S-Parameter Techniques: Test & Measurement

http://www.hp.
com/go/tmappnotes
H Test & Measurement
Application Note 95-1
S-Parameter
Techniques
for Faster, More Accurate Network Design

Contents
1. Foreword and Introduction
2. Two-Port Network Theory
3. Using S-Parameters
4. Network Calculations with Scattering Parameters
5. Amplifier Design using Scattering Parameters
6. Measurement of S-Parameters
7. Narrow-Band Amplifier Design
8. Broadband Amplifier Design
9. Stability Considerations and the Design of
Reflection Amplifiers and Oscillators
Appendix A. Additional Reading on S-Parameters
Appendix B. Scattering Parameter Relationships
Appendix C. The Software Revolution
Relevant Products, Education and Information
Contacting Hewlett-Packard
Copyright Hewlett-Packard Company, 1997. 3000 Hanover Street, Palo Alto California, USA.
H
T est & M easurement
A ppli cati on Note 95-1
S-P aramet er T echni ques

3
Foreword
H
A ppli cati on Note
95-1
February 1967 HP J ournal

Cover of issue in which
the original S-Parameters
Theory and Application,
written during Christmas
holiday 1966, first appeared.
HP J ournal is now online at:
www.hp.com/go/journal
HEWLETT-PACKARD J OURNAL
FEBRUARY 1967
A NEWMICROWAVE INSTRUMENT SWEEP
MEASURES GAIN, PHASE IMPEDANCE

WITHSCOPE ORMETERREADOUT; page 2
THE MICROWAVE ANALYZERIN THE
FUTURE; page 11
S-PARAMETERS THEORY AND
APPLICATIONS; page 13
Cover:
See Also:
T hi s appli cati on note i s based on an arti cle wri tten for the
February 1967 i ssue of the Hewlett-Packard J ournal, yet
i ts content remai ns i mportant today. S-parameters are an
essenti al part of hi gh-frequency desi gn, though much else
has changed duri ng the past 30 years. Duri ng that ti me,
HP has conti nuously forged ahead to help create today' s
leadi ng test and measurement envi ronment.
We conti nuously apply our capabi li ti es i n measurement,

communi cati on, and computati on to produce i nnovati ons
that help you to i mprove your busi ness results. I n wi reless
communi cati ons, for example, we esti mate that 85 percent
of the worlds GSM ( Groupe Speci ale M obi le) telephones
are tested wi th HP i nstruments. O ur accompli shments
30 years hence may exceed our boldest conj ectures.
T hi s i nteracti ve appli cati on note revi ses and updates

the1967 arti cle for onli ne electroni c medi a. I t reflects
the changes i n our i ndustry, whi le remi ndi ng us of the
underlyi ng sci enti fi c basi s for the technology, and takes
advantage of a potent new i nformati on di ssemi nati on capabi li ty,
the World Wi de Web. We hope you fi nd thi s tutori al useful.

Ri chard Anderson,
HP Vi ce Presi dent and General M anager,
M i crowave and Communi cati ons Group
4
Li near networks, or nonli near networks operati ng wi th si gnals
suffi ci ently small to cause the networks to respond i n a li near
manner, can be completely characteri zed by parameters
measured at the network termi nals ( ports) wi thout regard to
the contents of the networks. O nce the parameters of a
network have been determi ned, i ts behavi or i n any external
envi ronment can be predi cted, agai n wi thout regard to the
contents of the network.
S-parameters are i mportant i n mi crowave desi gn because they
are easi er to measure and work wi th at hi gh frequenci es than
other ki nds of parameters. T hey are conceptually si mple,
analyti cally conveni ent, and capable of provi di ng a great
i nsi ght i nto a measurement or desi gn problem.
To show how s-parameters ease mi crowave desi gn, and how
you can best take advantage of thei r abi li ti es, thi s appli cati on
note descri bes s-parameters and flow graphs, and relates them
to more fami li ar concepts such as transducer power gai n and
voltage gai n. Data obtai ned wi th a network analyzer i s used to
i llustrate ampli fi er desi gn.
Introduction
1
Maxwells equations
All electromagnetic
behaviors can ultimately be
explained by Maxwells four
basic equations:
However, it isnt always
possible or convenient to
use these equations directly.
Solving them can be quite
difficult. Efficient design
requires the use of
approximations such
as lumped and distributed
models.

+
D E
B
B H j
D
t
t
0
H
5
Although a network may have any number of ports,
network parameters can be explai ned most easi ly by
consi deri ng a network wi th only two ports, an i nput port
and an output port, li ke the network shown i n Fi gure 1.
To characteri ze the performance of such a network, any
of several parameter sets can be used, each of whi ch has
certai n advantages. Each parameter set i s related to a set of
four vari ables associ ated wi th the two-port model. Two of these
vari ables represent the exci tati on of the network ( i ndependent
vari ables) , and the remai ni ng two represent the response of
the network to the exci tati on ( dependent vari ables) . I f the
network of Fi g. 1 i s exci ted by voltage sources V
1
and V
2
, the
network currents I
1
and I
2
wi ll be related by the followi ng
equati ons ( assumi ng the network behaves li nearly) :
( 1)
( 2)
I y V y V
2 21 1 22 2
+

I y V y V
1 11 1 12 2
+
2
Two-Port Network Theory
Figure 1
General two-port network.
H
I
1
I
2
V
1
V
2
TWO - PORT
NETWORK
+ +

Port 2 Port 1
Why are models needed?
Models help us predict the
behavior of components,
circuits, and systems.
Lumped models are useful at
lower frequencies, where
some physical effects can be
ignored because they are so
small. Distributed models
are needed at RF frequencies
and higher to account for the
increased behavioral impact
of those physical effects.
6
I n thi s case, wi th port voltages selected
as i ndependent vari ables and port
currents taken as dependent vari ables,
the relati ng parameters are called
short-ci rcui t admi ttance parameters,
or y-parameters. I n the absence of
addi ti onal i nformati on, four
measurements are requi red to determi ne the four parameters
y
11
,
y
12
,
y
21
,
y
22
. Each measurement i s made wi th one port
of the network exci ted by a voltage source whi le the other
port i s short ci rcui ted. For example, y
21,
the forward
transadmi ttance, i s the rati o of the current at port 2 to the
voltage at port 1 wi th port 2 short ci rcui ted, as shown i n
equati on 3.
( 3)
y
I
V
V
21
2
1
0
2
( output short ci rcui ted)

2
Two-port models
Two-port, three-port, and
n-port models simplify the
input / output response of
active and passive devices
and circuits into black
boxes described by a set
of four linear parameters.
Lumped models use
representations such as
Y (conductances),
Z (resistances), and
h (a mixture of conductances
and resistances). Distributed
models use s-parameters
(transmission and reflection
coefficients).
H
7
I f other i ndependent and dependent vari ables had been
chosen, the network would have been descri bed, as
before, by two li near equati ons si mi lar to equati ons 1
and 2, except that the vari ables and the parameters
descri bi ng thei r relati onshi ps would be di fferent.
However, all parameter sets contai n the same
i nformati on about a network, and i t i s always possi ble
to calculate any set i n terms of any other set.
Scatteri ng parameters, whi ch are commonly referred
to as s-parameters, are a parameter set that relates to
the traveli ng waves that are scattered or reflected
when an n-port network i s i nserted i nto a transmi ssi on
li ne.
Appendi x B Scatteri ng Parameter Relati onshi ps
contai ns tables converti ng scatteri ng parameters to
and from conductance parameters ( y) ,
resi stance parameters( z) , and a mi xture of
conductances and resi stances parameters( h) .
2
H
8
3
Using S-Parameters
T he ease wi th whi ch scatteri ng parameters can be measured
makes them especi ally well sui ted for descri bi ng transi stors and
other acti ve devi ces. M easuri ng most other parameters calls for
the i nput and output of the devi ce to be successi vely opened
and short ci rcui ted. T hi s can be hard to do, especi ally at RF
frequenci es where lead i nductance and capaci tance make short
and open ci rcui ts di ffi cult to obtai n. At
hi gher frequenci es these measurements
typi cally requi re tuni ng stubs, separately
adj usted at each measurement frequency,
to reflect short or open ci rcui t condi ti ons
to the devi ce termi nals. Not only i s thi s
i nconveni ent and tedi ous, but a tuni ng
stub shunti ng the i nput or output may
cause a transi stor to osci llate, maki ng
the measurement i nvali d.
S-parameters, on the other hand, are usually measured wi th the
devi ce i mbedded between a 50 load and source, and there i s
very li ttle chance for osci llati ons to occur.
H

9
( 4) ( 5)

b
V Z I
Z
i
i i i
i
=
*
Re 2

a
V Z I
Z
i
i i i
i
=
+
2 Re
3
Transmission and Reflection
When light interacts with a
lens, as in this photograph,
part of the light incident on
the woman's eyeglasses is
reflected while the rest is
transmitted. The amounts
reflected and transmitted
are characterized by optical
reflection and transmission
coefficients. Similarly,
scattering parameters are
measures of reflection and
transmission of voltage
waves through a two-port
electrical network.
Using S-Parameters
H
Another i mportant advantage of s-parameters stems from
the fact that traveli ng waves, unli ke termi nal voltages and
currents, do not vary i n magni tude at poi nts along a lossless
transmi ssi on li ne. T hi s means that scatteri ng parameters can
be measured on a devi ce located at some di stance from the
measurement transducers, provi ded that the measuri ng devi ce
and the transducers are connected by low-loss transmi ssi on li nes.
Derivation
Generali zed scatteri ng parameters have been defi ned by
K . K urokawa [ Appendi x A] . T hese parameters descri be
the i nterrelati onshi ps of a new set of vari ables ( a
i
, b
i
) .
T he vari ables a
i
and b
i
are normali zed complex voltage waves
i nci dent on and reflected from the i
th
port of the network.
T hey are defi ned i n terms of the termi nal voltage V
i
, the
termi nal current I
i
, and an arbi trary reference i mpedance Z
i
,
where the asteri sk denotes the complex conj ugate:

For most measurements and calculati ons i t i s conveni ent
to assume that the reference i mpedance Z
i
i s posi ti ve
and real. For the remai nder of thi s arti cle, then, all
vari ables and parameters wi ll be referenced to a si ngle
posi ti ve real i mpedance, Z
0
.
T he wave functi ons used to defi ne s-parameters for a
two-port network are shown i n Fi g. 2.
10
3
Figure 2
Two-port network showi ng i nci dent waves
( a
1
, a
2
) and reflected waves ( b
1
, b
2
) used i n
s-parameter defi ni ti ons. T he flow graph for
thi s network appears i n Fi gure 3.
Using S-Parameters
Scattering parameters
relationship to optics
Impedance mismatches
between successive
elements in an RF circuit
relate closely to optics,
where there are successive
differences in the index of
refraction. A materials
characteristic impedance,
Z
0
, is inversely related to
the index of refraction, N:
The s-parameters s
11
and
s
22
are the same as optical
reflection coefficients;
s
12
and s
21
are the same
as optical transmission
coefficients.
N
Z
0
377
1
=
H
a
1
Z
S
Z
L
V
S
a
2
b
1
b
2
TWO - PORT
NETWORK
11
T he i ndependent vari ables a
1
and a
2
are normali zed
i nci dent voltages, as follows:
( 6)
( 7)
Dependent vari ables b
1
, and b
2
, are normali zed reflected voltages:
( 8)
( 9)

b
V I Z
Z Z
V
Z
r
2
2 2 0
0 0
2
0
2

voltage wave reflected from port 2

b
V I Z
Z Z
V
Z
r
1
1 1 0
0 0
1
0
2

voltage wave reflected from port 1

a
V I Z
Z Z
V
Z
i
2
2 2 0
0 0
2
0
2
+

voltage wave i nci dent on port 2

a
V I Z
Z Z
V
Z
i
1
1 1 0
0 0
1
0
2
+

voltage wave i nci dent on port 1

3
Using S-Parameters
H
T he li near equati ons descri bi ng the two-port network are
then:
( 10)
( 11)
T he s-parameters s
11
, s
22
, s
21
, and s
12
are:
( 12)
( 13)
( 14)
( 15)

s
b
a
a
12
1
2
0
1

s
b
a
a
21
2
1
0
2

s
b
a
a
22
2
2
0
1

s
b
a
a
11
1
1
0
2

b s a s a
2 21 1 22 2
+

b s a s a
1 11 1 12 2
+
12
3
= O utput reflecti on coeffi ci ent
wi th the i nput termi nated by a
matched load ( Z
S
= Z
0
sets V
s
= 0)
= Forward transmi ssi on ( i nserti on)
gai n wi th the output port
termi nated i n a matched load.
= Reverse transmi ssi on ( i nserti on)
gai n wi th the i nput port
termi nated i n a matched load.
Using S-Parameters
Limitations of
lumped models
At low frequencies most
circuits behave in a
predictable manner and
can be described by a
group of replaceable,
lumped-equivalent black
boxes. At microwave
frequencies, as circuit
element size approaches
the wavelengths of the
operating frequencies,
such a simplified type
of model becomes
inaccurate. The physical
arrangements of the
circuit components can
no longer be treated as
black boxes. We have to
use a distributed circuit
element model and
s-parameters.
= I nput reflecti on coeffi ci ent wi th
the output port termi nated by a
matched load ( Z
L
= Z
0
sets a
2
= 0)
H
13
Noti ce that
( 16)
and
( 17)
where i s the i nput i mpedance at port 1.
T hi s relati onshi p between reflecti on coeffi ci ent and i mpedance
i s the basi s of the Smi th Chart transmi ssi on-li ne calculator.
Consequently, the reflecti on coeffi ci ents s
11
and s
22
can be
plotted on Smi th charts, converted di rectly to i mpedance, and
easi ly mani pulated to determi ne matchi ng networks for
opti mi zi ng a ci rcui t desi gn.

Z
V
I
1
1
1

Z Z
s
s
1 0
11
11
1
1
( )
( )

s
b
a
V
I
Z
V
I
Z
Z Z
Z Z
11
1
1
1
1
0
1
1
0
1 0
1 0

+
3
S-parameters
S-parameters and
distributed models
provide a means of
measuring, describing,
and characterizing
circuit elements when
traditional lumped-
equivalent circuit
models cannot predict
circuit behavior to the
desired level of
accuracy. They are used
for the design of many
products, such as
cellular telephones.
Using S-Parameters
H

Smith Chart Transformation
T he movi e at the ri ght ani mates the
mappi ng between the complex i mpedance
plane and the Smi th Chart. T he Smi th
Chart i s used to plot reflectances, but the
ci rcular gri d li nes allow easy readi ng of
the correspondi ng i mpedance. As the
ani mati on shows, the rectangular gri d
li nes of the i mpedance plane are
transformed to ci rcles and arcs on the
Smi th Chart.
Verti cal li nes of constant resi stance on the i mpedance plane
are transformed i nto ci rcles on the Smi th Chart. Hori zontal
li nes of constant reactance on the i mpedance plane are
transformed i nto arcs on the Smi th Chart. T he transformati on
between the i mpedance plane and the Smi th Chart i s nonli near,
causi ng normali zed resi stance and reactance values greater
than uni ty to become compressed towards the ri ght si de of the
Smi th Chart.
14
3
Using S-Parameters
H
Animation 1
Transformation between
the impedance plane and
the Smith Chart. Click
over image to animate.
Showing transformations
graphically
To ease his RF design
work, Bell Labs Phillip H.
Smith developed increas-
ingly accurate and power-
ful graphical design aids.
One version, a polar co-
ordinate form, worked for
all values of impedance
components, but Smith
suspected that a grid with
orthogonal circles might
be more practical. In 1937
he constructed the basic
Smith Chart still used
today, using a transforma-
tion developed by co-
workers E.B. Ferrell and
J .W. McRae that accom-
modates all data values
from zero to infinity.

15
Advantages of S-Parameters
T he previ ous equati ons show one of the i mportant
advantages of s-parameters, namely that they are
si mply gai ns and reflecti on coeffi ci ents, both
fami li ar quanti ti es to engi neers.
By compari son, some of the y-parameters
descri bed earli er i n thi s arti cle are not so fami li ar.
For example, the y-parameter correspondi ng
to i nserti on gai n s
21
i s the forward
trans-admi ttance y
21
gi ven by
equati on 3. Clearly, i nserti on
gai n gi ves by far the
greater i nsi ght i nto
the operati on of
the network.
3
Digital pulses
Digital pulses are
comprised of high-order
harmonic frequencies
that determine the shape
of the pulse. A short
pulse with steep edges
has a signal spectrum
with relatively high
power levels at very high
frequencies. As a result,
some elements in
modern high-speed
digital circuits require
characterization with
distributed models and
s-parameters for accurate
performance prediction.
Using S-Parameters
H
16
3
Using S-Parameters
Another advantage of s-parameters spri ngs from the
si mple relati onshi p between the vari ables a
1
, a
2
, b
1
,
and b
2
, and vari ous power waves:

b
2
2
Power reflected from the output port of the network.

Power i nci dent on the load.
Power that would be deli vered to a Z load.
0

b
1
2
Power reflected from the i nput port of the network.

Power avai lable from a Z source mi nus the power
deli vered to the i nput of the network.
0

a
2
2
Power i nci dent on the output of the network.

Power reflected from the load.

a
Z
1
2
0
Power i nci dent on the i nput of the network

Power avai lable from a source i mpedance
.
.
Radar
The development
of radar, which
uses powerful signals
at short wavelengths to
detect small objects at
long distances, provided
a powerful incentive for
improved high frequency
design methods during
World War II. The design
methods employed at
that time combined
distributed measurements
and lumped circuit
design. There was an
urgent need for an
efficient tool that could
integrate measurement
and design. The Smith
Chart met that need.
H
S-P aramet er
T echni ques
17
T he previ ous four equati ons show that s-parameters are
si mply related to power gai n and mi smatch loss,
quanti ti es whi ch are often of more i nterest
than the correspondi ng voltage functi ons:

s
12
2
Reverse transducer power gai n wi th Z load and source
0

s
21
2
Power deli vered to a Z load

Power avai lable from Z source
T ransducer power gai n wi th Z load and source
0
0
0

s
22
2
Power reflected from the network output

Power i nci dent on the network output

s
11
2
Power reflected from the network i nput

Power i nci dent on the network i nput
3
Using S-Parameters
H
18
H
Signal Flow Graphs
Scatteri ng parameters turn out to be parti cularly conveni ent
i n many network calculati ons. T hi s i s especi ally true for power
and power gai n calculati ons. T he transfer parameters s
12
and
s
21
are a measure of the complex i nserti on gai n, and the
dri vi ng poi nt parameters s
11
and s
22
are a measure of the i nput
and output mi smatch loss. As di mensi onless expressi ons of
gai n and reflecti on, the s-parameters not only gi ve a clear and
meani ngful physi cal i nterpretati on of the network performance,
but also form a natural set of parameters for use wi th si gnal
flow graphs [ See references here and also i n Appendi x A ] .
O f course, i t i s not necessary to use si gnal flow graphs i n order
to use s-parameters, but flow graphs make s-parameter
calculati ons extremely si mple. T herefore, they are strongly
recommended. Flow graphs wi ll be used i n the examples
that follow.
4
References
J . K. Hunton, Analysis of
Microwave Measurement
Techniques by Means of
Signal Flow Graphs,
IRE Transactions on
Microwave Theory and
Techniques, Vol. MTT-8,
No. 2, March, 1960.
N. Kuhn, Simplified Signal
Flow Graph Analysis,
Microwave J ournal, Vol. 6,
No. 11, November, 1963.
Network Calculations with
Scattering Parameters
19
I n a si gnal flow graph, each port i s
represented by two nodes. Node a
n
represents the wave comi ng i nto the
devi ce from another devi ce at port n,
and node b
n
represents the wave
leavi ng the devi ce at port n. T he
complex scatteri ng coeffi ci ents are
then represented as multi pli ers on
branches connecti ng the nodes wi thi n
the network and i n adj acent
networks. Fi g. 3, ri ght, i s the flow
graph representati on of the system of Fi g. 2.
Fi gure 3 shows that i f the load reflecti on coeffi ci ent
L
i s
zero ( Z
L
= Z
0
) there i s only one path connecti ng b
1
to a
1
( flow graph rules prohi bi t si gnal flow agai nst the forward
di recti on of a branch arrow) . T hi s confi rms the defi ni ti on
of s
11
:

s
b
a
a b
L
11
1
1
0
2 2

s
11
s
22
s
21
s
12
a
1
a
2
b
2
b
1
1
S
Z Z
b
V Z
S
S
=
+
0
0
L
L
L
Z Z
Z Z
=
-
+
0
0
S
S
S
Z Z
Z Z
=
-
+
0
0
Figure 3
Flow graph for
two-port network
appearing in Figure 2.
4
H
20
T he si mpli fi cati on of network analysi s by flow graphs results
from the appli cati on of the non-touchi ng loop rule. T hi s rule
appli es a generali zed formula to determi ne the transfer
functi on between any two nodes wi thi n a complex system. T he
non-touchi ng loop rule i s explai ned below.
The Nontouching Loop Rule
T he nontouchi ng loop rule provi des a
si mple method for wri ti ng the soluti on of
any flow graph by i nspecti on. T he soluti on
T ( the rati o of the output vari able to the
i nput vari able) i s defi ned, where:
T
k
= path gai n of the k
th
forward path
= 1
(
all i ndi vi dual loop gai ns
)
+
(
loop gai n products of all possi ble
combi nati ons of 2 nontouchi ng loops
)

