EE 541, Fall 2009: Course Notes #1

Linear Two port Networks:

Theory, Models, and Applications




Dr. John Choma
Professor of Electrical Engineering
University of Southern California
Ming Hsieh Department of Electrical Engineering
University Park: Mail Code: 0271
Los Angeles, California 900890271
2137404692 [USC Office]
2137408677 [USC Fax]
[email protected]


ABSTRACT:
This report defines the commonly used two port parameters of generalized linear
electrical networks and discusses the theory that underpins two port parameter
models. It develops two port circuit analysis methods for evaluating the driving
point input impedance, the driving point output impedance, the transfer function,
the loop gain, and the stability characteristics of simple and relatively complex
active and passive two port network architectures. Basic measurement strategies
for the various two port parameters are also addressed.
Original: September 2002
Course Notes #1 University of Southern California John Choma

August 2006 2 Two Port Networks

1.0. INTRODUCTION
The majority of linear passive electrical and active electronic systems can be viewed as
two port networks. A network port consists of two electrical terminals, thereby suggesting that a
two port network has two readily accessible pairs of electrical terminals. One of these terminal
pairs is the input port, to which a known signal voltage or current is applied. The network of
interest processes this input voltage or current to produce a desired or designable response at the
second of the two network pairs, which is called the output port. The theories and analytical
techniques indigenous to basic circuit theory allow the output response of a two port network to
be related to its applied input energy as a function of the volt-ampere characteristics of the
individual branches embedded in the network. Although this direct formulation of the output
response -to- input signal relationship is systematic, three engineering issues limit the utility of
this straightforward analytical strategy.
One of the foregoing engineering issues derives from the sheer topological complexity
of practical circuits. A useful circuit, and particularly a useful electronic circuit, may contain
hundreds or even thousands of branch elements. The immediate effect of this ubiquity of electri-
cal elements is an explosion of the required number of equilibrium equations that must be solved
simultaneously to forge the desired input -to- output (I/O) transfer relationship. Manual solution
is therefore rendered daunting, if not impossible, thereby encouraging the use of the omnipresent
computer. But neither cumbersome manual analyses nor computationally efficient computer-
based numerical solutions are likely to inspire the engineering design insights whose assimilation
is, in fact, the fundamental goal of circuit analysis.
A second shortcoming of direct circuit analyses is that the electrical characteristics of
many internal network branches may not be well defined or clearly understood. This issue is
especially nontrivial when the network in question contains active elements, such as bipolar
junction transistors (BJTs) and metal-oxide-semiconductor field effect transistors (MOSFETs),
that are incorporated to process electrical signals at high speeds. In such an event, electrical
responses are vulnerable to stray capacitances, parasitic lead inductances, device processing
uncertainties, and undesirable electromagnetic coupling from proximately positioned circuits.
To be sure, mathematical models of active elements and the aforementioned second order electri-
cal phenomena can be constructed. But in the interest of analytical tractability, these macromod-
els are invariably simplified to an extent that virtually ensures inaccuracies in, or even
miscomprehension of, the fruits of analysis.
A final point of contention is the superfluous technical information generated as an
implicit byproduct of conventional mesh and nodal analyses. The application of the Kirchhoff
voltage law (KVL) and the Kirchhoff current law (KCL) produces, in addition to the desired out-
put response, all node and branch voltages and all mesh and loop currents intrinsic to the net-
work undergoing examination. But in many applications, such as active filters that exploit
commercially available operational amplifiers (op-amps) or application specific high perform-
ance amplifiers, the only engineering concern is the electrical relationship established between
the I/O ports of the filter. This relationship can be forged without an explicit awareness of the
voltages and currents established within the utilized amplifiers. Indeed, the overall I/O filter
transfer function and other performance metrics can be deduced as a function of only electrical
properties gleaned from measurements executed at the I/O ports of these amplifiers. These
electrical properties, or two port parameters, are admittedly nonphysical entities in that they
generally cannot be cast in terms of the physically sound phenomenology that underlie the
Course Notes #1 University of Southern California John Choma

August 2006 3 Two Port Networks

electrical nature of observable I/O characteristics. Nonetheless, they are useful because they do
derive from reproducible port measurements, and they do conduce the formulation of an analyti-
cally unambiguous overall network response.
2.0. FUNDAMENTAL TWO PORT CONCEPTS
Figure (1) abstracts a generalized linear two port electrical network. The input port is
formed by the terminal pairs, 12, while the output port, which is terminated in an arbitrary load
impedance, Z
l
, is the terminal pair, 34. It is presumed that no energy sources lie within the two
port network, which implies that if any energy storage elements are embedded therein, zero state
conditions apply. Energy is therefore applied to the two port system at only its input port. In
Figure (1a), this energy is represented by a Thvenin equivalent circuit comprised of the signal
source voltage, V
s
, and its internal series impedance, Z
s
. Alternatively, the applied energy can be
modeled as the Norton topology depicted in Figure (1b), where the Norton, or short circuit,
equivalent current, I
s
, is
Linear
Two Port
Network
Z
s
Z
l
+

V
1
I
1
+

V
s
I
2
+

V
2
(a).
Linear
Two Port
Network
Z
l
+

V
1
I
1
I
2
+

V
2
(b).
I
s
Z
s
1 3
2 4
1
2
3
4

Fig. (1). (a). A Linear Two Port Network For Which The Applied
Signal Source Is Modeled By A Thvenin Equivalent Cir-
cuit. (b). An Alternative Norton Representation Of The
Applied Signal Source.
s s s
I V Z = . (1)
Because of input signal excitation, a voltage, V
1
, is established across the input port, a current, I
1
,
flows into this port, a current, I
2
flows into the output port, and a voltage, V
2
, is developed across
the output port. Obviously, current I
2
and voltage V
2
are constrained by the Ohms law relation-
ship,
2 2 l
I V Z = . (2)
2.1. INPUT PORT MODEL
Because the two port abstraction in Figure (1) is a linear network, both the input port
defined by the terminal pair, 12, and the output port comprised of terminal pair 34 can be
modeled by Thvenin or Norton equivalent circuits. Since no energy sources appear in either the
Course Notes #1 University of Southern California John Choma

August 2006 4 Two Port Networks

linear network or in the load that terminates the output port, the Thvenin and Norton equivalent
circuits of the input port are identical. As is depicted in Figures (2a) and (2b), these equivalent
circuits consist of a two terminal impedance, say Z
in
, which is logically termed the input imped-
ance seen by the signal source circuit. Alternatively, Z
in
is often termed the driving point
input impedance. The phrase, driving point, signifies that the impedance properties of the
input port are modeled or measured under actual load conditions, as opposed to constraints that
stipulate a specific value, say 50 ohms, zero ohms (a short circuit), or infinity ohms (an open cir-
cuit), of the load impedance, Z
l
.
Z
s
Z
in
+

V
1
I
1
+

V
s
(a).
I
s
Z
s
1
2
Z
in
+

V
1
I
1
(b).
1
2

Fig. (2). (a). Input Port Modeling Of The Linear Two Port System Diagramed
In Fig. (1a). The Signal Source Is Represented Herewith As A
Thvenin Equivalent Circuit. (b). Same As (a), Except The Signal
Source Is Modeled As A Norton Equivalent Circuit.
An inspection of either of the diagrams in Figure (2) suggests that the input impedance
at hand is
in 1 1
Z V I = , (3)
which is equivalent to the stipulation,
s
s
in
Z 0
1
V
Z .
I
lim

=


(4)
Equation (4) is a potentially useful alternative to (3) in that a KVL relationship written around
the mesh containing the network input port is invariably required to deduce the transfer proper-
ties of the two port system. Accordingly, the input port equation of equilibrium written to dis-
cern the I/O transfer properties of the considered two port network can be massaged in
accordance with (4) to evaluate the driving point input impedance as well.
In terms of the applied signal excitation, the current, I
1
, flowing into the input network
port is
s s
1
s in s in
V Z
s
I I
Z Z Z Z


= =

+ +

, (5)
while the voltage, V
1
, developed across the input, port is
( )
in
1 s in
in s
Z
V V Z
Z Z


= =

+

s s
Z I . (6)
Course Notes #1 University of Southern California John Choma

August 2006 5 Two Port Networks

As expected, (5) confirms a null input current when the input port emulates an open circuit (Z
in
=
) and an input current identical to the Norton equivalent signal source current when the input
port behaves as a short circuit (Z
in
= 0). This is to say that the current drawn from the signal
source by the two port system is maximized at I
s
under a short circuit input port condition and
minimized to zero under an open circuit input port environment. On the other hand, (6) shows
that the input port voltage, V
1
, is maximized to a value that equals the Thvenin source voltage
when Z
in
is infinitely large, while V
1
reduces to zero when the input port is a short circuit.
It is interesting to note that at the extremes of a short circuited input port, for which the
input port voltage collapses to zero, and an open circuited input port, where the input port current
is null, zero power is transferred from the signal source to the subject input port. For all other
input port impedance conditions, where neither input port voltage V
1
nor input port current I
1
is
zero, the power delivered to the port is nonzero. It is hardly a leap of faith to surmise that for
some intermediate, nonzero and finite, value of the driving point input impedance, Z
in
, the power
transferred between the source and the input port is maximized. It can be demonstrated that
maximum signal power is indeed transferred to the input terminal pair of a linear two port net-
work when the driving point impedance is conjugately matched to the Thvenin source
impedance; that is, Z
in
= Z
s
*[1]
. This so-called match terminated condition, for which neither the
input port current nor the input port voltage is either maximal or minimal, is a routine design
criterion in high performance electronic systems implemented to process extremely low level
source signals. An example of such a system is the ubiquitous cellular telephone. The source
signal produced by the antenna in these telephones is often characterized by a very small
Thvenin equivalent voltage source. The anemic amplitude of this signal voltage places it at risk
of being masked by unavoidably present electrical noise or parasitic electromagnetic energy. To
minimize the risk of losing this signal and thus, to maximize the probability of signal capture,
the input port impedance of the two port amplifier to which the antenna is coupled is almost
always designed to match terminate to the antenna, thereby ensuring the transfer of maximum
signal power for subsequent amplification and other requisite processing.
2.2. OUTPUT PORT MODELS
The output port model of the network in Figure (1) is marginally more complicated
than is the input port model because energy is supplied to the output terminal pair, 34, by the
signal source voltage, V
s
. Specifically, the Thvenin equivalent output port circuit is not a sim-
ple impedance element. Instead, it consists of the series interconnection of the driving point out-
put impedance, Z
out
, and a voltage source, say V
2o
, whose value is the open circuit (meaning Z
l
=
) value of the voltage response evidenced across the output port as a result of the applied input
port signal energy. Equivalently, the output port can be represented as a Norton topology
consisting of Z
out
in shunt with the short circuit (meaning Z
l
= 0) value, say I
2o
, of the output port
current, I
2
. These assertions are effectively summarized by Figure (3), which confirms an output
port voltage, V
2
, given by
( )
l
2 2o l out 2o
l out
Z
V V Z Z
Z Z


= =

+

I , (7)
and an output port current, I
2
, of
Course Notes #1 University of Southern California John Choma

August 2006 6 Two Port Networks

Z
out
Z
l
+

V
2
I
2
+

V
2o
(a).
I
2o
Z
out
3
4
Z
l
+

V
2
I
2
(b).
3
4

Fig. (3). (a). Output Port Modeling Of The Linear Two Port System Dia-
gramed In Fig. (1a) With A Thvenin Equivalent Circuit. (b).
Norton Output Port Modeling Of The Linear Two Port System Of
Fig. (1a).
2o out
2
l out out l
V Z
2o
I I . (8)
Z Z Z Z


= =

+ +

Note that maximum output port voltage is achieved for Z
l
= and, as expected, this maximum
response is the Thvenin output voltage, V
2o
. Moreover, the maximum output current is the
Norton current, I
2o
, which is realized for Z
l
= 0.
Since the network undergoing modeling is a linear circuit, the Thvenin output voltage,
V
2o
, as well as the Norton output current, I
2o
, is proportional to every branch and node variable in
the considered two port system. Thus, for example, V
2o
is linearly related to the source voltage,
V
s
, the input port voltage, V
1
, the input port current, I
1
, the Norton source current, I
s
, and so forth.
An analogous statement applies to I
2o
. It is useful to capitalize on these observations through an
exploitation of gain and transfer relationships commonly used to define the electrical properties
of two port configurations. For example, and with reference to Figure (1), the voltage gain, A
v
,
of the system is
2
v
s
V
A
V
= , (9)
and the forward transimpedance, Z
f
, which defines the effect exerted on output port voltage by
input signal current, is
2
f
s
V
Z .
I
= (10)
Since the Thvenin output port voltage is the open circuit value of said port voltage, it is hardly
an analytical stretch to define a Thvenin voltage gain, A
vo
, and a Thvenin forward transim-
pedance, Z
fo
, in accordance with
l
2o 2
vo
s s
Z
V V
A
V V
=
= , (11)
and
Course Notes #1 University of Southern California John Choma

August 2006 7 Two Port Networks

l
2o 2
fo
s s
Z
V V
Z .
I I
=
= (12)
Similarly, if the forward transadmittance, Y
f
, and the current gain, A
i
, derive from the respective
transfer relationships,
2
f
s
I
Y
V
= (13)
and
2
i
s
I
A ,
I
= (14)
l
2o 2
fo
s s
Z 0
I I
Y , (15)
V V
=
=
is the Norton forward transadmittance, and
l
2o 2
io
s s
Z 0
I I
A .
I I
=
= (16)
Z
out
Z
l
+

V
2
I
2
+

A V
vo s
(a).
Y V
fo s
Z
out
3
4
Z
l
+

V
2
I
2
(c).
3
4
Z
out
Z
l
+

V
2
I
2
+

Z I
fo s
(b).
3
4
A I
io s
Z
out
Z
l
+

V
2
I
2
(d).
3
4

Fig. (4). (a). Thvenin Output Port Model With Thvenin Voltage Cast As A
Proportionality Of The Signal Source Voltage, V
s
. (b). Thvenin Output Port
Model With Thvenin Voltage Cast As A Proportionality Of The Signal
Source Current, I
s
. (c). Norton Output Port Model With Norton Current Cast
As A Proportionality Of The Signal Source Voltage, V
s
. (d). Norton Output
Port Model With Norton Current Cast As A Proportionality Of The Signal
Source Current, I
s
.
Course Notes #1 University of Southern California John Choma

August 2006 8 Two Port Networks

represents the Norton current gain. Equations (11), (12), (15), and (16) lead to the alternative
output port models delineated in Figure (4). One satisfying aspect of these alternate equivalents
is that they each underscore Z
out
as a bona fide Thvenin impedance established at the output port
of the generalized network in Figure (1). This statement follows from the fact that a Thvenin
impedance at any node pair of a linear network is always computed under the condition of null
independent sources of excitation. Since the original system in Figure (1) has but a single
independent source of energy in particular, V
s
or I
s
,that setting V
s
or I
s
to zero casts Z
out
as an
impedance directly incident with the output terminal pairs of the original two port network.
2.3. TWO PORT PARAMETERS OF IDEAL AMPLIFIERS
The preceding subsection of material leads directly to a compact disclosure of the volt-
ampere relationships of idealized two port amplifiers, whose characteristics are often exploited in
first order analyses of filters, cascaded gain stages, and other types of analog signal processors.
Moreover, the mathematical interpretation of the volt-ampere equations of these idealized elec-
tronic structures serve as an excellent means of introducing more general two port relationships,
which are addressed in subsequent sections of material. To these ends, four types of amplifiers
merit attention.
2.3.1. Ideal Voltage Amplifier
The ideal voltage amplifier in Figure (5a), which behaves electrically as a voltage con-
trolled voltage source (VCVS), is characterized by infinitely large input impedance, zero output
impedance, and constant, frequency independent, open circuit (or Thvenin) voltage gain. At
low signal frequencies, the electrical signature of commercially available operational amplifiers
emulates that of an ideal voltage amplifier. Recalling the generalized input port and output port
models of Figures (2a) and (4a), respectively, Figure (5b) is straightforwardly contrived as the
relevant two port equivalent circuit. In the latter diagram, the infinitely large input impedance
allows an exploitation of the fact that the input port voltage, V
1
, and the signal source voltage, V
s
,
are identical. The model at hand clearly shows that the input port current, I
1
, is zero and the out-
put port voltage, V
2
, is A
vo
V
1
. It follows that the port variables of the ideal voltage amplifier
interrelate in accordance with the matrix volt-ampere expression,
1
vo
2 2
I V
0 0
.
A 0
V I


=




1
(17)
It is worthwhile interjecting that the output voltage, V
2
, and the input port voltage, V
1
, are in
phase with one another if A
vo
> 0. For A
vo
< 0, the ideal voltage amplifier is said to be a phase
inverting amplifier.
In (17), observe that input port current I
1
and output port voltage V
2
are cast as depend-
ent electrical variables, while input port voltage V
1
and output port current I
2
serve as independ-
ent variables. In view of this codification of network variables, the four parameters (three of
which happen to be zero in this specific case) in the 2x2 matrix on the right hand side of (17) are
traditionally called the hybrid gparameters of the two port network under consideration. The
subject matrix itself is termed the gparameter matrix of the considered network.
Course Notes #1 University of Southern California John Choma

August 2006 9 Two Port Networks

Z
l
+

V
2
I
2
+

A V
vo 1
3
4
Ideal
Voltage
Amplifier
Z
s
Z
l
+

V
1
I
1
+

V
s
I
2
+

V
2
1 3
2 4
(a).
Z
s
+

V
1
I
1
+

V
s
1
2
(b).

