This action might not be possible to undo. Are you sure you want to continue?
com/go/ tmappnotes
H Test & Measurement
Appl i cati on Note 95 1
SParameter
Techniques
for Faster, More Accurate Network Design
Contents
1. Forew ord and I ntroducti on
2. Tw o Port Netw ork Theory
3. Usi ng S Parameters
4. Netw ork Cal cul ati ons w i th Scatteri ng Parameters
5. Ampl i fi er Desi gn usi ng Scatteri ng Parameters
6. M easurement of S Parameters
7. Narrow  Band Ampl i fi er Desi gn
8. Broadband Ampl i fi er Desi gn
9. Stabi l i ty Consi derati ons and the Desi gn of
Refl ecti on Ampl i fi ers and Osci l l ators
Appendi x A. Addi ti onal Readi ng on S Parameters
Appendi x B. Scatteri ng Parameter Rel ati onshi ps
Appendi x C. The Softw are Revol uti on
Rel evant Products, Educati on and I nformati on
Contacti ng Hew l ett Packard
© Copyright HewlettPackard Company, 1997. 3000 Hanover Street, Palo Alto California, USA.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
3
Foreword
H
Test & Measurement
Applicat ion Not e
951
SPar amet er Techniques
February 1967 HP Journal
Cover of issue in which
the original “SParameters
Theory and Application,”
written during Christmas
holiday 1966, first appeared.
HP Journal is now online at:
www.hp.com/ go/ journal
HEWLETTPACKARD J OURNAL
FEBRUARY 1967
A NEW MICROWAVE INSTRUMENT SWEEP
MEASURES GAIN, PHASE IMPEDANCE
WITH SCOPE OR METER READOUT; page 2
THE MICROWAVE ANALYZER IN THE
FUTURE; page 11
SPARAMETERS THEORY AND
APPLICATIONS; page 13
Cover:
See Also:
This applicat ion not e is based on an art icle writ t en for t he
February 1967 issue of t he HewlettPackard Jou rn al, yet
it s cont ent remains import ant t oday. Sparamet ers are an
essent ial part of highfrequency design, t hough much else
has changed during t he past 30 years. During t hat t ime,
HP has cont inuously forged ahead t o help creat e t oday' s
leading t est and measurement environment .
We cont inuously apply our capabilit ies in measurement ,
communicat ion, and comput at ion t o produce innovat ions
t hat help you t o improve your business result s. In wireless
communicat ions, for example, we est imat e t hat 85 percent
of t he world’s GSM ( Groupe Speciale Mobile) t elephones
are t est ed wit h HP inst rument s. Our accomplishment s
30 years hence may exceed our boldest conject ures.
This int eract ive applicat ion not e revises and updat es
t he1967 art icle for online elect ronic media. It reflect s
t he changes in our indust ry, while reminding us of t he
underlying scient ific basis for t he t echnology, and t akes
advant age of a pot ent new informat ion disseminat ion capabilit y,
t he World Wide Web. We hope you find t his t ut orial useful.
Richard Anderson,
HP Vice President and General Manager,
Microwave and Communicat ions Group
4
Linear net works, or nonlinear net works operat ing wit h signals
sufficient ly small t o cause t he net works t o respond in a linear
manner, can be complet ely charact erized by paramet ers
measured at t he net work t erminals ( port s) wit hout regard t o
t he cont ent s of t he net works. Once t he paramet ers of a
net work have been det ermined, it s behavior in any ext ernal
environment can be predict ed, again wit hout regard t o t he
cont ent s of t he net work.
Sparamet ers are import ant in microwave design because t hey
are easier t o measure and work wit h at high frequencies t han
ot her kinds of paramet ers. They are concept ually simple,
analyt ically convenient , and capable of providing a great
insight int o a measurement or design problem.
To show how sparamet ers ease microwave design, and how
you can best t ake advant age of t heir abilit ies, t his applicat ion
not e describes sparamet ers and flow graphs, and relat es t hem
t o more familiar concept s such as t ransducer power gain and
volt age gain. Dat a obt ained wit h a net work analyzer is used t o
illust rat e amplifier design.
Introduction
1
Maxwell’s equations
All electromagnetic
behaviors can ultimately be
explained by Maxwell’s four
basic equations:
However, it isn’t 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
ρ
∂
∂
∂
∂
t
t
0
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
5
Alt hough a net work may have any number of port s,
net work paramet ers can be explained most easily by
considering a net work wit h only t wo port s, an input port
and an out put port , like t he net work shown in Figure 1.
To charact erize t he performance of such a net work, any
of several paramet er set s can be used, each of which has
cert ain advant ages. Each paramet er set is relat ed t o a set of
four variables associat ed wit h t he t woport model. Two of t hese
variables represent t he excit at ion of t he net work ( independent
variables) , and t he remaining t wo represent t he response of
t he net work t o t he excit at ion ( dependent variables) . If t he
net work of Fig. 1 is excit ed by volt age sources V
1
and V
2
, t he
net work current s I
1
and I
2
will be relat ed by t he following
equat ions ( assuming t he net work behaves linearly) :
( 1)
( 2)
I y V y V
2 21 1 22 2
· +
I y V y V
1 11 1 12 2
· +
2
TwoPort Network Theory
Figure 1
General t woport net work.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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
In t his case, wit h port volt ages select ed
as independent variables and port
current s t aken as dependent variables,
t he relat ing paramet ers are called
short circuit admit t ance paramet ers,
or yparamet ers. In t he absence of
addit ional informat ion, four
measurement s are required t o det ermine t he four paramet ers
y
11
,
y
12
,
y
21
,
y
22
. Each measurement is made wit h one port
of t he net work excit ed by a volt age source while t he ot her
port is short circuit ed. For example, y
21,
t he forward
t ransadmit t ance, is t he rat io of t he current at port 2 t o t he
volt age at port 1 wit h port 2 short circuit ed, as shown in
equat ion 3.
