658 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 59, NO.

10, OCTOBER 2012

Extending Blackman’s Formula to Feedback

Networks With Multiple Dependent Sources
Eugene Paperno

Abstract—The Blackman’s formula for the impedance seen

from a pair of arbitrary terminals of a linear feedback network
with a single dependent source is extended to the case of linear
feedback networks with multiple dependent sources. To reach this
aim, we revisit the proof of the Blackman’s formula, proof it
in a similar manner for feedback networks with two dependent
sources, and then extend to the case of feedback networks with
multiple dependent sources.
Index Terms—Blackman’s formula, circuit and systems funda-
mentals, effect of feedback on network impedances, impedance
evaluation, linear feedback networks, multiple dependent sources,
networks with single, double, triple, and multiple dependent
sources, return ratio.

I. I NTRODUCTION

T O EXAMINE the effect of negative feedback [1]–[19]

on impedances seen from a pair of arbitrary terminals
of a linear feedback network, it is important to describe them
analytically as a function of feedback partial gains, such as
return ratios.
The impedance seen from a pair of arbitrary terminals of a
linear feedback network with a single dependent source can be
described by the Blackman’s formula [3]
1 + Tsc
Zt = Zt |a=0 (1)
1 + Toc
where Zt is the closed-loop impedance seen by a test source
vt [see Fig. 1(a)], a is the dependent source gain, Zt |a=0 is
the open-loop impedance, Tsc is the return ratio for the short-
circuited terminals, and Toc is the return ratio for the open-
circuited terminals
sε |vt =0
Tsc ≡ − (2)
sε
sε |it =0
Toc ≡ − . (3)
sε

As seen from (2) and (3), the return ratio is defined as the
contribution of a dependent source to its control signal, nor- Fig. 1. Finding the impedance seen by a test source connected to arbitrary
terminals of a generic feedback network with a single dependent source.

Manuscript received January 31, 2012; revised March 25, 2012 and May 29,
2012; accepted July 18, 2012. Date of publication September 10, 2012; date of
current version October 12, 2012. This work was supported by Analog Devices,
Inc. This brief was recommended by Associate Editor R. Martins.
The author is with the Department of Electrical and Computer Engineer-
ing, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel (e-mail:
paperno@ee.bgu.ac.il).
Color versions of one or more of the figures in this brief are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCSII.2012.2213355
malized to the total control signal sε appearing in the dependent
source value a sε (see Fig. 1).
Since the network independent sources are suppressed while
finding return ratios, it is necessary to substitute a dependent
source with an equivalent independent one having the same
value a sε to find the contributions sε |vt =0 and sε |it =0 of the
dependent source to its control signal sε .

1549-7747/$31.00 © 2012 IEEE

PAPERNO: EXTENDING BLACKMAN’S FORMULA TO FEEDBACK NETWORKS WITH MULTIPLE DEPENDENT SOURCES 659

The aim of this present work is to extend the Blackman’s Gv and Gi in (10) can be found from Fig. 1(c) and (d)
formula to the case of feedback networks with multiple de-
sε |a=0
pendent sources. To reach this aim, we revisit the proof of the Gv = for an arbitrary vt (11)
Blackman’s formula for feedback networks with a single de- vt
pendent source, proof it in a similar way for feedback net- sε |a=0 vt
Gi = for it |a=0 = . (12)
works with two dependent sources, and then show how it can it |a=0 Zt |a=0
be extended to the case of feedback networks with multiple
dependent sources. Note that it |a=0 = vt /Zt |a=0 in (12) keeps the same conditions
In feedback networks with two or more dependent sources, of the test source branch for both vt and it |a=0 sources and, as
each dependent source contributes in a general case not only to a result, sε |a=0 in (11) and (12) has the same value.
its own control signal but also to the control signals of other According to (11) and (12), Gi /Gv = Zt |a=0 . Thus, con-
dependent sources. For the sake of consistency, we refer in this sidering (10), the Blackman’s formula (1) can be obtained for
work to such cross contributions, normalized to the total control feedback networks with a single dependent source.
signal appearing in the dependent source value, as cross return
ratios. Although the signal may return to the control terminals
III. F EEDBACK N ETWORKS W ITH
of not the same source, but to the control terminals of other
T WO D EPENDENT S OURCES
dependent sources, it still belongs to the contributions of the
active part of the network, represented by all the dependent Following the approach introduced in the previous section,
sources. we extend in this section the Blackman’s formula to feedback
networks with two dependent sources (see Fig. 2).
Fig. 2(a) and (b) suggest the following equations for the
II. F EEDBACK N ETWORKS W ITH A control signals of the dependent sources
S INGLE D EPENDENT S OURCE