Fig. (5). (a). Voltage-Driven Ideal Voltage Amplifier Terminated At Its
Output Port In A Load Impedance, Z
l
. (b). The Two Port
Equivalent Circuit Of The Terminated Amplifier System In (a).
2.3.2. Ideal Transimpedance Amplifier
Figure (6a) abstracts an ideal transimpedance amplifier, the electrical characteristics of
which emulate those of a current controlled voltage source (CCVS). As is suggested by its
equivalent circuit in Figure (6b), the ideal transimpedance amplifier has zero input impedance,
zero output impedance, and a frequency invariant forward transresistance, R
fo
. Transimpedance
amplifiers are generally formed by coalescing a phase inverting voltage amplifier with appropri-
ate negative feedback. They are commonplace in optoelectronic sensor and communication net-
works, for which the signal source is a current produced by a reverse biased, and thus high
source impedance, photodiode.
Ideal
Amplifier
Transimpedance Z
l
+

V
1
I
1
I
2
+

V
2
(a).
I
s
Z
s
1
2
3
4
Z
l
+

V
2
I
2
+

R I
fo 1
3
4
+

V
1
I
1
I
s
Z
s
1
2
(b).

Fig. (6). (a). Current-Driven Ideal Transimpedance Amplifier Terminated
At Its Output Port In A Load Impedance, Z
l
. (b). The Two Port
Equivalent Circuit Of The Terminated Amplifier System In (a).
Note That Zero Input Impedance Forces The Input Port Current,
I
1
, To Be Identical To The Signal Source Current, I
s
.
Course Notes #1 University of Southern California John Choma

August 2006 10 Two Port Networks

The model in Figure (6b) confirms an input port voltage, V
1
, of zero and an output port
voltage, V
2
, of R
fo
I
1
. Accordingly, the port variables of the ideal transimpedance amplifier derive
from the matrix volt-ampere expression,
1
fo
2 2
V I 0 0
.
R 0
V I

=


1
(18)
With input voltage V
1
and output voltage V
2
serving as dependent variables and input current I
1

and output current I
2
used as independent electrical variables, the square matrix on the right hand
side of (18) is the z-parameter matrix of the network undergoing investigation. Moreover, the
four elements within this matrix are known as the open circuit impedance, or more simply, the z
parameters, of the network.
2.3.3. Ideal Transadmittance Amplifier
The ideal transadmittance amplifier offered in Figure (7a) functions as a voltage-con-
trolled current source (VCCS). As is confirmed by the model given in Figure (7b), it is effec-
tively the dual of the ideal transimpedance amplifier in that it is characterized by infinitely large
input impedance, infinitely large output impedance, and a forward transconductance, G
fo
, that is
independent of signal frequency. Transadmittance amplifiers are also known as transconductors
or operational transconductor amplifiers and find utility in wideband electronics realized in high-
speed current mode technology. They also comprise the circuit foundation for analog integra-
tors, which, in turn, are used to realize active biquadratic filters, inductorless oscillators, and
gyrators.
Ideal
Amplifier
Transadmittance
Z
s
Z
l
+

V
1
I
1
+

V
s
I
2
+

V
2
(a).
1 3
2 4
G V
fo 1
Z
l
+

V
2
I
2
3
4
Z
s
+

V
1
I
1
+

V
s
1
2
(b).

Fig. (7). (a). Ideal Transadmittance Amplifier Terminated At Its Output
Port In A Load Impedance, Z
l
. (b). The Two Port Equivalent
Circuit Of The Terminated Amplifier In (a).
From the equivalent circuit in Figure (7b), it is apparent that the I/O port volt-ampere
equations subscribe to
1
fo
2 2
I V 0 0
.
G 0
I V

=


1
(19)
Course Notes #1 University of Southern California John Choma

August 2006 11 Two Port Networks

With input voltage V
1
and output voltage V
2
serving as independent variables and input current I
1

and output current I
2
exploited as dependent electrical variables, the square matrix on the right
hand side of (19) becomes the y-parameter matrix of the network undergoing investigation.
Moreover, the four elements within this matrix are known as the short circuit admittance, or y
parameters, of the system. Observe that the model in Figure (7b) reflects the subcircuit of Figure
(4c), subject to the proviso that the infinitely large input impedance constrains the input port
voltage, V
1
, to be identical to the source voltage, V
s
.
2.3.4. Ideal Current Amplifier
The ideal current amplifier abstracted in Figure (8a) operates as a current-controlled
current source (CCCS). This two port network features zero input impedance, infinitely large
output impedance, and a forward current gain, A
io
, that is independent of signal frequency. Its
resultant electrical model, which is predicated on the structure in Figure (4d), appears in Figure
(8b). The most common application of a current amplifier is current buffering (also known as
cascoding) in conjunction with the output port of transadmittance amplifiers. In a cascode
application, a current amplifier can significantly increase overall circuit bandwidth, particularly
if the pragmatic implementation of a transadmittance amplifier results in significant capacitive
feedback between the output and input ports of the transadmittance unit.
Ideal
Current
Amplifier
Z
l
+

V
1
I
1
I
2
+

V
2
(a).
I
s
Z
s
1
2
3
4
A I
io 1
Z
l
+

V
2
I
2
3
4
+

V
1
I
1
I
s
Z
s
1
2
(b).

Fig. (8). (a). Ideal Current Amplifier Terminated At Its Output Port In A
Load Impedance, Z
l
. (b). The Two Port Equivalent Circuit Of
The Current Amplifier In (a).
The model in Figure (8b) leads to the volt-ampere matrix relationship,
1
io
2 2
V I
0 0
.
A 0
I V


=




1
(20)
With input voltage V
1
and output current I
2
serving as dependent electrical variables and input
current I
1
and output voltage V
2
identified as independent variables, the square matrix on the
right hand side of (20) becomes the hybrid h-parameter matrix of the considered amplifier.
Moreover, the four elements within this matrix are known as the hybrid h-parameters of the sys-
Course Notes #1 University of Southern California John Choma

August 2006 12 Two Port Networks

tem. Because of the short circuit at the network input port, note that the input port current, I
1
, is
identical to the signal source current, I
s
.
2.4. GENERALIZED TWO PORT PARAMETERS
The abstraction in Figure (1) and the specialized cases summarized by Figures (5) -
through- (8) proffer two lessons. The first is that the two types of driving point specifications
(input and output impedance or admittance) and the four types of transfer properties (voltage
gain, transimpedance, transadmittance, and current gain) can derive from the measurable input
and output port voltages (V
1
and V
2
) and the input and output port currents (I
1
and I
2
). The sec-
ond lesson is that these observable performance metrics can be quantified analytically even if the
circuit architecture implicit to the two port configuration of Figure (1) is unknown, inaccessible,
or simply too intricate for straightforward mesh or nodal circuit analyses. The upshot of the mat-
ter is that if jumping into the two port box to execute appropriate analyses is impossible or
impractical, only two equilibrium equations can be written. One of these equations focuses on
the input port, where the source energy, source impedance, and the input port variables, V
1
and
I
1
, reside, while the other addresses the output port, where the load impedance and the output
port variables, V
2
and I
2
, prevail. Since only two equations in the four variables, V
1
, I
1
, V
2
, and
I
2
, can be written, the formulation of a unique network solution mandates that two of these vari-
ables be viewed as independent, and the resultant two be interpreted as dependent, variables. A
viable solution also requires that V
2
and I
2
be related through the branch properties of the load
termination, while V
1
and I
1
must be uniquely constrained by the source excitation and source
impedance. The selection of the independent and dependent variable sets is arbitrary, subject to
the proviso that the corresponding two port parameters that define the electrical properties of the
network are meaningfully definable and measurable.
2.4.1. Hybrid hParameters
A hybrid h-parameter model of a linear two port network derives from choosing input
port current I
1
and output port voltage V
2
as independent electrical variables. Resultantly, input
port voltage V
1
and output port current I
2
are dependent variables. Because the two port network
undergoing scrutiny is linear, each dependent variable is a linear superposition of the effects of
each independent variable. This observation gives rise to the symbolic volt-ampere relation-
ships,
1 11 1 12 2
2 21 1 22 2
V h I h V
,
I h I h V
= +
= +
(21)
or in matrix format,
1 11 12 1
2 21 22
V h h I
.
I h h V

=

2
(22)
Observe that the ideal current amplifier relationship in (20) is a special case of (22); specifically,
it is the special case of h
11
= 0, h
12
= 0, h
22
= 0, and h
21
= A
io
. The h-parameters are viewed as
hybrid because of their differing dimensional units. To this end, note in (21) and (22) that
dimensional consistency requires that h
11
be an impedance, h
22
be an admittance, and h
12
and h
21

be dimensionless.
Course Notes #1 University of Southern California John Choma

August 2006 13 Two Port Networks

A well-known fact of basic linear circuit theory is that an analysis of a linear circuit
produces a set of linear equations. It follows that a set of linear equations, such as is postulated
by (21) and (22), corresponds to a linear circuit, which becomes known as an electrical model of
the phenomenology addressed by the equation set. Circuit modeling is therefore the inverse of
circuit analysis; that is, meaningful equations derive from circuit analysis, while ostensibly use-
ful electrical models are premised on pertinent equilibrium equations.
In the case of (21) and (22), the so called h-parameter model, or h-parameter equivalent
circuit, of the linear two port network in Figure (1) is the topological structure shown in Figure
(9). The subject model mirrors basic linear circuit theory, which stipulates that any port of a lin-
ear circuit can be modeled by either a Thvenin or a Norton equivalent circuit. Recall that
Thvenin and Norton models for a simple one port configuration are cast in terms of the
independent electrical variables that excite that port. In Figure (9), observe that the Norton
equivalent output port current, h
21
I
1
, and the Thvenin equivalent input port voltage, h
12
V
2
, are
respectively proportional to the variables, I
1
and V
2
, which have been selected as independent
electrical variables in the course of formulating the two port equations of equilibrium.
Linear
Two Port
Network
+

V
1
+

V
1
I
1
I
1
I
2
I
2
+

V
2
+

V
2
1
1
3
3
2
2
4
4
+

h
22
h
11
h I
21 1
h V
12 2

Fig. (9). The h-Parameter Equivalent Circuit Of A Linear Two Port Net-
work. Parameter h
11
Is An Impedance, h
22
Is An Admittance,
And h
12
And h
21
Are Dimensionless Parameters.
Measurement procedures for the hybrid h-parameters derive directly from (21) or (22).
For example, if the output port voltage, V
2
, is clamped to zero, which corresponds to the short
circuited output port drawn in Figure (10a),
2
2
1
11
1
V 0
2
21
1
V 0
V
h
I
.
I
h
I
=
=
=
=
(23)
It follows that h
11
represents the short circuit (meaning that the output port is a short circuit)
input impedance of the two port undergoing examination, while h
21
designates a forward short
circuit current gain. Note, therefore, that h
11
is a particular value of the driving point input
Course Notes #1 University of Southern California John Choma

August 2006 14 Two Port Networks

impedance addressed in conjunction with Figure (2); specifically, h
11
is the driving point input
impedance under the special case of a short circuited termination of the network output port. On
the other hand, h
21
is a measure of the forward gain of the network since it defines a value for the
output port current corresponding to a given input port current. More correctly, h
21
defines the
maximum possible forward current gain in view of the fact that a short circuited load termination
is certainly conducive to maximal output port current.
Linear
Two Port
Network
+

V = h I
1 11 1
I
1
I = h I
2 21 1
+

V = 0
2
1 3
2 4
T
e
s
t
C
u
r
r
e
n
t

S
o
u
r
c
e
S
h
o
r
t
C
i
r
c
u
i
t
(a).
Linear
Two Port
Network
+

V = h V
1 12 2
I = 0
1
I = h V
2 22 2
+

V
2
1 3
2 4
O
p
e
n
C
i
r
c
u
i
t
T
e
s
t
V
o
l
t
a
g
e
S
o
u
r
c
e
(b).
+

Fig. (10). (a). Measurement Of The Short Circuit Hybrid h-Parameters. (b). Measure-
ment Of The Open Circuit Hybrid h-Parameters.
With I
1
= 0, which reflects the open circuited input port diagrammed in Figure (10b),
(21) or (22) yield
1
1
1
12
2
I 0
2
22
2
I 0
V
h
V
.
I
h
V
=
=
=
=
(24)
The parameter, h
12
, is termed the reverse voltage gain, or the h-parameter feedback factor, of a
two port network. It is literally the maximum possible reverse voltage gain since the condition,
I
1
= 0, is identical to a no load condition at the input port. A two port network, and particularly
an active two port network, is naturally thought of as a system capable of delivering very large
h
21
so that maximal output signal is generated in response to input port excitation. But a portion
of the output response can be returned, or fed back, to the input port because of the electrical
nature of the devices implicit to the two port network and the manner in which the elements of
the two port are laid out or interconnected. Feedback can also be manifested by the electrical
nature of the package in which the electronic circuit is embedded. Feedback can be an undesir-
able phenomenon, as is the case with packaging anomalies and when bipolar and MOS technol-
ogy transistors are operated at high signal frequencies. It can also be a specific design objective,
as when feedback signal flow paths are appended around active subcircuits to optimize overall
Course Notes #1 University of Southern California John Choma

August 2006 15 Two Port Networks

circuit response. Regardless of the source of network feedback, parameter h
12
is its measure in
an h-parameter emulation of linear network I/O performance.
Yet another useful interpretation of h
12
is that of an isolation factor between output and
input ports, with the understanding that h
12
= 0 represents perfect isolation and large h
12
infers
poor isolation, and hence good coupling, of the output port -to- the input port. In an attempt to
clarify this assertion, observe in Figure (10b) that
1 11 1 12 2
V h I h V = + , (25)
whence V
1
/I
1
, which is literally the driving point input impedance of the network, is
1
11 12
1 1
V V
h h
2
I I


= +


. (26)
In (26), the ratio, V
2
/I
1
, is the forward transimpedance of the circuit undergoing examination.
For a fixed input current, I
1
, this transimpedance function is certainly influenced by the load
termination, which supports the output port voltage V
2
. For example, a short circuited load
necessarily renders V
2
= 0, whereby (26) confirms an input impedance that is identical to h
11
.
This result is certainly synergistic with the definition of h
11
in (23). But note that the same
impedance result, V
1
/I
1
= h
11
, is obtained if h
12
= 0. Evidently h
12
= 0 decouples, or isolates, the
output port from its input counterpart in the sense that the input port does not respond to any out-
put port voltage perturbation induced by load fluctuations, parasitic signal coupling, or other
phenomena.
In an active two port system, it is generally essential to ensure |h
21
| >> |h
12
|; that is, the
magnitude of the maximum possible forward current gain is desirably much larger than the
magnitude of the maximum possible reverse voltage gain. This design requirement is clearly
satisfied when the I/O ports are perfectly isolated, in which case the subject two port becomes
known as a unilateral network. The term, unilateral, refers to an ability of a network to trans-
mit signal between I/O ports in only one direction; in this case, from the input port -to- the output
port. A passive network can be shown to have h
21
= h
12
, which implies |h
21
| |h
12
|. Any linear
network for which h
21
= h
12
is said to be bilateral, which means that signal can be propagated
equally well from input -to- output ports and from output -to- input ports.
Finally, h
22
in (24) is the open circuit (meaning that the input port is open circuited)
output admittance. Parameter h
22
is the infinitely large source impedance value of the driving
point output admittance, 1/Z
out
. For h
22
= 0, the output port in Figure (9) emulates an ideal cur-
rent source.
A measurement complication arises when the two port network undergoing investiga-
tion is an active system. This complication stems from the fact that the active devices embedded
within an active topology are inherently nonlinear. Fortunately, these devices behave as approxi-
mately linear circuit elements when appropriate static voltages and currents, separate and apart
from the input signal source, are applied to bias them within the linear regime of their volt-
ampere characteristics. Unfortunately, a short circuited output port, which is required in the
measurement of parameters h
11
and h
21
, and an open circuit input port, which is a prerequisite for
the determination of parameters h
12
and h
22
, are likely to upset the requisite biasing. But on the
assumption that the test sources appearing in Figure (10) are sinusoids having zero average
value, the biasing dilemma can be circumvented by implementing a short circuit with a suffi-
Course Notes #1 University of Southern California John Choma