( 3)
y
I
V
V
21
2
1
0
2
·
· ( output short circuited)
2
Twoport models
Twoport, threeport, and
nport 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 sparameters
(transmission and reflection
coefficients).
TwoPort Network Theory
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
7
If ot her independent and dependent variables had been
chosen, t he net work would have been described, as
before, by t wo linear equat ions similar t o equat ions 1
and 2, except t hat t he variables and t he paramet ers
describing t heir relat ionships would be different .
However, all paramet er set s cont ain t he same
informat ion about a net work, and it is always possible
t o calculat e any set in t erms of any ot her set .
“Scat t ering paramet ers,” which are commonly referred
t o as sparamet ers, are a paramet er set t hat relat es t o
t he t raveling waves t hat are scat t ered or reflect ed
when an nport net work is insert ed int o a t ransmission
line.
Appendix B “Scat t ering Paramet er Relat ionships”
cont ains t ables convert ing scat t ering paramet ers t o
and from conduct ance paramet ers ( y) ,
resistance parameters ( z) , and a mixture of
conductances and resistances parameters ( h) .
2
TwoPort Network Theory
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
8
3
Using SParameters
The ease wit h which scat t ering paramet ers can be measured
makes t hem especially well suit ed for describing t ransist ors and
ot her act ive devices. Measuring most ot her paramet ers calls for
t he input and out put of t he device t o be successively opened
and short circuit ed. This can be hard t o do, especially at RF
frequencies where lead induct ance and capacit ance make short
and open circuit s difficult t o obt ain. At
higher frequencies t hese measurement s
t ypically require t uning st ubs, separat ely
adjust ed at each measurement frequency,
t o reflect short or open circuit condit ions
t o t he device t erminals. Not only is t his
inconvenient and t edious, but a t uning
st ub shunt ing t he input or out put may
cause a t ransist or t o oscillat e, making
t he measurement invalid.
Sparamet ers, on t he ot her hand, are usually measured wit h t he
device imbedded bet ween a 50 Ω load and source, and t here is
very lit t le chance for oscillat ions t o occur.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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 twoport
electrical network.
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Anot her import ant advant age of sparamet ers st ems from
t he fact t hat t raveling waves, unlike t erminal volt ages and
current s, do not vary in magnit ude at point s along a lossless
t ransmission line. This means t hat scat t ering paramet ers can
be measured on a device locat ed at some dist ance from t he
measurement t ransducers, provided t hat t he measuring device
and t he t ransducers are connect ed by lowloss t ransmission lines.
De r i va t i on
Generalized scat t ering paramet ers have been defined by
K. Kurokawa [Appendix A]. These paramet ers describe
t he int errelat ionships of a new set of variables ( a
i
, b
i
) .
The variables a
i
and b
i
are normalized complex volt age waves
incident on and reflect ed from t he i
t h
port of t he net work.
They are defined in t erms of t he t erminal volt age V
i
, t he
t erminal current I
i
, and an arbit rary reference impedance Z
i
,
where t he ast erisk denot es t he complex conjugat e:
For most measurement s and calculat ions it is convenient
t o assume t hat t he reference impedance Z
i
is posit ive
and real. For t he remainder of t his art icle, t hen, all
variables and paramet ers will be referenced t o a single
posit ive real impedance, Z
0
.
The wave funct ions used t o define sparamet ers for a
t woport net work are shown in Fig. 2.
10
3
Figure 2
Twoport net work showing incident waves
(a
1
, a
2
) and ref lect ed waves (b
1
, b
2
) used in
sparamet er def init ions. The f low graph f or
t his net work appears in Figure 3.
Using SParameters
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 material’s
characteristic impedance,
Z
0
, is inversely related to
the index of refraction, N:
The sparameters 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
ε
=
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
a
1
Z
S
Z
L
V
S
a
2
b
1
b
2
TWO  PORT
NETWORK
11
The independent variables a
1
and a
2
are normalized
incident volt ages, as follows:
( 6)
( 7)
Dependent variables b
1
, and b
2
, are normalized 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 incident on port 2
a
V I Z
Z Z
V
Z
i
1
1 1 0
0 0
1
0
2
·
+
· ·
voltage wave incident on port 1
3
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
The linear equat ions describing t he t woport net work are
t hen:
( 10)
( 11)
The sparamet ers 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
= Out put reflect ion coefficient
wit h t he input t erminat ed by a
mat ched load ( Z
S
=Z
0
set s V
s
=0)
= Forward t ransmission ( insert ion)
gain wit h t he out put port
t erminat ed in a mat ched load.