sε1 = vt G1v − sε1 T1sc − sε2 T21sc
Let us consider a linear feedback network with a single (13)
sε2 = vt G2v − sε2 T2sc − sε1 T12sc
dependent source (see Fig. 1). To find the impedance seen from

a pair of arbitrary terminals, we connect to them a voltage sε1 = it G1i − sε1 T1oc − sε2 T21oc
(14)
test source in Fig. 1(a) and find the control signal by applying sε2 = it G2i − sε2 T2oc − sε1 T12oc
superposition
where
sε = vt Gv − sε Tsc (4)
sε1 | vt =0 sε1 | it =0
T1sc ≡ − T1oc ≡ −
a2 =0 a2 =0

where the input transmission for the voltage test source (15)
sε1 sε1
sε |a=0 sε2 | vt =0 sε2 | it =0
Gv ≡ . (5)
T12sc ≡ − T12oc ≡ −
a2 =0 a2 =0
vt (16)
sε1 sε1
From (4), vt can be obtained as
sε2 | vt =0 sε2 |i vt =0
T2sc ≡ − T2oc ≡ −
a1 =0 a1 =0
sε (1 + Tsc ) (17)
vt = . (6) sε2 s2
Gv
Let us now replace in Fig. 1(b) source vt with it , such sε1 | vt =0 sε1 | it =0
T21sc ≡ − T21oc ≡ −
a1 =0 a1 =0
. (18)
that it = vt /Zt , to keep the same conditions of the test source sε2 sε2
branch. Keeping the same branch voltage and current leaves the
signal sε unchanged. As a result Equations (13) and (14) can be solved for vt and it
1 + T1sc + T2sc + T1sc T2sc − T12sc T21sc
sε = it Gi − sε Toc (7) vt = sε2 (19)
G2v + G2v T1sc − G1v T12sc
where the input transmission for the current test source 1 + T1oc + T2oc + T1oc T2oc − T12oc T21oc
it = sε2 . (20)
sε |a=0 G2i + G2i T1oc − G1i T12oc
Gi ≡ . (8)
it Dividing (19) by (20) gives
From (7), it can be obtained as
G2i 1 + T1oc − G1i
G
2i
T12oc 1 + TΣsc
Zt = (21)
sε (1 + Toc ) G2v 1 + T1sc − G T12sc 1 + TΣoc
G 1v
it = . (9) 2v
Gi
where
Hence
vt Gi 1 + Tsc TΣsc = T1sc + T2sc + T1sc T2sc − T12sc T21sc
Zt = = . (10)
it Gv 1 + Toc TΣoc = T1oc + T2oc + T1oc T2oc − T12oc T21oc . (22)
660 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 59, NO. 10, OCTOBER 2012

Fig. 3. Generic feedback network with two dependent sources, where vt = 1,

sε1 = 1, and sε2 = 0.

To find the second term in (21), let us consider Fig. 3, where

vt = 1, sε1 = 1, and sε2 = 0, and hence

 sε1 | av1t =1
 =1
= G1v − T1sc
a2 =0
(24)
 sε1 | av1t =1
 =1
= it | vt =1 G1i − T1oc
a1 =1
a2 =0 a2 =0

 sε2 | av1t =1
 =1
= G2v − T12sc
a2 =0
(25)
 sε2 | av1t =1
 =1
= it | vt =1 G2i − T12oc .
a1 =1
a2 =0 a2 =0

From (24) and (25)