(
loop gai n products of all possi ble
combi nati ons of 3 nontouchi ng loops
)
+
. . .
k
= T he value of not touchi ng the k
th
forward path.
4
Better Smith Charts
On the copyrighted Smith
Chart, curves of constant
standing wave ratio,
constant attenuation, and
constant reflection
coefficient are all circles
coaxial with the center of
the diagram. Refinements
to the original form have
enhanced its usefulness.
In an article published in
1944, for example, Smith
described an improved
version and showed how
to use it with either
impedance or admittance
coordinates. More recent
improvements include
double Smith Charts for
impedance matching and
a scale for calculating
phase distance.
H

T
T
k k
k

21
A path i s a conti nuous successi on of branches, and a forward
path i s a path connecti ng the i nput node to the output node,
where no node i s encountered more than once. Path gai n i s
the product of all the branch multi pli ers along the path. A loop
i s a path that ori gi nates and termi nates on the same node, no
node bei ng encountered more than once. Loop gai n i s the
product of the branch multi pli ers around the loop.
For example, i n Fi gure 3 there i s only one forward path from
b
s
to b
2,
and i ts gai n i s s
21
. T here are two paths from b
s
to b
1
;
thei r path gai ns are s
21
s
12
L
and s
11
respecti vely. T here are
three i ndi vi dual loops, only one combi nati on of two
nontouchi ng loops, and no combi nati ons of three or more
nontouchi ng loops. T herefore, the value of for thi s network
i s
T he transfer functi on from b
s
to b
2
i s therefore
b
b
s
s
2
21
=
4
H
S-parameters & Smith Charts
Invented in the 1960s,
S-parameters are a way
to combine distributed
design and distributed
measurement. s
11
and s
22
,
the two s-parameters
typically represented using
Smith Charts, are similar to
lumped models in many
respects because they are
related to the input
impedance and output
impedance, respectively.
The Smith Chart performs
a highly useful translation
between the distributed
and lumped models and is
used to predict circuit and
system behavior.
( )
s s
s s
11 12
21 22
= + + + 1
11 21 12 22 11 22
( s s s s s s
S L S L L S
)

22
Transducer Power Gain
Usi ng scatteri ng parameter flow-graphs and the non-touchi ng
loop rule, i t i s easy to calculate the transducer power gai n wi th an
arbi trary load and source. I n the followi ng equati ons, the load and
source are descri bed by thei r reflecti on coeffi ci ents
L
and
S
,
respecti vely, referenced to the real characteri sti c i mpedance Z
0
.
T ransducer power gai n:
G
P
P
P P P
b
P
b
G
b
b
T
L
avS
L
L
avS
S
S
T
S
S L
= =
=
=
=
=
Power deli vered to the load
Power avai lable from the source
i nci dent on load reflected from load ( ) ( )
( )
( ) ( )
2
2 2
2
2
2
2
2 2
1
1
( )
1 1

4
H
23
4
Obtaining maximum
performance
S-parameters are used
to characterize RF and
microwave components
that must operate
together, including
amplifiers, transmission
lines, and antennas (and
free space). Because
s-parameters allow the
interactions between
such components to be
simply predicted and
calculated, they make it
possible to maximize
performance in areas
such as power transfer,
directivity, and
frequency response.
Usi ng the non-touchi ng loop rule,
( 18)
Two other parameters of i nterest are:
1) I nput reflecti on coeffi ci ent wi th the output
termi nati on arbi trary and Z
s
= Z
0
.
( 19)

+
s
b
a
s s s s
s
s
s s
s
L L
L
L
L
11
1
1
11 22 21 12
22
11
21 12
22
1
1
1
( )

b
b
s
s s s s s s
s
s s s s
G
s
s s s s
S S L L S S L
S L L S
T
S L
S L L S
2
21
11 22 21 12 11 22
21
11 22 21 12
21
2 2 2
11 22 21 12
2
1
1 1
1 1
1 1

( ) ( )
( ) ( )
( ) ( )
H
24
2) Voltage gai n wi th arbi trary source
and load i mpedances
( 20)
Appendi x B contai ns formulas for calculati ng many often-used
network functi ons ( power gai ns, dri vi ng poi nt characteri sti cs,
etc.) i n terms of scatteri ng parameters. Also i ncluded are
conversi on formulas between s-parameters and h-, y-, and
z-parameters, whi ch are other parameter sets used very often
for speci fyi ng transi stors at lower frequenci es.

A
V
V
V a b Z V V
V a b Z V V
a b
b s a
A
b
a s
s
s s
V i r
i r
L
V
L L
L
+ +
+ +
+
+

+
+
2
1
1 1 1 0 1 1
2 2 2 0 2 2
2 2
1 11 1
2
1 11
21
22 11
1
1
1
1 1
( )
( )
( )
( )
( )
( ) ( )
4
Waveguides
A radar system delivers a
large amount of energy
from a microwave (W)
source to the transmitting
antenna. The high field
strengths cause short
circuits in standard wires,
cabling, and coax, so
waveguides are used.
These hollow metal tube
constructions conduct W
energy much like a
plumbing system. In the
design of waveguides, we
can test for signal reflec-
tions and transmission
quality with s-parameters.
H

Traveling wave amplifier
S-parameters are extensively
used for designing RF/W
circuits such as the HP TC702
distributed traveling wave
amplifier (TWA) enlarged
in the photograph below.
The frequency-dependent
impedances (or dispersion)
in this integrated circuit can
not be modeled by lumped-
equivalent circuit elements,
but s-parameters can
accurately characterize the
amplifier's response.
25
Using Scattering Parameters
T he remai nder of thi s appli cati on note wi ll show wi th several
examples how s-parameters are used i n the desi gn of transi stor
ampli fi ers and osci llators. To keep the di scussi on from becomi ng
bogged down i n extraneous detai ls, the emphasi s i n these
examples wi ll be on s-parameter desi gn methods,
and mathemati cal mani pulati ons wi ll
be omi tted wherever possi ble.
5
Amplifier Design
H
The HP 83017A microwave
system amplifier achieves
0.526.5 GHz bandwidth by
incorporating the HP TC702
GaAs MESFET TWA IC.

26
M ost desi gn problems wi ll begi n wi th a
tentati ve selecti on of a devi ce and the
measurement of i ts s-parameters.
Fi gures 4a 4e, whi ch appear to the
ri ght and on the next two pages, are a
set of osci llograms showi ng complete
s-parameter data btween 100 M Hz and
1.7 GHz for a 2N3478 transi stor i n the
common-emi tter confi gurati on.
T hese graphs are the results of swept-
frequency measurements made wi th the
classi c HP 8410A mi crowave network
analyzer. T hey were ori gi nally publi shed
as part of the 1967 HP J ournal arti cle.
M easurements made wi th a modern
network analyzer are presented at the
end of thi s secti on. Whi le the
measurement tools have changed over
the past 30 years, the basi c
measurement techni ques have not.
6
Measurement of
S-Parameters
Figure 4a
s
11
of a 2N3478 transistor measured with the
classic HP 8410A network analyzer. Outermost
circle on Smith Chart overlay corresponds to
|s
11
| = 1. The movement of s
11
with frequency
is approximately along circles of constant
resistance, indicative of series capacitance and
inductance.
H
1700 MHz
100 MHz
S
11

Figure 4c
Magnitude and phase of s
12
. While the
phase of s
12
is relatively insensitive to the
frequency, the magnitude of s
12
increases
about 6dB/octave.
27
6
Measurement of
S-Parameters
Figure 4b
Displayed on the same scale as Figure 4a,
s
22
moves between the indicated frequencies
roughly along circles of constant conductance,
characteristic of a shunt RC equivalent circuit.
H
1700 MHz
100 MHz
S
22
30
dB
110
1700 MHz 100 MHz
|
S
12
|
10 dB/cm
S
12
50/cm

Figure 4e Removing Linear Phase Shift.
Magnitude and phase of s
21
measured
with a line stretcher adjusted to remove
the linear phase shift above 500 MHz.
28
6
Measurement of
S-Parameters
H
Figure 4d Magnitude and phase of s
21
.
The magnitude of s
21
decays with a slope
of about 6 dB/octave, while the phase
decreases linearly above 500 MHz.
0 dB
+
90
1700 MHz 100 MHz
|
S
21
|
10 dB/cm
S
21
50/cm
0 dB

+
20
1700 MHz 100 MHz
|
S
21
|
10 dB/cm
S
21
50/cm

29
I n Fi g. 4f, the magni tude of s
21
from Fi g. 4d i s
replotted on a logari thmi c frequency scale,
along wi th addi ti onal data on s
21
below
100 M Hz, measured wi th a vector voltmeter.
T he magni tude of s
21
i s essenti ally constant
to 125 M Hz, and then i t rolls off at a slope of
6 dB/octave.
T he phase of s
21
, as seen i n
Fi g. 4d, vari es li nearly wi th
frequency above about 500 M Hz.
By adj usti ng a cali brated li ne
stretcher i n the network
analyzer, a compensati ng li near
phase shi ft was i ntroduced, and
the phase curve of Fi g. 4e
resulted. To go from the phase
curve of Fi g. 4d to that of Fi g.4e
requi red 3.35 cm of li ne, that
i s equi valent to a pure ti me
delay of 112 pi coseconds.
6
Measurement of
S-Parameters
10 MHz 100 MHz 1 GHz 10 GHz
20
10
0
10
20
30
10
3.162
1
.3162
.03162
.1
Frequency
M
a
g
n
i
t
u
d
e

(
d
B
)
M
a
g
n
i
t
u
d
e
u
|s
21
|
H
Figure 4f
Top curve: |s
21
| from Fig. 4 is replotted
on a logarithmic frequency scale. Data
below 100 MHz was measured with an
HP 8405A vector voltmeter. The bottom
curve u, the unilateral figure of merit,
calculated from s-parameters.
30
After removal of the constant-delay, or li near-phase,
component, the phase angle of s
21
for thi s transi stor ( Fi g. 4e)
vari es from 180 at dc to +90 at hi gh frequenci es, passi ng
through +135 at 125 M Hz, the -3 dB poi nt of the magni tude
curve. I n other words, s
21
behaves li ke a si ngle pole i n the
frequency domai n, and i t i s possi ble to wri te a closed
expressi on for i t. T hi s expressi on i s
( 21)
where
T he ti me delay T
0
= 112 ps i s due pri mari ly to the
transi t ti me of mi nori ty carri ers ( electrons) across
the base of thi s npn transi stor.
T ps
f
s
0
0
21
112
2
2 125
112 21
0

MHz
dB .