= Reverse t ransmission ( insert ion)
gain wit h t he input port
t erminat ed in a mat ched load.
Using SParameters
Limitations of
lumped models
At l ow f requenci es most
ci rcui t s behave i n a
predi ct abl e manner and
can be descri bed by a
group of repl aceabl e,
l umpedequi val ent bl ack
boxes. At mi crowave
f requenci es, as ci rcui t
el ement si ze approaches
t he wavel engt hs of t he
operat i ng f requenci es,
such a si mpl i f i ed t ype
of model becomes
i naccurat e. The physi cal
arrangement s of t he
ci rcui t component s can
no l onger be t reat ed as
bl ack boxes. We have t o
use a di st ri but ed ci rcui t
el ement model and
sparamet ers.
= Input reflect ion coefficient wit h
t he out put port t erminat ed by a
mat ched load ( Z
L
=Z
0
set s a
2
=0)
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
13
Not ice t hat
( 16)
and
( 17)
where is t he input impedance at port 1.
This relat ionship bet ween reflect ion coefficient and impedance
is t he basis of t he Smit h Chart t ransmissionline calculat or.
Consequent ly, t he reflect ion coefficient s s
11
and s
22
can be
plot t ed on Smit h chart s, convert ed direct ly t o impedance, and
easily manipulat ed t o det ermine mat ching net works for
opt imizing a circuit design.
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
Sparameters
Sparamet ers and
di st ri but ed model s
provi de a means of
measuri ng, descri bi ng,
and charact eri zi ng
ci rcui t el ement s when
t radi t i onal l umped
equi val ent ci rcui t
model s cannot predi ct
ci rcui t behavi or t o t he
desi red l evel of
accuracy. They are used
f or t he desi gn of many
product s, such as
cel l ul ar t el ephones.
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Smi t h Cha r t Tr a nsf or ma t i on
The movie at t he right animat es t he
mapping bet ween t he complex impedance
plane and t he Smit h Chart . The Smit h
Chart is used t o plot reflect ances, but t he
circular grid lines allow easy reading of
t he corresponding impedance. As t he
animat ion shows, t he rect angular grid
lines of t he impedance plane are
t ransformed t o circles and arcs on t he
Smit h Chart .
Vert ical lines of const ant resist ance on t he impedance plane
are t ransformed int o circles on t he Smit h Chart . Horizont al
lines of const ant react ance on t he impedance plane are
t ransformed int o arcs on t he Smit h Chart . The t ransformat ion
bet ween t he impedance plane and t he Smit h Chart is nonlinear,
causing normalized resist ance and react ance values great er
t han unit y t o become compressed t owards t he right side of t he
Smit h Chart .
14
3
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Animation 1
Transf ormat i on bet ween
t he i mpedance pl ane and
t he Smi t h Chart . Cl i ck
over i mage t o ani mat e.
Showing transformations
graphically
To ease hi s RF desi gn
work, Bel l Lab’s Phi l l i p H.
Smi t h devel oped i ncreas
i ngl y accurat e and power
f ul graphi cal desi gn ai ds.
One versi on, a pol ar co
ordi nat e f orm, worked f or
al l val ues of i mpedance
component s, but Smi t h
suspect ed t hat a gri d wi t h
ort hogonal ci rcl es mi ght
be more pract i cal . In 1937
he const ruct ed t he basi c
Smi t h Chart st i l l used
t oday, usi ng a t ransf orma
t i on devel oped by co
workers E.B. Ferrel l and
J.W. McRae t hat accom
modat es al l dat a val ues
f rom zero t o i nf i ni t y.
15
Adva nt a ge s of S Pa r a me t e r s
The previous equat ions show one of t he import ant
advant ages of sparamet ers, namely t hat t hey are
simply gains and reflect ion coefficient s, bot h
familiar quant it ies t o engineers.
By comparison, some of t he yparamet ers
described earlier in t his art icle are not so familiar.
For example, t he yparamet er corresponding
t o insert ion gain s
21
is t he ‘forward
t ransadmit t ance’ y
21
given by
equat ion 3. Clearly, insert ion
gain gives by far t he
great er insight int o
t he operat ion of
t he net work.
3
Digital pulses
Di gi t al pul ses are
compri sed of hi ghorder
harmoni c f requenci es
t hat det ermi ne t he shape
of t he pul se. A short
pul se wi t h st eep edges
has a si gnal spect rum
wi t h rel at i vel y hi gh
power l evel s at very hi gh
f requenci es. As a resul t ,
some el ement s i n
modern hi ghspeed
di gi t al ci rcui t s requi re
charact eri zat i on wi t h
di st ri but ed model s and
sparamet ers f or accurat e
perf ormance predi ct i on.
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
16
3
Using SParameters
Anot her advant age of sparamet ers springs from t he
simple relat ionship bet ween t he variables a
1
, a
2
, b
1
,
and b
2
, and various power waves:
b
2
2
·
·
·
Power reflected from the output port of the network.
Power incident on the load.
Power that would be delivered to a Z load.
0
b
1
2
·
·
Power reflected from the input port of the network.
Power available from a Z source minus the power
delivered to the input of the network.
0
a
2
2
·
·
Power incident on the output of the network.