 
G1i G1v G1i G1i
T1oc − T12oc − G2v − = T1sc − T12sc .
G2i G2v G2i G2i
(26)
Considering that, in (26), in accordance with (23)
G1v G2v G1v G1i
Fig. 2. Finding the impedance seen by a test source connected to arbitrary = = Zt | a1 =0 ⇒ = (27)
terminals of a generic feedback network with two dependent sources. G1i G2i a2 =0 G2v G2i
(26) can be rewritten as follows:
The first term in (21) represents the open-loop impedance
seen by vt G1i G1v
T1oc − T12oc = T1sc − T12sc . (28)
G2i G2v
G2i sε2 | a1 =0 vt vt
≡
a2 =0
= Considering (21), (23), and (28), Zt can eventually be ob-
G2v it | a1 =0 sε2 | a1 =0 it | a1 =0
a2 =0 a2 =0 a2 =0 tained as
vt
= Zt | a1 =0 for it | a1 =0 = . (23) Zt = Zt | a1 =0
1 + TΣsc
a2 =0 a2 =0 Zt | a1 =0 a2 =0 1 + TΣoc
. (29)
a2 =0
PAPERNO: EXTENDING BLACKMAN’S FORMULA TO FEEDBACK NETWORKS WITH MULTIPLE DEPENDENT SOURCES 661

IV. F EEDBACK N ETWORKS W ITH M ULTIPLE from a pair of arbitrary terminals of a linear feedback network
D EPENDENT S OURCES with three dependent sources can be obtained by (29), where
the open-loop impedance is given by (37) and TΣsc and TΣoc
Following the approach given in the previous sections, the
are given by (35) and (36), respectively.
impedance seen from a pair of arbitrary terminals of feedback
networks with multiple dependent sources can be obtained in
accordance with (29).
V. C ONCLUSION
For example, for a feedback network with three dependent
sources, the control signals of the dependent sources It has been shown that the Blackman’s formula can be
 extended to the case of feedback networks with multiple depen-
 sε1 = vt G1v − sε1 T1cs − sε2 T21cs − sε3 T31cs
dent sources. The obtained equations provide exact solutions
sε2 = vt G2v − sε2 T2cs − sε1 T12cs − sε3 T32cs (30)
 for feedback networks with two and three dependent sources.
sε3 = vt G3v − sε3 T3cs − sε1 T13cs − sε2 T23cs
 Following the proposed approach, equations for a greater num-
 sε1 = it G1i − sε1 T1oc − sε2 T21oc − sε3 T31oc ber of dependent sources can easily be developed.
sε2 = it G2i − sε2 T2oc − sε1 T12oc − sε3 T32oc (31) Transistor small-signal models in many cases include only

sε3 = it G3i − sε3 T3oc − sε1 T13oc − sε2 T23oc . one dependent source. Double and multiple transistor feedback
circuits are used widely. As a result, the proposed theory could
Equations (30) and (31) can be solved for Zt = vt /it serve a useful reference to do quick calculations by hand
G3i Boc 1 + TΣsc and, thus, to obtain a better intuitive insight into the effect of
Zt = (32) feedback on the circuit impedances.
G3v Bsc 1 + TΣoc
In contrast to the canonical approach [11], [19], the described
where theoretical treatment involves no approximations to simplify the
analysis.
Boc = 1 + T1oc + T2oc + T1oc T2oc − T12oc T21oc To validate the proposed theory, theoretical results were com-
Gi1 pared against simulations and experimental measurements. In
+ (T12oc T23oc − T13oc − T2oc T13oc ) both the theoretical calculations and simulations, the transistors
Gi3
were modeled by equivalent small-signal circuits with a single
Gi2
+ (T13oc T21oc − T23oc − T1oc T23oc ) (33) dependent source hf e ib . A perfect matching has been obtained
Gi3 between the theory and simulations. The matching between
Bsc = 1 + T1sc + T2sc + T1sc T2sc − T12sc T21sc the theoretical and experimental results was very close. The
difference between these results is due to the approximate
Gv1
+ (T12sc T23sc − T13sc − T2sc T13sc ) transistor models and the measurement uncertainties of the
Gv3
transistor small-signal parameters.
Gv2
+ (T13sc T21sc − T23sc − T1sc T23sc ) (34)
Gv3
TΣsc = T1sc − T1sc T23sc T32sc + T2sc − T2sc T13sc T31sc ACKNOWLEDGMENT