s
s e
j
j T
21
21
0
0
0
1
6
Measurement of
S-Parameters
H
Importance of simple
approximations
Using first-order
approximations such as
equation 21 is an important
step in circuit design. The
intuitive sense that
designers gain from
developing an under-
standing of these
approximations can
eliminate much frustration.
The acquired insight can
save hours of time that
otherwise might be wasted
generating designs that
cannot possibly be realized
in the lab, while also
decreasing development
costs.
31
6
Measurement of
S-Parameters
Figure 5a
S-parameters of 2N3478 transistor
in common-emitter configuration,
measured by an HP 8753 network
analyzer. This plot shows s
11
and
s
22
on a Smith Chart. The marker set at 47 MHz represents
the -3 dB gain roll off point of s
21
. The frequency index of
this point is referenced in the other plots.
H
S
11
S
22
T he s-parameters of an 2N3478
transi stor shown i n Fi gures 4a
through 4f were measured wi th the
classi c HP 8410A network analyzer.
I n Fi gures 5a through 5e, the
s-parameters of an 2N3478 transi stor
are shown re-measured wi th a
modern HP 8753 network analyzer.
Fi gures 5a through 5e represent the
actual s-parameters of thi s transi stor
between 0.300 M Hz and 1.00 GHz.
32
6
Measurement of
S-Parameters
S
12
Figure 5b
A plot of the magnitude and phase of s
12
. While the
phase of s
12
depends only weakly on the frequency,
the magnitude increases rapidly at low frequencies.
Figure 5c
A polar plot of s
21
. The frequency
marker shown is at the -3 dB point.
Both the phase angle and magnitude
decrease dramatically as the
frequency is increased.
H
S
21
Frequency
P
h
a
s
e

(
d
e
g
)
M
a
g
n
i
t
u
d
e

(
d
B
)
33
6
Measurement of
S-Parameters
H
S
21
Figure 5d
The magnitude of s
21
plotted on a log
scale showing the 6 dB/octave roll-off
above 75 MHz.
Figure 5e
The phase angle, in degrees, of s
21
. A time
delay of 167 ps was de-embedded from the
measured data using the analyzer's electrical-
delay feature to get a response with a single-
pole transfer characteristic. Removing this
time delay allows the phase distortion to be
viewed with much greater resolution.
S
21
Frequency (Hz)
Frequency (GHz)
M
a
g
n
i
t
u
d
e

(
d
B
)
P
h
a
s
e

(
d
e
g
)

Vector Network Analyzers
Most modern design
projects, RF through
lightwave, use
sophisticated simulation
software to model system
performance from
components through
subsystems. These
programs require complete
s-parameter data on each
component. Measurements
are made with a VNA,
Vector Network Analyzer,
an instrument that
accurately measures the
s-parameters, transfer
function, or impedance
characteristic of linear
networks across a broad
range of frequencies.
34
I n Fi g. 5d, the magni tude of s
21
from Fi g. 5c i s replotted
on a logari thmi c frequency scale. T he magni tude of s
21
i s
essenti ally constant to 75 M Hz, and then rolls off at a slope
of 6 dB/octave. T he phase angle of s
21
as seen i n Fi g. 5e
vari es li nearly wi th frequency above 500 M Hz. T o better
characteri ze phase di storti on, a compensati ng li near phase
shi ft was i ntroduced electroni cally i n the network analyzer.
T hi s establi shed an accurate cali brati on for measuri ng the
devi ce, resulti ng i n the second phase curve of Fi gure 5e.
T o go from the fi rst phase curve of Fi g. 5e to the second phase

curve requi red removi ng a pure ti me delay of 167 pi coseconds.
T hi rty years ago thi s operati on was accompli shed by
de-embeddi ng 5.0 cm of li ne usi ng a cali brated li ne stretcher.
T oday i ts performed by software i n the network analyzer.
After removal of the constant-delay, or li near-phase,

component, the phase angle of s
21
for thi s transi stor
( Fi g. 5e) vari es from 180 at dc to +90 at hi gh frequenci es,
passi ng through +135 at 75 M Hz, the -3 dB poi nt of the
magni tude curve. I n other words, s
21
behaves li ke a si ngle
pole i n the frequency domai n.
6
Measurement of
S-Parameters
H

6
35
Si nce s
21
behaves li ke a si ngle pole i n the frequency domai n,
i t i s possi ble to wri te a closed expressi on for i t. T hi s expressi on
i s the same as equati on 21, repeated here.
T he ti me delay T
0
= 167 ps i s due pri mari ly to the transi t ti me
of mi nori ty carri ers ( electrons) across the base of thi s npn
transi stor. Removi ng thi s ti me delay usi ng the electri cal-delay
feature of the vector network analyzer allows the phase
di storti on to be vi ewed wi th much greater resoluti on.

Usi ng the fi rst-order, si ngle-pole approxi mati on for s
21
i s an
i mportant step i n ci rcui t desi gn. T oday, however, we have desi gn
technology undreamed of i n 1967. Subsequently, through the
process of electroni c desi gn automati on ( EDA) , computer-ai ded
engi neeri ng ( CAE) tools now can be used i terati vely to si mulate
and refi ne the desi gn. T hese tools combi ne accurate models
wi th performance-opti mi zati on and yi eld-analysi s capabi li ti es.

where
T ps
f
s
0
0
21
0
167
2
2 75
84 185
=
=
=
= =

MHz
dB . .
s
s e
j
j T
21
21
0
0
0
1
=
Measurement of
S-Parameters
Complete network
characterization
Vector Network Analyzers
(VNA) are ideal for applica-
tions requiring complete
network characterization.
They use narrow-band
detection to achieve wide
dynamic range and provide
noise-free data. VNAs are
often combined with powerful
computer-based electronic
design automation (EDA)
systems to both measure
data, and simulate and
optimize the performance
of the complete system
implementation being
developed.
H
36
H
6
Measurement of
S-Parameters
Explanation of Measurement Discrepancies
You may have noti ced a di fference between the measured and
caclulated data from the 1967 HP J ournal arti cle and the data
obtai ned for thi s updated appli cati on note. Both sets of data are
fundamentally correct. T wo maj or sources account for these
di fferences:
Measurement techniques Early network analyzers di d not
have onboard computers, an HP-I B standard, or hi gh-resoluti on
graphi cs to perform cali brati on, extract preci si on numeri cal
data, or di splay electroni c markers. Cali brati on techni ques
used i n 1967 were procedurally and mathemati cally si mpler
than those used today. M odern network anaylzers contai n
sophi sti cated automated techni ques that enhance measurement
processi ng capabi li ti es and reduce operator errors.
Device differences Semi conductor manufacturi ng processes
evolve over ti me. Devi ce engi neers attempt to produce i denti cal
transi stors wi th di fferent processes. Nevertheless, successi ve
generati ons of parts li ke the 2N3478 can exhi bi t uni ntenti onal,
and someti mes unavoi dable, performance di fferences, especi ally
i n characteri sti cs not guaranteed on the datasheet.
37
Suppose now that thi s 2N3478 transi stor i s to be
used i n a si mple ampli fi er, operati ng between a
50 source and a 50 load, and opti mi zed
for power gai n at 300 M Hz by means of
lossless i nput and output matchi ng
networks. Si nce reverse gai n s
12
for thi s
transi stor i s qui te small 50 dB smaller
than forward gai n s
21
, accordi ng to
Fi g. 4 there i s a possi bi li ty that i t can be
neglected. I f thi s i s so, the desi gn problem wi ll
be much si mpler, because setti ng s
12
equal to zero wi ll make
the desi gn equati ons much less compli cated.
I n determi ni ng how much error wi ll be i ntroduced by assumi ng
s
12
= 0, the fi rst step i s to calculate the uni lateral fi gure of
meri t u, usi ng the formula gi ven i n Appendi x B. T hat i s,
( 22)

u
s s s s
s s

11 12 21 22
11
2
22
2
1 1 ( ) ( )
7
Narrow-Band
Amplifier Design
H
Unilateral figure of merit
All two-port models are
bilateral, so both the
forward and reverse signal
flow must be considered.
If the signal flow in the
reverse direction is much
smaller than the flow in
the forward direction,
it is possible to make the
simplification that the
reverse flow is zero.
The unilateral figure of
merit is a quick calculation
that can be used to
determine where this
simplification can be
made without significantly
affecting the accuracy of
the model.
38
A plot of u as a functi on of frequency, calculated from
the measured parameters, appears i n Fi g. 4f. Now i f G
T u
i s the transducer power gai n wi th s
12
= 0 and G
T
i s the
actual transducer power gai n, the maxi mum error
i ntroduced by usi ng G
T u
i nstead of G
T
i s gi ven by the
followi ng relati onshi p:
( 23)
From Fi g. 4f, the maxi mum value of u i s about 0.03, so the
maxi mum error i n thi s case turns out to be about +0.25 dB
at 100 M Hz. T hi s i s small enough to j usti fy the assumpti on
that s
12
= 0.
I nci dentally, a small reverse gai n, or feedback factor, s
12
, i s an
i mportant and desi rable property for a transi stor to have, for
reasons other than i t si mpli fi es ampli fi er desi gn. A small feedback
factor means that the i nput characteri sti cs of the completed
ampli fi er wi ll be i ndependent of the load, and the output wi ll be
i ndependent of the source i mpedance. I n most ampli fi ers,
i solati on of source and load i s an i mportant consi derati on.

1
1
1
1
2 2
( ) ( ) +
< <
u
G
G
u
T
T u
7
Narrow-Band
Amplifier Design
High-frequency transistors
Discrete transistors were
the mainstay of high-
frequency system design in
1967 when Dick Anderson
wrote the article on which
this application note is
based. Thirty years later,
discrete devices are still
available, manufactured
and selected for specific,
often exceptional,
performance characteristics.
Discrete transistors remain
the best choice for many
applications, such as
sensitive first-stage
amplifiers in satellite
TV receivers.
H
39
Returni ng now to the 300-M Hz ampli fi er desi gn, the uni lateral
expressi on for transducer power gai n, obtai ned ei ther by
setti ng s
12
= 0 i n equati on 18 or looki ng i n Appendi x B, i s
( 24)
When
|
s
11
|
and
|
s
22
|
are both less than one, as they are i n
thi s case, maxi mum G
T u
occurs for
S
= s*
11
and
L
= s*
22
( Appendi x B) .
T he next step i n the desi gn i s to synthesi ze matchi ng networks
that wi ll transform the 50 load and source i mpedances to
the i mpedances correspondi ng to reflecti on coeffi ci ents of s*
11
and s*
22
, respecti vely. Si nce thi s i s to be a si ngle-frequency
ampli fi er, the matchi ng networks need not be compli cated.
Si mple seri es-capaci tor, shunt-i nductor networks wi ll not only
do the j ob, but wi ll also provi de a handy means of bi asi ng the
transi stor vi a the i nductor and of i solati ng the dc bi as from
the load and the source.