Power reflected from the load.
a
Z
1
2
0
·
·
Power incident on the input of the network
Power available from a source impedance
.
.
Radar
The devel opment
of radar, whi ch
uses powerf ul si gnal s
at short wavel engt hs t o
det ect smal l obj ect s at
l ong di st ances, provi ded
a powerf ul i ncent i ve f or
i mproved hi gh f requency
desi gn met hods duri ng
Worl d War II. The desi gn
met hods empl oyed at
t hat t i me combi ned
di st ri but ed measurement s
and l umped ci rcui t
desi gn. There was an
urgent need f or an
ef f i ci ent t ool t hat coul d
i nt egrat e measurement
and desi gn. The Smi t h
Chart met t hat need.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er
Techniques
17
The previous four equat ions show t hat sparamet ers are
simply relat ed t o power gain and mismat ch loss,
quant it ies which are oft en of more int erest
t han t he corresponding volt age funct ions:
s
12
2
· Reverse transducer power gain with Z load and source
0
s
21
2
·
·
Power delivered to a Z load
Power available from Z source
Transducer power gain with Z load and source
0
0
0
s
22
2
·
Power reflected from the network output
Power incident on the network output
s
11
2
·
Power reflected from the network input
Power incident on the network input
3
Using SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
18
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Si gna l Fl ow Gr a phs
Scat t ering paramet ers t urn out t o be part icularly convenient
in many net work calculat ions. This is especially t rue for power
and power gain calculat ions. The t ransfer paramet ers s
12
and
s
21
are a measure of t he complex insert ion gain, and t he
driving point paramet ers s
11
and s
22
are a measure of t he input
and out put mismat ch loss. As dimensionless expressions of
gain and reflect ion, t he sparamet ers not only give a clear and
meaningful physical int erpret at ion of t he net work performance,
but also form a nat ural set of paramet ers for use wit h signal
flow graphs [See references here and also in Appendix A ].
Of course, it is not necessary t o use signal flow graphs in order
t o use sparamet ers, but flow graphs make sparamet er
calculat ions ext remely simple. Therefore, t hey are st rongly
recommended. Flow graphs will be used in t he examples
t hat 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. MTT8,
No. 2, March, 1960.
N. Kuhn, ‘Simplified Signal
Flow Graph Analysis,’
Microwave Journal, Vol. 6,
No. 11, November, 1963.
Network Calculations with
Scattering Parameters
19
In a signal flow graph, each port is
represent ed by t wo nodes. Node a
n
represent s t he wave coming int o t he
device from anot her device at port n,
and node b
n
represent s t he wave
leaving t he device at port n. The
complex scat t ering coefficient s are
t hen represent ed as mult ipliers on
branches connect ing t he nodes wit hin
t he net work and in adjacent
net works. Fig. 3, right , is t he flow
graph represent at ion of t he syst em of Fig. 2.
Figure 3 shows t hat if t he load reflect ion coefficient Γ
L
is
zero ( Z
L
= Z
0
) t here is only one pat h connect ing b
1
t o a
1
( flow graph rules prohibit signal flow against t he forward
direct ion of a branch arrow) . This confirms t he definit ion
of s
11
:
s
b
a
a b
L
11
1
1
0
2 2
·
· · Γ
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
Γ
L
L
L
Z Z
Z Z
=

+
0
0
Γ
S
S
S
Z Z
Z Z
=

+
0
0
Figure 3
Fl ow graph f or
t woport net work
appeari ng i n Fi gure 2.
4
Network Calculations with
Scattering Parameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
20
The simplificat ion of net work analysis by flow graphs result s
from t he applicat ion of t he “nont ouching loop rule.” This rule
applies a generalized formula t o det ermine t he t ransfer
funct ion bet ween any t wo nodes wit hin a complex syst em. The
nont ouching loop rule is explained below.
The Nontouchi ng Loop Rul e
The nont ouching loop rule provides a
simple met hod for writ ing t he solut ion of
any flow graph by inspect ion. The solut ion
T ( t he rat io of t he out put variable t o t he
input variable) is defined, where:
T
k
= pat h gain of t he k
t h
forward pat h
∆ = 1 –
Σ (
all individual loop gains
)
+
Σ (
loop gain product s of all possible
combinat ions of 2 nont ouching loops
)
–
Σ (
loop gain product s of all possible
combinat ions of 3 nont ouching loops
)
+
. . .
∆
k
= The value of ∆ not t ouching t he k
t h
forward pat h.
4
Network Calculations with
Scattering Parameters
Better Smith Charts
On t he copyri ght ed Smi t h
Chart , curves of const ant
st andi ng wave rat i o,
const ant at t enuat i on, and
const ant ref l ect i on
coef f i ci ent are al l ci rcl es
coaxi al wi t h t he cent er of
t he di agram. Ref i nement s
t o t he ori gi nal f orm have
enhanced i t s usef ul ness.
