplIfier stages With a gam of 1000 that is accurate to
within +30%. Provide a design that uses a number of these
gain stages In cascade, \nth each stage employmg negative
feedback of an appropriate amount. ObViously, your design
should use the lowest possible number of stages while meet-
ing specification.
10.14 Consider an amplifier havmg a midband gam A "and a
low-!Tequency response characterized by a pole at s = -WL
and a zero at s ::; O. Let the amplifier be connected in a nega-
tlve.feedback loop with a feedback factor p Find an expres-
'"
:a
w
...J
III
o
a:
e.
o
...
II:
III
....
e.
<C
::r::
V
sion for the mIdband gain and the lower 3-dB trequenc} of
the closed-loop amplifier By what factor have both changed"
o 10.15 It IS reqUIred to deSIgn an amplifier to have a
nommal closed-loop gam of lOY Y us 109 a battery-oper-
ated amplifier whose gain reduces to half its normal full-bat-
tery value o,er the life of the battery. If only 2% drop in
closed-loop gam is desired, what nommal open-loop ampli-
fier gain must be used 10 the deSIgn' (Note that smce the
change 10 is large, it is inaccurate to use dlfl'erentials.)
What value of pshould be chosen' If component-value vari-
ation In the p nen,\ ork may produce as much as a I % varia-
tion 10 p. to \\ hat value must A be raised to ensure the
reqUITed minimum gain'}
10.16 A capacltl\ely coupled amplifier has a midband
gam of 1000 \ Y. a single hlgh-trequency pole at 10kHz,
and a single low-trequency pole at 100 Hz. Negatl\e feed-
back IS employed so that the midband gam is reduced to 10.
What are the upper and 10\\ er 3-dB trequenCles of the
closed-loop gam'
0 10.17 Low-cost audIO power amplifiers often aVOId
dlfect coupling of the loudspeaker to the output stage
because any resultmg dc b,as current in the speaker can use
up (and thereby waste) ItS limIted mechanical dynamic
range. Unfortunately, the coupling capacItor needed can be
large! But feedback helps. For example, for an 8-Q loud-
speaker and {L = 100 Hz. what size capacItor is needed?
Now, if feedback is arranged around the amplifier and the
speaker so that a closed-loop gain AI = lOY V IS obtamed
from an amplifier whose open-loop gam IS 1000 \' V, \\ hat
value of {Lf results? If the ultimate product-design speCIfi-
cation requires a 50-Hz cutoff, what capacitor can be used'>
o 1018 It is reqUIred to design a dc amplifier WIth a low-
frequency gam of 1000 and a 3-dB frequency of O.S MHz.
You have available gam stages with a gain of 1000 but WIth
a dommant hIgh-frequency pole at 10 kHz. Provide a deSIgn
that employs a number of such stages 10 cascade, each WIth
negative feedback of an appropriate amount. Use identical
stages. [Hill/. When negative feedback of an amount
(I + Ap) is employed around a gam stage, ItS x-dB li-e-
quency IS mcreased by the factor (I + API.]
o 10.19 DeSIgn a supply-npple-reduced power amplifier
for which the output stage can be modeled by the block dia-
gram of Fig. lOA, where A 0.9 Y,Y, and the power-sup-
ply npple J's + I Y. A closed-loop gam of 10 YOV IS
desired. What is the gam of the low-ripple preamplIfier
needed to reduce the output rIpple to IOO mY' To IO
mY? To I mY' For each case, specify the value required
for the feedback factor P-
O 10.20 Design a feedback amplifier that has a closed-
loop gam of 100 YIV and is relatIvely msenSItlve to change
10 baSIc-amplifier gain. In particular it should p .
. _ " ' rovlde a
reduction 10 A, to 99 y , V lor a reduction In to One t h
. '. ent Its
nom mal value. What IS the reqUIred loop gain? What
nal value of A i reqUIred" What value of p should be
What would the closed-loop gain become if' .
. 1
mcreased tenfold" If A were made infinIte?
o 10,21 A feedback amplifier is to be deSIgned using,
feedback loop connected around a two-stage amplifier. Th
first stage is a dlfect-couplcd. small-signal amplifier with e
high upper 3-dB frequency. The second stage is a
output stage \\ith a midband gam of 10 YIV and upper and
10\\ er 3-dB trequencies of 8 kllz and 80 Hz, respectIVely.
The leedback amplifier should have a midband gain of IIIIJ
V V and an upper 3-dB li-equcncy of 40 kHz. What is the
requlfed gain of the small-signal amplifier? What value ofP
should be used" What does the lower 3-dB frequency of the
overall amplifier become"
' 10.22 The complementary BJT follov,er sho", In
Fig. PIO.22(al has the approXImate transfer charactenslle
tI'
"10--.. v
\ '
(" )
0.7 0 t o.7 l',
(b)
Figure Pl0.22
shown 10 Fig. PIO.22(b). Observe that for -0.7 Y $ v, < +0.7
V the output IS zero. This "dead band" leads to crossover
,
dIStortIOn (see ecllon 11 .3). ConSIder this follower to be
dm en by the output of a differential amplifier of gam 100
whose positive-input terminal is connected to the input sig-
nal source tis and whose negative-input terminal is con-
nected to the cmllters of the follower. Sketch the transfer
characteristic lIo versus Vs of the resulting feedback ampli-
fier. What are the limits of the dead band, and what are the
gams outSIde the dead band?
o 10.23 A particular amplifier has a nonlmear transfer
charactenstlc that can be approximated as follows .
(a) For small input SIgnals, iVA $ 10 mY, V
O
' VI = 10'
(b) For mtermediate input signals, 10 mY $ lvll $ 60 mY,
,
6"('0, 6 v, =10
(c) For large input signals, > 60 mY, the output saturates
If the amplifier IS connected 10 a negallve-feedback loop,
find the feedback factor p that reduces the factor-of-IO
change in gam (occurnng at IVII = 10 mY) to only a 10%
change. What is the transfer characteristic Vo versus Vs of the
amplifier WIth feedback?
Section 10,3: The Four Basic Feedback
Topologies
0 10.24 For the feedback voltage amplifier of Fig. 10.7(a)
let the op amp have an infinite input resistance. a zero output
resIStance, and a fimte open-loop gam A = 10' y,v. If
R, = I kn, find the value of R, that results in a closed-
loop gain of 100 YN. What does the gain become if R, IS
removed?
10,25 ConSIder the feedback voltage amplifier of Fig.
10.7(c). Neglect I' a and assume that (R, + R,) RD'
(a) Find expressIons for A and p and hence the amount of
feedback.
(b) Notmg that the feedback can be eliminated by removing
R I and R, and connecting the gate of Q to a constant dc
voltage (signal ground) give the mput resistance R, and the
output resIstance Ra of the open-loop amplifier
(c) Usmg standard circuit analysis (i.e, WIthout invokmg the
feedback approach), find the input resistance R'f and the
output resistance Raj of the CIfCUIt 10 Fig. 10.7(b). How
does R" relate to R, ' and Raf to Ra?
10.26 The feedback current amplifier 10 Fig. PI 0.26 uti-
li7es an op amp with an mput differential resistance R"I' an
open-loop gain JI. and an output resistance roo The output
current 10 that IS delivered to the load resistance Rr is
sensed by the feedback network composed of the twO reSIS-
tances R'I and R,., and a fraction If IS fed back to Ihe
amplifier input node. Find expressions for A = 1(/ I,.
Problems 893
P = 1/ I a' and A f '" I a' I , assuming that the feedback
causes the voltage at the input node to be near ground. Ifthe
loop gam IS large, what does the closed-loop current gam
become? State precisely the condition under which this is
obtained. For f.1 = 10' YIV R I = I Mn I' = 100 n
'II '0 '
RL = 10 kn , R\{ = 100 n, and RF = 10 kn, find A,
p,and AI'
,----+
-
-
I , t
- -
- -
Figure Pl0.26
10,27 Figure PIO.27 shows a feedback transconductance
amplifier utilizing an op amp WIth open-loop gain f.1, very
large input resistance. and a very small output resistance,
and an MOS transi stor Q. The amplifier delivers its output
current to RL The feedback network, composed of resistor
R. senses the equal current in the source tenninal of Q and
delivers a proportional voltage ' f to Ihe negative input ter-
mmal of the op amp.
V, +
-
-
Figure Pl0.27
+
V,
-
-+
J.A.
\ R
-
-
(a) Show that the feedback is negative.
(b) Open the feedback loop by breaking the connection of R
to the negative input of the op amp and groundmg the nega-
tive input tenninal. Find an expression for A == 101 V,
VI