G
s
s s
T u
S L
S L

21
2 2 2
11
2
22
2
1 1
1 1
( ) ( )

7
Narrow-Band
Amplifier Design
H
Satellite Broadcast Signals
Satellites provide broad
geographical signal
coverage over a wide band
of frequencies by using
high power vacuum tubes,
called Traveling Wave
Tubes (TWTs), which are
best characterized by
s-parameters.
40
Values of L and C to be used i n the matchi ng networks for the
300-M Hz ampli fi er are determi ned usi ng the Smi th Chart of
Fi g. 6, whi ch i s shown on the next page. Fi rst, poi nts
correspondi ng to s
11
, s*
11
, s
22
, and s*
22
at 300 M Hz are
plotted. Each poi nt represents the ti p of a vector leadi ng away
from the center of the chart, i ts length equal to the magni tude
of the reflecti on coeffi ci ent bei ng plotted, and i ts angle equal
to the phase of the coeffi ci ent.
Next, a combi nati on of constant-resi stance and constant-
conductance ci rcles i s found, leadi ng from the center of the
chart, representi ng 50 , to s*
11
and s*
22
. T he ci rcles on the
Smi th Chart are constant-resi stance ci rcles; i ncreasi ng seri es
capaci ti ve reactance moves an i mpedance poi nt counter-
clockwi se along these ci rcles.
7
Narrow-Band
Amplifier Design
Matching networks
Matching networks are
extra circuit elements
added to a device or
circuit to cancel out
or compensate for
undesired characteristics
or performance variations
at specified frequencies.
To eliminate reflections in
an amplifier, one matching
network is carefully
designed to transform
the 50 load impedance
to s*
11
. Another matching
network transforms the
50 source impedance to
s*
22
.
H
You will find an interactive Impedance Matching Model at the
HP Test & Measurement website listed below. Challenge
yourself to impedance matching games based on principles
and examples discussed in this application note!
Click on the URL below or type the address in your browser.
http://www.hp.com/go/tminteractive
41
7
s
11
*
s
22
*
s
22
s
11
L
1

1
C
L
2
2
C
Narrow-Band
Amplifier Design
Figure 6
Smith Chart for 300-MHz
amplifier design example.
I n thi s case, the ci rcle to be used for
fi ndi ng seri es C i s the one passi ng
through the center of the chart, as
shown by the soli d li ne i n Fi g. 6.
I ncreasi ng shunt i nducti ve
susceptance moves i mpedance
poi nts clockwi se along constant-
conductance ci rcles. T hese
ci rcles are li ke the constant-
resi stance ci rcles, but they
are on another Smi th Chart,
one that i s j ust the reverse
of the chart shown i n Fi g. 6.
T he constant-conductance
ci rcles for shunt L all pass
through the leftmost poi nt of
the chart rather than the ri ghtmost
poi nt. T he ci rcles to be used are
those passi ng through s*
11
and s*
22
,
as shown by the dashed li nes i n Fi g. 6.
H
42
O nce these ci rcles have been located,
the normali zed values of L and C
needed for the matchi ng networks are
calculated from readi ngs taken from the
reactance and susceptance scales of the
Smi th Charts.
Each elements reactance or
susceptance i s the di fference between
the scale readi ngs at the two end poi nts
of a ci rcular arc. Whi ch arc corresponds
to whi ch element i s i ndi cated i n Fi g. 6.
T he fi nal network and the element
values, normali zed and unnormali zed,
are shown i n Fi g. 7.
7
Figure 7
A 300-MHz amplifier with
matching networks for
maximum power gain.
Narrow-Band
Amplifier Design
H
50 50 L
1
L
2
C
1
C
2
2N3478
X
L
C
L
2
50
0.32
156
2
2
= =
C
1
L 26nH
1
x
nH
156
2 03 10
83
9
= =
( . )
x
pF
1
2 03 10 35 50
3
9
= =
( . )( . )( )
pF
1
25 = =
x 2 03 10
9
( . )(0.42)(50)
50
= =
x 2 03 10
9
( . )(1.01)
Calculations:

T he ani mati on to the ri ght demonstrates how to use a
Smi th Chart to desi gn a matchi ng network between a transi stor
output and a resi sti ve load. As previ ously descri bed, to
maxi mi ze the power deli vered to the load, the s*
22
parameter of
the transi stor must be matched to the load i mpedance, 50 i n
thi s 300-M Hz ampli fi er example. T hi s matchi ng i s achi eved
usi ng the LC ci rcui t shown at the ri ght.
Starti ng from the 50 load, the seri es capaci tance i s vari ed to
move the i mpedance poi nt along the ci rcle of constant 50
resi stance on the Smi th Chart. T he capaci tance i s adj usted unti l
i t i ntersects the constant conductance ci rcle on whi ch s*
22
i s
si tti ng. Varyi ng the shunt i nductance then moves the i mpedance
poi nt along thi s constant
conductance ci rcle as
i ndi cated by the admi ttance
Smi th Chart. To reach s*
22
,
the shunt i nductance i s
adj usted unti l the i mpedance
poi nt reaches s*
22
.
43
7
Narrow-Band
Amplifier Design
50 L
C
2N3478
Animation 2
Impedance matching
using the Smith Chart
for the matching
network shown at the
left. Click over the chart
to start animation.
http://www.hp.com/go/tminteractive
H
C = 1000 pF
L = 10000 nH
S
22
*
44
Desi gni ng a broadband ampli fi er, that i s, one
whi ch has nearly constant gai n over a
prescri bed frequency range, i s a matter of
surroundi ng a transi stor wi th external
elements i n order to compensate for the
vari ati on of forward gai n,
|
s
21
|
wi th
frequency.
T hi s can be done i n ei ther of two ways
fi rst, negati ve feedback, or second, selecti ve
mi smatchi ng of the i nput and output
ci rcui try. We wi ll use the second method.
When feedback i s used, i t i s usually
conveni ent to convert to y- or z-parameters
( for shunt or seri es feedback, respecti vely)
usi ng the conversi on equati ons gi ven i n
Appendi x B and a di gi tal computer.
Equati on 24 for the uni lateral transducer
power gai n can be factored i nto three parts,
as shown to the ri ght:
8
Broadband
Amplifier Design

G
Tu
G
0
G
1
G
2
G
0
s
21
2
G
1
1
s
2
1 s
11
s
2
G
2
1
L
2
1 s
22 L
2
H
45
When a broadband ampli fi er i s desi gned by selecti ve
mi smatchi ng, the gai n contri buti ons of G
1
and G
2
are vari ed to
compensate for the vari ati ons of G
0
= |s
21
|
2
wi th frequency.
Suppose that the 2N3478 transi stor whose s-parameters are
gi ven i n Fi g. 4 i s to be used i n a broadband ampli fi er that wi ll
operate from 300 MHz to 700 MHz. T he ampli fi er i s to be
dri ven from a 50 source and i s to dri ve a 50 load.
Accordi ng to Fi gure 4f,
To reali ze an ampli fi er wi th a constant gai n of 10 dB, source
and load matchi ng networks must be found that wi ll decrease
the gai n by 3 dB at 300 MHz, leave the gai n the same at
450 MHz, and i ncrease the gai n by 4 dB at 700 MHz.

s
21
2
13 dB at 300 MHz
10 dB at 450 MHz
6 dB at 700 MHz
8
Broadband
Amplifier Design
H
46
Although i n the general case both a source
matchi ng network and a load matchi ng
network would be desi gned, G
1max
( i .e.,
G
1
for
s
= s*
11
) for thi s transi stor i s
less than 1 dB over the frequenci es of
i nterest, whi ch means there i s li ttle to
be gai ned by matchi ng the source.
Consequently, for thi s example, only
a load-matchi ng network wi ll be
desi gned. Procedures for desi gni ng
source-matchi ng networks are
i denti cal to those used for desi gni ng
load-matchi ng networks.
T he fi rst step i n the desi gn of the
load-matchi ng network i s to plot s*
22
over the requi red frequency range on the
Smi th Chart, Fi g. 8a.
8
Broadband
Amplifier Design
s
22
*
700
450
300
H
Figure 8a
A plot of s*
22
over the frequency
range from 300 MHz to 700 MHz.
47
8
Broadband
Amplifier Design
H
Next, a set of constant-gai n ci rcles i s drawn. As shown i n Fi g. 8b,
each ci rcle i s drawn for a si ngle frequency; i ts center i s on a li ne
between the center of the Smi th Chart and the poi nt
representi ng s*
22
at that frequency. T he di stance from
the center of the Smi th Chart to the center of the
constant gai n ci rcle i s gi ven by the followi ng
equati ons, whi ch also appear i n Appendi x B:
where
T he radi us of the constant-gai n ci rcle i s:

2
2 22
2
22
2
2
1 1
1 1

g s
s g
( )
( )

g
G
G
G s
2
2
2
2 22
2
1
max
( )

r
g s
s g
2
2 22
22
2
2
1
1 1

( )
2
2
r
2
2
r
2
r
G dB
at MHz
2
4
700
=+
G dB
at MHz
2
0
450
=
G dB
at MHz
2
3
300
=
Figure 8
Constant-gain circles.
48
For thi s example, three ci rcles wi ll be drawn, one for G
2
= -3 dB
at 300 M Hz, one for G
2
= 0 dB at 450 M Hz, and one for G
2
=
+4 dB at 700 M Hz. Si nce |s
22
| for thi s transi stor i s constant at
0.85 over the frequency range [ see Fi gure 4( b) ] , G
2max
for all
three ci rcles i s ( 0.278)
-1
, or 5.6 dB. T he three constant-gai n
ci rcles are i ndi cated i n Fi g. 8b.
T he requi red matchi ng network must transform the center of the
Smi th Chart, representi ng 50 , to some poi nt on the -3 dB
ci rcle at 300 M Hz, to some poi nt on the 0 dB ci rcle at 450 M Hz,
and to some poi nt on the +4 dB ci rcle at 700 M Hz. T here are
undoubtedly many networks that
wi ll do thi s. O ne sati sfactory
soluti on i s a combi nati on of two
i nductors, one i n shunt and one i n
seri es, as shown i n Fi g. 9.
Z
L
= 50
50
L
series
L
shunt
2N3478
8
Broadband
Amplifier Design
Figure 9
A combination of shunt
and series inductances
is a suitable matching
network for the
broadband amplifier.
H
49
8
Broadband
Amplifier Design
H
Shunt and seri es elements move i mpedance poi nts
on the Smi th Chart along constant-conductance
and constant-resi stance ci rcles, as explai ned i n
the narrow-band desi gn example. As shown i n
Fi g. 10a, the shunt i nductance transforms the
50 load alng a ci rcle of constant conductance
and varyi ng ( wi th frequency) i nducti ve
susceptance. T he seri es i nductor transforms
the combi nati on of the 50 load and the
shunt i nductance along ci rcles of constant
resi stance and varyi ng i nducti ve reactance.
O pti mi zi ng the values of shunt and seri es L
i s an i terati ve process wi th two goals:
the transformed load reflecti on
termi nates on the ri ght gai n ci rcle at
each frequency, and
the susceptance component decreases
wi th frequency and the reactance component
i ncreases wi th frequency. ( T hi s rule appli es to
i nductors; capaci tors would behave i n the
opposi te way.)
3
0
0
M
H
z
4
5
0
M
H
z
7
0
0
M
H
z
Constant
conductance
circle-Transformation
due to L
shunt
Constant resistance
circles-Transformation
due to L
series
L
locus
Figure 10a
Matching paths
for shunt and
series L.
50
O nce appropri ate constant-
conductance and constant-
resi stance ci rcles have been found,
the reactances and susceptances
of the elements can be read
di rectly from the Smi th Chart.
T hen the element values are
calculated, the same as they were
for the narrow-band desi gn.
Fi gure 10b i s a schemati c di agram
of the completed broadband
ampli fi er, wi th unnormali zed
element values.
Figure 10b
Broadband ampli fi er wi th constant gai n
of 10 dB from 300 M Hz to 700 M Hz.
50
Z
S
= 50
20.4 nH
36.4 nH
2N3478
Inductance Calculations:
j L
Z
j j
Z
j L
j
series
shunt
0
0
364 044 32
13
= - =
=-
( . . ) .
L nH nH
series