In an art i cl e publ i shed i n
1944, f or exampl e, Smi t h
descri bed an i mproved
versi on and showed how
t o use i t wi t h ei t her
i mpedance or admi t t ance
coordi nat es. More recent
i mprovement s i ncl ude
doubl e Smi t h Chart s f or
i mpedance mat chi ng and
a scal e f or cal cul at i ng
phase di st ance.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
T
T
k k
k
·
∑
∆
∆
21
A pat h is a cont inuous succession of branches, and a forward
pat h is a pat h connect ing t he input node t o t he out put node,
where no node is encount ered more t han once. Pat h gain is
t he product of all t he branch mult ipliers along t he pat h. A loop
is a pat h t hat originat es and t erminat es on t he same node, no
node being encount ered more t han once. Loop gain is t he
product of t he branch mult ipliers around t he loop.
For example, in Figure 3 t here is only one forward pat h from
b
s
t o b
2,
and it s gain is s
21
. There are t wo pat hs from b
s
t o b
1
;
t heir pat h gains are s
21
s
12
Γ
L
and s
11
respect ively. There are
t hree individual loops, only one combinat ion of t wo
nont ouching loops, and no combinat ions of t hree or more
nont ouching loops. Therefore, t he value of ∆ for t his net work
is
The t ransfer funct ion from b
s
t o b
2
is t herefore
b
b
s
s
2
21
=
∆
4
Network Calculations with
Scattering Parameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Sparameters & Smith Charts
Invent ed i n t he 1960’ s,
Sparamet ers are a way
t o combi ne di st ri but ed
desi gn and di st ri but ed
measurement . s
11
and s
22
,
t he t wo sparamet ers
t ypi cal l y represent ed usi ng
Smi t h Chart s, are si mi l ar t o
l umped model s i n many
respect s because t hey are
rel at ed t o t he i nput
i mpedance and out put
i mpedance, respect i vel y.
The Smi t h Chart perf orms
a hi ghl y usef ul t ransl at i on
bet ween t he di st ri but ed
and l umped model s and i s
used t o predi ct ci rcui t and
syst em behavi or.
( )
s s
s s
11 12
21 22
∆ Γ Γ Γ Γ Γ Γ = − + + + 1
11 21 12 22 11 22
( s s s s s s
S L S L L S
)
22
Tr a nsduc e r Pow e r Ga i n
Using scat t ering paramet er flowgraphs and t he nont ouching
loop rule, it is easy t o calculat e t he t ransducer power gain wit h an
arbit rary load and source. In t he following equat ions, t he load and
source are described by t heir reflect ion coefficient s Γ
L
and Γ
S
,
respect ively, referenced t o t he real charact erist ic impedance Z
0
.
Transducer power gain:
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 delivered to the load
Power available from the source
incident on load reflected from load ( ) ( )
( )
( ) ( )
2
2 2
2
2
2
2
2 2
1
1
( )
1 1
Γ
Γ
Γ Γ
4
Network Calculations with
Scattering Parameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
23
4
Network Calculations with
Scattering Parameters
Obtaining maximum
performance
Sparamet ers are used
t o charact eri ze RF and
mi crowave component s
t hat must operat e
t oget her, i ncl udi ng
ampl i f i ers, t ransmi ssi on
l i nes, and ant ennas (and
f ree space). Because
sparamet ers al l ow t he
i nt eract i ons bet ween
such component s t o be
si mpl y predi ct ed and
cal cul at ed, t hey make i t
possi bl e t o maxi mi ze
perf ormance i n areas
such as power t ransf er,
di rect i vi t y, and
f requency response.
Using t he nont ouching loop rule,
( 18)
Two ot her paramet ers of int erest are:
1) Input reflect ion coefficient wit h t he out put
t erminat ion arbit rary 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
( ) Γ Γ
Γ
Γ
Γ
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
·
− − − +
·
− − −
·
− −
− − −
Γ Γ Γ Γ Γ Γ
Γ Γ Γ Γ
Γ Γ
Γ Γ Γ Γ
( ) ( )
( ) ( )
( ) ( )
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
24
2) Volt age gain wit h arbit rary source
and load impedances
( 20)
Appendix B cont ains formulas for calculat ing many oft enused
net work funct ions ( power gains, driving point charact erist ics,
et c.) in t erms of scat t ering paramet ers. Also included are
conversion formulas bet ween sparamet ers and h, y, and
zparamet ers, which are ot her paramet er set s used very oft en
for specifying t ransist ors at lower frequencies.
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
( )
( )
( )
( )
( )
( ) ( )
Γ
Γ Γ
Γ
4
Waveguides
A radar syst em del i vers a
l arge amount of energy
f rom a mi crowave (µW)
source t o t he t ransmi t t i ng
ant enna. The hi gh f i el d
st rengt hs cause short
ci rcui t s i n st andard wi res,
cabl i ng, and coax, so
wavegui des are used.
These hol l ow met al t ube
const ruct i ons conduct µW
energy much l i ke a
pl umbi ng syst em. In t he
desi gn of wavegui des, we
can t est f or si gnal ref l ec
t i ons and t ransmi ssi on
qual i t y wi t h sparamet ers.
Network Calculations with
Scattering Parameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Traveling wave amplifier
Sparameters are extensively
used for designing RF/µW
circuits such as the HP TC702
distributed traveling wave
amplifier (TWA) enlarged
in the photograph below.
The frequencydependent
impedances (or dispersion)
in this integrated circuit can
not be modeled by lumped
equivalent circuit elements,
but sparameters can
accurately characterize the
amplifier' s response.