August 2006 16 Two Port Networks

ciently large capacitance and an open circuit with a sufficiently large inductance. Figure (11)
portrays these topological modifications to the h-parameter measurement strategy. Note in this
figure that the load and source impedances are not removed since the low frequency values of
these elements may be exploited in the network biasing scheme.
Linear Active
Two Port
Network
+

V = h I
1 11 1
I
1
I = h I
2 21 1
+

V 0
2

1 3
2 4
T
e
s
t
C
u
r
r
e
n
t

S
o
u
r
c
e
L
a
r
g
e
(a).
Linear Active
Two Port
Network
+

V = h V
1 12 2
I 0
1
I = h V
2 22 2
+

V
2
1 3
2 4
T
e
s
t
V
o
l
t
a
g
e
S
o
u
r
c
e
(b).
+

Z
l
Z
l
Z
s
Z
s
Biasing
Large
Biasing

Fig. (11). (a). Measurement Of The Short Circuit Hybrid h-Parameters Of A Linear Active Two Port Net-
work. (b). Measurement Of The Open Circuit Hybrid h-Parameters Of A Linear Active Two
Port.
Although the test fixturing in Figure (11) is conceptually correct from the perspective
of mitigating biasing difficulties, it is unfortunately impractical when applied to most linear cir-
cuits, and it is especially inappropriate when applied to broadband or high frequency linear
active two port networks. The fundamental problem is that very few linear active networks, and
almost no broadband active networks, are unconditionally stable for all possible source and load
terminations. This is to say, that the majority of linear active networks display a propensity for
parasitic oscillations for certain values of source and load impedances. Since the amplitudes and
frequencies of these oscillations are independent of signal source energy, the two port network in
question ceases to be a linear network, thereby invalidating all two port analytical disclosures at
this juncture. Open circuit source and load ports are especially troublesome, but even short cir-
cuit I/O terminations can induce oscillations. For this reason, the h-parameters and other two
port parameter sets for active two port networks are rarely measured directly. Instead, these met-
rics are usually measured indirectly by calculating them as a function of the measured scattering,
or S-, parameters
[2]-[4]
. Like the h-parameters, the S-parameters quantify the driving point, for-
ward transfer, and feedback properties of a two port system. But unlike h- and other two port
parameters to be discussed shortly, S-parameters derive under the operating condition of finite,
nonzero, and equal source and load impedances that are chosen to ensure that the network under
test operates as a stable entity. Invariably, this equal source and load test termination, which is
called the measurement reference impedance, is usually 50, 75, or 300 ohms.
Course Notes #1 University of Southern California John Choma

August 2006 17 Two Port Networks

EXAMPLE #1:
Figure (12) depicts the low frequency, small signal equivalent circuit of a common
emitter bipolar junction transistor amplifier that utilizes an emitter degeneration resis-
tance, R
e
, to achieve a forward gain that is nominally independent of the transistor
current gain parameter, . With reference to Figure (1), note that electrical ground is
common between the input and output ports so that the port terminals, 2 and 4, are
one and the same ground node. Let the amplifier input resistance, r
i
, be 2.2 K, the
transistor output resistance, r
o
, is 25 K, = 90, and R
e
= 80 . Derive general
expressions for, and numerically evaluate, the four h-parameters, h
ij
, of the emitter
degenerated configuration.
r
i
R
e
I
1
r
o
V
1
I
1
I
2
1 3
2
V
2
4

Fig. (12). The Small Signal Equivalent Circuit Of
The Amplifier Addressed In Example #1.
SOLUTION:
(1). Figure (13a) is the equivalent circuit appropriate for evaluating the short circuit h-
parameters, h
11
and h
21
. In particular, the subject figure is Figure (12) drawn
under the condition of a short circuited output port. A KVL equation written for
the output port delivers
( ) (
o 2 1 e 1 2
0 r I
)
I R I I = + + ,
whence
e o 2
h 89.71 amps / amp .
21
1 e o
1 - R r I
I 1 R r
= = =



+


Armed with this result, KVL around the input loop yields
( ) ( )
e o
V r I R R ,
1 i 1 e 1 2 i e 1 e
e o
1 R r
1
I I r R I
1 + R r
= + =

+ + +


I
from which
( )
( )
1
h = = r = 9.46 K .
11 i o e
1
V
1 r R
I
+ +

(2). Figure (13b) is the diagram appropriate for the computation of h-parameters h
12

and h
22
. With I
1
= 0, it is clear that
( )
e
V
2 o e 2
V R
1 2
r R I
I
=
=
+

Course Notes #1 University of Southern California John Choma

August 2006 18 Two Port Networks

r
i
R
e
I
1
r
o
V
1
I
1
I
2
1 3
2
V = 0
2
4
I I
2 1

I + I
1 2
(a).
r
i
R
e
I
1
r
o
V
1
I = 0
1
I
2
1 3
2
V
2
4
I
2
I
2
(b).

Fig. (13). (a). Model Used To Calculate h-Parame-
ters h
11
and h
21
In The Circuit Of Figure
(12). (b). Model Appropriate For The
Calculation Of h-Parameters h
12
and h
22
In
Figure (12).
and thus,

2
h = = = 39.87 S
22
2 e o
e 1
h = = = 3.19 mVolts/Volt .
12
2 e o
I
1
V R r
R V
V R r
+
+

COMMENTS: In addition to expediting amplifier modeling and circuit analysis, the
h-parameters are seen herewith as highlighting the electrical effects
of an appended circuit element (in this case, the emitter degeneration
resistance, R
e
). For example, by inspecting the R
e
= 0 (no emitter
degeneration) values of the four h-parameters, the following
conclusions are drawn. First, emitter degeneration minimally
impacts the forward short circuit current gain, h
21
and the open circuit
output port conductance, h
22
, provided R
e
<< r
o
. On the other hand,
the emitter degeneration resistance substantively increases the open
circuit input resistance, h
11
, and is literally the cause of nonzero isola-
tion, h
12
, between the input and output amplifier ports.
Course Notes #1 University of Southern California John Choma

August 2006 19 Two Port Networks

2.4.2. Hybrid gParameters
The independent and dependent electrical variable sets used to define the hybrid g-
parameters are the converse of those used in conjunction with the hybrid h-parameters. Specifi-
cally, the independent variables for g-parameter modeling are the input port voltage, V
1
, and the
output port current, I
2
, thereby rendering the input port current, I
1
, and the output port voltage,
V
2
, dependent electrical quantities. From superposition theory,
1 11 12
2 21 22
I g g V
,
V g g I

=


1
2
22 21 vo
(27)
which gives rise to the g-parameter equivalent circuit depicted in Figure (14). Dimensional
consistency requires g
11
to be an admittance, g
22
a resistance, and g
12
and g
21
both dimensionless.
Observe that for the special case of an ideal voltage amplifier, whose volt-ampere equations are
given by (17), g
11
= 0, g
12
= 0, g = 0, and g = A .
Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
+

V
2
I
2
I
1
+

V
1
3 1
4 2
+

g
11
g
22
g I
12 2
g V
21 1

Fig. (14). The g-Parameter Equivalent Circuit Of A Linear Two Port
Network. Parameter g
11
Is An Admittance, g
22
Is An
Impedance, And g
12
And g
21
Are Dimensionless Parameters.
As with the hybrid h-parameters, the measurement strategy for the hybrid g-parameters
derives directly from the defining volt-ampere modeling relationships. To wit, (27) suggests that
parameters g
11
and g
21
are measured or evaluated under an open circuited output port condition,
while g
12
and g
22
evolve from a short circuited input port constraint. Specifically,
2
2
1
11
1
I 0
2
21
1
I 0
I
g
V
,
V
g
V
=
=
=
=
(28)
which casts g
11
as an open circuit (meaning that the output port is open circuited) input port
admittance and g
21
as an open circuit forward voltage gain of the two port network. With refer-
Course Notes #1 University of Southern California John Choma

August 2006 20 Two Port Networks

ence to Figure (1), g
11
is the inverse of the input impedance, Z
in
, under the special case of I
2
= 0,
while g
21
is the input port -to- output port Thvenin voltage gain. Continuing in (27),
1
1
1
12
2
V 0
2
22
2
V 0
I
g
I
.
V
g
I
=
=
=
=
(29)
Thus, g
22
is the short circuit (meaning that the input port is short circuited) output impedance of
the two port. It is, in fact, the output impedance, Z
out
, in Figure (1) under the special case of a
short circuit source impedance termination at the input port. Parameter g
12
, like h-parameter h
12
,
is a measure of feedback from the output port -to- the input port. Equivalently, g
12
measures the
degree of isolation between input and output network ports. The salient implications of (28) and
(29) are summarized topologically by Figure (15).
Linear
Two Port
Network
+

V
1
I = g V
1 11 1
I = 0
2
+

V = g V
2 21 1
1 3
2 4
S
h
o
r
t
C
i
r
c
u
i
t
(a).
Linear
Two Port
Network
+

V = 0

1
I = g I
1 12 2
I
2
+

V = g I
2 22 2
1 3
2 4
O
p
e
n
C
i
r
c
u
i
t
T
e
s
t
V
o
l
t
a
g
e

S
o
u
r
c
e
(b).
+

T
e
s
t
C
u
r
r
e
n
t

S
o
u
r
c
e

Fig. (15). (a). Measurement Of The Open Circuit Hybrid g-Parameters. (b). Measurement
Of The Short Circuit Hybrid g-Parameters.
At this juncture, two alternatives for the modeling and ultimate analysis of a linear two
port network have been documented. If h-parameters are selected as the modeling vehicle, the
applicable equivalent circuit is the structure appearing in Figure (9), while if g-parameters are
preferred, the applicable model is the topology of Figure (14). When either modeling approach
is adopted for a given two port system, both must necessarily predict identical driving point, for-
ward transfer, and feedback properties. This homogeneity constraint implies that the h- and g-
parameters are not mutually independent network metrics; that is, the individual h- and g-
parameters must be related.
The nature of the interrelationship between the h- and g-parameters of linear two port
network is rendered transparent by a comparison of (27) with (22). In particular, the g-parameter
matrix on the right hand side of (27) is the inverse of the h-parameter matrix on the right hand
side of (22); that is,
Course Notes #1 University of Southern California John Choma

August 2006 21 Two Port Networks

1
11 12 11 12
21 22 21 22
g g h h
,
g g h h

=


(30)
and conversely,
1
11 12 11 12
21 22 21 22
h h g g
.
h h g g

=


(31)
With
h 11 22 12
h h h h =
21
(32)
representing the determinant of the h-parameter matrix, (30) requires
22
11
h
11
22
h
12
12
h
21
21
h
h
g

h
g

.
h
g

h
g

=
=
=
=
(33)
From (31),
22
11
g
11
22
g
12
12
g
21
21
g
g
h

g
h

,
g
h

g
h

=
=
=
=
(34)
where
g 11 22 12 21
g g g g = (35)
is the determinant of the g-parameter matrix.
Course Notes #1 University of Southern California John Choma

August 2006 22 Two Port Networks

It is interesting to note that although in principle, the selection of the type of parameters
exploited to model a two port network is arbitrary, the actual choice may be dictated by network
pragmatics. For example, observe in (33) that the g-parameters are not finite, and thus are not
numerically deterministic, if the determinant of the h-parameter matrix is zero. Zero determinant
defines matrix singularity and thus, g-parameters cannot be defined if the corresponding h-
parameter matrix is singular. Moreover, (34) shows that h-parameters cannot be evaluated
numerically if the g-parameter matrix is singular, that is if
g
= 0. It follows that some two port
networks, and particularly active two ports, may have an h-parameter modeling representation in
lieu of a g-parameter equivalent circuit and vice versa. Actually, it is possible that neither an h-
nor a g-parameter model exists, in which case other two port parameter sets must be exploited.
Finally, note that the g-parameter feedback factor, g
12
, is nominally proportional to its
h-parameter counterpart, h
12
. Similarly, the feedforward parameter, g
21
, is nominally propor-
tional to h
21
. Accordingly, both the g-parameter and the h-parameter equivalent circuits are
analytically consistent with respect to disclosing the degree of feedback, or isolation, evidenced
and also with respect to the degree of feedforward transmission observed.
EXAMPLE #2:
Figure (16) is a simplified low frequency, small signal equivalent circuit of a voltage
amplifier that exploits feedback via the resistance, R
2
. Let R
1
= 9 K, and R
2
= 1
K, and assume an amplifier output resistance, r
o
, of 100 , and an amplifier open
loop gain (gain in the absence of feedback), A
o
, of 100. Derive general expressions
for, and numerically evaluate, the four g-parameters, g
ij
, of the feedback amplifier.
R
1
R
2
+

r
o
V
1
I
1
I
2
1 3
2
V
2
4
A V
o

+
V

Fig. (16). The Small Signal Equivalent Circuit Of
The Amplifier Studied In Example #2.
SOLUTION:
(1). Figure (17a), which is the circuit diagram in Figure (16) with an open circuited
output port invoked, is the equivalent circuit for evaluating the g-parameters, g
11

and g
21
. By inspection of the input port mesh,
( )
1 1 2
V R R = +
1
I ,
so that
1
g 100 S .
11
1 1 2
I
1
V R R
= = =
+

At the output port of Figure (17a),
Course Notes #1 University of Southern California John Choma

August 2006 23 Two Port Networks

R
1
R
2
+

r
o
V
1
I
1
I
1
I = 0
2
1 3
2
V
2
4
A V
o

+
V
(a).
R
1
R
2
+

r
o
V
= 0
1
I
1
I + I
1 2
I
2
1 3
2
V
2
4
A V
o

+
V
(b).

Fig. (17). (a). Model Used To Calculate g-Parame-
ters g
11
and g
21
In The Circuit Of Figure
(16). (b). Model Appropriate For The
Calculation Of g-Parameters g
12
and g
22
In
Figure (16).
( )
1
V A V R I A R I R I
2 o 2 1 o 1 1 2 1 o 1 2
1 2
V
A R R .
R R
= + = + =
+

It follows that
2 o 1 2
g A 89.90 volts/volt
21 o
1 2 1
1 R A R V
V 1 R R
= = =

+


(2). Figure (17b) is the relevant equivalent circuit for computing g-parameters g
12
and
g
22
. With V
1
= 0, the I/O port currents are constrained by
( )
0 R I R
1 1 2 1 2
, I I = + +
from which,

1 2
g 0.10 amps/amp
12
2 1 2
I R
I R R
= = =
+

Moreover
.
( )
V r I A R I R
2 o 2 o 1 1 2 1 2
I I = + +
Course Notes #1 University of Southern California John Choma

August 2006 24 Two Port Networks

Using the preceding result for the current ratio, I
1
/I
2
, it is a simple matter to show
that
( )( )
2
g r 91.0 K .
22 o o 1 2
2
V
A 1 R R
I
= = + = +
COMMENTS: The g-parameter model underscores the fact that for the amplifier at
hand, feedback resistance R
2
results in an enormous boost of the short
circuit output resistance. It also lowers the open circuit input conduc-
tance. As expected, a feedback resistance of zero incurs no short
circuit feedback within the amplifier. Finally, R
2
has little impact on
the open circuit voltage gain of the amplifier.
2.4.3. Short Circuit yParameters
When the input port voltage, V
1
, and the output port voltage, V
2
, are chosen as
independent variables in the course of analyzing a linear two port network, the relevant volt-
ampere relationships assume the form,
1 11 12
2 21 22
I y y V
.
I y y V

=


1
2
(36)
In (36), for which the circuit-level interpretation is Figure (18), the y
ij
are called the admittance,
or y-, parameters of the two port system undergoing characterization. The y
ij
are also called the
short circuit admittance parameters because each element in the matrix on the right hand side of
(36) is an admittance function and is measured under the condition of a short circuited network
port. Specifically,
Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
+

V
2
I
2
I
1
+

V
1
3 1
4 2
y
11
y V
12 2
y V
21 1
y
22

Fig. (18). The Short Circuit Admittance Parameter Model Of A Linear
Two Port Network. Each of the y-parameters, y
ij
, Have
Units Of Admittance.
Course Notes #1 University of Southern California John Choma

August 2006 25 Two Port Networks

2
2
1
11
1
V 0
2
21
1
V 0
I
y
V
,
I
y
V
=
=
=
=
(37)
and
1
1
2
2
1
12
V 0
2
22
V 0
I
y
V
.
I
y
V
=
=
=
=
(38)
It follows that parameter y
11
is the short circuit (meaning that the output port is short circuited)
input admittance of the two port network, while y
21
, is the corresponding short circuit forward
transadmittance. Parameter y
11
is therefore related to the driving point input admittance or
impedance of the network undergoing scrutiny. On the other hand, y
21
is a measure of the for-
ward gain. In (38) y
12
is seen as a feedback factor, measured herewith as a reverse short circuit
(meaning that the input port is short circuited) transadmittance, and y
22
is the short circuit output
admittance.
Because a two port network study predicated on y-parameters must produce results that
mirror those that derive from investigations based on either h- or g-parameters, the y
ij
of a linear
network are necessarily related to both the h
ij
and the g
ij
of the same network. For example,
compare (22) with (36) to deduce that with a short circuit imposed at the network output port (V
2
= 0),
2
2
1 11
21
1 11
1
1
11
h
V 0
h
2
21
h
V 0
I
y
V
.
I
y
V
=
=
= =
= =
(39)
As expected, y
21
, which is the y-parameter measure of forward signal transmission through a lin-
ear network, is proportional to h
21
, which is the h-parameter counterpart to this network gain
metric. Moreover, recall that h
11
is the short circuit input impedance of a linear two port
network. It is therefore only fair that y
11
, which represents the short circuit input admittance, be
the inverse of h-parameter h
11
. Continuing the comparison between (22) and (36),
Course Notes #1 University of Southern California John Choma

August 2006 26 Two Port Networks

1
1
12
2 11
12 21
22
2 1
h
1
12
h
V 0
h h
2
22
h
V 0
I
y
V
.
I
y h
V
=
=
= =
= =
1
(40)
Also as expected, the y-parameter feedback parameter, y
12
, is directly proportional to the h-
parameter feedback measure, h
12
. Recalling that h
12
= h
21
for a bilateral two port network, (39)
and (40) imply that bilateralness requires y
12
y
21
. Despite the fact that both h
22
and y
22
measure
network output port admittance, y
22
does not equal h
22
. The reason underlying y
22
h
22
is that
h
22
monitors the output port admittance with the input port open circuited, while y
22
is the output
admittance under the condition of a short circuited input port.
An interesting, and often utilitarian alternative to the basic y-parameter equivalent cir-
cuit in Figure (18) results when (36) is rewritten in the form
3-Terminal
Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
+

V
2
I
2
I
1
+

V
1
3 1
4 2
y
i
y V
f 1
y
r
y
o
(a).
+

V
2
I
2
I
1
+

V
1
3 1
4 2
y
i
y
r
y
o
(b).