32 50
2 07
364 = =
( . )( )
( . )
.
.
L nH
shunt

50
13 2 03
204 = =
( . )( )( . )
.
From 700 MHz data,
From 300 MHz data,
8
Broadband
Amplifier Design
H
51
Design of Reflection Amplifiers
and Oscillators
When the real part of the i nput i mpedance of a
network i s negati ve, the correspondi ng i nput
reflecti on coeffi ci ent ( Equati on 17) i s greater
than one, and the network can be used as the
basi s for two i mportant types of ci rcui ts,
reflecti on ampli fi ers and osci llators. A reflecti on
ampli fi er ( Fi g. 11) can be reali zed wi th a
ci rculator a nonreci procal three-port devi ce
and a negati ve-resi stance devi ce.
T he ci rculator i s used to separate the i nci dent
( i nput) wave from the larger wave reflected by
the negati ve-resi stance devi ce. T heoreti cally, i f
the ci rculator i s perfect and has a posi ti ve real
characteri sti c i mpedance Z
0
, an ampli fi er wi th
i nfi ni te gai n can be bui lt by selecti ng a negati ve-
resi stance devi ce whose i nput i mpedance has a
real part equal to -Z
0
and an i magi nary part
equal to zero ( the i magi nary part can be set
equal to zero by tuni ng i f necessary) .
Figure 11
Reflection amplifier
consists of circulator and
transistor with negative
input resistance.
9
Stability Considerations
Circulator
Input Output
Two port with
s'
11
> 1
(Real part of input
impedance is
negative)
H

H
52
Ampli fi ers, of course, are not supposed to osci llate, whether they
Fi gure 3, the rati o
of the reflected voltage wave b
1
to the i nput voltage wave b
s
i s
where s
11
i s the i nput reflecti on coeffi ci ent wi th
s
= 0 ( Z
2
= Z
0
)
and an arbi trary load i mpedance Z
L
, as defi ned i n Equati on 19.
I f at some frequency
( 25)
the ci rcui t i s unstable and i t wi ll osci llate at that frequency.
On the other hand, i f
the devi ce i s uncondi ti onally stable and wi ll not osci llate, whatever
the phase angle of
s
mi ght be.

< s
s
11
1
s
s =
11
1

b
b
s
s
s s
1
11
11
1
=

9
Computer Aided
Engineering tools (CAE)
CAE software tools are
used in the design process
to simulate actual device
and circuit behavior so
designs can be evaluated
before they're built. The
CAE approach is faster,
produces accurate results,
and is easier to follow
than manual methods
using graphical design aids.
CAE tools are part of the
total engineering solution.
are reflecti on ampli fi ers or some other ki nd. T here i s a conveni ent
cri teri on based upon scatteri ng parameters for determi ni ng whether
a devi ce i s stable or potenti ally unstable wi th gi ven source and load
i mpedances. Referri ng agai n to the flow graph of
53
To see how these pri nci ples of
stabi li ty are appli ed i n desi gn
problems, consi der the
transi stor osci llator desi gn
i llustrated i n Fi g. 12. I n thi s
case the i nput reflecti on
coeffi ci ent s
11
i s the reflecti on
coeffi ci ent looki ng i nto the
collector ci rcui t, and the
source reflecti on coeffi ci ent
s
i s one of the two tank-ci rcui t
refecti on coeffi ci ents,
T 1
or
T 2
. From equati on 19,

+
s s
s s
s
L
L
11 11
12 21
22
1
9
150

T
1
0
, Q
T
2
200 Z
L
,
L
s'
11
15
0
, Q
Parallel
Tank
Circuit
Series
Tank
Circuit
Gain
Element
H
Figure 12 The transistor
oscillator is designed by
choosing a tank circuit
such that

s
s
11
1

54
T o make the transi stor osci llate, s
11
and
s
must be adj usted so
that they sati sfy equati on 25. T here are four steps i n the desi gn
procedure:
Measure the four scatteri ng parameters of the transi stor as
functi ons of frequency.
Choose a load reflecti on coeffi ci ent
L
that makes s
11
greater
than uni ty. I n general, i t may also take an external feedback
element that i ncreases s
12
s
21
to make s
11
greater than one.
Plot 1/s
11
on a Smi th Chart. ( I f the network analyzer i s
bei ng used to measured the s-parameters of the transi stor,
1/s
11
can be measured di rectly by reversi ng the reference and
test channel connecti ons between the reflecti on test uni t and
the harmoni c frequency converter. T he polar di splay wi th a
Smi th Chart overlay then gi ves the desi red plot i mmedi ately.)
Connect ei ther the seri es or the parallel tank ci rcui t to the
collector ci rcui t and tune i t so that
T 1
or
T 2
i s large enough
to sati sfy equati on 25. ( T he tank ci rcui t reflecti on coeffi ci ent
plays the role of
s
i n thi s equati on.)
9
H
55
Fi gure 13 shows a Smi th Chart plot of
1/s
11
for a hi gh frequency transi stor
i n the common-base confi gurati on.
Load i mpedance Z
L
i s 200 ,
whi ch means that
L
referred to
50 i s 0.6. Reflecti on
coeffi ci ents
T 1
and
T 2
are
also plotted as functi ons of the
resonant frequenci es of the
two tank ci rcui ts. O sci llati ons
occur when the locus of
T 1
or
T 2
passes through the shaded
regi on. T hus, thi s transi stor
would osci llate from 1.5 to
2.5 GHz wi th a seri es tuned ci rcui t,
and from 2.0 to 2.7 GHz wi th a
parallel tuned ci rcui t.
Di ck Anderson, 1967 and 1997
9
s'
11
0 0
2.5 GHz
T
1
T
2
2.7 GHz
2.0 GHz
1.5 GHz
Figure 13 The transistor will oscillate in the shaded area between 1.5 and 2.5 GHz
with a series-tuned circuit and between 2.0 and 2.7 GHz with a parallel-tuned circuit.
H
56
H
I n addi ti on to previ ous references li sted earli er and repeated
agai n here, the followi ng papers and books were li sted i n the
1967 HP J ournal arti cle as sources for i nformati on on
s-parameter desi gn procedures and flow graphs. Current
references are also menti oned.
J. K . Hunton, Analysi s of M i crowave M easurement
Techni ques by M eans of Si gnal Flow Graphs,
I RE T ransacti ons on M i crowave T heory and Techni ques,
Vol. M T T- 8, No. 2, M arch, 1960.
D.C Youla, O n Scatteri ng M atri ces Normali zed to Complex
Port Numbers, Proc. I RE, Vol. 49, No. 7, July, 1961.
J.G. Li nvi ll and J.F. Gi bbons, T ransi stors and Acti ve
Ci rcui ts, M cGraw-Hi ll, 1961. ( No s-parameters, but good
treatment of Smi th Chart desi gn methods.)
N. K uhn, Si mpli fi ed Si gnal Flow Graph Analysi s,
M i crowave Journal, Vol. 6, No, 11, November, 1963.
K . K urokawa, Power Waves and the Scatteri ng M atri x,
I EEE T ransacti ons on M i crowave T heory and Techni ques,
Vol. M T T-13, No. 2, M arch, 1965.
A
Additional Reading
on S-Parameters

57
H
F. Wei nert, Scatteri ng Parameters Speed Desi gn of
Hi gh-Frequency T ransi stor Ci rcui ts,
Electroni cs, Vol. 39, No. 18, Sept. 5, 1966.
G. Fredri cks, How to Use S-Parameters for T ransi stor
Ci rcui t Desi gn, EEE Vol. 14, No. 12, Dec., 1966.
Among many modern reference sources on the subj ect, the
followi ng book, fi rst publi shed i n 1969, i s defi ni tely a classi c:
Smi th, Phi lli p H., Electroni c Appli cati ons of the Smi th
Chart i n Wavegui de, Ci rcui t and Component Analysi s,
Noble Publi shi ng Classi c Seri es, T ucker, Georgi a, 1995,
I SBN-1-884932-39-8, 237 pp.
We also menti on a useful textbook contai ni ng 5 chapters,
2 appendi ces, and problem sets. T hi s text presents a uni fi ed
treatment of the analysi s and desi gn of mi crowave transi stor
ampli fi ers usi ng scatteri ng parameters techni ques:
G. Gonzalez, M i crowave T ransi stor Ampli fi ers:
Analysi s and Desi gn, Prenti ce Hall, 1984,
I SBN 0-13-581646-7, 240 pp.
A
Additional Reading
on S-Parameters
Keep up to date
This book by P. H. Smith
referenced here must be
considered the ultimate
source on Smith Charts.
Many excellent articles on
the use of Smith charts
have appeared in trade
publications such as the
Microwave J ournal, and
educational Smith Chart
software is sold over the
internet.