25
Usi ng Sc a t t e r i ng Pa r a me t e r s
The remainder of t his applicat ion not e will show wit h several
examples how sparamet ers are used in t he design of t ransist or
amplifiers and oscillat ors. To keep t he discussion from becoming
bogged down in ext raneous det ails, t he emphasis in t hese
examples will be on sparamet er design methods,
and mat hemat ical manipulat ions will
be omit t ed wherever possible.
5
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
The HP 83017A mi crowave
system ampl i fi er achi eves
0.5–26.5 GHz bandwi dth by
i ncorporati ng the HP TC702
GaAs MESFET TWA IC.
26
Most design problems will begin wit h a
t ent at ive select ion of a device and t he
measurement of it s sparamet ers.
Figures 4a – 4e, which appear t o t he
right and on t he next t wo pages, are a
set of oscillograms showing complet e
sparamet er dat a bt ween 100 MHz and
1.7 GHz for a 2N3478 t ransist or in t he
commonemit t er configurat ion.
These graphs are t he result s of swept 
frequency measurement s made wit h t he
classic HP 8410A microwave net work
analyzer. They were originally published
as part of t he 1967 HP Jou rn al art icle.
Measurement s made wit h a modern
net work analyzer are present ed at t he
end of t his sect ion. While t he
measurement t ools have changed over
t he past 30 years, t he basic
measurement t echniques have not .
6
Measurement of
SParameters
Figure 4a
s
11
of a 2N3478 t ransi st or measured wi t h t he
cl assi c HP 8410A net work anal yzer. Out ermost
ci rcl e on Smi t h Chart overl ay corresponds t o
s
11
 = 1. The movement of s
11
wi t h f requency
i s approxi mat el y al ong ci rcl es of const ant
resi st ance, i ndi cat i ve of seri es capaci t ance and
i nduct ance.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
1700 MHz
100 MHz
S
11
Figure 4c
Magni tude and phase of s
12
. Whi l e the
phase of s
12
i s rel ati vel y i nsensi ti ve to the
frequency, the magni tude of s
12
i ncreases
about 6dB/octave.
27
6
Measurement of
SParameters
Figure 4b
Di spl ayed on the same scal e as Fi gure 4a,
s
22
moves between the i ndi cated frequenci es
roughl y al ong ci rcl es of constant conductance,
characteri sti c of a shunt RC equi val ent ci rcui t.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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.
Magni t ude and phase of s
21
measured
wi t h a l i ne st ret cher adj ust ed t o remove
t he l i near phase shi f t above 500 MHz.
28
6
Measurement of
SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Figure 4d – Magnitude and phase of s
21
.
The magni t ude of s
21
decays wi t h a sl ope
of about 6 dB/oct ave, whi l e t he phase
decreases l i nearl y 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
In Fig. 4f, t he magnit ude of s
21
from Fig. 4d is
replot t ed on a logarit hmic frequency scale,
along wit h addit ional dat a on s
21
below
100 MHz, measured wit h a vect or volt met er.
The magnit ude of s
21
is essent ially const ant
t o 125 MHz, and t hen it rolls off at a slope of
6 dB/oct ave.
The phase of s
21
, as seen in
Fig. 4d, varies linearly wit h
frequency above about 500 MHz.
By adjust ing a calibrat ed line
st ret cher in t he net work
analyzer, a compensat ing linear
phase shift was int roduced, and
t he phase curve of Fig. 4e
result ed. To go from t he phase
curve of Fig. 4d t o t hat of Fig.4e
required 3.35 cm of line, t hat
is equivalent t o a pure t ime
delay of 112 picoseconds.
6
Measurement of
SParameters
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Figure 4f
Top curve: s
21
 f rom Fi g. 4 i s repl ot t ed
on a l ogari t hmi c f requency scal e. Dat a
bel ow 100 MHz was measured wi t h an
HP 8405A vect or vol t met er. The bot t om
curve u, the unilateral figure of merit,
cal cul at ed f rom sparamet ers.
30
Aft er removal of t he const ant delay, or linearphase,
component , t he phase angle of s
21
for t his t ransist or ( Fig. 4e)
varies from 180° at dc t o +90° at high frequencies, passing
t hrough +135° at 125 MHz, t he 3 dB point of t he magnit ude
curve. In ot her words, s
21
behaves like a single pole in t he
frequency domain, and it is possible t o writ e a closed
expression for it . This expression is
( 21)
where
The t ime delay T
0
= 112 ps is due primarily t o t he
t ransit t ime of minorit y carriers ( elect rons) across
t he base of t his npn t ransist or.
T ps
f
s
0
0
21
112
2
2 125
11 2 21
0
·
·
·
· ·
MHz
dB .
s
s e
j
j T
21
21
0
0
0
1
·
+
6
Measurement of
SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Importance of simple
approximations
Using firstorder
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
SParameters
Figure 5a
Sparamet ers of 2N3478 t ransi st or
i n commonemi t t er conf i gurat i on,
measured by an HP 8753 net work
anal yzer. Thi s pl ot shows s
11
and
s
22
on a Smi t h Chart . The marker set at 47 MHz represent s
t he 3 dB gai n rol l of f poi nt of s
21
. The f requency i ndex of
t hi s poi nt i s ref erenced i n t he ot her pl ot s.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
S
11
S
22
The sparamet ers of an 2N3478
t ransist or shown in Figures 4a
t hrough 4f were measured wit h t he
classic HP 8410A net work analyzer.