...
....
CD
o
a:;
no
o
....
(c) Find an expression for /3= J'r' l o
(d) Fmd an expressIon for AI = 10 J',
(e) What IS the cond,tIOn to obtam 10 =
J' R ' ,
10.28 Figure PIO.28 shows a feedback transconductance
amplifier Implemented usmg an op amp WIth open-loop gain
J.1. a very large mput resIstance. and an output resistance 1"0 '
The output current 10 that IS delivered to the load resistance
RL is sensed by the feedback network composed of the three
resistances R". R, . and R" and a proportional voltage If
is fed back to the negative-input term mal of the op amp. Find
expressions for A = 1/ v. /3 = Vr 10, and A
J
,,10 1',
If the loop gain IS large. fmd an approXImate expression for
A I and state precisely the condItIOn for which this applies.
+
\
v, -
-
-
R,
\
-
-
-
-
Figure Pl0.28
10.29 For the feedback transresistance amphfier in Fig.
10.11 (d), use small-SIgnal analYSIS to find the open-loop
gain A = Voll,. the feedback factor /3 = 1/ J'Q' and the
closed-loop gain AI = V.lI, . Neglect "0 of each of Q,
and Q
2
and assume that Rc <;; /3,R and RE <;; R
F
and that
the feedback causes the signal voltage at the mput node to
be nearly zero. Evaluate V/ I, for the following compo-
nent values: /3, = /32 = 100. Rc = R = 10 kn. and
R
F
= 100kn.
10.30 For the feedback transreSlstance amphfier m Fig.
PIO.30. let RF Rc and ro Rc. and assume that the
feedback causes the signal voltage at the input node to be
nearly zero. Derive expressions for A = V /1 /3" I I r
. 0" f u'
and AI" V.l1 Ftnd the value of A J for the case of
Rc = 10 kn. RF = 100 kn. and the tranSIStor current
gain /3 = 100
Section 10.4: The Feedback Voltage Amplifier
(Series- Shunt)
10.31 A series-shunt feedback ampltfier employs a basic
amplifier with input and output reSIstances each of 2 kn and
I,
-
-
Figure Pl0.30
I,

I.

R,
\'
-
-
gam.4 = 1000 VN The feedback factor p= 0.1 YN. Find
the gain .4
r
the Input R'I' and the output re'iistance
R.if of the closed-loop amplifier
10.32 For a partIcular amphfier connected In a feedbac,
loop m which the output voltage IS >ampled. measurement
of the output remtance before and after the loop is con
nected shows a change by a factor of 100. Is the resistance
with feedback higher or lower? What IS the value of the loop
gain AjJ? If R." IS 100 n. what IS R. without feedback'
10.33 The formulas for R'J and RoJ m Eqs. (10.19) and
(10.22). respectively. also apply for the case in "h,ch A is!
function of frequency. In th,s case. the resulting Impedances
Z'I and Zol will be functIons of frequency. Consider the
case of a series4hunt amplifier that has an Input resistance
R,. an output resistance Ro. and openloop gam
A = Ao/ ( 1+ (s l (J)H' and a feedback factor P that IS
independent of frequency. Find Z" and Zo/ and gil< an
eqUIvalent circuit for each. together with the values of all the
elements in the equivalent cirCUIts.
10.34 A series-shunt feedback amplifier utilizes the feed
back circuit shown In Fig. PIO.34
(a) Find expressions for the h paramete" of the feedback
CIrCUit (see Fig. I O.I4b).
(b) IfR,= I knand/3=O.Ol,whatarethevalue,ofallfour
h parameters? Give the Units of each parameter,
(ci For the case R = I kn and R, = I kn. sketch and label
an eqUIvalent circuit followtng the model In Fig. 1O.14(c).
CD
Figure Pl0.34
10.35 A feedback amplifier utilizing voltage sampling and
employmg a basic voltage amplifier with a gain of 1000 VN
and an mput resistance of 1000 n has a closed-loop mput
resistance of 10 Idl What is the closed-loop gam? If the
basic amplifier IS used to implement a unity-gai n voltage
buffer. what input reSIstance do you expect?
' 10.36 In the senes- shunt feedback amplifier shown in
Fig. PI 0.36. the transIstors are biased with ideal current-
sources 1, = 0. 1 rnA and 12 = I rnA, the deVIces operate
with V
B
= 0.7 V and have /3, = /3
2
= 100. The input
signal v" has a zero de component. Resistances
R, = 100 n .R, = Ikn .R2= 10kn, andR
L
= I kn.
(a) If the loop gain is large, what do you expect the c1osed-
loop gam V/ V, to be? Give both an expression and its
approximate value.
(b) Find the dc emlner current in each of Q, and Q2' Also
find the dc voltage at the emitter of Q2
(c) Sketch the A circuit without the dc sources. Derive
expressions for A, R, . and Ro. and find their values.
(d) Give an expression for /3and find its value.
(e) Find the closed-loop gam V/ V,. the mput resistance
R". and the output resistance Rom' By what percentage
does the value of A I differ from the approximate value
found m (a)?
, 0 10.37 Figure P10.37 shows a series-shunt ampli-
fier WIth a feedback factor /3 = I. The amplifier IS designed
so that "0 = 0 for "s = O. with small deviatIons in "0 from 0 V
dc being mmlmized by the negative-feedback action. The
technology ut tl ized has k; = 2k; = 120 J.1AV' . I V,I = 0.7 V.
and IV:I = 24 V'iJ.I1l.
(b) With the feedback loop opened at the gate of Q,. and the
gate terminals of Q, and Q, grounded. find the dc current and
the overdrive voltage at which each of Q, to Q, is operatmg.
Ignore the Early effect. Also find the dc voltage at the output.
Problems 895
(c) Findg. and r of each of the five transIstors.
(d) Find the expressIOns and values of A and R Assume
that the biaS current sources are ideal.
(e) Find the gam with feedback. A,. and the output resis-
lance Raul'
(I) How would you modify the circuit to realize a closed
loop voltage gain of 5 VN? What is the value of output
resistance obtained?
o 10.38 Figure PI 0.38 shows a seriesshunt amplifier in
whIch the three MOSFETs are sized to operate at
IV o,j = 0.2 V. Let V,I = 0.5 V and IVAI = 10 V. The
current sources utilize single transistors and thus have out-
put resistances equal to r o'
(b) Assummg the loop gam to be large. what do you expect
the closed-l oop voltage gam V
o
' V, to be approXImately'
(c) If V, has a zero dc component. find the dc voltages at
nodes S,. G,. SJ. and G
J
. Verify that each of the current
sources has the minimum required de voltage across it for
proper operation.
(d) Find the A circUIt. Calculate the gain of each of the three
stages and the overall voltage gam. A .
[Him: A CS amplifier with a resistance R, in the source
lead has an effective transconductance gn/( I + gmR,) and
an output resistance rot I + gmR, ) 1
(e) Find /3 .
(I) Find At = V/ V,. By what percentage does this value
differ from the approxImate value obtamed m (bJ?
(g) Find the output resistance Rom'
o 10.39 The active-loaded dIfferential amplifier in
Fig. PI O. 39 has a feedback network consIsting of the volt-
age divider (R,. R
2
) . with R, + R, = I Mn The
deVIces are sized to operate at II' od = 0.2 V. For all
devices. I VAl = 10 V. The mput signal source has a zero dc
component.
R,
-
00
----<>\
R,
I,
-
v,
-
- -
-
Rn
R.
-
-
Figure Pl0.36
....
CD
o
a:

Q
...
+2.5 V
Q,
(40, I)---l
Q,
-
(20/ I)
-
-
Figure Pl0.37
/,=0.1 rnA
c,
Q,
-'(120, I)
-2.5 V
V
DO
= 1.8V
Q,
-,
(
Q,
/, = 0.1 rnA
-
/, = 0.1 rnA
-
V
-
R,
18 kil
V
DC
= 0.9 V
- -
- -
lure PIO.38
(a) Sho\\ that the feedback 's negat,ve
(b) What do you expect the dc voltage at the gate of Q, to
be') At the output? (Neglect the Early effect.)
(c) Find the A ClTcutt. Derive an expression for A and find ItS
value.
(d) Select values for R, and R2 to obtain a closed-loop
voltage gain Vol V, = 5 VN.
(e) Find the value of R
o
,,'
Qs
(20/ I)
v
"1
Ii
Q]
s
-
-
,
I
R ,
(f) Utilizing the open-CITCult. closed-loop gam (5 \'V) and
the value of R found In (e), find the value 01 gam
oul ('). neeted \(l
obtained when a reSistance RL = 10 b. IS con
the output. I 511
. f) b edo thean
a
)
(g) As an altematrve approach to ( a ove, r f R
of the A ClTcuit including R I . . Then utilize the val:' th'
and R2 found m (d) to deteIT",ne j3 and AI' Co P
value of AI to that found in (f).
Problems 897
- -1_Q,
I
R,
R,
Q, Q,
R,
R
-
200
-
- -
-
Figure Pl0.39
V ? 5 V DO :;;:+....
Q,
80 kfi
Q,
+ In 0
I
Q, Q,
I
0 In
Out
,-----<l Out
Q,
Q7