Fig. (19). (a). Alternative Form Of y-Parameter Equivalent
Circuit For A Three Terminal, Linear Two Port
Network Operated With A Common Terminal
Between Input And Output Ports. (b). The Model
Of (a) For The Special Case Of A Bilateral Three
Terminal Network.
Course Notes #1 University of Southern California John Choma

August 2006 27 Two Port Networks

( ) ( )
( ) ( ) ( )
1 11 1 12 2 11 12 1 12 1 2
2 21 1 22 2 21 12 1 22 12 2 12 2 1
I y V y V y y V y V V
I y V y V y y V y y V y V V
.
= + = +
= + = + +
(41)
Upon introducing the subsidiary admittance parameters,
i 11 12
r 12
f 21 12
o 22 12
y y y
y y
y y y
y y y
,
+

(42)
(41) becomes
( )
( )
1 i 1 r 1 2
2 f 1 o 2 r 2 1
I y V y V V
I y V y V y V V
,
= +
= + +
(43)
which implies the alternative y-parameter model topology offered in Figure (19a). This -type
structure requires that the original linear two port network operate with a common terminal
between its input and output ports; that is, the original configuration is effectively a three termi-
nal, two port system. The alternative structure requires only one dependent generator, portrays y
i

as a shunting input port branch admittance, y
o
as a shunting output port branch admittance, y
r
as
an I/O port coupling admittance, and y
f
as an effective forward transimpedance. For the special
case of a bilateral two port network, for which y
12
= y
21
, y
f
is zero, and the model in Figure (19a)
collapses to the true -type architecture shown in Figure (19b).
EXAMPLE #3:
g V
m 1
r
o
C
db
C
gs
C
gd
V
1
V
2
I
1
I
2
1

(
G
a
t
e
)
2

(
D
r
a
i
n
)
3 (Source)

Fig. (20). Small Signal Model Of A MOSFET Configured To
Operate As A Common Source Amplifier. The Model
Is The Focus Of Attention In Example #3.
Figure (20) is an approximate small signal model of a metal-oxide-semiconductor
field-effect transistor (MOSFET) configured to operate ultimately as a common
source amplifier. Terminal 1 represents the gate of the transistor, terminal 2 is the
transistor drain terminal, and terminal 3 is the source terminal of the device. In this
model, C
gs
represents the gate -to- source capacitance of the transistor, C
gd
is the gate
-to- drain capacitance, C
db
is the drain -to- bulk substrate capacitance (the substrate
Course Notes #1 University of Southern California John Choma

August 2006 28 Two Port Networks

terminal is presumed grounded for small signals), r
o
designates the source-drain
channel resistance, and finally, g
m
is the forward transconductance of the device.
Find general expressions for each of the four common source y-parameters, y
ij
. Also,
derive a general expression for the short circuit (meaning the drain terminal is short
circuited to the source terminal) current gain and the frequency at which the
magnitude of this short current gain degrades to unity. If g
m
= 8 mS, r
o
= 8 K, C
gs

= 15 fF, C
gd
= 3 fF, and C
db
= 12 fF, numerically evaluate this unity gain frequency,
which is commonly symbolized as f
T
.
SOLUTION:
(1). The good news about this problem is that the model in Figure (20) is identical to
that of the generalized topology shown in Figure (19a). By inspection, therefore,
and with g
o
signifying the conductance value of the resistance, r
o
,
i gs
r gd
o o db
f m
y sC
y sC
.
y g sC
y g
=
=
= +
=

It follows from (42) that
( )
( )
12 r gd
11 i 12 gs gd
22 o 12 o db gd
21 f 12 m gd
y y sC
y y y s C C
.
y y y g s C C
y y y g sC
= =
= = +
= = + +
= + =

(2). Extensive circuit analysis is not required for an evaluation of the short circuit cur-
rent gain. Instead, it is necessary only that h-parameter h
21
be recalled as the
short circuit current gain of a linear two port network. From (39),
21
21
11
y
h ,
y


and using the results of the preceding analytical step,
( )
m gd
21
gs gd
g sC
h .
s C C

=
+

(3). The short circuit current gain, h
21
, computed in the preceding solution step indi-
cates the presence of a right half plane zero at a frequency of g
m
/C
gd
, which is
2(424.4 GHz). This extremely large critical frequency is potentially beyond the
frequency range of validity of the model in Figure (20) and is likely to be signifi-
cantly larger than the unity gain frequency, which is to be computed. Accord-
ingly, if the unity gain frequency is indeed much smaller than the frequency of
the right half plane zero,
( ) ( )
m gd
m
21
gs gd gs gd
g sC
g
h .
s C C s C C

=
+ +

Course Notes #1 University of Southern California John Choma

August 2006 29 Two Port Networks

Under steady state sinusoidal signal conditions, s is replaced by j in the preced-
ing disclosure, with the result that
( )
m
21
gs gd
g
h .
j C C

+

Clearly, the magnitude of h
21
decreases progressively with increasing signal and
becomes one at
( )
m
T
gs gd
g
f 70.7 GHz .
2 C C
= =
+

COMMENTS: In this particular example, note that two port parameter theory leads
expediently to the analytical and numerical definitions of a com-
monly invoked figure of merit for MOSFETs; namely, the unity gain
frequency, f
T
. The computed f
T
value of better than 70 GHz is quite
large and indicative of deep submicron technology devices. How-
ever, it is worthwhile interjecting that although f
T
commonly serves
as a reasonably useful metric for comparing the high frequency
attributes of one transistor against another, it is a less than meaning-
ful metric for MOSFET circuits and systems. The unity gain
frequency of an actual circuit that exploits MOSFET technology is
invariably much smaller than the transistor f
T
rating. A principle rea-
son for this disparity is that the short circuit basis for enumerating f
T

inherently ignores drain-bulk capacitance. It is also oblivious to the
potentially dominant energy storage elements parasitic or other-
wise embedded within the circuit.
2.4.4. Indefinite Admittance Parameters
An interesting and useful extension to the two port y-parameter concept entails the
consideration of the three terminal linear network abstracted in Figure (21). The network at hand
is operated as a three port configuration, with each of the three ports supporting a voltage refer-
enced to system ground. At first blush, the subject analytical extension is seemingly trivial since
superposition theory allows the matrix expression,
11 12 13 1 1
2 21 22 23 2
3 3 31 32 33
y y y I V
I y y y V
I V y y y


=



, (43)
to be postulated as the pertinent condensed volt-ampere relationship. To be sure, (43) is a valid
generality, but it incorrectly implies that each of the admittance parameter, y
ij
, is independent of
one another. In truth, only four of the nine indicated admittance metrics can be measured or
computed independently.
A clarification of the preceding assertion begins be observing that the subject three port
network functions essentially as one giant happy node in the sense that the three port currents, I
1
,
I
2
, and I
3
, must sum to zero for any and all port voltages, V
1
, V
2
, and V
3
. From (43), this con-
straint means that
Course Notes #1 University of Southern California John Choma

August 2006 30 Two Port Networks

Linear
Three Port
Network
+

V
1
I
1
I
3
I
2
1 2
3
V
2
+

V
2
+

V
3

Fig. (21). Schematic Abstraction Of A Lin-
ear Three Port Network. Note
That The Second Terminal Of
Each Of The Three Ports Is Sys-
tem Ground.
11 21 31
12 22 32
13 23 33
y y y 0
y y y 0
y y y 0
+ +
+ +
+ +
, (44)
which is tantamount to requiring that the sum of the elements in each of the three columns in the
3x3 matrix on the right hand side of (43) be zero. An additional parametric constraint is mani-
fested through the consideration of the special circumstance in which all three network port volt-
ages are identical. In this case, no potential difference exists and thus no energy is applied,
between any two of the three network terminals. Under the condition of null differential applied
energy implied by V
1
= V
2
= V
3
, I
1
= I
2
= I
3
= 0 if and only if
11 12 13
21 22 23
31 32 33
y y y 0
y y y 0
y y y 0
+ +
+ +
+ +
,
2
(45)
Accordingly, the admittance elements in each of the three rows of the 3x3 matrix in (43) must,
like the elements in each column, sum to zero. The immediate ramification of this disclosure is
that the generalized volt-ampere relationship in (43) is expressible as
( )
( )
( ) ( ) ( )
11 12 11 12
1 1
2 21 22 21 22
3 3
11 21 12 22 11 12 21 22
y y y y
I V
I y y y y
I V
y y y y y y y y

+




=




+ + + + +



V + . (46)
The 3x3 matrix on the right hand side of this relationship is termed the indefinite admittance
matrix of the three port shown in Figure (21). For an nterminal linear network, an nxn indefi-
Course Notes #1 University of Southern California John Choma

August 2006 31 Two Port Networks

nite admittance matrix can be constructed straightforwardly in terms of the admittance parame-
ters of the reduced (n-1)terminal configuration simply by invoking the constraints of zero row
and column sums in the resultant nxn matrix.
In an attempt to garner appreciation for the indefinite admittance matrix concept, con-
sider the case in which the four admittance parameters of the two port network resulting from V
3

= 0 have been measured or otherwise evaluated. The resultant volt-ampere expression, is (36),
where it is tacitly assumed that the port voltages, V
1
and V
2
, are referenced to electrical ground.
It follows that volt-ampere characteristics when V
3
is nonzero is (46), where again, all port volt-
ages are presumed referenced to electrical ground. Suppose now that the network in Figure (21)
is operated with terminal 2 grounded and with nonzero V
1
and V
3
. Since V
2
= 0, the second col-
umn of matrix elements on the right hand side of (46) can be discarded without loss of any
information. Moreover, since (I
1
+ I
2
+ I
3
) = 0 is guaranteed by the construction of the indefinite
admittance matrix, the analytical information surrounding current I
2
, which flows into the
grounded second port, is superfluous. It follows that the second row of matrix elements in (46)
can also be discarded, thereby rendering
( )
( ) ( )
11 11 12 1 1
3 3
11 21 11 12 21 22
y y y I V
I V
y y y y y y

+

+ + + +

(47)
as the applicable system of volt-ampere equations. Note that this system of equilibrium equa-
tions results exclusively from the characterization of the original system operated with terminal 3
grounded and does not require the execution of any substantive new analyses.
EXAMPLE #4:
Recall Figure (20) as a small signal model of a MOSFET operated as a common
source amplifier. Consider now the schematic diagram of Figure (22a), which
depicts the same MOSFET operated in a common drain (also referred to as a source
follower) configuration. Determine general expressions for each of the four common
drain y-parameters, y
11
, y
13
, y
31
, and y
33
. The MOSFET model parameters remain g
m

= 8 mS, r
o
= 8 K, C
gs
= 15 fF, C
gd
= 3 fF, and C
db
= 12 fF.
SOLUTION:
(1). The parameters, y
11
, y
12
, y
21
, and y
22
, for the common source circuit have been
determined in Example #2. For convenience, they are repeated herewith. Recall
that g
o
= 1/r
o
.
( )
( )
11 gs gd
12 gd
21 m gd
22 o db gd
y s C C
y sC
.
y g sC
y g s C C
= +
=
=
= + +

(2). The parameters for the common drain circuit in Figure (22a) derive directly from
the foregoing results and (47). Specifically,
Course Notes #1 University of Southern California John Choma

August 2006 32 Two Port Networks

g V
m 1
r
o
C
db
C
gs
C
gd
V
1
V
3
I
1
I
3
1

(
G
a
t
e
)
2

(
D
r
a
i
n
)
3 (Source)
(a).
(g +sC )V
m gs 1
r
o
C
db
C
gs
V
1
I
1
1

(
G
a
t
e
)
3

(
S
o
u
r
c
e
)
2 (Drain)
(b).
I
3
V
3
C
gd
sC V
gs 3
1/g
m
C
gs

Fig. (22). (a). Small Signal Equivalent Circuit Of A MOSFET Oriented Toward Common Drain Signal
Processing. The Circuit Is Addressed In Example #4. (b). The Resultant y-Parameter Model Of
The Common Drain Circuit In (a).
( )
( )
( )
( )
( )
11
13
31
33
gs gd
11 12 gs
11 21 m gs
11 12 21 22 m o gs db
y s C C
y y y sC
.
y y y g sC
y y y y y g g s C C
= +
= + =
= + = +
= + + + = + + +

COMMENTS: The resultant y-parameter equivalent circuit is offered in Figure
(22b). This model shows that the output impedance of the considered
source follower is small, even at low signal frequencies, because of
the resistance, 1/g
m
, that appears directly across the output port.
2.4.5. Open Circuit zParameters
The use of open circuit impedance parameters is premised on selecting the input port
current, I
1
, and the output port current, I
2
, as independent electrical variables in the two port sys-
tem of Figure (1). The upshot of this selection is the volt-ampere matrix expression
1 11 12 1
2 21 22 2
V z z I
,
V z z I

=


(48)
where the z
ij
are the network open circuit impedance parameters, or zparameters. The equiva-
lent circuit corresponding to (48) is shown in Figure (23) and is seen to exploit Thvenins theo-
rem at both the input and output ports.
Course Notes #1 University of Southern California John Choma

August 2006 33 Two Port Networks

Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
+

V
2
I
2
I
1
+

V
1
3 1
4 2
z
11
z I
12 2
z I
21 1
+

z
22

Fig. (23). The Open Circuit Impedance Parameter Model Of A Linear
Two Port Network. Each of the z-parameters, z
ij
, Have
Units Of Impedance.
The designation of the zparameters as open circuit impedances follows from the
parametric measurement strategy implied by (48). For example
2
2
1
1
1
11
I 0
2
21
I 0
V
z
I
,
V
z
I
=
=
=
=
(49)
whence z
11
is the open circuit (meaning an open circuited output port) input impedance of the lin-
ear two port network. Moreover, z
21
is seen as the open circuit transimpedance of the two port
system. Thus, like h
11
, g
11
, and y
11
, z
11
influences the input impedance of the network undergo-
ing investigation, and like h
21
, g
21
, and y
21
, z
21
is a measure of the achievable forward gain.
Equation (48) also confirms that
1
1
2
2
1
12
I 0
2
22
I 0
V
z
I
.
V
z
I
=
=
=
=
(50)
Accordingly, z
12
is the feedback transimpedance under the condition of an open circuited input
port, and z
22
is the open circuit output impedance measured under the same input port open cir-
cuit.
Course Notes #1 University of Southern California John Choma