In Figures 5a t hrough 5e, t he
sparamet ers of an 2N3478 t ransist or
are shown remeasured wit h a
modern HP 8753 net work analyzer.
Figures 5a t hrough 5e represent t he
act ual sparamet ers of t his t ransist or
bet ween 0.300 MHz and 1.00 GHz.
32
6
Measurement of
SParameters
S
12
Figure 5b
A pl ot of t he magni t ude and phase of s
12
. Whi l e t he
phase of s
12
depends onl y weakl y on t he f requency,
t he magnitude i ncreases rapi dl y at l ow f requenci es.
Figure 5c
A pol ar pl ot of s
21
. The frequency
marker shown i s at t he 3 dB poi nt .
Bot h t he phase angl e and magni t ude
decrease dramat i cal l y as t he
f requency i s i ncreased.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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
SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
S
21
Figure 5d
The magnitude of s
21
pl ot t ed on a l og
scal e showi ng t he 6 dB/oct ave rol l of f
above 75 MHz.
Figure 5e
The phase angl e, i n degrees, of s
21
. A t i me
del ay of 167 ps was deembedded f rom t he
measured data usi ng t he anal yzer' s el ect ri cal 
del ay f eat ure t o get a response wi t h a si ngl e
pol e t ransf er charact eri st i c. Removi ng t hi s
t i me del ay al l ows t he phase di st ort i on t o be
vi ewed wi t h much great er resol ut i on.
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
sparameter data on each
component. Measurements
are made with a VNA,
Vector Network Analyzer,
an instrument that
accurately measures the
sparameters, transfer
function, or impedance
characteristic of linear
networks across a broad
range of frequencies.
34
In Fig. 5d, t he magnit ude of s
21
from Fig. 5c is replot t ed
on a logarit hmic frequency scale. The magnit ude of s
21
is
essent ially const ant t o 75 MHz, and t hen rolls off at a slope
of 6 dB/oct ave. The phase angle of s
21
as seen in Fig. 5e
varies linearly wit h frequency above 500 MHz. To bet t er
charact erize phase dist ort ion, a compensat ing linear phase
shift was int roduced elect ronically in t he net work analyzer.
This est ablished an accurat e calibrat ion for measuring t he
device, result ing in t he second phase curve of Figure 5e.
To go from t he first phase curve of Fig. 5e t o t he second phase
curve required removing a pure t ime delay of 167 picoseconds.
Thirt y years ago t his operat ion was accomplished by
deembedding 5.0 cm of line using a calibrat ed line st ret cher.
Today it ’s performed by soft ware in t he net work analyzer.
Aft er removal of t he const ant delay, or linearphase,
component , t he phase angle of s
21
for t his t ransist or
( Fig. 5e) varies from 180° at dc t o +90° at high frequencies,
passing t hrough +135° at 75 MHz, t he 3 dB point of t he
magnit ude curve. In ot her words, s
21
behaves like a single
pole in t he frequency domain.
6
Measurement of
SParameters
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
6
35
Since s
21
behaves like a single pole in t he frequency domain,
it is possible t o writ e a closed expression for it . This expression
is t he same as equat ion 21, repeat ed here.
The t ime delay T
0
= 167 ps is due primarily t o t he t ransit t ime
of minorit y carriers ( elect rons) across t he base of t his npn
t ransist or. Removing t his t ime delay using t he elect ricaldelay
feat ure of t he vect or net work analyzer allows t he phase
dist ort ion t o be viewed wit h much great er resolut ion.
Using t he first order, singlepole approximat ion for s
21
is an
import ant st ep in circuit design. Today, however, we have design
t echnology undreamed of in 1967. Subsequent ly, t hrough t he
process of elect ronic design aut omat ion ( EDA) , comput eraided
engineering ( CAE) t ools now can be used it erat ively t o simulat e
and refine t he design. These t ools combine accurat e models
wit h performanceopt imizat ion and yieldanalysis capabilit ies.
where
T ps
f
s
0
0
21
0
167
2
2 75
8 4 18 5
=
=
= ×
= =
ω π
ω π MHz
dB . .
s
s e
j
j T
21
21
0
0
0
1
=
−
+
− ω
ω
ω
Measurement of
SParameters
Complete network
characterization
Vector Network Analyzers
(VNA) are ideal for applica
tions requiring complete
network characterization.
They use narrowband
detection to achieve wide
dynamic range and provide
noisefree data. VNAs are
often combined with powerful
computerbased electronic
design automation (EDA)
systems to both measure
data, and simulate and
optimize the performance
of the complete system
implementation being
developed.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
36
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
6
Measurement of
SParameters
Ex pl a na t i on of M e a sur e me nt Di sc r e pa nc i e s
You may have not iced a difference bet ween t he measured and
caclulat ed dat a from t he 1967 HP Jou rn al art icle and t he dat a
obt ained for t his updat ed applicat ion not e. Bot h set s of dat a are
fundament ally correct . Two major sources account for t hese
differences:
Me as ure me nt t e chnique s – Early net work analyzers did not
have onboard comput ers, an HPIB st andard, or highresolut ion
graphics t o perform calibrat ion, ext ract precision numerical
dat a, or display elect ronic markers. Calibrat ion t echniques
used in 1967 were procedurally and mat hemat ically simpler
t han t hose used t oday. Modern net work anaylzers cont ain
sophist icat ed aut omat ed t echniques t hat enhance measurement
processing capabilit ies and reduce operat or errors.