-j
Q,
IltlkI!
- In o--...
R,
-
If --?5V 5S - _.
-
(a)
Figure Pl0.40
o 10.40 The CMOS op amp in Fig. PI 040(a) is fabri-
cated In a technology for which I',,, = I 'p = 0.75 v.
P,Co, = 2/1pCo, = 100 and I, = 10 All
transistors In the circuit have L I Ilm.
(a) It IS reqUITed to perform a dc b13S deSign of the ClTCUIl
For thIS purpose, let the two mput termmals be at lero volts
dc and neglect channel-length modulatIOn (I.e, let I I =
Design to obtam 101 = I D2 = 50 los = 250 and
(bl
10 = O. and operate all transistors except for the source
follower Q, at 1'0' = 0.25 V. Assume that Q, and Q,
are perfect I} matched. and smlliarly for Q .. and Q, For
each transistor. find /0 and II L.
(b) What IS the allowable range of input common-mode
voltage')
(c) Find gm for each of Q" Q" and Q,.
(d) For each transistor, calculate roo
CI
...
898 .hapter 10 Feedback
(e) The I potentiometer shown In Fig. 10.40(b) IS con-
nected between the output terminal (Out) and the inverting
Input terminal (-In) to prOVIde negative feedback whose
amount is controlled by the setting of the wiper. A voltage
Signal " ., IS applIed between the nonInvertIng Input (+In) and
ground. A load resistance R L = 100 ill IS connected
between the output terminal and ground. The potentiometer tS
adjusted to obtain a closed-loop gain A f " V.I V, = 10 V N.
Specify the required setting of the potentIOmeter by giv-
Ing the values of RI and R,. Toward this end. find the A
circuit (supply a circuit diagram), the value of A, the j3 Cir-
CUit (supply a ClrcUlt diagram), and the value of P .
(t) What is the output resistance of the feedback amplIfier,
excluding R L ?
o Figure PI 0.41 shows a senes-shunt feedback
amplIfier without details of the bias circuIt.
(a) Sketch the A circUit and the circuIt for determining p.
Q,
v, QI
-
-
\
- -
- -
Figun Pl0.
R,
Al A,
+
v
-
- -
-
-
-
R,
-
/I -
-
RI
-
-
Figure Pl0.4;>
(b) Show that If AP IS large then the closed-loop
gain IS given approximately by ,
A
= _ RF+ R
f- -
, , R
(c) If R IS selected equal to 50 fl, find RF that will result I,
a closed-loop gain of approximately 25 VN.
(d) If QI is biased at I rnA, Q, at 2 rnA, and Q, at 5 rnA, and
assuming that the transistors have "" = I 00, find approXI.
mate val ues for Rn and RC2 to obtain gains from the
of the A circuit as follows: a voltage gain of Q
I
of about-IO
and a vol tage gain of Q, of about -50.
(e) For your design, what is the closed-loop voltage lrun
reali zed?
(f) Calculate the Input and output resistances of the closed.
loop amplIfier designed.
10.42 Figure PI0.42 shows a three-stage feedbackamph.
fier:
A I has an 82-kfl differential Input reSIStance, a 20VN
open-circuit differential voltage gain, and a 3.2-kQ output
resistance.
A 2 has a 5-kQ Input resistance, a 20-mAN short-clICUiI
transconductance, and a 20-kQ output resistance.
A 3 has a 20-kQ Input resistance, Unity open-circuit
gain, and a l-kQ output resistance.
The feedback amplIfier feeds a I-k fl load reSlStanceand IS
fed by a signal source with a 9-k fl resistance. The feedback
networkhas R
I
= 10 kfl andR
2
= 90 kfl
(a) Show that the feedback IS negative.
(b) Supply the small-signal equivalent Circuli.
(c) Sketch the A circuit and determine A
(d) Find j3 and the amount of feedback.
(e) Find the closed-loop gain Ar = V jV,
(t) Find the feedback amplIfier's Input resistanceR".
(g) Find the feedback amplIfier's output resistance Root' .
(h) If the high-frequency response of the open-loop gatnA IS
dominated by a pole at 100 Hz, what is the upper J-d8 fre
quency of the closed-loop gain?
R ,
J
\
-
-
(i) If for some reason A I drops to half lis nominal value,
what is the percentage change to AI?
Section 10.5: The Feedback Transconductance
Amplifier (Series-Series)
10.43 A series-series feedback amplifier employs a
transconductance amplifier having a short-circuit transcon-
ductance G. of 0.5 AN, input resistance of 10 kfl, and out-
put resistance of 100 kfl. The feedback network has P =
100 fl, an input resistance (with pon I open-circuited) of
100 fl, and an input resistance (with pon 2 open-circuited)
of 10 kfl . The amplIfier operates with a signal source having
a resistance of 10 kfl and with a load resistance of 10 kfl.
Find AI' R,n' and Root
10.44 ReconSider the circuit in Fig. 10.23(a), analyzed in
Example 10.6, this time with the output voltage taken at the
emitter of Q3 . In this case the feedback can be conSidered
to be of the senes-shunt type. Note that R 2 should now be
conSidered part of the basic amplIfier and not of the feed-
back network.
(a) Detenmne j3.
(b) Find an approximate value for AI" V,/ V, assuming
that the loop gain remains large (a safe assumption, since the
loop In fact does not change).
[Nole: If you continue with the feedback analYSIS, you'll
find that A j3 In fact changes somewhat; this IS a result of the
different approximatIons made In the feedback analysis
approach.)
ROI
10 kfl
0>----11 -t Q I
+
v,
-
-
-
Figure Pl0.45
Rm
10 kn
Q,
-
-
\
Ii
Problems 899
o 10.45 Figure PIO.45 shows a feedback triple uttlizmg
MOSFETs. All three MOSFETs are biased and sized to
operate at gm = 4 mAN. You may neglect thetr "0 's
(except for the calculation of R"" as indicated below).
(a) Considermg the feedback amplifier as a transconduc-
tance amplifier With output current 1
0
, find the value of RF
that results in a closed-loop transconductance of approxI-
mately 100 mAN.
(b) Sketch the A circuit and find the value of A = 101 V,.
(c) Find I + AP and Af = V/ I,. Compare to the value of
Af you designed for. What is the percentage difference?
What resistance can you change to make A! exactly 100
mAN, and in which direction (increase or decrease)?
(d) Assuming that 1'03 = 20 kfl, find the output resistance
R"tl . Since the current sampled by the feedback network is
exactly equal to the output current, you can use the feedback
formula.
(e) If the voltage Vo is taken as the output, in which case
the amplifier becomes series-shunt feedback, what is the
value of the closed-loop voltage gam V/ V, ? Assume that
RF has the original value you selected in (a). Note that In
this case RS2 should be considered pan of the amplifier
and not the feedback network. The feedback analysis will
reveal that A P changes somewhat, which may be puzzlIng
given that the feedback loop did not change. The change is
due to the different approximation used.
(t) What is the closed-loop output resistance R
O
"1l of the
voltage amplifier in (e) above?
o
..
10.46 Consider the CIrcuit in Fig. P10.46 as a transconduc-
tance amplifier wIth input ,. and output 10 The transistor
IS specified In tenns of ItS gm and ro
0>--111-
+
\
-
-
-
Figure P10.46
(a) Skelch the small-signal equIvalent circuit and convIOce
yourself that the feedback cirCUli is composed of reSI stor
RF
(b) Find the A circuit and the fJ CirCUIt
(c) Derive expresSIons forA, fJ, (I +AfJ), A
j
, R
o
' and
R Oj
0 10.47 The transconductance amplifier 10 Fig. P10.47
utilizes a differenual amplifier WIth garn Ii and a very high
input resistance. The dIfferential amplifier dnves a transistor
Q characterized by its gm and r , A reSIstor RF senses the
output current 1
0
,
+
I"
v +
-
-
-
-
F,gure Pl0.47
(a) For AfJ p i, find an approximate expressIon for the
closed-loop transconductance A/ = I., V Hence, select a
value for R f that results 10 A, 10 mNV
(b) Find the A circuit and derive an expresSion for A Evalu-
ate A for the case Ii = 1000 VN, gm = 2 mAN,
1'" = 20 kQ, and the value of RF you selected in (a).
(c) Give an expression for AfJ and evaluate Its .. I
;a ue and
that of I + AfJ
(d) Find the closed-loop gam A/ and compare to th aI
.. ' eVue
you antIcIpated 10 (a) above.
<
e) Find expressIOns and values for Rand R
o 0/
10.48 It IS reqUIred to show that the output resistance f
the BJT cirCUIt 10 Fig. PI 0.48 is gIven by 0
Ro= 1'0 + [R,.II (1',,+ R
h
)] (I + gmr rn )
rn + Rb
To derive thIS expresSIon, set V,= 0, replace the BJT wiili
ItS small-signal, hybrid-ll'model, apply a test voltage V to
the collector. and find the current I , drawn from' and
hence Ro as V,. I ,. Note that the bias arrangement is 1101
shown. Forthe case of Rh = 0, find the maxImum possible
value for Ro' ote that this theoretical maxImum IS
obtained when R, IS so large that the signal current in ilie
emitter IS nearly zero. In this case, with V, applied and
J' = 0, what is the current in the base. in the gmI', genera.
tor, and in '-0' all in terms of I x') Show these currents on a
sketch of the equivalent Circuit with R, set to 00.
1
R"
+
v, R,
-
-
-
Figure Pl0.48
10.49 As we found out 10 Example 10.6, whenever the
feedback network senses the emitter current of Ihe Bn, the
feedback output resIStance formula cannot predict the output
resistance 100klOg IOto the collector. To understand thIS
ISsue more clearly, consider the feedback transconductance
amplifier shown 10 Fig. PI 0.49(a). To determine the output
resistance. we set V, = 0 and apply a test voltage V, torh'
collector, as shown 10 Fig. PI0.49(b). NOli, leI J1 be
increased to the POlOt where the feedback signal aero" R,
equals the input to the pOSItIve terminal of the dlfferenn,l
amp Ii fier, now zero. Thus the SIgnal currenl through R,
will be zero. By replacing the BJT WIth its hybrid IT model,
show that
Rout = rJr+ (h
ft
+ 1)1"0 = hfJo
r I of
where h,. IS the tranSIstor fJ Thus ,or arge
" fl' rndepe"
feedback, R IS l,m,ted to a maxImum 0 '{,' . d
0", . Id b expeel< ,
dent of the amount of feedback. ThIS shou e R I
smce no current flows through the feedback nelwo
rk
F
-->---+
I"
--
v,
-
-
-
-
<a)
Figure P10.49
This phenomenon does 1101 occur 10 the MOSFET verSion of
this circuit
10,50 For the feedback transconductance ampli fier of Fig.
10.10(c) derive expressions for A, fJ, AfJ, A/ , R
o
' and R
ol
.
Evaluate A/ and RoJ for the case of gml = gm2 = 5 mAN,
RD = 10 ill, " 02 = 20 kQ, RF = 100 1.1 , and RL = I
kQ For Slmpircity, neglect " 0 1 and take " 02 into account
only when calculating output resistances.
0 10,51 For the feedback transconductance amplifier in
Fig. PIO.5I, denve an approximate expresSIon for the
closed-loop transconductance A / " 1
0
, V, for the case of
A f3 'P \. Hence select a value for R2 to obtarn A / = 100
mAN If Q IS bIased to obtain gm = I mAN, specify the
value of the gain Ii of the differentral amplifier to obtain an
amount of feedback of 60 dB. If Q has "0 = 50 kQ, find
the output resistance Rout .
---->---+
v
-
-
R,
lOon
Figure P10.5l
-
-
-
I"
R,
-
-
Problems 901
I