August 2006 34 Two Port Networks

A comparison of (48) with (36) suggests immediately that the z-parameter matrix of a
two port network is the inverse of the y-parameter matrix of the same system. Assuming that the
y-parameter matrix is non-singular and letting
y 11 22 12
y y y y =
21
(51)
denote the determinant of the y-matrix in (36), the z-parameters relate to their y-parameter
counterparts in accordance with
22
11
12 21
y
11
22
11
22
12 21
y
22
22
12
12
y
21
21
y
y
1
z
y y

y
y
y
1
z
y y

y
.
y
y
z

y
z

= =

= =

=
=
(52)
Note that for a bilateral two port, which has y
12
= y
21
, z
12
is identical to z
21
.
2.4.6. Chain Parameters
The chain parameters, c
ij
, which are often referred to as the transmission parameters, of
the linear two port network in Figure (1) are implicitly defined by the matrix volt-ampere
description,
1 11 12 2
1 21 22 2
V c c V
.
I c c I

=

(53)
Note that in the c-parameter representation of the volt-ampere characteristics of a linear two port
system, the selected independent variables are output port voltage and the negative of output port
current, thereby remanding input port voltage and input port current to dependent variable status.
In contrast to the four traditional two port parameter representations, the chain parameters are
rarely used for circuit modeling purposes. Instead, the c-parameters are used directly in the
analysis and design of networks and are especially useful in the design of passive filters com-
prised of cascaded passive networks.
As usual, the strategy underlying the measurement of the c-parameters follows directly
from the defining volt-ampere characteristics. To this end, let the output port of the considered
system be open circuited so that the output port current, I
2
, is nulled. Simultaneously, let the
input port be voltage-driven to establish an input port voltage, V
1
, and an input port current, I
1
.
The resultant test fixturing, shown in Figure (24a), is identical to that of Figure (15a), which is
Course Notes #1 University of Southern California John Choma

August 2006 35 Two Port Networks

used to measure the open circuit hybrid g-parameters, g
11
and g
21
. From (53), I
2
= 0 suggests
that
Linear
Two Port
Network
+

V
1
I = c V
1 21 2
I = 0
2
+

V = V /c
2 1 11
1 3
2 4
S
h
o
r
t
C
i
r
c
u
i
t
(a).
Linear
Two Port
Network
+

V = c I

1 12 2
I
1
I = I /c
2 1 22

V = 0

2
1
2
O
p
e
n
C
i
r
c
u
i
t
T
e
s
t
V
o
l
t
a
g
e

S
o
u
r
c
e
(b).
3
4
+

T
e
s
t
C
u
r
r
e
n
t

S
o
u
r
c
e

Fig. (24). (a). Circuit For The Measurement Of The Open Circuit Transmission, Or
Chain, Parameters. (b). Measurement Of The Short Circuit Transmission
Parameters.
2
2
2
11 1
21 1
1
c
I 0
1
2
c
I 0
V
V
.
V
I
=
=
=
=
(54)
Thus, parameter c
11
is dimensionless, and its inverse is the Thvenin voltage gain of the subject
two port network. On the other hand, c
21
is in units of admittance, and its inverse is the open cir-
cuit (meaning that the output port is open circuited), or Thvenin, forward transimpedance of the
network. With the output port short circuited, as indicated in Figure (24b), V
2
= 0 in (53) gives
2
2
2
12 1
22 1
1
c
V 0
1
2
c
V 0
I
V
.
I
I
=
=

=
(55)
Clearly, the negative inverse of dimensionless parameter c
22
is the Norton current gain of the net-
work. Moreover, the negative inverse of parameter c
12
is the Norton forward transadmittance,
which renders c
12
itself in dimensions of impedance.
As is the case with any other set of two port parameters, the chain parameters can be
related to the h-, g-, y-, or z-parameters. Since the open circuit impedance parameters are the last
Course Notes #1 University of Southern California John Choma

August 2006 36 Two Port Networks

parameters examined and thus still relatively fresh in personal memory, compare (53) with (48).
Specifically, observe that
2
2
1 11
2 21
1
21
z
11
z
I 0
1
21
z
2
I 0
V
c
V
.
I
c
V
=
=
= =
= =
(56)
Moreover,
2
2
1 22
2 21
1 11 22 12 21 z
2 21
z
22
z
V 0
z z z z
12
z z
V 0
I
c
I
.
V
c
I
=

=
= =
= = =
21
(57)
From (56) and (57), the determinant,
c
, of the c-parameter matrix is
12
c 11 22 12 21
21
z
c c c c .
z
= = (58)
It follows that the determinant of the chain matrix is unity if the network undergoing study is
bilateral; that is if z
12
= z
21
. Interestingly enough, the c-matrix is singular if the network it repre-
sents is unilateral, which corresponds to a network divorced of feedback (z
12
= 0).
2.4.6.1 Input and Output Impedances
Highlighting the fundamental attributes of chain parameters begins by applying the
chain matrix concept to the problem of determining the driving point input and output imped-
ances of the original two port system in Figure (1). The input impedance, Z
in
, derives from (53)
as
1 11 2 12 2
in
1 21 2 22 2
V c V c I
Z .
I c V c I

= =

(59)
But, since V
2
= Z
l
I
2
, (59) becomes
11 l 12
in
21 l 22
c Z c
Z .
c Z c
+
=
+
(60)
Equation (60) suggests that the driving point input impedance reduces to c
11
/c
21
for the special
case of an open circuit load port. When the load termination is a short circuit, Z
in
is c
12
/c
22
.
Course Notes #1 University of Southern California John Choma

August 2006 37 Two Port Networks

In order to derive an expression for the driving point output impedance, Z
out
, the signal
source voltage, V
s
, is reduced to zero, thereby forcing V
1
= Z
s
I
1
. This fact and (53) confirm that
22 s 12 2
out
2 21 s 11
c Z c V
Z .
I c Z c
+
= =
+
(61)
In the limit of an ideal current source signal drive (Z
s
), the output impedance approaches
c
22
/c
21
, while for an ideal voltage source signal excitation (Z
s
0), Z
out
collapses to c
12
/c
11
.
2.4.6.2 Voltage Transfer Function
The voltage transfer function, V
2
/V
1
, from the input -to- the output port can be found
by using (53) to write
2
1 11 2 12 2 11 12
l
V
V c V c I c c
Z


= =


, (62)
whence
l 2
1 11 l 12
Z V
.
V c Z c
=
+
(63)
Note that (63) corroborates with (54) in the sense that an infinitely large load impedance delivers
a port -to- port voltage gain of 1/c
11
. Since the input port voltage, V
1
, relates to the signal source
voltage, V
s
, as the simple voltage divider,
in 1
s in s
Z V
,
V Z Z
=
+
(64)
(63) and (59) combine to deliver an overall system voltage gain of
( ) ( )
l 2
s
11 21 s l 12 22 s
Z V
.
V
c c Z Z c c Z
=
+ + +
(65)
As expected, (65) reduces to (63) when the source impedance, Z
s
, is zero.
2.4.6.3 Cascade Interconnection
The chain parameters and its associated c-parameter matrix are particularly amenable
to ascertaining the transfer and driving point characteristics of networks whose architecture is
comprised of a cascade of passive or active circuit sections. To this end, consider Figure (25),
which depicts a two stage cascade interconnection of two linear two port networks. In terms of
the indicated port voltages and port currents, the volt-ampere characteristics of Network A are
given by the matrix relationship,
x 1
1 o
V V
,
I I

=

A
C
1
(66)
Course Notes #1 University of Southern California John Choma

August 2006 38 Two Port Networks

Linear
Two Port
Network A
Linear
Two Port
Network B
+

V
1
I
1
I
i2
I
o1
I
2
+

V
x
+

V
2
1 3
2 4
5
6

Fig. (25). A Cascade Interconnection Of Two Linear Two Port Networks. Net-
work A Is Presumed To Have A Chain Matrix, C
A
, While C
B
Is
The Chain Matrix Of Network B.
where C
A
is the 2x2 chain matrix for Network A. Similarly for Network B,
x 2
i2 2
V V
,
I I

=

B
C (67)
with C
B
denoting the 2x2 chain matrix for Network B. But since the input port current, I
i2
, to
Network B is obviously the negative of the output port current, I
o1
, of Network A, (66) and (67)
combine to produce
x x 1
1 o1 i2 2
V V V V
,
I I I I

= = =



A A A B
C C C C C
2 2
2
V
I
(68)
where matrix C, which is given as,
=
A B
C C C (69)
is properly interpreted as the 2x2 chain matrix for the overall cascade interconnection. Obvi-
ously, (69) can be extended to embrace a cascade of n network sections, wherein the overall
chain matrix becomes simply the ordered product of the chain matrices of the individual mem-
bers of the cascade. Once the overall chain matrix is evaluated, its elements can be used in
conjunction with (60), (61), (63), and (65) to determine the I/O impedances and transfer function
of the cascade. Since matrix multiplication is non-commutative (C
A
C
B
C
B
C
A
), the chain
matrix, and thus the performance attributes, of a cascade interconnection of circuits is dependent
on the ordering of the individual sections of the cascade. This observation mathematically con-
firms the transparent engineering fact that the observable network performance depends on the
manner in which the individual elements and subsections of the network are connected.
2.4.6.4 Series and Shunt Elements
Equation (69) offers an analytical foundation for the systematic evaluation of the
transfer and driving point impedance functions of networks comprised of cascade interconnec-
tions of passive subcircuits. To wit, consider the impedance, Z, connected as shown in Figure
(26a) as a branch element in series with an input port having port voltage V
1
and port current I
1

and an output port, where the pertinent voltage and current variables are V
2
and I
2
, respectively.
An application of (54) and (55) to the subject diagram gives a chain matrix, say C
z
, of
1 Z
,
0 1

=


z
C (70)
Course Notes #1 University of Southern California John Choma

August 2006 39 Two Port Networks

+

V
2
I
2
I
1
+

V
1
Z Z
1
Z
3
(a).
+

V
2
I
2
I
1
+

V
1
Y
(b).
+

V
2
I
2
I
1
+

V
1
Y
2
(c).

Fig. (26). (a). Impedance Branch Placed In Series Between Input And Output Ports. (b). Admittance
Branch Connected In Shunt With The Input And Output Ports. (c). Tee Network Comprised Of
Cascade Interconnections Of Subcircuits Having The Topological Forms In (a) And (b).
where the determinant of C
z
is seen to be unity, which is as expected in light of the presumed
passivity of impedance Z. For Figure (26b), which depicts a branch admittance, Y, connected as
an element that simultaneously shunts both input and output ports,
1 0
.
Y 1

=


y
C (71)
It follows that for the tee configuration of Figure (26c), whose structure is a series impedance,
Z
1
, connected in cascade with a shunt admittance, Y
2
, which in turn is cascaded with a second
series impedance, Z
3
, the network chain matrix, say C
n
, is
( )
2 1 3 1 2 3 3 1
2
2 2 3
1 Y Z Z Z 1 Y Z 1 Z 1 Z 1 0
.
Y 1
0 1 0 1 Y 1 Y Z

+ + +


= =


+


n
C (72)
EXAMPLE #5:
The lowpass filter in Figure (27) is terminated at its output port in a resistance, R
l
.
The Thvenin source resistance can either be presumed to be zero or otherwise
absorbed into the resistance, R. Use chain parameters to derive expressions for the
voltage transfer function, A
v
(s) = V
2
/V
1
, the driving point input impedance, Z
in
(s), and
the driving point output impedance, Z
out
(s).
V
2
I
2
I
1
R
R
l C
L
V
1
Z (s)
in
Z (s)
out

Fig. (27). The Lowpass RLC Filter Addressed In Exam-
ple #5.
SOLUTION:
(1). The chain matrix of the filter follows directly from (72), where Z
1
= R, Y
2
= sC,
and Z
3
= sL.
( )
2
2
1 sRC sL R 1 s LC
.
sC 1 s LC

+ + +

=

+

n
C

Course Notes #1 University of Southern California John Choma

August 2006 40 Two Port Networks

(2). Using (65) with Z
l
= R
l
and Z
s
= 0, the voltage gain evaluates as
( )
( )
( )
l 2
v
2
1
l
l
l
2
l
l l
R V
A (s)
V
R 1 sRC sL R 1 s LC
R
R R
L R
1 s R R C s L
R R R R
= =
+ + + +
+
=


+ + +

+ +

C

(3). Using (60), the driving point input impedance is
( ) ( )
2
l l
in
2
l
R R s L R RC s RLC
Z (s) .
1 sR C s LC
+ + + +
=
+ +

(4). From (61) and recalling that Z
s
= 0, the driving point output impedance is found
to be
2
out
R sL s RLC
Z (s) .
1 sRC
+ +
=
+

COMMENTS: It is notable that the chain parameter concept allows for the formula-
tion of the foregoing gain and I/O impedance relationships without
need of laborious circuit analysis. As a check on the propriety of the
results, observe that as expected, the zero frequency voltage gain is
the resistive divider, (R
l
/R
l
+ R), the zero frequency value of the driv-
ing point input impedance is (R
l
+ R), and the zero frequency output
impedance is simply R.
3.0. TWO PORT METHODS OF CIRCUIT ANALYSIS
The two port parameter sets introduced in the preceding section of material allow for an
unambiguous volt-ampere characterization of any linear two port network in terms of the electri-
cal properties that are observable and measurable at the input and output ports of the network.
The equivalent circuits respectively corresponding to the hybrid h-, hybrid g-, short circuit y-,
and open circuit z-parameters exploit superposition theory and comprise a modest extension of
the classic one port formulation of the Thvenin and Norton theorems. These models are the
fundamental tools that expedite a computationally efficient analysis of the transfer and driving
point characteristics of terminated electrical and electronic systems whose intrinsic circuit
topologies are either unknown or too complicated for straightforward circuit analyses predicated
on the Kirchhoff laws.
3.1. CIRCUIT ANALYSIS IN TERMS OF hPARAMETERS
Recalling Figure (9), the h-parameter equivalent circuit of the original, voltage-driven,
terminated two port system in Figure (1) is the topology provided in Figure (28). Although
either version of the system diagrammed in Figure (1) can be selected for analysis, the voltage-
Course Notes #1 University of Southern California John Choma

August 2006 41 Two Port Networks

driven version is chosen because the Thvenin nature of the h-parameter input port model is
synergistic with the application of the Kirchhoff voltage law to said input port. In particular,
+

V
1
I
1
I
2
+

V
2
1
2 4
3
+

h
22
h
11
h I
21 1
h V
12 2
Linear
Two Port
Network
Z
s
Z
s
Z
l
Z
l
+

V
1
I
1
+

V
s
+

V
s
I
2
+

V
2
1 3
2 4
Z
in
Z
out

Fig. (28). The h-Parameter Equivalent Circuit Of The Linear Two Port Network In Figure (1). The Volt-
age-Driven Version Of Figure (1) Is Chosen Because Of The Thvenin Nature Of The h-
Parameter Model Input Port.
( )
s s 11 1 12 2
V Z h I h V = + + , (73)
while at the output port,
21 1
2
22 l
h I
V , (74)
h Y
=
+
where Y
l
is the admittance corresponding to the terminating load impedance, Z
l
. If (74) is
inserted into (73),
s 12 21
s 11
1 2
V h h
2 l
Z h .
I h Y
= +
+
(75)
Equations (74) and (75) now yield a system voltage gain, A
v
, of
( )( )
( )( )
21
11 s 22 l
2
v
12 21
s
11 s 22 l
h
h Z h Y
V
A
h h
V
1
h Z h Y

+ +
= =

+ +
. (76)
3.1.1. Open Loop and Loop Gain Concepts
The rather cumbersome algebraic form of the voltage gain expression in (76) inspires
neither analytical confidence nor relevant insights about the dynamic behavior of the system in
Figure (28). A more illuminating format begins to materialize from the substitution,
Course Notes #1 University of Southern California John Choma

August 2006 42 Two Port Networks

( )( )
21
vo
11 s 22 l
h
A
h Z h Y

+ +
, (77)
whereupon (76) becomes
vo 2
v
s 12 vo
A V
A
V 1 h A
= =
+
.
)
,
(78)
Equation (78) implies
(
2 vo s 12 2
V A V h V = (79)
which can be represented algebraically by the block diagram shown in Figure (29). This diagram
underscores h-parameter h
12
as the factor of the output voltage response that is fed back to the
system input port. Since feedback is such a critically important phenomenon in virtually all
electrical and electronic circuits and systems, the parameters implicit to the architecture in Figure
(29) are worthy of further exploration.
A
vo
h
12
V
s
V
2
+

h V
12 2

Fig. (29). Block Diagram Model Of The Circuit In Figure
(28). The Diagram Underscores The Feedback
Inherent To The Subject Circuit.
To begin this complementary exploration, note in Figure (29) that the signal path from
the summing unit through the block labeled, A
vo
, and thence through the h
12
block, constitutes a
closed loop, which is doubtlessly the reason that the gain expression in (78) is traditionally
termed the closed loop gain of the system undergoing study. Since setting h
12
to zero reduces A
v

in (78) to zero, and since h
12
= 0 effectively opens the aforementioned closed loop, A
vo
, in (77) is
logically termed the open loop gain. Note that the circuit in Figure (28) also verifies that A
v
=
A
vo
when h
12
= 0 since by inspection,
12
s
1
h 0
s 11
V
I ,
Z h
=
=
+
(80)
whence by (74), the resultant gain is A
vo
, in accordance with (77).
The product, h
12
A
vo
, appearing in the denominator on the right hand side of (78) is seen
in Figure (29) as the multiplication of the open loop gain and the feedback factor. It is effec-
tively the gain around the closed loop and is often termed the loop gain of the considered system.
This h-parameter loop gain, say T
h
(Z
s
, Y
l
), is functionally dependent on the source impedance, Z
s
,
and the load admittance, Y
l
, and is defined by
Course Notes #1 University of Southern California John Choma

August 2006 43 Two Port Networks

( )
12 21
h s l 12 vo
11 s 22 l
h h
T Z , Y h A
h Z h Y


=

+ +

,

(81)
whence (78) is expressible as
( )
vo 2
v
s
h s l
A V
A
V
1 T Z , Y
= =
+
. (82)
It is interesting to note that the second parenthesized factor on the right hand side of
(81) is the negative ratio of the output port voltage, V
2
, -to- the input port current, I
1
, where V
2

and I
1
are recalled as the independent variables of h-parameter modeling. In other words, for an
assumed input port current, I
1
, the subject second factor is the negative forward transimpedance,
V
2
/I
1
, as is suggested in Figure (30a). On the other hand, and as is portrayed by Figure (30b),
the first factor comprising the loop gain in (81), inclusive of the negative sign, is the reverse
transadmittance ratio, I
1
/V
2
, for the case of zero signal source excitation, V
s
= 0. In other words,
the negative first factor on the right hand side of (81) defines the input port current resulting
exclusively from a presumed output port voltage. The loop gain concept is now apparent. In
particular, the loop gain, which is always dimensionless, can be viewed conceptually as
+

V
1
I
1
I
2

+
V
2
1
3
2
4
+

h
22
h
11
h I
21 1
h V
12 2
Z
s
Z
l
(a).
(b).