+ T3sc − T3sc T12sc T21sc + T1sc T2sc + T1sc T3sc The author would like to thank Prof. S. Ben-Yaakov for very
fruitful and inspiring discussions.
+ T2sc T3sc − T12sc T21sc − T13sc T31sc − T23sc T32sc
+ T1sc T2sc T3sc + T12sc T23sc T31sc R EFERENCES
+ T13sc T21sc T32sc (35) [1] H. Nyquist, “Regeneration theory,” Bell Syst. Techn. J., vol. 11, pp. 126–
147, Jan. 1932.
[2] H. S. Black, “Stabilized feedback amplifiers,” Elect. Eng., vol. 53,
TΣoc = T1oc − T1oc T23oc T32oc + T2oc − T2oc T13oc T31oc pp. 114–120, Jan. 1934.
[3] R. B. Blackman, “Effect of feedback on impedance,” Bell Syst. Techn. J.,
+ T3oc − T3oc T12oc T21oc + T1oc T2oc + T1oc T3oc vol. 22, pp. 269–277, Oct. 1943.
[4] H. W. Bode, Network Analysis and Feedback Amplifier Design. New
+ T2oc T3oc − T12oc T21oc − T13oc T31oc − T23oc T32oc York: Van Nostrand, 1945.
[5] F. H. Blecher, “Transistor multiple loop feedback amplifiers,” in Proc.
+ T1oc T2oc T3oc + T12oc T23oc T31oc Nat. Elect. Conf., 1957, vol. 13, pp. 19–34.
[6] C. Belove and D. L. Schilling, “Feedback made easy for the undergradu-
+ T13oc T21oc T32oc . (36) ate,” IEEE Trans. Educ., vol. E-12, no. 2, pp. 97–103, Jun. 1969.
[7] S. Rosenstark, “A simplified method of feedback amplifier analysis,”
According to (23) and (27), the first term in (32) is IEEE Trans. Educ., vol. E-17, no. 4, pp. 192–198, Nov. 1974.
[8] A. M. Davis, “General method for analyzing feedback amplifiers,” IEEE

G3i  Trans. Educ., vol. E-24, no. 4, pp. 291–293, Nov. 1981.
= Zt  . (37) [9] K. S. Yeung, “An alternative approach for analyzing feedback amplifiers,”
G3v aj =0 IEEE Trans. Educ., vol. E-25, no. 4, pp. 132–136, Nov. 1982.
j=1,2,3
[10] S. Rosenstark, Feedback Amplifier Principles. New York: MacMillan,
1986.
As in the previous section, it can also be shown that the term [11] K. R. Laker and W. M. C. Sansen, Design of Analog Integrated Circuits
Boc /Bsc , in (32), equals unity. As a result, the impedance seen and Systems. New York: McGraw-Hill, 1994.
662 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 59, NO. 10, OCTOBER 2012
[12] P. J. Hurst, “A comparison of two approaches to feedback circuit analysis,”
IEEE Trans. Educ., vol. 35, no. 3, pp. 253–261, Aug. 1992.
[13] S. Franco, Design With Operational Amplifiers and Analog Integrated
Circuits. New York: McGraw-Hill, 1997.
[14] B. Nikolic and S. Marjanovic, “A general method of feedback amplifier
analysis,” in Proc. IEEE ISCAS, 1998, vol. 3, pp. 415–418.
[15] F. Corsi, C. Marzocca, and G. Matarrese, “On impedance evaluation
in feedback circuits,” IEEE Trans. Educ., vol. 45, no. 4, pp. 371–379,
Nov. 2002.
[16] A. S. Sedra and K. C. Smith, Microelectronic Circuits, 5th ed. London,
U.K.: Oxford Univ. Press, 2003.
[17] G. Palumbo and S. Pennisi, Feedback Amplifiers: Theory and Design.
Norwell, MA: Kluwer, 2003.
[18] P. R. Gray, P. J. Hurst, S. H. Lewis, and R. G. Meyer, Analy-
sis and Design of Analog Integrated Circuits. Hoboken, NJ: Wiley,
2009.
[19] B. Pellegrini, “Improved feedback theory,” IEEE Trans. Circuits Syst. I,
Reg. Papers, vol. 56, no. 9, pp. 1949–1959, Sep. 2009.