\ +
P-
-
-
-
-
ov
-
-
(b)
Iii" 10,52 All the MOS tranSIstors 10 the feedback
transconductance amplifier (series-senes) of FIg. P10.52
are Sized to operate at 1 V = 0.2 V For all transistors.
IV,I = 0.4 V and IVAI = 20 V.
(al If V has a zero dc component, find the dc voltage at the ,
output, at the dram of Q, ,and at the drain of Q2'
(b) Find an approximate expresSIon and value for
A, =I/V, for the case AfJ 'P I
(c) Use feedback analysis to obtam a more precise value for
A/ .
(d) Find the value of Roo<
<e) If the voltage at the source of Q, is taken as the output,
find the voltage gain using the value of 1
0
,- 1'\ obtained in
<c). Also find the output resistance of this senes-shunt volt-
age amplifier.
Section 10,6: The Feedback Transresistance
Amplifier (Shunt- Shunt)
10,53 For the transresistance amplifier analyzed 10 Exam-
ple 10.7, use the formulas denved there to evaluate AI'
R and R when Ii is one-tenth the value used 10 the
m' out J
example. That is, evaluate for Ii = 10 VN, R,d =
,. = 100Q,R
F
= 10kQ,andR, = R
L
= I kQ .Com-
p:re to the corresponding values obtamed in Example
10.7.
10,54 Use the formulas denved 10 Example 10.7 to solve
the problem in Exercise 10.15.
10,55 The CE BJT ampirfier in Fig. P10.55 employs
shunt-shunt feedback: Feedback resistor RF senses the out-
put voltage Vo and provides a feedback current to the base
node.
.....
a:l
o
a:
Q.
Q
...
CIC
w
I-
Q.
CC
:z:
v
+I.2 v
R, = 10 kfl
+
\ ,
-
0.2 mA
-
-
-I.2 V
Figure P'O.52
+12 V
'-----0\
R=IOkfl
i ..
R
\
R
- -
- -
gUI P10.55
(a) If V, has a zero dc component, find the dc collector cur-
rent of the BlT Assume the transistor fJ = 100.
(b) Find the small-signal equIvalent circuli of the ampltfier
wllh the signal source represented by liS Norton equIValent
(as we usually do when the feedback connectton at the mput
IS shunt).
(c) Find the A clTcuit and determine the value of4. R, and
Ro
(d) Find fJ and hence AfJ and I + AfJ
(e) Find A
f
R'I' and R , and hence Rand R
In out
(I) What voltage gam V/ V. is realtzed? How does thIS
value compare 10 the ideal value obtamed If the loop gam is
very large and thus the signal voltage at the base becomes
almost 7ero (like what happens 10 an mverting op-amp cir-
cuit). Note that thIS single-transistor poor-man's op amp IS
not that bad'
R,
'I' /"
0.8 mA
-
-
o 10.56 The circUit in Fig. PI 0.56 uttlizes a voltageampli.
fier wIth gain J1 in a shunt-shunt feedback topology
the feedback network composed of resistor R F ' In order to
be able to use the feedback equations, you should first con
vert the signal source to its Norton representatton. You will
then see that all the formulas derived in Example 10.7 apply
here as well
Rr
R
Il
<
1
,
V +
-
-
R.
,
- R
-
Figure Pl0.56
(a) If the loop gam IS very large. what approximate closed
loop voltage gam 1'0 r is realtzed? If R, = I k(L gtl'
the value of RF that wIll result 10 1'0/ r'. = -10 \'\.
J . t
(b) If the ampltfier J1 has a dc gam of 10 VN. an IO
PU
,
reSIstance R ,J = 100 k n. and an output reslS
ta
"'
I' = I kn find the actual r' , r realtzed. Also find R,
0' 0
1
, 'ust'
and R
o
", (indicated on the CIfCUlt dIagram). You rna)
formulas derived 10 Example 10.7 kH
fl' 1
(c) If the ampltfier J1 has an upper 3-dB frequency 0 dB
and a uniform -20 -dB/decade gain rolloff, what IS the J.
frequency of the gain I V,/ V,I?
10.57 The feedback transresistance ampltfier 10 Fig. PIO.57
utiltzes two identtcal MOSFETs bIased by ideal current
sources / = 0.5 rnA. The MOSFETs are sized to operate at
VOl' = 0.2 V and have V, = 0.5 V and VA = 10 V. The
feedback resistance RF = 10 kn .
(a) If f, has a zero dc component, find the dc voltage at the
rnput, at the dram of Q, ' and at the output.
(b) Find gm and 1'0 of Qt and Q2'
(c) provide the A ClTcuit and derive an expressIon for A In
termsofgm!. r ot. gm2' r o2, and RFo
(d) What is fJ? Give an expression for the loop gain AfJ
and the amount of feedback ( I + A j3).
(e) Derive an expression for AI
(I) Derive expressions for R
i
, R
m
, Ro. and R
oul
'
(g) Evaluate A, fJ, AfJ, AI ' Ri , Ro ' Rm , and ROUI forthe
component values given.
I
Q,
/,
t -
I
-
R.
-
- -
-
Figure P10.57
10.58 Analyze the circuit in Fig. E I 0. 15 from first pnncl-
pies (I.e., do not use the feedback approach) and hence show
that
(R.II RAgm - )(1'0 II RI )
Af = _ I
f , I + (R, II Rf )( gm - )<1'0 II RF RI
l
Companng this expression to the one given 10 Exercise
10.15, part (b), you will note that the only difference IS that
gm has been replaced by (gm - 1/ R
I
) Note that -1/ Rf
represents the forward transmiSSIOn in the feedback net-
work, which the feedback-analysis method neglects. What is
the condllion then for the feedback-analysis method to be
reasonably accurate for this cirCUit?
Problems 903
10.59 For the shunt-shunt feedback ampltfier of Fig.
10.II(c). derive expressions for A, fJ. AfJ, AI ' R
i
R'I'
Ro. and ROllO terms of gml' gm2' R
Dl
, R
02
' and R
F
.
Neglect " 0 I and r 02' Present your expressions in a fonnat
that makes them easy to mterpret (e.g .. like those derived 10
Example 10.7 or those asked for in Exercise 10. 15).
10.60 For the feedback transresistance amplifier In Fig.
10.II(d) let Vcc=-VE=5V. Rc =R=R
F
=
10 kn . The transistors have V
B
= 0.7 V and fJ = 100.
(a) If f, has a zero dc component, show that Q, and Q
2
are
operating at dc collector currents of approxImately 0.35 rnA
and 0.58 rnA, respeCltvely. What IS the de voltage at the out-
put?
(b) Find the A clfcuit and the value of A. R" and R
o
'
(c) Find the value of fJ, the loop gain, and the amount of
feedback.
(d) Find AI = Vo, f" the input resistance R'f' and Ihe out-
put resistance R
oJ
.
D ' 10.61 (a) Show that for the CITCUll In Fig. PIO.61(a).
if the loop gain is large, the voltage gain f; is given
approxImately by
I 0 _ I!.J.
- --
V, R\
(b) Using three cascaded stage, of the type shown in
Fig. PIO.61(b) to implement the amplifier fl. design a feedback
amplifier with a voltage gain of approximately -100 VN. The
amplifier is to operate between a source R, = 10 ill
and a load resistance R, = I ill. Calculate the actual value of
realized. the input resistance (excludmg R,). and the
output resIstance (excluding R,). A"ume that the BITs have
h" of 100. [Note. In practice. the three amplifier stages are nOl
made identical. for stability reasons. I
D 10.62 Negative feedback is to be used to modify Ihe
characteristics of a particular amplifier for various purposes.
Idenufy the feedback topology to be used If:
(a) Input resistance is to be lowered and output resistance
raised.
(b) Both input and output resistances are to be raised.
(c) Both input and output are to be lowered.
Section 10.7: The Feedback Current Amplifier
(Shunt- Series)
10.63 For the feedback current amplifier in Fig. 10.8(b):
(a) ProVide the A cirCUIt and deTive expressions for R, and
A. Neglect 1'0 of both transistors.
(b) Provide the fJ cirCUIt and an expression for fJ
(c) Find an expression for A fJ
...J
a:I
o
0::
0..
Q
..
0::
w
t-
o..
<C
:z:
V
\
-
-
x
10 kfl
R,
Vohage amplifier
< a)
+15 V
15 kfl
4.7kfl
- -
- -
(b)
75 kfl
x
-
-
Figure Pl0.61
-
-
\
R,
(d) For g., = g .. = 5 rnAIV. R, = 20 kn, R" )0 kQ and RF
90 ill, find the values of A, /3, A/3. A
f
R,. and R,J'
<e) If 1'0' = 20 kQ and RL = I kQ. find the output reSIS-
tance as seen by R L
o 1 .64 Design the feedback current amplifier of Fig.
I 0.3 1(a) to meet the following specifications
(I) A, " l / f = -100 AA
(II) amount of feedback = 40 dB
