265

IEEE TRANSACTIONS ON EDUCATION, VOL. 31. NO. 4, NOVEMBER 1988

A Network Theorem Dual to Millers Theorem

MARIAN K. KAZIMIERCZUK

Abstract-Millers theorem is an important tool commonly used in

analysis and design of amplifiers with voltage-shunt feedback. A network theorem presented and proved here is dual to Millers theorem
and can be utilized in the analysis and design of amplifiers with current-series feedback. Three examples are given to illustrate the applicability of the theorem.

I. INTRODUCTION
ILLERS theorem [1]-[4] is one of the basic network theorems presented in almost all electronic
circuits textbooks, e.g., [ 3 ] , [4].It simplifies the analysis
and design of amplifiers with voltage-shunt feedback. The
theorem is especially useful for amplifier analysis at high
frequencies where the voltage-shunt feedback due to the
base-to-collector capacitance C,, in BJTs or the gate-todrain capacitance Cgdin FETs should be taken into account. Applying Millers theorem, it is possible to determine the input impedance, the voltage and current gains,
and the upper 3 dB frequency of amplifiers. The increase
in the input capacitance over the base-to-emitter capacitance or the gate-to-source capacitance due to the feedback is called the Miller effect. The purpose of this paper
is to present a network theorem which is dual to Millers
theorem. This theorem simplifies the analysis and design
of amplifiers with current-series feedback.

t+
I-

VL

(b)
Fig. 1. Illustration of the theorem dual to Millers theorem. (a) The amplifier with current-series feedback. (b) Its equivalent circuit.

The circuits of Fig. l(a) and (b) are equivalent at the

input if the voltages VI and the currents I , are the same in
both circuits. It can be seen from (3) and (4) that this
condition is satisfied if

Zl = Zf( 1 - K I ) .
Similarly, for the circuit of Fig. 1(a),

11. THEOREM
Consider the amplifier with current-series feedback,
shown in Fig. l(a). The current gain KI is defined as
=

Assume that it is possible to find the value of KI by some

independent means. The current If through the feedback
impedance Zf is given by

From (1) and (2),

v, = ZfIf = Zf(I,
=

I,)

IIZf(1 - K I ) .

I,Zf 1

);

*,Zf( 1 -

For the circuit of Fig. l(b),

v, = I,&.

(7)
The circuits of Fig. l(a) and (b) are equivalent at the output if the voltages V, and the currents I2 are the same in
both circuits. It follows from (6) and (7) that this condition is fulfilled if

z,

Zf(1 -

i).

(3)

For the circuit of Fig. l(b),

v, = ZIZ,.

(5)

(4)

Manuscript received July 30, 1987; revised March 30, 1988.

The author is with the Department of Electrical Systems Engineering,
Wright State University, Dayton, OH 45435.
IEEE Log Number 88231 15.

111. COMMENTS
A . Applicability of the Theorem
It should be emphasized that the above theorem, similarly as the original Millers theorem, can only be used to
determine the input impedance and the forward gain of
the amplifier with feedback. In general, the theorem can-

0018-9359/88/1100-0265$01.OO 0 1988 IEEE

266

IEEE TRANSACTIONS ON EDUCATION. VOL. 31. NO. 4. NOVEMBER 1988

not be applied for calculating the output impedance and

the reverse gain. This is because the conditions that exist
in the network when K1 is determined are different from
those when the output impedance is determined [2], [3].
The mutual equivalence of the circuits shown in Fig. l(a)
and (b) is valid only for the forward (from input to output)
transmission. Therefore, the output impedance and the reverse gain should be determined from the original circuit
of Fig. l(a), using, e.g., a classical nodal or mesh analysis [3] or Blackmans impedance formula [4].