Fig. (30). (a). Output Port Model Of The Equivalent Circuit
In Figure (28), Which Is Used To Compute the
Negative Input Port -To- Output Port Forward
Transimpedance, V
2
/I
1
, Component Of The Cir-
cuit Loop Gain. (b). Input Port Model Of The
Equivalent Circuit In Figure (28), Which Is Used
To Compute the Output Port -To- Input Port
Reverse Transadmittance, I
1
/V
2
, Component Of
The Circuit Loop Gain Function, T
h
(Z
s
, Z
l
).
( )
s
12 21 1 2
h s l
11 s 22 l 2 1
V 0
h h I V
T Z , Y
h Z h Y V I
=

= =

+ +

, (83)
Course Notes #1 University of Southern California John Choma

August 2006 44 Two Port Networks

which is a product of two gains formed of the independent modeling variables. In particular, it is
product of the phase-inverted forward gain (transimpedance) from input port -to- output port and
the zero signal value of the reverse gain (transadmittance) from output port -to- input port.
The loop gain of a linear circuit or system is a critically important metric for several
reasons, two of which can be comprehended immediately. The first of these reasons derives
from the presumption that over a stipulated range of signal frequencies, the magnitude of the
loop gain is significantly larger than unity. Under this presumption, (81) and (82) combine to
deliver a closed loop, or actual, voltage gain of
( )
vo 2
v
s 12
h s l
A V
1
A
V h
1 T Z , Y
= =
+
, (84)
which is dependent on only the feedback h-parameter, h
12
, as opposed to a dependence on four h-
parameters, a source impedance, and a load admittance. To the extent that h
12
can be physically
implemented as a network function that is dependent on a ratio of controllable passive compo-
nents, the achievable voltage gain is rendered reliable and analytically predictable. This laudable
attribute is particularly germane to integrated electronics whose active components are invariably
plagued by model parameters that are either ill-defined or poorly controlled during monolithic
processing. In other words, a sufficiently large loop gain magnitude delivers an actual gain that
is rendered relatively insensitive to both the h-parameters, h
11
, h
21
, and h
22
, and the source and
load terminations, Z
s
, and Y
l
, whose precise values are also somewhat unclear in broad varieties
of practical system applications. Note, however, that since closed loop gains larger than unity
are generally desirable, h
12
is necessarily less than unity, which forces the open loop gain, A
vo
to
be very large magnitude. Unfortunately, large open loop gains, such as those that prevail in
commercially available operational amplifiers, portend of potentially serious bandwidth limita-
tions. The upshot of the matter is that while large loop gains are assuredly advantageous in
active electronics from the perspective of gain desensitization with respect to parametric
uncertainties and tolerances, they are generally difficult to achieve in broadband systems.
A second reason underlying the importance of the loop gain is the picture that this met-
ric conveys with respect to the stability of active networks. In Figure (30b), consider the case in
which over a range of signal frequencies, the algebraic sign of h
12
V
2
is such as to increase the
input port current, I
1
. This so-called positive feedback would mean that feedback alone
divorced of any source excitation increases I
1
, which results in an increase in the magnitude of
the output port voltage, V
2
, which results in further increases in I
1
, and so forth. In other words,
the circuit or system under investigation is potentially unstable and, pending the extent by which
I
1
increases, may be outright unstable. Although a definitive stability analysis
[5]
is a complicated
frequency domain task that is beyond the scope of the present discourse, a clue as to the loop
gain circumstances surrounding potential instability derives from an elementary consideration of
low frequencies, where all parameters and terminations are real numbers. Specifically, I
1

increases and is therefore a positive current with V
s
= 0 if h
12
V
2
is negative. Because of (74),
h
12
V
2
< 0 requires h
12
h
21
/(h
22
+Y
l
) > 0. Since (83) yields
( ) (
12 21
11 s h s l
22 l
h h
h Z T Z , Y
h Y
= +
+
)
, (85)
Course Notes #1 University of Southern California John Choma

August 2006 45 Two Port Networks

the undesirable instability situation requires that the loop gain, T
h
(Z
s
, Z
l
), be negative, assuming
the routine case of positive real impedances, h
11
and positive Z
s
. In other words, a negative loop
gain (at least at low signal frequencies) is tantamount to network potential instability. In con-
trast, positive loop gain assures network stability. Since positive loop gain requires h
12
V
2
>0,
input current I
1
decreases, the output response remains forever bounded for bounded inputs, and
the system under consideration is said to be characterized by negative feedback.
The phrase, potential instability, is used in the foregoing paragraph in the context of
stipulating T
h
(Z
s
, Y
l
) < 0 as a necessary, but not sufficient, condition for circuit instability. For
example, if the loop gain is negative, but has a magnitude less than unity, the circuit remains sta-
ble. If on the other hand, the loop gain is negative with a magnitude exceeding one, there must
exist at least one frequency where the loop gain is precisely negative one. At these frequencies,
the considered circuit is unstable in the sense that its gain is boundless, which means that output
voltages are established even when the signal source is nulled. If T
h
(Z
s
, Y
l
) = 1 at precisely one
value of frequency, say
o
, the circuit is technically unstable but nonetheless useful in
communications and other systems as a sinusoidal oscillator whose steady state radial frequency
of oscillation is
o
[6]
.
3.1.2. I/O Impedances
The loop gain is also instrumental with respect to stipulating the magnitude and general
nature of network impedances established at input and output ports. To this end, consider Figure
(31a), which relies on the model in Figure (28) to delineate the equivalent circuit pertinent to a
determination of the driving point input impedance, Z
in
. A straightforward circuit analysis read-
ily confirms,
+

V
x
I
x
I
2
+

V
2
1
2
3
4
+

h
22
h
11
h I
21 x
h V
12 2
Z
l
Z
in
I
x
(a).
+

V
1
I
1
I
2
+

V
y
1
2 4
3
+

h
22
h
11
h I
21 1
h V
12 y
Z
out
I
y
(b).
Z
s

Fig. (31). (a). Equivalent Circuit For The Evaluation Of The Driving Point Input Impedance, Z
in
, Of
the Circuit Modeled In Figure (28). (b). Equivalent Circuit For Calculating The Driving
Point Output Admittance Of The Circuit Of Figure (28).
Course Notes #1 University of Southern California John Choma

August 2006 46 Two Port Networks

( )
x 12 21 12 21
in 11 11
x 22 l
11 22 l
V h h h h
Z h h 1
I h Y
h h Y
,


= = =

+
+


(86)
where I
x
is a test independent input port current, V
x
is the input port response to this current, and
V
x
/I
x
exploits Ohms law to determine the input impedance. Observe an input impedance of h
11

in the absence of feedback (h
12
= 0), which means that h
11
can be thought of as an open loop
input impedance. Moreover, (83) allows (86) to be expressed as
( )
12 21
in 11 11 h l
22 l
h h
Z h h 1 T
h Y
0, Y ,

= = +

+
(87)
where T
h
(0, Y
l
) symbolizes the h-parameter loop gain under the condition of zero source imped-
ance (Z
s
= 0). Since the loop gain is proportional to the open loop gain, and since the magnitude
of the open loop voltage gain for zero source impedance is surely larger than it is for nonzero
source impedance, |T
h
(0, Y
l
)| > |T
h
(Z
s
, Y
l
)|, which confirms
( ) ( )
in 11 h l 11 h s l
Z h 1 T 0, Y h 1 T Z , Y .

= + > +


(88)
And since the magnitude of the loop gain is desirably large for acceptable gain desensitization
with respect to parametric vagaries, the last result indicates that the system in Figure (28) is
capable of a driving point input impedance that is significantly larger than the h-parameter
prediction of its counterpart open loop value.
Because of the Norton nature of the h-parameter output port, the driving point output
admittance, Y
out
, is more conveniently evaluated than is the corresponding output impedance.
Figure (31b) can be used to demonstrate that
( )
y
12 21
out 22 22 h s
y 11 s
I
h h
Y h h 1 T Z
V h Z
, 0 ,

= = = +

+
(89)
where (83) has been used once again, and T
h
(Z
s
, 0) is the Y
l
= 0 value of the h-parameter loop
gain. Comments analogous to those made in regard to the input impedance can be proffered for
the output admittance. In particular, (89) shows that the driving point output admittance can be
considerably larger than its open loop value, which is h
22
. It follows that the closed loop output
impedance can be substantively smaller than the open loop output impedance predicted by h-
parameters.
EXAMPLE #6:
Figure (32) is an extension of the model given in Figure (12) in that it addresses an
emitter-degenerated bipolar transistor amplifier whose input and output ports are
terminated. The signal source applied to the amplifier input port has a Thvenin
resistance, R
s
, of 75 , while the amplifier output port drives a resistive load, R
l
, of 3
K. Recall that the amplifier parameters are r
i
= 2.2 K, r
o
= 25 K, = 90, and R
e

= 80 . The results of Example #1 are the h-parameters, h
11
= 9.46 K, h
21
= 89.71
amps/amp, h
12
= 3.19 mV/V, and h
22
= 39.87 S. Determine the open loop voltage
Course Notes #1 University of Southern California John Choma

August 2006 47 Two Port Networks

gain, A
vo
, of the amplifier, the loop gain, T
h
(R
s
, G
l
), the closed loop gain, A
v
, the driv-
ing point input resistance, R
in
, and the driving point output resistance, R
out
.
r
i
R
e
I
1
r
o
V
1
I
1
I
2
1
2
V
2
4
3
+

V
s
R
l
R
s
R
in
R
out

Fig. (32). Small Signal Equivalent Circuit Of The Emitter-Degener-
ated Common Emitter Amplifier Considered In Example #6.
The Model And Circuit Parameters Are r
i
= 2.2 K, r
o
= 25
K, = 90, R
e
= 80 , R
s
= 75 , and R
l
= 3 K.
SOLUTION:
(1). Using (77) with Z
s
= R
s
= 75 and Y
l
= 1/R
l
= 1/3 K, the open loop voltage
gain of the subject amplifier is found to be A
vo
= 25.21, or a magnitude of 28.03
dB. From (81), the loop gain is T
h
(R
s
, G
l
) = 80.42(10
3
), which is only 21.89
dB. It follows from (78) that the closed loop, or actual, voltage gain is A
v
=
27.42, or a closed loop gain magnitude of 28.76 dB.
(2). The R
s
= 0 value of the loop gain is, by (83), T
h
(0, G
l
) = 81.06(10
3
), while the
G
l
= 0 value of this metric is T
h
(R
s
, 0) = 752.77(10
3
). Using (88) and (89), the
driving point input resistance is calculated to be R
in
= 8.69 K, while, the driving
point output conductance is G
out
= 9.86 S, whence an output resistance of R
out
=
101.45 K.
COMMENTS: The single stage emitter-degenerated bipolar amplifier is so ubiqui-
tous in state of the art electronics that rarely is it perceived as a posi-
tive feedback architecture. Yet, the loop gain, as calculated above, is
a negative number but fortunately, it is a negative number whose
magnitude is considerably less than one. Indeed, this loop gain
magnitude is so much smaller than one that it is often tacitly ignored
in gain and input impedance calculations, which probably explains
why its positive feedback nature is typically unnoticed. But because
of this small loop gain, the amplifier at hand is a terrible voltage sig-
nal processor, principally because its extremely high output imped-
ance renders the closed loop voltage gain vulnerable to even modest
load termination uncertainties.
3.2. CIRCUIT ANALYSIS IN TERMS OF gPARAMETERS
The circuit analyses executed in terms of a g-parameter model of a linear two port net-
work are similar to those conducted in conjunction with h-parameters, and indeed, with y- and z-
parameters. In the case of g-parameters, the model in Figure (14) allows the system of Figure (1)
to be represented by the equivalent circuit offered in Figure (33). The shunt nature of the g-
parameter model input port renders a Norton representation of the signal source more expedient
Course Notes #1 University of Southern California John Choma

August 2006 48 Two Port Networks

than a Thvenin architecture. Moreover, the series nature of the g-parameter output port model
encourages a view of the load termination as an impedance, as opposed to an admittance.
Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
+

V
2
I
1
+

V
1
3 1
4 2
+

g
11
g
22
g I
12 2
g V
21 1
Z
l
Z
l
I
s
I
s
Y
s
Y
s
Y
in
I
2
Z
out

Fig. (33). The g-Parameter Equivalent Circuit Of The Linear Two Port Network In Figure (1). The Cur-
rent-Driven Version Of Figure (1) Is Chosen Because Of The Norton Nature Of The g-Parame-
ter Model Input Port.
A conventional circuit analysis of the equivalent circuit in Figure (33) yields a current
gain, A
i
, of
( )( )
( )( )
21
11 s 22 l
2
i
12 21
s
11 s 22 l
g
g Y g Z
I
A , (90)
g g
I
1
g Y g Z

+ +
= =

+ +
which is expressible as
( )
io 2
i
s
g s l
A I
A
I
1 T Y , Z
= =
+
.
)
(91)
In the last expression
( )(
21
io
11 s 22 l
g
A
g Y g Z
=
+ +
(92)
is the open loop current gain; that is, it is the current gain of the two port system in Figure (1)
under the condition that the feedback g-parameter, g
12
, is nulled. Moreover,
( )
12 21
g s l 12 io
11 s 22 l
g g
T Y , Z g A
g Y g Z


=

+ +

(93)
is the g-parameter loop gain of the two port network. Note that this loop gain is expressed as an
explicit function of the Thvenin source admittance, Y
s
, and the terminating load impedance, Z
l
.
Course Notes #1 University of Southern California John Choma

August 2006 49 Two Port Networks

Moreover, observe the striking similarity in form of (90) -through- (93) with their respective h-
parameter counterpart relationships.
The aforementioned functional similarity with corresponding h-parameter expressions
applies equally well insofar as the driving point input admittance, Y
in
, and the driving point out-
put impedance, Z
out
, in Figure (33) are concerned. Specifically, it can be demonstrated that
( )
12 21
in 11 11 g l
22 l
g g
Y g g 1 T 0, Z ,
g Z

= = +


+
(94)
and
( )
12 21
out 22 22 g s
11 s
g g
Z g g 1 T
g Y
Y , 0 .