(Iii) Ron = I kQ
Specify the values of R, R, and /1 Assume that the ampli-
fier /1 has infimte input resistance and that R = ~ First
obtaIn an approximate value of /1 uti liz 109 the approXImate
formulas derived in Example 10.8. Then with the knowl-
edge that for the MOSFET. gm = 5 mAN and
1'0 = 20 kQ, modify the value of /1 to meet the design
speci ficatlOns. What Rout is obtained?
10.65 The feedback current amplifier 10 Fig. PI 0.65 Utl-
117es two Identical NMOS transistors sized so that at
fD = 0.2 mA they operate at VOl = 0.2 V. Both deVices
have I', = 0.5 Vand VA = 10 V
1 = 0.2 rnA
ff
r
R
Q,
Q,
-
-
R,
-
I
14 kfl R,
3.5 kO
R
-
- -
-
Figure Pl0.65
(a) If 1, has zero dc component, show that both Q, and Q,
are operating at 1 D= 0.2 mA What IS the dc voltage at the
mput?
(b) Find gm and 1'0 for each of Q, and Q,
(c) Find the A ClTcult and the value of R,. A. and Ro'
(d) Find the value of /3.
(e) Find A/3 and AI
(f) Find Ron and Roo,
'10.66 The feedback current amplifier 10 Fig. PIO.66(a)
can be thought of as a "super" CG transistor Note thai rather
than connectIng the gate of Q, to signal ground. an ampli
fier is placed between source and gate.
(a) If J.l is very large, what IS the Signal voltage at the mpUi
terminal? What is the input resistance' What IS the curren!
gain 10 I
J
?
(b) For finite J.l but assummg that the mput reSistanceoflhe
amplifier /1 IS very large, find the A circuit and deme
expressions for A. R" and Ro
(c) What is the value of /3?
(d) Find A/3 and AI If /1 IS large, what is the value of A/
(e) Find Rand R assuming the loop gam is large.
m out od
(f) The "super" CG transistor can be utilized m the case e
configuration shown In Fig. PI 0.66(b), where J'G isa de bias
voltage. ReplaCIng Q, by ItS smail-Signal model, use th,
analogy of the resulting CITCUlt to that m Fig. PIO.66(a))O
find f 0 and R
o
", '
10.67 Figure PI 0.67 shows an mterestmg and very useful
application of feedback to Improve the performance of the
h nn""
current mirror formed by Q, and Q, . Rather t an co
. leeUf'
109 the dram of Q I to the gate, as IS the case m Simp
rent mirrors, an amplifier of gain +J.l1S connected between
. k I does nol
the dram and the gate. Note that the leedbac oop h
tha' t e
mclude transistor Q, The feedback loop ensures
-
-
t(
rR,
+
Q,
Ii
-
-
I, R,
R,
- -
-
-
(a)
FIgure Pl0.66
value of the gate-to-source voltage of Q I is such that f 0 I
equals 1,. ThIS regulated V
g
IS also applied to Q, Thus.
If WIL of Q, is n times WfL of Q" 10 , = nfol = nf,. This
current track 109, however, is not regulated by the feedback
loop.
f
I,
,
,
,
yR.
,
,
ti ,
+
Q, ~
Ii
-
-y
- -
- -
Figure Pl0.67
VG
(b) If /1 is very large and the input resistance of the ampli-
fier /1 is infinite, what dc voltage appears at the dram of
Q I ? If Q I is to operate at an overdnve voltage of 0.2 V.
what is the minimum value that V BIAS must have?
(c) Replacing Q, by Its small-signal model, find an expres-
Sion for the small-signal input resistance R In assummg
fiOlte gain but mfimte input resistance for the amplifier /1.
Note that here it is much easier to do the analysis directly
than to use the feedback-analysis approach.
(d) What IS the output resistance Ro", ?
' 10.68 The ci rcuit 10 Fig. PI 0.68 IS an Implementation of a
particular circuit bUilding block known as second-
generat ion current convoyer (CCII). It has three termmals
besides ground: x, )'. and :. The heart of the ClTcuit is the
feedback amplifier ~ o n s i s t i n g of the differential amplifier /1
Problems 905
-
-
t
l
rR"
+
Q,
I f-
Q,
-
-
(b)
and the complementary source follower ( Q,. Q p). (Note
that this feedback circuit is one we have encountered a num-
ber of times in this chapter, albett with only one source fol-
lower transistor.) In the follow 109, assume that the
differential amplifier has a very large gain /1 and infinite
differential input resistance. Also. let the two current mirrors
have unity current-transfer ratios.
\
>--
+
p.
Figure Pl0.68
-
- Q,
,
Qp
Q,
(a) If a reSistance R is connected between.l' and ground. ~
voltage signal I'f is connected between x and ground. and_
is short-cirCUited to ground. Find the current f, through the
Co
...
906 lIapter 0 Feedback
shorl circuli. ho\\ ho\\ Ihls currenl IS developed and lIs
palh for ", poslllve and for V, negallve.
(b) If x is connecled to ground, a current source I .. IS con-
nected to input termmal .1'. and : IS connected to ground,
what voltage appears at y and what is the input resIstance
seen by I .. ? What IS the current I, that nows through the
OUlput shorl circuit? Also. explam Ihe current now Ihrough
the CirCUIt for I, positive and for I .. negallve.
(c) What is the output resistance at ': ?
IJ:I For the amphfier cirCUli in Fig. PIO.69.
assummg that " . has a zero dc component. find the dc volt-
ages at all nodes and the dc emItter currents of Q, and Q,.
LeI the BJTs have fJ = 100. Use feedback analysis to find
'0 , andR". leI ".c =0.7V.
Q,
-
-
R - 10 kn
\
-
-
r
+ t5 V
200 p.A
Q,
-
-
\
R,
140 n
10. The feedback amplifier of Fig. PIO.70 con",ls of a
common-gale amplifier formed by Q, and R
D
, and a feed-
back Circuli formed by Ihe capacilive diVIder (C C) d
h
i ' 2 an
I e common-source transIstor Q Note that the b' .
_ r Circuit
gl,:' nOI shown. It is required to denve expressions for
1 0 1 I.. R,n and Rool' Assume that C. and C
2
are suffi-
cIently small that their loading effecI on the baSIC amphfier
can be neglected. Also neglect, . Find the values of A R
and for the case in which g., = 5 mAN, RD = 10 kQ I'C ':
0.9 pF, C = 0.1 pF. and g.l= I mAN . ,-
Figure PI 0.71 shows a feedback amphfier ulJlizmg
the shunt-senes topology. All transIstors have fJ = 100
and VBE = 0.7 V. Neglect ' 0 except 10 (I).
(a) Perform a dc analysis to find the dc emllter currents 10
and Q, and hence determine their small-signal param t Q
1
e ers.
\ ,
Q,
C,
II, ,
I,
C,
-
-
-
- -
- -
Fi gure Pl0.70
(b) Replacmg the BJTs with their hybrid- Jr models, gIVe the
eqUIvalent CirCUIt of the feedback amplifier.
(c) Give the A cirCUli and determme A. R, . and R,. NOle
that Ro IS the resistance determmed by breaking the eminer
loop of Q, and measunng the resIstance between the lenm
nals thus created.
(d) Find the /3 Circuit and determme the value of P
(e) Find A/3. I + A /3. AI ' R'I' and Rol NOle Ihal Rot rep-
resents the reSIstance Ihat in effect appears In the emlller of
Q, as a result of the feedback.
(I) Determine ' 0' / '''' R", and Ro,, ' To determine R""
use VA' = 75 V and recall that the maxImum pOSSI ble
outpul resistance looking Into Ihe collector of a BJT is
approxlmalely /3r
o
where /3 is the BJT's fJ (see Problem
10.49).
Section 10.9: Determining the Loop Gain
10.72 Derive an expression for the loop gam AP oflhe
feedback amplifier m Fig. 10.22 (a) (Example 10.5). Sel
V, = O. break the loop at the gate of Q2' apply a lesl
voltage VI to the gate of Q, ' and delermme Ihe vohage
V, thaI appears at the output of amplifier A I' PUI your
expression m the form in Eq (10.36) and Indicale Ihe
difference.
10.73 It " reqUired to determme lhe loop gain of the
amplifier cirCUIt shown m Fig. PI 0.41 The mosl convemenl
place 10 break the loop IS al the base of Q,. Thus. connecl a
resislance equal lO r between the collector of Q, and
'"
ground. apply a lest voltage V to the base of Q" and deler
mme Ihe returned vohage at the coll ector of Q, (with 1', sel
to zero, of course). Show that
Problems 907
+ 12 V
R,
-
-
FIgure Pl0,71
to k!!