R, = 2k0

(a)

B. Use@lness of the Theorem

In the original Millers theorem, an added equivalent
impedance, placed in parallel with the load, is [3], [4]

z,=

Zf
~

(9)

- -

KV

where Kv = V2/ V I .
The original Millers theorem and its dual version given
here are based on the assumptions that it is possible to
determine KV or K1 by independent means, respectively.
The usefulness-or lack thereof-of Millers theorem
in either its original or dual version can only be put in
proper perspective by evaluating, in any given case, the
effect of 1) Z2 on Kv and KV on Z 2 , or 2) Z2 on K, and K I
on Z,, whichever applies. The result invokes one of three
levels of acceptable approximation.
1) The effect of Z2 on KV or K1 is negligible, in which
case one can just ignore it. Then the open-loop gain KV
or K1 may be used very conveniently to compute the altered input impedance and the forward gain of the amplifier with feedback.
2) The effect of Z2 on KV or K1 is not negligible, but
the effect of KV or K1on Z2 is negligible, the case for large
gains (then Z2 = Zr). For acceptable accuracy in using
Millers theorem to compute the input impedance and the
forward gain, one must now include Z2 in computing KV
or K1. Although this is a little more trouble, it still leaves
Millers theorem a viable and insightful approach to getting equivalent input impedance.
3) The effect of KV or K1 on Z, is not negligible and the
effect of Z2 on KV or K1 is not negligible either. If Millers
theorem were now employed, one would be faced with
the circular procedure of having to know, say, K1 in order
to calculate Z2 in order to calculate K1. In such a case,
Millers theorem, although still valid, becomes virtually
useless. Therefore, a classical nodal or mesh analysis
would be advised to get the input impedance and the forward gain.

IV. EXAMPLES
A. Example I
Fig. 2(a) shows a common-emitter amplifier with current-series feedback. The small-signal parameters of the
BJT are h,, = 2.1 kR, hfp = 100, h,, = 0, and h,, = 0.
Determine the input impedance R , , the current gain Al =

I(C)

Fig. 2. The common-emitter amplifier with current-series feedback. (a) The

circuit. (b) The small-signal model. (c) The equivalent small-signal
model.

I n / I s , and the voltage gain A ,

frequency range.

V O / v sfor the midband

B. Solution
Fig. 2(b) shows the small-signal model of the amplifier
of Fig. 2(a). In this circuit, R,, = R , R 2 / (R I + R , ) = 90
kR, Rf = R,, and the gain K , is

Using the theorem given above, the equivalent circuit of

Fig. 2(b) is as shown in Fig. l(c). From ( 5 ) and (8),

RI = Rf(1

KI)

R,(1

+ h k ) = 101 kR,

The input impedance R, is

R , =v,
-=R

+ h,, = R , ( 1 + hf,) + h,, = 103.1 kR.

Ib

The input impedance of the feedback amplifier is

The current gain of the amplifier is

A1

In =
I,

10
__
1, Is

-h f;r-

RI,

+ R:

-46.6.

261

KAZIMIERCZUK: NETWORK THEOREM DUAL TO MILLER'S THEOREM

+l:c

The resistance R2 is connected in series with the ideal current source and therefore does not affect the circuit operation.
Since

=5kn

and

the voltage gain is

A vs = -Vo= - -Vo=V,
V,
V, V,

hf,Rc
R,(1 + hf,)

I+

Ri

VO

+ hi, Ri + R,
1

Rp = RE(l +-)

R~l1+
hfel

C. Example 2
Fig. 3(a) shows a common-emitter amplifier with the
emitter bypass capacitor. The small-signal parameters of
the BJT are hi, = 1 kQ, hf, = 100, h,, = 0 , and h,, = 0.
Determine the low 3 dB frequency fL.

D. Solution

(C)

Fig. 3(b) shows the small-signal model of the amplifier

for the low-frequency range. The current gain KI is

Fig. 3 . Common-emitter amplifier with a bypass capacitor. (a) The circuit.

(b) The small-signal low-frequency model. (c) The equivalent circuit.

where
The impedance of the parallel combination of RE and CE
in the s domain is

and

1
RE -

RE
From ( 5 ) and (8),

SCE
1

+SCE

Fig. 3(c) shows the equivalent circuit of the amplifier.

The impedance Z1 is greater than the impedance ZE by a
1 = 101 and consists of the resistance R I
factor of hf,
in parallel with the capacitance C 1 . Analogously, the
impedance Z, is the parallel combination of the resistance
R2 = RE( 1
l / h f , ) 2: RE and the capacitance C2 =
CE/( 1 + l / h f , ) = CE. However, this impedance does
not affect the circuit operation because it is connected in
series with the ideal current source. The input impedance
of the amplifier is

Hence,
Since

and

IEEE TRANSACTIONS ON EDUCATION. VOL. 31, NO. 4, NOVEMBER 1988

268

where

the voltage gain is

RI

1 + hfe +

($

and

+ W,
+

= AV,, ___
s
Wp

where the midband voltage gain is

the frequency of the real zero is

and the frequency of the real pole is

+ :,l++2]

=-

f
P

Sincefi

2 ~ R CE
E

<< fp, f L = &

32.4 Hz.

<

WH.