= = +


+
(95)
In (94), the source admittance (not the source impedance) is nulled in the course of evaluating
the input admittance. In (95), the load impedance (not the load admittance) is set to zero for the
output impedance computation.
3.3. CIRCUIT ANALYSIS IN TERMS OF yPARAMETERS
If the two port system of Figure (1) is modeled by short circuit admittance, or y-,
parameters, the model in Figure (18) applies, and the equivalent circuit in Figure (34) results.
The shunt nature of the y-parameter model input port suggests the prudence of a Norton
representation of the signal source, while the similar Norton topology of the y-parameter output
port directs an admittance representation of the load admittance. With port voltage V
2
as the
independent output variable and Norton signal current I
s
as the known independent energy
source, the closed loop gain is a transimpedance, Z , which can be shown to be
f
Linear
Two Port
Network
+

V
1
I
1
I
2
+

V
2
1 3
2 4
Y
l
Y
l
I
s
I
s
Y
s
Y
s
Y
in
Y
out
+

V
2
I
2
I
1
+

V
1
3 1
4 2
y
11
y V
12 2
y V
21 1
y
22

Fig. (34). The y-Parameter Equivalent Circuit Of The Linear Two Port Network In Figure (1). The
Current-Driven Version Of Figure (1) Is Chosen Because Of The Norton Nature Of The y-
Parameter Model Input Port.
Course Notes #1 University of Southern California John Choma

August 2006 50 Two Port Networks

( )
fo
2
f
s
y s l
Z
V
Z ,
I
1 T Y , Y
= =
+
(96)
where
( )
12 21
y s l 12 fo
11 s 22 l
y y
T Y , Y y Z
y Y y Y


=

+ +

)
(97)
is the loop gain expressed in terms of y-parameters, and
( )(
21
fo
11 s 22 l
y
Z
y Y y Y
=
+ +
(98)
is the open loop gain; that is, the closed loop gain value under the condition of y
12
= 0. A
conventional Ohms law type of circuit analysis confirms driving point input and output admit-
tances, Y
in
and Y
out
, respectively, of
( )
12 21
in 11 11 y l
22 l
y y
Y y y 1 T 0, Y
y Y

= = +


+
(99)
and
( )
12 21
out 22 22 y s
11 s
y y
Y y y 1 T Y , 0 .
y Y

= = +


+
(100)
3.4. CIRCUIT ANALYSIS IN TERMS OF zPARAMETERS
The open circuit impedance, or z-, parameter model of a linear two port network is
shown in Figure (23). The use of this model in conjunction with the system of Figure (1) deliv-
ers the equivalent circuit in Figure (35). The series nature of both the input and output ports of
this equivalent circuit encourages an impedance characterization of the load termination and a
Thvenin representation of the applied signal source. Accordingly, the most expediently evalu-
ated forward gain is the transadmittance, I
2
/V
s
, where I
2
is the output port current and, of course,
V
s
is the Thvenin signal voltage.
A traditional analysis of the model in Figure (35) delineates the aforementioned
transadmittance as
( )
fo
2
f
s
z s l
Y
I
Y , (101)
V
1 T Z , Z
= =
+
)
where
( )(
21
fo
11 s 22 l
z
Y
z Z z Z
=
+ +
(102)
Course Notes #1 University of Southern California John Choma

August 2006 51 Two Port Networks

+

V
2
I
2
I
1
+

V
1
3 1
4 2
z
11
z I
12 2
z I
21 1
+

z
22
Linear
Two Port
Network
Z
s
Z
s
Z
l
Z
l
+

V
1
I
1
+

V
s
+

V
s
I
2
+

V
2
1 3
2 4
Z
in
Z
out

Fig. (35). The z-Parameter Equivalent Circuit Of The Linear Two Port Network In Figure (1). The
Voltage-Driven Version Of Figure (1) Is Chosen Because Of The Thvenin Nature Of The
z-Parameter Model Input Port.
is the open loop transadmittance value for the zero feedback condition, z
12
= 0. Moreover
( )
12 21
z s l 12 fo
11 s 22 l
z z
T Z , Z z Y
z Z z Z


=

+ +

(103)
is the z-parameter prediction of the system loop gain. The indicated input and output imped-
ances, Z
in
and Z
out
, respectively, can be shown to be
( )
12 21
in 11 11 z l
22 l
z z
Z z z 1 T 0, Z
z Z

= = +

+
(104)
and
( )
12 21
out 22 22 z s
11 s
z z
Z z z 1 T
z Z
Z , 0 .

= = +

+
(105)
EXAMPLE #7:
The equivalent circuit given in Figure (36) is an approximate, low frequency, com-
mon source, model of a MOSFET amplifier that can be designed to realize match
terminated impedance characteristics. With reference to the diagram in Figure (36),
match termination means that for a Thvenin source resistance, R
s
, equal to the
terminating load resistance, R
l
, the driving point input resistance, R
in
, and the driving
point output resistance, R
out
, are equal and, in fact, are equal to R
s
. In other words, if
R symbolizes the numerical value of the source and load resistances, R
s
= R
in
= R
out
=
R
l
R under match terminated operating conditions. Use two port parameter tech-
niques to determine the values of the circuit resistances, R
x
and R
f
, commensurate
with a match terminated condition. For this situation, derive an expression for the
resultant voltage gain, A
v
= V
2
/V
s
. For the purpose of this problem, assume that the
resistance, R
i
, which represents the static resistance seen between the gate and source
Course Notes #1 University of Southern California John Choma

August 2006 52 Two Port Networks

terminals of the utilized MOSFET, is very large. No presumption should be made
about the degeneration resistance, R
g
, in the source terminal of the MOSFET circuit
model.
R
i
R
f
R
g
g V
m
V
1
I
1
I
2
V
2
+

V
s
R
l
R
s
R
in
R
out

V

+

R
x

Fig. (36). The Low Frequency, Small Signal Equivalent Circuit Of
The Match Terminated Amplifier Considered In Example
#7.
SOLUTION:
(1). The first step in the solution procedure is a determination of the appropriate two
port parameters of the network that couples the signal source to the load termina-
tion in Figure (36). The short circuit admittance parameters are somewhat
arbitrarily selected herewith and to this end, Figure (37a), which depicts the sub-
ject network under the condition of a short circuited output port, is pertinent to
the evaluation of the y-parameters, y
11
and y
21
. Note in this figure that the current
conducted by resistance R
g
is
R
i
R
f
R
g
g V
m
V
1
I
1
I
2
V
= 0
2

V

+

R
x
V
R
i
(1 + g R )
m i
(a).
R
i
R
f
R
g
g V
m
V
= 0
1
I
1
I
2
V
2

V

+

R
x
V
R
i
(1 + g R )
m i
(b).

Fig. (37). (a). Circuit Model Used To Evaluate The y-Parameters, y
11
and y
21
, in Example #7. (b). Equivalent
Circuit For Calculating The y-Parameters, y
12
and y
22
, in Example #7.
( )
m m
i i
V V
i
g V 1 g R
R R
+ = +

whence the input port voltage, V
1
, is expressible as
( )
1 i m i
i
V
V R 1
g
g R R .
R

= + +



In Figure (37a), it follows that
Course Notes #1 University of Southern California John Choma

August 2006 53 Two Port Networks

1
1
i f
V
V
I ,
R R
= +

from which
( )
2
1
11
1 f
i m i g
V 0
I
1 1
f
1
y .
V R
R 1 g R R
=
= = +
+ +
R

Moreover, since
1
2 m
f
V
I g V ,
R
=

the short circuit forward transadmittance, y
21
, is
( )
2
m i 2
21 me
1 f
i m i g
V 0
g R I
1
y g ,
V R
R 1 g R R
=
= =
+ +

where
m
me
m g
g
g
1 g R +

represents an effective forward transconductance. Note that the impact of the
source terminal degeneration resistance, R
g
, is a degradation of the original value
of the MOSFET forward transconductance, g
m
.
(2). Figure (37b), offers the equivalent circuit appropriate to the computation of the y-
parameters, y
12
and y
22
. In this figure that the current conducted by resistance R
g
,
is identical to the current flowing through R
g
in Figure (37a). Hence, the input
port voltage, V
1
, has the same algebraic expression for both circuit models, which
means that with V
1
= 0, voltage V is also zero. It follows immediately that
1
1
12
2 f
V 0
I
1
y
V R
=
= =

and
1
2
22
2 f
V 0
I
1 1
x
y .
V R
=
= = +
R

(3). If the Thvenin source impedance and the terminating load impedance are purely
resistive and each equal to the stipulated value, R, Y
s
= Y
l
= 1/R in (99) and (100).
A further inspection of these two relationships indicates that the driving point
input admittance, Y
in
(conductance in this case), and the driving point output
admittance, Y
out
(also presently a conductance), are equal to one another for Y
s
=
Y
l
= 1/R if and only if y
11
y
22
. Using the results of the preceding two computa-
tional steps, y
11
= y if the shunting output port resistance, R
x
, is chosen to satisfy
22
( )
x g m g i
R R 1 g R R = + + ,
which, for large R
i
, reduces to
Course Notes #1 University of Southern California John Choma

August 2006 54 Two Port Networks

( )
x m g i
R 1 g R R + .
Note, therefore that large R
i
commands a proportionately large R
x
.
(4). Returning to either (99) or (100),
12 21
11
11
y y
1
y .
1
R
y
R
=
+

Using the y-parameter results of Steps #1 and #2 and the result for R
x
in Step #3,
this requirement can be shown to imply
( )
( )
me x x
f
2
x
2 g R R
R .
R R 1
+
=

Since R
x
, is very large by virtue of its direct proportionality to the very large
model resistance, R
i
,
2
2 m
f me
m g
g R
R g R .
1 g R
=
+

(5). The transimpedance of the network in Figure (36) is given by (96) -through- (98).
Transimpedance Z
f
relates to the network voltage gain, A
v
, in accordance with
f
2 2
v
s s
Z
V V
A ,
V I R R
= = =

where the source resistance, R
s
, is equated to its match terminated value of R.
With y
11
= y
22
and Y
s
= Y
l
= R
s
R, the preceding result and the aforementioned
gain equations produce
2 21
v
2
s
11 12 21
V y R
A .
V
1
y y
R
= =

+


y

The substitution of the pertinent computational results of Steps #1-#4 into this
expression, followed by a bit of algebraic acrobatics, yields
me 2
v
s
g R 1 V
A .
V 2

=



COMMENTS: Match terminated amplifiers are pivotal to state of the art
communication circuits. These circuits are plagued by signal
strength levels that are comparable to electrical noise floors, thereby
rendering reliable signal detection problematic. Matched termina-
tions at a given port assure maximum power transfer therein, which
ensures minimal, if any, signal power loss between the effective
source and load of this port. Although the traditional engineering
design approaches that underpin the realization of a match terminated
amplifier can prove to be a cumbersome undertaking, this example
points out the relevant propriety of two port parameters. In particu-
lar, the two port parameters, and particularly the y-parameters
exploited herein, support the design objective by highlighting the
Course Notes #1 University of Southern California John Choma

August 2006 55 Two Port Networks

fundamental requirements supportive of, and streamlining the design-
oriented computation steps that result in, match terminated circuit
operation.
4.0. SYSTEMS OF INTERCONNECTED TWO PORTS
The closed loop gain (voltage gain, current gain, transimpedance, or transadmittance)
of a linear two port network is nominally independent of the source and load terminations and
nominally inversely dependent on the two port feedback parameter (h
12
, g
12
, y
12
, or z
12
) if the
magnitude of the network loop gain is sufficiently large. For most active networks divorced of
purposefully applied external feedback and almost all passive circuits, the internal feedback
parameter is too small to deliver the loop gain magnitude required for this parametric
desensitization.
To ensure reliable and predictable closed loop performance, the internal feedback
parameter of an active two port network is commonly augmented by incorporating a second
feedback network extrinsic to the active cell. This appended feedback loop is routinely, but not
universally, formed of passive components for at least two reasons. First, the values of passive
elements, and especially the ratios of passive element component values, are more predictable
than are the parameters of active devices. Thus, to the extent that the incorporated external feed-
back network is the dominant vehicle for determining the effective feedback parameter of the
overall closed loop system, the resultant closed loop gain becomes a predictable system perform-
ance metric. A second reason underpinning the use of passive external feedback structures stems
from the fact that passive circuits are reciprocal architectures that are incapable of providing
gain. They are thus stable structures that do not substantially alter the forward gain magnitude
and phase characteristics of the active cells with which they interact. In particular, the magni-
tude of the forward transmission parameter of a passive feedback cell is likely to be significantly
smaller than the magnitude of the corresponding forward transmission parameter of the active
cell to which it is interconnected.
When the appended feedback loop is connected from the output port of the original
open loop two port network to its input port, the nature of the feedback is said to be global.
There are four commonly exploited forms of global feedback: series-shunt feedback, shunt-series
feedback, shunt-shunt feedback, and series-series feedback. Assuming that the two port parame-
ters of the original open loop structure and those of the appended feedback subcircuit are unal-
tered by their input and output port electrical interconnections, the analysis of the resultant global
architecture can be straightforwardly formulated in terms of a mere superposition of their
appropriate respective two port parameters
[7]
. That which constitutes appropriate two port
parameters in the analysis of global feedback systems is largely a matter of technical common
sense, as the following subsections illustrate.
4.1. SERIES-SHUNT FEEDBACK
Series-shunt feedback materializes by interconnecting two networks in such a way as to
ensure the satisfaction of two operational constraints. The first of these constraints is that the
overall input port voltage of the global interconnection is the sum of the input port voltages of
the individual two networks. Second, the net output port current is precisely the sum of the
respective output port currents for the two interconnected structures. The situation at hand is
abstracted in Figure (38a), which shows an uncompensated linear two port network, Network
Course Notes #1 University of Southern California John Choma

August 2006 56 Two Port Networks

A, that is compensated by a second linear two port network, Network B, that interconnects
with the first network as a series-shunt feedback structure. The series nature of the intercon-
nected input ports of these two networks is confirmed by the observation that the net input port
voltage, V
1
, relates to the input port voltage, V
1a
, of Network A and the input port voltage, V
1b
, of
Network B in accordance with
Linear
Two Port
Network A
Z
s
Z
l
+

V
1a
I
1a
+

V
s
I
2a
+

V
2a
Linear
Two Port
Network B
+

V
1b
I = k I
1b h 1a
I
2b
+

V
2b
I
2
V
2
I
1
V
1
+

V
1
I
1
I
2
+

V
2
+

h
22e
h
11e
h I
21e 1
h V
12 e 2
Z
l
Z
s
+

V
s
(a).
(b).

Fig. (38). (a). Series-Shunt Feedback Interconnection Of Two Linear Two Port Networks. (b). The h-
Parameter Equivalent Circuit Of The System In (a).
1 1a 1b
V V V = + . (106)
Moreover, the net output port current, I
2
, conducted by the extrinsic load impedance, Z
l
is seen to
be the sum of the output currents, I
2a
and I
2b
, of Networks A and B, respectively; that is,
2 2a 2b
I I I = + .
.
(107)
The shunt interconnection nature of the two network output ports is further confirmed by the
observation that the overall output port voltage, V
2
, is identical to the output port voltage V
2a
, of
Network A and the output port voltage, V
2b
, of Network B. Specifically,
2 2a 2b
V V V = = (108)
Although a strictly series input port architecture mandates that the net input current, I
1
, is identi-
cal to both of the individual input port currents, I
1a
and I
1b
, such current equality cannot generally
Course Notes #1 University of Southern California John Choma

August 2006 57 Two Port Networks

be guaranteed, particularly when Network A is an active configuration. However, the linearity of
the interconnected closed loop system allows the conjecture that the Network B input port cur-
rent, I
1b
, is proportional to I
1a
, the input port current of Network A. Accordingly, let
1b h 1a
I k I , = (109)
where k
h
is a constant, independent of all network voltages and currents, determined by the
electrical properties of Network A.
Because the h-parameter equivalent circuit is effectively a series-shunt topology evok-
ing input port voltage V
1
and output port current I
2
, both of which are sums of respective port
variables as per (106) and (107), as dependent variables, h-parameters are used to model each of
the two subsystems in Figure (38a). Accordingly,
1a 11a 12a 1a 11a 12a 1
2a 21a 22a 2a 21a 22a 2
V h h I h h I
,
I h h V h h V

= =


(110)
where (108) and the fact that I
1a
is the overall input port current, I
1
, are exploited. For Network
B,
1b 11b 12b 1b 11b 12b h 1a h 11b 12b 1
2b 21b 22b 2b 21b 22b 2b h 21b 22b 2
V h h I h h k I k h h I
,
I h h V h h V k h h V

= = =


(111)
where (108), (109), and the rules of matrix algebra, have been applied. Noting that the input port
current, I
1
, and the output port voltage, V
2
, appear as common independent variables in both of
the preceding two relationships, (106) and (107) combine with (110) and (111) to deliver
( ) ( )
( ) ( )
11a h 11b 12a 12b
1
2 2
21a h 21b 22a 22b
h k h h h
V I
.
I V
h k h h h

+ +

+ +



1

(112)
This elegantly simple result conduces a straightforward delineation of the h-parameter equivalent
circuit for the series-shunt feedback network of Figure (38a). This model appears in Figure
(38b), where from (112), it is understood that the effective h-parameters of the overall feedback
system are
11e 11a h 11b
12e 12a 12b
21e 21a h 21b
22e 22a 22b
h h k h
h h h
.
h h k h
h h h
= +
= +
= +
= +
(113)
The performance metrics of the series-shunt feedback structure now derive directly from the
pertinent equations given in Section (3.1), subject to the proviso therein that, h
ij
is replaced by
the effective parameter, h
ije
.
Course Notes #1 University of Southern California John Choma

August 2006 58 Two Port Networks

EXAMPLE #8:
Figure (39a) is a classic example of series-shunt feedback implemented in bipolar
junction transistor (BJT) technology. As is suggested in this figure, active Network
A is the two stage transistor amplifier, while feedback Network B is the voltage
divider comprised of the resistances, R
1
and R
2
. Assume that both transistors are
identical and are biased identically (biasing is not shown) and that Figure (38b) is a
sufficiently accurate low frequency model of each transistor. In this model, take r
i
=
2 K and = 120. Additionally, let R
1
= 150 , and R
2
= 900 . Assume that the
signal source resistance, R
s
, is 75 , while the terminating load resistance, R
l
, is 1.2
K. Calculate the overall voltage gain, A
v
= V
2
/V
s
, the driving point input resistance,
R
in
, and the driving point output resistance, R
out
. Additionally, deduce appropriately
simplified analytical relationships for each of these three performance indices.
R
1
r
i
R
l
R
2
R
s
Network B
Network A
+

V
1a
I
1a
I
( +1)I
I
2a
I
1b
+

V
s
+

V
1b
I
1
V
1
V
2
I
2b
I
2
I
Base
Collector Collector
Emitter
Emitter
Base
(a).
(b).