I,
tOO
R
to
Q,
R., 15k!!
- -
- -
10.74 Show that the loop gain of the amplifier cirCUli 10
Fig. PI 0.52 is
AfJ - r I' RF II 'os
- gml.2( 02 I 04)( R II ) + I I
F r 05 g",5
where g.,) IS the g. of each of Q, and Q,.
10.75 Derive an expression for the loop gain of Ihe feed-
back CirCUIt shown In Fig. PI0.26. Assume that the op amp
is modeled by an Input resistance R"" an open-circulI voltage
gam P. and an output resistance rD.
10.76 Fi nd the loop gain of the feedback amphfier shown
in Fi g. P1 0.37 by breaking the loop at the gate of Q, (and. of
course. setting Vs = 0), Use the values given in the statement
of Problem 10.37. Determine the value of
10.77 Denve an expression for the loop gam of the feed-
back amphfi er shown In Fig. 10.27(a} (Example 10.7). Eval-
uale AfJ fo r the componenl values given in Example 10.7
and compare to the value determined there.
10.78 Deri ve an expression for the loop gam of the feed-
back amphfier m Fig. 10.3 1(a) (Exampl e 10.8). Evaluate
Ap for Ihe component values given 10 Example 10.8 and
compare to Ihe result found Ihere.
10.79 For Ihe feedback amphfier 10 Fig. PIO.70, sel I, = 0
and denve an expression for Ihe loop gam by breakmg Ihe
loop al Ihe gale lermmal of lran>lslOr Q
r
Refer 10 Problem
10.70 for more delail s.
87011
-
-
...J--- I-
Q,
R"
14 kll :::;:
-
-
,
I