= 32.4 Hz.

ACKNOWLEDGMENT

E. Example 3

KI=

hfe
7

+;- W H

where f H = w H / ( 2 a ) = 1 MHz. Determine the impedance Z , for the high-frequency range.

F. Solution
From ( 5 ) ,

Thus, the impedance ZI consists of the resistance R I

shunted by the capacitance CI. Both R I and C , are functions of frequency.
For w / w H = 1 , R I = 100Re = 100 kQ and CI = 1.56
pF; for w / w H = 10, RI = 51.25R, = 51.25 kCl and CI
= 1.55 pF; and for w / w H = 100, R I = 2Re = 2 kQ and
Cl = 0.79 pF.
It can be seen that the resistance R I decreases from
Re ( hfe 1 ) to R, as the frequency f increases from zero
to infinity. The capacitance CI is very small throughout
the entire frequency range and can therefore be neglected.
Thus, the advantage of the high input impedance due to
the current-series feedback can be fully obtained only for

REFERENCES
111 J . M. Miller, Dependence of input impedance of three-electrode vacuum tube upon the load in the plate circuit, Sri. Papers, Nut. Bureau
Stundards, no. 3 5 1 , pp. 362-285, 1919.
121 A. B . Macnee. On the presentation of Millers theorem. IEEE
Trans. Educ., vol. E-28, pp. 92-93, May 1985.

131 A. S . Sedra and K. C. Smith, Microelecrronic Circuits. 2nd ed. New

York: Holt, Rinehart and Winston, 1987, pp. 52-56.
141 J . M. Millman and A. Grabel, Mirroc,/ertronics, 2nd ed.
McCraw-Hill, 1987, p. 849.

NewYork:

Marian K. Kazimierczuk was born in Poland on

March 3, 1948 He received the M S , Ph D , and

-- _
RI

+;wc,

Habilitate Doctorate degrees in electronics engineering from the Department of Electronics,

Technical University of Warsaw, Warsaw. Poland, in 1971, 1978, and 1984, respectively
In 1972 he joined the Institute of Radio Electronics, Department of Electronics, Technical
University of Warsaw. Warsaw, Poland, where he
was employed as an Instructor from 1972 and 1978
and as an Assistant Professor from 1978 to 1984
He headed the Radio Electronics Laboratory and the Electronic Apparatus
Laboratory from 1978 to 1984. His teaching, research, and development

KAZIMIERCZUK: NETWORK THEOREM DUAL TO MILLERS THEOREM

269

activities were in the areas of RF power technology, radio transmitters,

electronic circuits and systems, semiconductor device modeling, electromagnetic field theory, microwave theory and technique, electronic measurements, circuit theory, communications, and computer-aided design. In
1984 he worked as a Project Engineer at Design Automation, Inc., Lexington, MA, where he was responsible for designing Class E high-efficiency switching-mode dcidc converters. In 1984-1985 he was a Visiting
Professor with the Department of Electrical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, where he taught electromagnetic fields and electronic circuits and systems. and his research activity was in the area of power electronics. Since 1985 he has been an
Assistant Professor with the Department of Electrical Systems Engineer-

ing, Wright State University, Dayton, OH, where he has been working in
the areas of analog and digital electronics, integrated circuits, electronic
devices, and power electronics. He is the author of 6 0 scientific papers. of
which 19 were published in IEEE TRANSACTIONS
and JOURNALS.He also
holds six patents related to the new concepts of high-efficiency switchingmode tuned power amplifiers and oscillators.
Dr. Kazimierczuk is a member of the Association of Polish Engineers
and the Polish Society of Theoretical and Applied Electrical Sciences. He
received seven awards from the President of the Technical University of
Warsaw, three awards from the Polish Ministry of Science, University Education, and Technology in 1981, 1982, and 1985, and an award from the
Polish Academy of Sciences in 1983 for scientific achievements.