Fig. (39). (a). Series-Shunt Feedback Amplifier Realized In Bipolar
Junction Transistor (BJT) Technology. (b). Simplified Small
Signal, Low Frequency Model Of A BJT.
SOLUTION:
(1). The equivalent circuit of the overall feedback configuration is depicted in Figure
(40a), where use has been made of the bipolar model suggested in Figure (39b).
Since feedback Network B is connected in series-shunt with active Network A, h-
parameters are the appropriate modeling vehicle for Network B. To this end,
Course Notes #1 University of Southern California John Choma

August 2006 59 Two Port Networks

R
1
r
i
R
l
R
2
R ||R
1 2
( +1)(R ||R )
1 2
R
s
Network B
Network A
+

V
1a
I
1a
( +1)I
1a
I
2a
I
1b
I
1b
V
1b
+

V
s
+

V
1b
I
1
V
1
V
2
I
2b
I
2
I
1a
r
i
I
( +1)I
I
(a).
+

R V
1 2
R V
1 2
R + R
1 2
R + R
1 2
R I
1 1b
R + R
1 2
R + R
1 2
( +1)R I
1 1
R + R
1 2
R + R
1 2
I
2b
V
2
(b).
r
i
R
l
R
s
R
in
+

V
s
I
1
V
1
V
2
I
2

2
I
1
(c).
R
iout

Fig. (40). (a). Small Signal, Low Frequency Equivalent Circuit Of The Feedback Amplifier In
Figure (39a). (b). The h-Parameter Model Of Feedback Network B. (c). Simplified
Version Of The Equivalent Circuit In (a). Note That The Topology Of This Architec-
ture Reflects That Of The Classic h-Parameter Equivalent Circuit Of A Linear Net-
work.
Course Notes #1 University of Southern California John Choma

August 2006 60 Two Port Networks

2b
1b
1b
1b
11b 1 2
1b
V 0
2b
22b
2 1 2
I 0
1b 1
12b 21b
2 1 2
I 0
V
h R R
I
I
1
h .
V R R
V R
h h
V R R
=
=
=
= =
= =
+
= =
+

This result establishes Figure (40b) as the h-parameter equivalent circuit of feed-
back Network B. If the feedback network in the amplifier of Figure (40a) is
replaced by its h-parameter model of Figure (40b), the equivalent circuit of the
overall amplifier can be configured as illustrated in Figure (40c). In the latter dia-
gram, use is made of the fact that current I in Figure (40a) is I
1a
= I
1
, whence
I in the same figure becomes
2
I
1
. Moreover, observe that the input current,
I
1b
, of feedback Network B is (+1)I
1a
= (+1)I
1
, which implies that the constant,
k
h
, introduced in (109) is k
h
= (+1). It follows that h
21b
I
1b
= (+1)h
21b
I
1
.
(2). By comparing the model developed in Figure (40c) with the generalized h-
parameter equivalent circuit in Figure (28), the hybrid h-parameters of the overall
feedback amplifier can be deduced by mere inspection. In particular,
( )
( )
11e i 1 2
h r 1 R R 17.56 K , = + + =
( )
2
1
21e
1 2
R
h 1 14.42 KA/ A
R R


= + + =

+

,

1
12e
1 2
R
h 1
R R
/ = =
+
7 V/V ,

22e
1 2
1
h 9
R R
52.4 S . = =
+

(3). The closed loop performance of the subject amplifier can now be determined
directly from pertinent relationships in Section (3.1). From (77), the open loop
voltage gain is A
vo
= 457.9 = 53.22 dB, and from (78), the corresponding closed
loop voltage gain is A
v
= 6.90 = 16.77 dB. Note a loop gain of T
h
(R
s
,G
l
) = 65.41
= 38.31 dB.
(4). Using (87), the closed loop input resistance is R
in
= 1.17 Meg , while (89)
yields a closed loop output conductance of G
out
= 117.8 mS, which equates to an
output resistance of R
out
= 8.49 .
(5). Since the loop gain in this example is very large, several simplified analytical
relationships can be forged. Recalling (78) once again, the closed loop gain is
approximately
Course Notes #1 University of Southern California John Choma

August 2006 61 Two Port Networks

2
v
12e 1
R
1
A 1 7
h R
= + = .0 ,
which is only 1.45% higher than the precisely computed voltage gain.
(6). Because of large loop gain, the driving point input resistance, R
in
, is expressible
as
in 11e h l
R h T ( 0, G ) ,
which for large is equivalent to
2 l
in 1
l 1 2
R
R R 1.15 Meg .
R R R



=

+ +


This result is lower than that computed in Step #4 by only 1.54%. Similarly, the
output conductance, G
out
= 1/R
out
, is
out 22e h s
G h T ( R , 0 ) ,
or
2
out
R
R 7.5 .

=
When compared to the calculation in Step #4, this result is seen to be in error by
11.7%.
COMMENTS: Very large input resistance and very small output resistance combine
to make the series-shunt feedback amplifier an excellent voltage
amplifier. In particular, its low output resistance allows small load
resistances to be driven without rendering the voltage gain
significantly dependent on actual load resistance. Similarly, its large
input resistance effectively desensitizes the voltage gain to uncertain-
ties in source resistances. Best of all, the resultant approximate
voltage gain is dependent on a resistance ratio, which is relatively
easy to control accurately during monolithic processing. It is impor-
tant to underscore the fact that these design-oriented observations
derive directly from the fruits of a straightforward two port parameter
analysis of the subject feedback configuration. In the absence of this
two port methodology, a conventional circuit analysis of the ampli-
fier at hand would likely have generated the documented numerical
results without the insights that implicitly underpin innovative circuit
design.
4.2. SHUNT-SERIES FEEDBACK
The shunt-series feedback amplifier, whose system level diagram is abstracted in Fig-
ure (41a) is exceptionally well-suited for current amplification because of the low input imped-
ance and high output impedance it delivers. Accordingly, the signal source is represented as a
Norton structure, and it is understood that the relevant output response variable of interest is the
overall output port current, I
2
. The shunt nature of the respective input port connections is under-
scored by the fact that the overall input port voltage, V
1
, the input port voltage, V
1a
, of Network A
(which is generally the active unit of the feedback system), and the input port voltage, V
1b
, of
Network B are identical. Moreover, the overall input port current, I
1
, is the sum of currents I
1a

Course Notes #1 University of Southern California John Choma

August 2006 62 Two Port Networks

and I
1b
, which respectively symbolize the currents flowing into the input ports of Networks A
and B. On the other hand, the overall output port voltage, V
2
, is the sum of the output port volt-
age, V
2a
, of Network A and V
2b
, the voltage developed across the output port of Network B.
Although this output port voltage relationship suggests a series topology, the current, I
2b
,
conducted by the output port of Network B is not necessarily the same as the output port current,
I
2a
, of Network A because of electrical constraints implicit to the circuit architecture of Network
A. However, system linearity compels that I
2b
be proportional to I
2a
and to this end, a
proportionality constant, k
g
, is introduced, as is indicated in Figure (41a).
Linear
Two Port
Network A
Z
l
+

V
1a
I
1a
I
s
I
s
I
2a
+

V
2a
Linear
Two Port
Network B
I = k I
2b g 2a
+

V
2b
I
2
V
2
I
1
V
1
Z
l
(a).
+

V
2
I
2
I
1
+

V
1
+

g
11e
g
22e
g I
12e 2
g V
21e 1
+

V
1b
I
1b
Z
s
Z
s
(b).

Fig. (41). (a). Shunt-Series Feedback Interconnection Of Two Linear Two Port Networks. (b). The g-
Parameter Equivalent Circuit Of The Network In (a).
Because the g-parameter equivalent circuit is a shunt-series topology, g-parameters can
be gainfully exploited to effect an efficient analysis of the shunt-series feedback amplifier. For
Network A,
1a 11a 12a 1a 11a 12a 1
2a 21a 22a 2a 21a 22a 2
I g g V g g V
,
V g g I g g I

= =


(114)
and for Network B,
11b g 12b
1b
1b 11b 12b 1b 11b 12b 1
g 2a
2b 21b 22b 2b 21b 22b 21b g 22b 2
g k g
V
I g g V g g V
.
k I
V g g I g g g k g I





= = =






(115)
Since I
1
= I
1a
+ I
1b
and V
2
= V
2a
+ V
2b
, it follows that
Course Notes #1 University of Southern California John Choma

August 2006 63 Two Port Networks

( )
( )
( )
( )
11a 11b 12a g 12b
1
2 2
21a 21b 22a g 22b
g g g k g
I V
,
V I
g g g k g

+ +


=

+ +

1

(116)
whence the effective g-parameters in the g-parameter equivalent circuit shown in Figure (41b)
are
11e 11a 11b
12e 12a g 12b
21e 21a 21b
22e 22a g 22b
g g g
g g k g
.
g g g
g g k g
= +
= +
= +
= +
(117)
The performance merits derived in Section (3.2) can now be applied directly to characterize the
shunt-series feedback amplifier.
Shunt-series feedback generally entails the interconnection of an active structure, Net-
work A, with a passive unit, Network B. In this case, |g
21a
| >> |g
21b
| and |g
12a
| << |k
g
g
12b
|. This
is to say that the input -to- output, or forward, signal path is dominated by the gain afforded by
the active structure, while the feedback subcircuit is designed to ensure that it dominates the out-
put -to- input, or feedback signal flow path. From Section (3.2), a large loop gain delivers a
resultant closed loop current gain, A
i
, of
2
i
s 12e g 12b
I
1 1
A
I g k g
= , (118)
which is nominally independent of parametric vagaries inevitably encountered in active circuits.
Moreover, large loop gain produces a driving point input admittance, Y
in
, of
g 12b 21a
in
22a g 22b l
k g g
Y
g k g Z

+ +
(119)
and a driving point output impedance, Z
out
, given approximately by
g 12b 21a
in
11a 11b s
k g g
Z .
g g Y

+ +
(120)
The negative algebraic signs appearing on the right hand sides of the preceding two expressions
are not a concern in unconditionally stable shunt-series systems, which have k
g
g
12b
g
21a
< 0 for at
least low signal frequencies.
4.3. SHUNT-SHUNT FEEDBACK
In the shunt-shunt feedback architecture of Figure (42a), y-parameters comprise the
most expeditious analytical vehicle. The structure at hand produces low input impedance and
low output impedance, thereby rendering it suitable for transimpedance applications. Armed
Course Notes #1 University of Southern California John Choma

August 2006 64 Two Port Networks

with the experiences gleaned from the preceding two considered feedback architectures, it is eas-
ily demonstrated that the pertinent y-parameter equivalent circuit is the model in Figure (42b),
where the effective y-parameters, y
ije
, are
Linear
Two Port
Network A
Z
l
+

V
1a
I
1a
I
s
I
s
I
2a
+

V
2a
Linear
Two Port
Network B
+

V
2b
I
2
V
2
I
1
V
1
Z
l
(a).
+

V
2
I
2
I
1
+

V
1
y
22e
y V
12e 2
y V
21e 1
+

V
1b
I
1b
Z
s
Z
s
(b).
I
2b
y
11e

Fig. (42). (a). Shunt-Shunt Feedback Interconnection Of Two Linear Two Port Networks.
(b). The y-Parameter Equivalent Circuit Of The System In (a).
11e 11a 11b
12e 12a 12b
21e 21a 21b
22e 22a 22b
y y y
y y y
.
y y y
y y y
= +
= +
= +
= +
(121)
If Network A is an active circuit and Network B is a passive structure, |y
21a
| >> |y
21b
| and |y
12a
|
<< |y
12b
| for reasons analogous to those proffered in conjunction with the shunt-series amplifier.
Under these conditions and for large loop gain, the closed loop transimpedance can be shown to
be given by
2
f
s 12e 12b
V
1 1
Z .
I y y
= (122)
Moreover, the driving point input and output admittances respectively become
Course Notes #1 University of Southern California John Choma

August 2006 65 Two Port Networks

12b 21a
in
22a 22b l
y y
Y ,
y y Y

+ +
(123)
and
12b 21a
out
11a 11b s
y y
Y .
y y Y

+ +
(124)
4.4. SERIES-SERIES FEEDBACK
In the series-series feedback system of Figure (43a), Thvenin equivalent I/O port mod-
els, and thus z-parameter modeling is appropriate. The considered system at hand produces high
input and output driving point closed loop impedances, whence the system functions as a
transadmittance amplifier. As is the case with the series components of the series-shunt and
shunt-series amplifiers, note that the input and output port currents of Network B are indicated as
respectively proportional to the input and output port currents of the presumably active subcir-
cuit, Network A. The applicable z-parameter equivalent circuit is provided in Figure (43b),
where the effective z-parameters, z
ije
, are
Linear
Two Port
Network A
Z
s
Z
l
+

V
1a
I
1a
+

V
s
I
2a
+

V
2a
Linear
Two Port
Network B
+

V
1b
I = k I
1b i 1a
+

V
2b
I
2
V
2
I
1
V
1
+

V
1
I
1
+

z
11e
z I
12e 2
Z
s
+

V
s
(a).
(b).
I = k I
2b o 2a
Z
l
+

V
2
I
2
+

z
22e
z I
21e 1

Fig. (43). (a). Series-Series Feedback Interconnection Of Two Linear Two Port Networks. (b).
The z-Parameter Equivalent Circuit Of The System In (a).
Course Notes #1 University of Southern California John Choma
11e 11a i 11b
12e 12a o 12b
21e 21a i 21b
22e 22a o 22b
z z k z
z z k z
.
z z k z
z z k z
= +
= +
= +
= +
(125)
Assuming large loop gain, |z
21a
| >> |k
i
z
21b
|, and |z
12a
| << |k
o
z
12b
|, the closed loop transadmittance
is
2
f
s 12e o 12b
I
1 1
Y
V z k z
= , (126)
the closed loop input impedance is
o i 12b 21a
in
22a o 22b l
k k z z
Z ,
z k z Z

+ +
(127)
and the closed loop output impedance is
o i 12b 21a
out
11a i 11b s
k k z z
Z .
z k z Z

+ +
(128)
5.0. REFERENCES
[1]. A. M. Davis, Linear Circuit Analysis. Boston, Massachusetts: PWS Publishing Company,
1998, pp. 589-590.
[2]. J. Choma, Jr., Electrical Networks: Theory and Analysis. New York: Wiley-Interscience, 1985,
chap. 9.
[3]. S-Parameter Techniques for Faster, More Accurate Network Design, Application Note #95-1,
Hewlett-Packard Company, Feb. 1967.
[4]. S-Parameter Design, Application Note #154, Hewlett-Packard Company, Apr. 1972.
[5]. G. Palumbo and J. Choma, Jr., An Overview Of Single And Dual Loop Analog Feedback; Part
I: Basic Theory, Journal of Analog Integrated Circuits and Signal Processing, vol. 17, pp.
175-194, Nov. 1998.
[6]. T. H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits. Cambridge, United
Kingdom: Cambridge University Press, 1998, chap. 16.
[7]. A. J. Cote, Jr. and J. B. Oakes, Linear Vacuum-Tube and Transistor Circuits. New York:
McGraw-Hill Book Company, 1961, pp.40-46.


August 2006 66 Two Port Networks