A
V
V
s
s s
V
L
L

+
+
2
1
21
22 11
1
1 1
( )
( )( )
s s
s s
s
S
S
22 22
12 21
11
1

s s
s s
s
L
L
11 11
12 21
22
1
58
Input reflection coefficient with
arbitrary Z
L
Output reflection coefficient with
arbitrary Z
s
Voltage gain with
arbitrary Z
L
and Z
S
Scattering Parameter
Relationships

b s a s a
b s a s a
1 11 1 12 2
2 21 1 22 2
+
+
B
a
1
V
2
V
1
a
2
b
1
b
2
TWO - PORT
NETWORK
+ +

H

59
B
Relationships

Power Gain
G
Power delivered to load
Power input to network
= =
s
s s D
L 21
2 2
11
2 2
22
2 2
1
1
L
2Re( N)
S
2Re( M)

Available Power Gain
G
A
Power available fromnetwork
Power available fromsource
= =
s
s s D
S
S
21
2 2
22
2 2
11
2 2
1
1

Transducer Power Gain
G
T
Power delivered to load
Power available fromsource
= =
( )
( )
s
s s s s
S L
S L L S
21
2 2
2
11 22 12 21
2
1 1
1 1

)
H
60
B
Relationships

Unilateral Transducer Power Gain
MaximumUnilateral Transducer Power Gain when s
and s Maximumobtained for and
11
22
s
G
s
s s
G G G
s s
G
s
s s
Tu
S L
S L
S L
u
12
21
2 2 2
11
2
22
2
0 1 2
11 22
21
2
11
2
22
2
0
1 1
1 1
1
1
1 1
( )
_
,
_
,
<
<
_
,
_
,

.
* *
G G G
0 1 2 max max

G s
G
s
G
s
G
s
i
S
S
L
L
i
ii
0 21
2
1
2
11
2
2
2
22
2
2
1
1
1
1
1
1
1 2
max
,
H
61
Constant Gain Circles (Unilateral case: s
12
=0)
B
Relationships
i
i
r
S
ii
*
Constant Gain Circle
center of constant gain circle is on line between center
of Smith Chart and point representing s*
ii
distance of center of circle from center
of Smith Chart:
radius of circle:
where i = 1, 2, and

g
G
G
G s
i
i
i
i ii

max
( ) 1
2

i
i ii
ii i
g s
s g

1 1
1 1
2
2
( )
( )

r
g s
s g
i
i ii
ii i
1 1
2
( )
H
62
B
Relationships

Unilateral Figure of Merit
Error Limits on Unilateral Gain Calculations
u
s s s s
s s
u
G
G
u
T
Tu

+
< <
11 22 12 21
11
2
22
2
2 2
1 1
1
1
1
1
( )( )
( ) ( )
H
Unilateral figure of merit
All two-port models are
bilateral, so both the
forward and reverse signal
flow must be considered.
If the signal flow in the
reverse direction is much
smaller than the flow in
the forward direction, its
possible to make the
simplification that the
reverse flow is zero.
The unilateral figure of
merit is a quick calculation
that can be used to
determine where this
simplification can be
made without significantly
affecting the accuracy of the
model.

63
B
Relationships

a s s
b
s s M
s D
c
s s N
s D
. ,
.
.
*
*
11 22
12 21
11
2 2
12 21
22
2 2
1 1
1
1
< <
>
>

Conditions for Absolute Stability:
No passive source or load will cause a network to
oscillate, if conditions a, b, and c are all satisfied.
Condition that a two-port network can be simultaneously
matched with a real source and load:
> or < where = Linvill C Factor and K C C
C K
K
D s s
s s
1 1
1
2
1
2
11
2
22
2
12 21
=
=
+

D s s s s
M s Ds
N s Ds
=
=
=
11 22 12 21
11 22
22 11
*
*
H
64

B s s D
B s s D
1 11
2
22
2 2
2 22
2
11
2 2
1
1
+
+
B
Relationships

(Use minus sign
when is positive,
plus sign when
is negative.)
B
B
1
1
H
Source and load for Simultaneous Match
MaximumAvailable Power Gain
If ,
where
mS
mL
A
M
B B M
M
N
B B N
N
K G
s
s
K K
K C
1
]
1
1
1
1
]
1
1
1
> t
_
,

*
*
max
1 1
2
2
2
2 2
2
2
2
21
12
2
1
4
2
4
2
1 1
65
s-parameters
in terms of z-parameters
z-parameters
in terms of s-parameters
B
Relationships

z
s s s s
s s s s
z
s
s s s s
z
s
s s s s
z
s s s s
s s s s
11
11 22 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
22 11 12 21
11 22 12 21
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+ +

+ +

( )( )
( )( )
( )( )
( )( )
( )( )
( )( )
s
z z z z
z z z z
s
z
z z z z
s
z
z z z z
s
z z z z
z z z z
11
11 22 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
11 22 12 21
11 22 12 21
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+
+ +
+ +
+ +
+
+ +
( )( )
( )( )
( )( )
( )( )
( )( )
( )( )
H
66
s-parameters
in terms of y-parameters
y-parameters
B
Relationships

s
y y y y
y y y y
s
y
y y y y
s
y
y y y y
s
y y y y
y y y y
11
11 22 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
11 22 12 21
11 22 12
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+ +
+ +

+ +

+ +
+ +
+ +
( )( )
( )( )
( )( )
( )( )
( )( )
( )( )
21 21

y
s s s s
s s s s
y
s
s s s s
y
s
s s s s
y
s s s s
s s s s
11
22 11 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
11 22 12 21
11 22 12
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+ +
+ +

+ +

+ +
+ +
+ +
( )( )
( )( )
( )( )
( )( )
( )( )
( )( )
21 21
H
67
s-parameters
in terms of h-parameters
h-parameters
B
Relationships

s
h h h h
h h h h
s
h
h h h h
s
h
h h h h
s
h h h h
h h h h
11
11 22 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
11 22 12 21
11 22 12 21
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+
+ +
+ +

+ +
+ +
+ +
( )( )
( )( )
( )( )
( )( )
( )( )
( )( )

h
s s s s
s s s s
h
s
s s s s
h
s
s s s s
h
s s s s
s s s s
11
11 22 12 21
11 22 12 21
12
12
11 22 12 21
21
21
11 22 12 21
22
22 11 12 21
11 22 12 21
1 1
1 1
2
1 1
2
1 1
1 1
1 1
+ +
+ +
+ +

+ +

+ +
( )( )
( )( )
( )( )
( )( )
( )( )
( )( )
H
68
The h, y, and zparameters listed in
previous tables are all normalized to Z
0
.
If h, y, z are the actual parameters, then:
B
Relationships

z z Z
z z Z
z z Z
z z Z
11 11 0
12 12 0
21 21 0
22 22 0

y y Z
y y Z
y y Z
y y Z
11 11 0
12 12 0
21 21 0
22 22 0
/
/
/
/

h h Z
h h
h h
h h Z
11 11 0
12 12
21 21
22 22 0
/
Parameter Normalization
The various scattering
parameters are all
normalized by the
reference impedance, Z
0
.
This impedance is usually
the characteristic
impedance of the
transmission line in which
the network of interest is
embedded. Normalizing
the scattering parameters
makes the Smith Chart
readily applicable to
transmission lines of any
impedance. In addition,
impedance and
admittance values can be
plotted on the same
chart.
H
69
Electronic Design Automation (EDA)
The Software Revolution
I n the 30 years that have elapsed si nce the
publi cati on of Di ck Anderson' s arti cle,
computer ai ded engi neeri ng ( CAE) tools have
been developed for the hi gh-frequency desi gn
methods that were tradi ti onally i mplemented
usi ng penci l and paper. T hese computer
software programs run on UNI X workstati ons
and PCs, and do much more than merely
assi st i n computati on-i ntensi ve desi gn tasks.
M odern CAE tools for hi gh-frequency
desi gn eli mi nate the need for si mpli fyi ng
assumpti ons ( such as, s
12
= 0) and can
accurately si mulate actual devi ce, ci rcui t, or
system behavi or. T hey enable broadband
soluti ons, offer opti mi zati on and yi eld-analysi s capabi li ti es, and
provi de answers to What i f? questi ons. CAE tools also speed the
analysi s of a wi de range of RF and mi crowave devi ces, ci rcui ts,
and systems, for a shorter ti me to market, whi le loweri ng costs.
C
H
CAE tools for High-
Frequency Design

C
H
CAE tools for High-
Frequency Design
70
T he CAE tools of i nterest to RF and
mi crowave desi gners i nclude those
summari zed below:
Small-signal (s-parameter) simulation
Small-si gnal analysi s CAE tools si mulate
response over a range of frequenci es, so actual i mplementati ons
perform more closely to desi gn parameters. M atchi ng ci rcui ts are
easi ly determi ned, and can be readi ly opti mi zed, savi ng ti me.
A yi eld-analysi s feature allows the selecti on of components i n
matchi ng networks for the best producti on yi eld, savi ng costs.
Large-signal simulation T hi s powerful analysi s tool
i ncludes the harmoni c balance i mplementati on, useful for
osci llator desi gn and many other problems.
Circuit Envelope simulation Effi ci ently analyzes ci rcui ts
and feedback loops i n the presence of modulated or transi ent
hi gh-frequency si gnals.
Circuit Envelope Simulation
The waveform above
typifies the modulated
and transient signals that
can be efficiently analyzed
using circuit envelope
simulator software. Circuit
envelope simulation is
orders of magnitude faster
than traditional SPICE
simulation software if the
envelope bandwidth of the
RF carrier frequency is
much smaller than the
carrier frequency itself.
This is the case in many
communications, and radar
circuits and subsystems.

71
C
CAE tools for High-
Frequency Design
H
Time-domain analysis A CAE tool that i s especi ally
useful for si mulati ng the response of di gi tal systems at
hi gh clock rates.
Planar electromagnetic analysis T hi s si mulator
accurately computes the s-, y-, or z-parameters of arbi trari ly
shaped, multi layer planar structures such as stri pli nes.
3D electromagnetic analysis T hi s CAE tool accurately
computes the s-parameters for passi ve, three-di mensi onal,
multi port structures.
System analysis System and board-level si mulators offer
di screte-ti me and frequency-domai n capabi li ti es; can analyze
and opti mi ze compli cated system topologi es; handle complex
waveforms; and perform physi cal layout desi gn.
Modeling systems Hardware and software are combi ned to
extract parameters needed for accurate acti ve devi ce modeli ng.
http://www.hp.com/go/hpeesof
Vi si t the HP EEsof websi te for the latest i n
CA E news, products, software, and soluti ons.
High Flying Software
The pattern of a horn
antenna, displayed using
HP High-Frequency
Structure Simulator
3D electromagnetic
visualization software,
shows beam shapes
in both azimuth and
elevation in a single plot.
Relevant Products
Features of HP 8720D VNAs
Built-in synthesized source
with 1 Hz resolution
Allows measuring all four
s-parameters with a
single connection
Continuous updates for
two-port error correction
http://www.hp.com/go/tmdatasheets
H
Vector network analyzers with built-in synthesized
sources cover 50 MHz to up to 40 GHz
T he HP 8720D vector network analyzer fami ly
characteri zes RF and mi crowave components
from 50 M Hz up to 40 GHz. T hey combi ne a
fast synthesi zed source, tuned recei ver, and
S-parameter test set i n a si ngle i nstrument.
T he devi ces have the performance and
flexi bi li ty to solve di ffi cult measurement
problems and cut test ti mes, all at an
attracti ve cost. Use them to qui ckly and
accurately measure magni tude and phase of
all four s-parameters, as well as group delay,
plus the absolute output power of mi crowave
components.
Producti vi ty i s enhanced wi th pass/fai l testi ng, di rect pri nter/
plotter output of results, advanced marker functi ons, save/
recall of test confi gurati ons to i nternal memory or a bui lt-i n
floppy di sk dri ve, and test sequenci ng for automati on. O pti ons
allow hi gh-power tests, frequency offset mi xer testi ng, and
hi gh-accuracy noncoaxi al and on-wafer measurements.
HP 8720D Series Vector Network Analyzers
Microwave network analyzers
with coverage up to 110 GHz
T he HP 8510C mi crowave vector network analyzer fami ly
provi de complete soluti ons for characteri zi ng the li near
behavi or of ei ther acti ve or passi ve networks over the
45 M Hz to 110 GHz range. A complete system consi sts of
the HP 8510C network analyzer, an s-parameter test set,
and a compati ble RF source. Also avai lable are fully
i ntegrated systems, tested and veri fi ed pri or to shi pment.
T he HP 8510C di splays measurement results i n log/li near
magni tude, phase, or group delay format on a large, color CRT
wi th two i ndependent, yet i denti cal, channels. T he i mpact of
systemati c errors i s removed by vi rtually real-ti me error
correcti on, so a test devi ce can be adj usted whi le i ts bei ng
measured. Effecti ve di recti vi ty and source match can be i mproved
to as much as 60 dB. T he HP 85161B software leads the operator
one step at a ti me, from setup and cali brati on to hardcopy results.
Vi si t the HP T est & M easurement websi te and fi nd more than
1, 000 up to date product datasheets. ( A dobe A crobat users wi th
the Webli nk plug-i n may cli ck di rectly on the URL below. )
Relevant Products
HP 8510C Microwave Network Analyzers
Features of the HP 8510C
45 MHz to 110 GHz
frequency range
Real-time error-corrected
measurements
60 dB effective directivity
and source match
Up to 100 dB dynamic
range
0.001 dB, 0.01 degree,
0.01 ns resolution
Optional time domain
and pulsed RF
measurement capability
H