Section 10.10: The Stability Problem
R, I kO
-
-
10.80 An op amp designed 10 have a low-frequency gam
of 10' and a hIgh-frequency response dommaled by a single
pole at 100 radls, acquires, through a manufactunng error, a
pair of additional poles al 10.000 radis. AI whal frequency
does Ihe lOlal phase shift reach 180' ? AI thIS frequency. for
whal value of fJ, assumed 10 be frequency independent. does
the loop gain reach a value of unity? What the corre-
sponding value of closed-loop gain at low frequencies?
10.81 For the siluation described 10 Problem 10.80. skelch
Nyquist plOls for P = 1.0 and 10 ' (Plot for OJ = 0 radis,
100 radis, 10' radis, 10' radis. and radis.)
10.82 An op amp havmg a low-frequency gam of 10' and a
smgle-pole rolloff at let radls connected in a negallve-
feedback loop via a feedback network having a transmission
k and a Iwo-pole rolloff at 10' rad/s. Find the value of k
above which the closed-loop amplifier becomes unslable.
10.83 Consider a feedback amplifier for which the open-
loop gam A(s) is given by
1000
A(s) = ---'-=---:;
10')(1+; 10' )' ( I + s
If the feedback faclor PIS mdependenl of frequency. find the
frequency at which the phase 180, and find the criti-
cal value of pat which oscillation will commence.
Section 10.11 : Effect of Feedback on the
Amplifier Poles
10.84 A de amphfier having a SlOg Ie-pole response with
pole frequency 10 Hz and unily-gam frequency of I MHz IS
opera led in a loop whose frequencY-Independent feedback
faclor is 0.0 I Find Ihe low-frequency gam. Ihe 3-dB
...
....J
aI
o
a::
Co.
o
...
frequency. and the unit)-gam frequency of the closed-loop
amplifier B) \\ hat factor does the pole shih"
'10.85 An amplifier hanng a low-frequency gam of 10'
and poles at 10' Hz and 10' Hz "operated in a closed nega-
tive-feedback loop with a frequency-independent p.
(al For what value of P do the closed-loop poles become
coincident? At what frequency?
(bl What is the low-frequency gam correspondmg to the sit-
ualton 10 (ai' What is the value of the closed-loop gam at
the frequency of the coincident pole,>
(cl \\-'hat is the value of Q corresponding to the sltualton in
(al?
(dl If Pis mcreased by a factor of 10. what are the new pole
localtons' What is the correspondmg pole Q"
o 0.86 A dc amplifier has an open-loop gain of 1000 and
two poles. a dOminant one at I h.Hz and a hlgh-frequenc)
one whose location can be controlled It is required to con-
nect this amplifier in a negatiye-feedback loop that provides
a de closed-loop gain of 10 and a maxlmall) flat response
Find the reqUired value of P and the frequency at which the
second pole should be placed.
10.87 ReconSider Example 10.9 With the circull in Fig.
10AO modified to incorporate a so-called tapered network,
10 which the components immediately adjacent to the ampli-
tier input are raised 10 Impedance to CliO and lOR Find
expressIOns for the resulting pole frequency ltl, and Q factor
For what value of K do the poles coinCide? For what value
of K does the response become maXimally flat? For what
value of A' doe .... the circuit o,cillale?
10.88 Three identical inverT 109 amplifier stages each char-
acterized by a low-frequency gain K and a smgle-pole
response \\ llh f
ldB
= 100 kHz are connected in a feedback
loop with P = I. What IS the mmlmum value of /\ at which
the clrcull oscillates? What would the frequency of oscilla-
tion be>
Section 10.12: Stability Study Using Bode
Plots
10.89 ReconSider Exercise 10.24 for the case of the op
amp wired as a unity-gain buffer At what frequency 1\
IApi = I" What is the corr<>pondmg phase margm'
10.90 Reconsider ExerCise 10.24 for the case of a manu-
fac turi ng error mtroducing a second pole at 10- HI What is
now the frequency for which IAPI = I" What is the corre-
spondmg phase margm' For what values of P IS the phase
margin 45 or more')
10.' 1 For what phase margin does the gam peaking have a
val ue of 5'lP Of 100/(> Of 0. 1 dB> Of I dB" II/lilt Use Ihe
result 10 Eq. 10. 105 )
10.92 An amplifier has a dc gain of 10' and pol '
, .
Hz, 3.16 x 10 Hz. and 10 Hz Find the value of{1 d
d I d I
. r
correspon 109 case - oop gam, ,or which a phase
of 45' IS obtamed. mar!"
10.93 A two-pole amplifier for which A = 10' andh .
. 0 aVlOg
poles at I MHz and 10 MHz IS to be connected as , diffe",.
lIator. On the baSIS of the rate-or-clo,ure rule. what I,
smaJlest dtfferentlator time constant for which Operation IS
stable> What are the corresponding gain and phase margl ,
nI .
10.94 For the amplifier described by Fig. 10.43 and
frequency-independent feedback, what IS the minimum
closed-loop voltage gain that can be obtamed for phase mar-
gms of 90' and 45 '?
Section 10.13: Frequency Compensation
0 10.95 A multipole amplifier havlOg a first pole '1 3
MHz and a dc open-loop gain of 60 dB is to be compenlated
for closed-loop gams as low as uOity by the introduction of a
new dominant pole. At what frequency must the new pole be
placed>
o 10.96 For the amplifier deSCribed 10 Problem 10.95,
rather than introducing a new dominant pole we can
addillonal capacitance at the circun node at which the pole
IS formed to reduce the frequency of the first pole. If the fre-
quency of the second pole IS 15 MHz and if 11 remam,
unchanged while additIOnal capacitance IS mtroduced as
mentIOned. find the frequency to which the first pole mUll
be lowered so that the resulting ampliller IV stable for
closed-loop gams as low as uOity By what faclor I' the
capacitance at the conlrolling node Increased?
10.97 Contemplate the effecr- of pole splitting by cOnlld
enng Eqs. (10. I 121. (10.1 16). and (10.1171 under theco,dl'
R =R.C (,1I0=C.C,,,,C,andg.=IOO'
R, by calculating W"I' w
n
' and (V;l' w;'"
0 10.98 An op amp With open-loop voltage gain of 10'
and po lev at 10" HI, 10 HI, and 10' Hz IS to be compe'-
sated by the addllion of a fourth dommant pole to oper.1I<
stably with uOity feedback Cp = I I. What is the frequency of
the required dominant pole') The compensation
to consist of an RC low-pass networJ.. placed in the negau\e-
feedback path of the or amp. The dc blJS condllioOl are
. ." 1m
such that a I-MQ resistor can be tolerated In ..,ene\
each of the negative and pm,itivc Input term mal". What
. . h' . put and ,round
capacitor IS reqUired between ( c negative In e
to implement the reqUired fourth pole"
O I
e gain of
10.99 An op amp wllh an open-loop 10 tag
80 dB and poles at 10' HI, 10" HI. and 2 x 10' Hz 1\ to be
compenvated to be stable for unity p. Assume thai the op
amp incorporates an amplifi er equivalent to (hat in Fig.
10.46, with C, = ISO pF, C, = 5 pF, and g. = 40 mAN. and
that J", IS caused by the input circuit and I" by the output
circuit of this amplifier Find the reqUired value of the com-
pensating Miller capacitance and the new frequency of the
output pole.
leml " 10.100 The op amp in the cirCUIl of Fig. PI O. IOO
has an open-loop gam of 10' and a single-pole rolloff with
(J) = 10 radls.

(a) Sketch a Bode plot for the loop gam.
(b) Find the frequency at which IAPI = I , and fi nd the cor-
responding phase margm.
(c) Find the closed-loop transfer function, mcludi ng lls zero
and poles. Sketch a pole-zero pial. Sketch the magnitude of
the transfer funCll on versus frequency, and label the Impor-
tant parameters on your sketch.
Problems 909
R, = 100 kn
+
J
-
+
-
100 kn
-
-
0.01 ",F
Figure Pl0.l00

CH 10

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

CH 10

Caricato da

Copyright:

Formati disponibili

CHAPTER 10

808 Chapter 10 Feedback

amI R l' ohtaincd from

884 Chapter 10 Feedback

Potrebbero piacerti anche