Relevant Products
H
RF network analyzer with integrated s-parameter test set
performs characterizations from30 kHz to 6 GHz
T he HP 8753D RF vector network analyzer wi ll
si mpli fy and speed your devi ce, component, or
network measurements i n the 30 kHz to 6 GHz
range. A 1-Hz resoluti on swept synthesi zed source,
s-parameter test set, and sensi ti ve recei ver are
i ntegrated i nto thi s compact i nstrument, whi ch i s
si mple to set up and use i n the lab or on the
producti on li ne. T he HP 8573D provi des magni tude
and phase i nformati on, offers up to 110 dB dynami c
range, makes group delay and ti me domai n
measurements, and uses vector accuracy
enhancement to mi ni mi ze measurement uncertai nty.
T o i ncrease your throughput i n producti on, the HP 8753D offers
features such as the test sequence functi on, whi ch allows you to
make a measurement once from the front panel and automati cally
save the keystrokes wi thout an external computer. T he analyzers
fast CPU clock rate, LI F and DO S formats for output to the
bui lt-i n di sk dri ve or an external di sk dri ve, and a 512 K B
nonvolati le memory also help i mprove your producti vi ty.
HP 8753D RF Network Analyzer

Features of the HP 8753D
Built-in s-parameter test
set, synthesized source
Optional time-domain
and swept-harmonic
measurements
Up to 110 dB dynamic range
Superb accuracy, with
comprehensive calibration
Save/recall to built-in
disk drive
T i me to market i s cri ti cal to product success i n todays competi ti ve

envi ronment. CAE tools from Hewlett-Packards EEsof Di vi si on
gi ve compani es a competi ti ve edge by si mpli fyi ng and expedi ti ng
the development of RF and mi crowave ci rcui ts and systems. For
cost savi ngs and
manufacturi ng
yi eld i mprove-
ments, desi gns
can be accurately
modeled, si mu-
lated, and opti -
mi zed before
they are actually
produced.
A comprehensi ve
set of CAE tools
from HP EEsof i ncludes hi gh-frequency ci rcui t, electro-magneti c,
and system si mulators, layout tools, devi ce modeli ng systems and
li brari es, and li nks to i nstrumentati on and thi rd-party desi gn
software. Vi si t the HP EEsof websi te for a conti nuously updated
li st of news, products, software, and soluti ons.
Relevant Products
CAE solutions for RF and microwave design
ELECTRONIC DESIGN AUTOMATION (EDA)
SOLUTIONSFROMHP EEsof
HIGH-FREQUENCY
H

Fast, accurate electromagnetic simulation of 3D
high-frequency structures saves design time
Wi th todays computer technology, the best method
for desi gni ng hi gh frequency ci rcui ts and machi ned structures
i s to compute thei r behavi or wi th 3D electromagneti c ( EM )
si mulator software, maki ng i mprovements through successi ve
si mulati ons, then bui ldi ng desi gns to the refi ned speci fi cati ons.
T hi s approach i s more cost effecti ve and takes less ti me than
bui ldi ng and testi ng prototype after prototype i n the lab.
T he HP Hi gh Frequency Structure Si mulator ( HP HFSS) ,

Release 5.0, models arbi trari ly-shaped, passi ve 3D structures
such as antennas, machi ned components, and RF and di gi tal
ci rcui ts. Usi ng accuracy-dri ven adapti ve soluti on refi nement,
the software produces accurate results ten ti mes faster, usi ng
half the memory of previ ous releases. I ts drawi ng envi ronment
and parts li brary si mpli fy speci fyi ng complex structures.
For comprehensi ve evaluati ons, ani mated EM fi elds, surface
currents, vector plots, antenna polar patterns and tabular,
generali zed s-parameters and Smi th charts can be vi ewed.
Relevant Products
95-1
Features of the HP 85180A
Reduces "cut and try"
prototyping, for lower
development costs and a
quicker time to market
Yields accurate results ten
times faster, using half the
memory of previous releases
Requires minimal user
knowledge of EM field theory
For PC and UNIX platforms
H
HP 85180A High Frequency Structure Simulator
A ppli cati on Note

HP offers an extensi ve curri culum of educati on servi ces at
locati ons worldwi de. To help hi gh-frequency desi gn
engi neers learn how to use CAE tools qui ckly, we provi de
trai ni ng courses taught by the experts from HPs EEsof
Di vi si on. Advanced RF/mi crowave CAE courses are also
conducted to help extend the capabi li ti es and hone the
ski lls of experi enced CAE users.
HP educati on classes are scheduled
regularly. T he curri culum can be tai lored
to a companys speci al needs, and trai ni ng
sessi ons can be conducted at i ts si te.
Vi si t HPs World Wi de Websi te for the complete and
conti nuously updated li st of Test & M easurement class
schedules and locati ons around the world. Cli ck
on the URL below or type the address i n your
browser. Regi ster di rectly onli ne!
http://www.hp.com/go/tmeducation
Relevant Services
RF and microwave education and training
H
Worldwide Contacts
Hewlett-Packard Test & Measurement offices
Asia-Pacific
Hewlett-Packard Asia Pacific Ltd.
Hewlett-Packard Hong Kong Ltd.
17-21/F Shell Tower,Times Square
1 Matheson Street, Causeway Bay
Hong Kong
85 2 599 7777
For Japan: 0120-421-345
Australia/New Zealand
Hewlett-Packard Australi a Ltd.
31-41 Joseph Street
Blackburn, Vi ctori a 3130
Australi a
1 800 627 485
Canada
Hewlett-Packard Ltd.
5150 Spectrum Way
Mi ssi ssauga, Ontari o L4W 5G1
905 206 4725
Europe/Middle East/Africa
Hewlett-Packard S.A.
150, Route du Nant-dAvri l
1217 Meyri n 1
Geneva, Swi tzerland
For Europe:
41 22 780 81 11
For Mi ddle East/Afri ca:
41 22 780 41 11
Latin America
Lati n Ameri can Regi on
Headquarters
Hewlett-Packard Company
Waterford Bui ldi ng, Sui te 950
5200 Blue Lagoon
Mi ami , FL 33126
305-267-4245
United States of America
Hewlett-Packard Company
800-452-4844
Vi si t HP s World Wi de Websi te for the complete and up
to date li sti ng of T est & M easurement sales offi ce
addresses and call center phone numbers. Cli ck on the
UR L below or type the address i n your browser.
http://www.hp.com/go/tmdir
H
Copyrights and Credits

Authors, contributors and producers
H
Authors
Dick Anderson ori gi nal author
of thi s appli cati on note.
Lee Smith S-parameter guru,
mathemati cal programmi ng,
techni cal i llustrati ons, updated &
revi sed content, and ani mati ons.
J eff Gruszynski expert domai n
consultant, content perspecti ves,
si de-bars text, and source i mages.
Contributors
Walt Patstone techni cal
content and si de-bar edi ti ng.
Chuck McGuire, Mike
Cdebaca new s-parameter
measurements & devi ce output.
Graphic &
Interaction Design
Ev Shafrir ori gi nal concept,
art di recti on, i nteracti on desi gn,
di gi tal publi shi ng & producti on,
content supervi si on, overall
di recti on, & proj ect management.
Christina Bangle free-style
i llustrati ons.and graphi c revi ews.
Kathy Cunningham page
layout desi gn and Quark guru.
Leann Scully source i mages.
Business Manager
Nyna Casey fundi ng, support,
exci tement, and encouragement.
Copyri ght Hewlett-Packard Company 1996-1997 All Ri ghts Reserved.
Adaptati on, reproducti on, translati on, extracti on, di ssemi nati on or di s-
assembly wi thout pri or wri tten permi ssi on i s prohi bi ted except as allowed
under the copyri ght laws. S-Parameter Techni ques for Faster, More
Accurate Network Desi gn i s electroni cally publi shed as part of the HP
Test & Measurement Di gi tal Appli cati on Note Li brary for the World Wi de
Web, November 1996. Ori gi nal pri nted publi cati on Number 5952-1130.
Richard W. Anderson
Holds a 1959 BSEE degree
from Utah State University
and a 1963 MSEE degree
from Stanford University.
During his 37 years with HP
Dick has contributed to the
development of numerous
microwave and other T&M
instruments. Currently, he
is HP Vice President, and
General Manager of the
T&M Microwave and
Communications Group.

S-Parameter Techniques: Test & Measurement

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

S-Parameter Techniques: Test & Measurement

Uploaded by

Copyright:

Available Formats

http://www.hp.

A ppli cati on Note

S-P aramet er T echni ques

February 1967 HP J ournal

A NEWMICROWAVE INSTRUMENT SWEEP

MEASURES GAIN, PHASE IMPEDANCE

We conti nuously apply our capabi li ti es i n measurement,

T hi s i nteracti ve appli cati on note revi ses and updates

( output short ci rcui ted)

Power reflected from the output port of the network.

Power reflected from the i nput port of the network.

Power i nci dent on the output of the network.

Power i nci dent on the i nput of the network

Power deli vered to a Z load

Power reflected from the network output

Power reflected from the network i nput

T o go from the fi rst phase curve of Fi g. 5e to the second phase

After removal of the constant-delay, or li near-phase,

HP 8753D RF Network Analyzer

T i me to market i s cri ti cal to product success i n todays competi ti ve

T he HP Hi gh Frequency Structure Si mulator ( HP HFSS) ,

Copyrights and Credits

You might also like