De vice diffe re nce s – Semiconduct or manufact uring processes
evolve over t ime. Device engineers at t empt t o produce ident ical
t ransist ors wit h different processes. Nevert heless, successive
generat ions of part s like t he 2N3478 can exhibit unint ent ional,
and somet imes unavoidable, performance differences, especially
in charact erist ics not guarant eed on t he dat asheet .
37
Suppose now t hat t his 2N3478 t ransist or is t o be
used in a simple amplifier, operat ing bet ween a
50 Ω source and a 50 Ω load, and opt imized
for power gain at 300 MHz by means of
lossless input and out put mat ching
net works. Since reverse gain s
12
for t his
t ransist or is quit e small—50 dB smaller
t han forward gain s
21
, according t o
Fig. 4—t here is a possibilit y t hat it can be
neglect ed. If t his is so, t he design problem will
be much simpler, because set t ing s
12
equal t o zero will make
t he design equat ions much less complicat ed.
In det ermining how much error will be int roduced by assuming
s
12
= 0, t he first st ep is t o calculat e t he unilat eral figure of
merit u, using t he formula given in Appendix B. That is,
( 22)
u
s s s s
s s
·
− −
11 12 21 22
11
2
22
2
1 1 ( ) ( )
7
NarrowBand
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Unilateral figure of merit
All twoport 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 funct ion of frequency, calculat ed from
t he measured paramet ers, appears in Fig. 4f. Now if G
Tu
is t he t ransducer power gain wit h s
12
= 0 and G
T
is t he
act ual t ransducer power gain, t he maximum error
int roduced by using G
Tu
inst ead of G
T
is given by t he
following relat ionship:
( 23)
From Fig. 4f, t he maximum value of u is about 0.03, so t he
maximum error in t his case t urns out t o be about +0.25 dB
at 100 MHz. This is small enough t o just ify t he assumpt ion
t hat s
12
= 0.
Incident ally, a small reverse gain, or feedback fact or, s
12
, is an
import ant and desirable propert y for a t ransist or t o have, for
reasons ot her t han it simplifies amplifier design. A small feedback
fact or means t hat t he input charact erist ics of t he complet ed
amplifier will be independent of t he load, and t he out put will be
independent of t he source impedance. In most amplifiers,
isolat ion of source and load is an import ant considerat ion.
1
1
1
1
2 2
( ) ( ) +
< <
− u
G
G
u
T
Tu
7
NarrowBand
Amplifier Design
Highfrequency 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 firststage
amplifiers in satellite
TV receivers.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
39
Ret urning now t o t he 300MHz amplifier design, t he unilat eral
expression for t ransducer power gain, obt ained eit her by
set t ing s
12
= 0 in equat ion 18 or looking in Appendix B, is
( 24)
When

s
11

and

s
22

are bot h less t han one, as t hey are in
t his case, maximum G
Tu
occurs for Γ
S
= s*
11
and Γ
L
= s*
22
( Appendix B) .
The next st ep in t he design is t o synt hesize mat ching net works
t hat will t ransform t he 50 Ω load and source impedances t o
t he impedances corresponding t o reflect ion coefficient s of s*
11
and s*
22
, respect ively. Since t his is t o be a singlefrequency
amplifier, t he mat ching net works need not be complicat ed.
Simple seriescapacit or, shunt induct or net works will not only
do t he job, but will also provide a handy means of biasing t he
t ransist or—via t he induct or—and of isolat ing t he dc bias from
t he load and t he source.
G
s
s s
Tu
S L
S L
·
− −
− −
21
2 2 2
11
2
22
2
1 1
1 1
( ) ( ) Γ Γ
Γ Γ
7
NarrowBand
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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
sparameters.
40
Values of L and C t o be used in t he mat ching net works for t he
300MHz amplifier are det ermined using t he Smit h Chart of
Fig. 6, which is shown on t he next page. First , point s
corresponding t o s
11
, s*
11
, s
22
, and s*
22
at 300 MHz are
plot t ed. Each point represent s t he t ip of a vect or leading away
from t he cent er of t he chart , it s lengt h equal t o t he magnit ude
of t he reflect ion coefficient being plot t ed, and it s angle equal
t o t he phase of t he coefficient .
Next , a combinat ion of const ant resist ance and const ant 
conduct ance circles is found, leading from t he cent er of t he
chart , represent ing 50 Ω, t o s*
11
and s*
22
. The circles on t he
Smit h Chart are const ant resist ance circles; increasing series
capacit ive react ance moves an impedance point count er
clockwise along t hese circles.
7
NarrowBand
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 Ω load impedance
to s*
11
. Another matching
network transforms the
50 Ω source impedance to
s*
22
.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
You wi l l f i nd an i nt eract i ve Impedance Mat chi ng Model at t he
HP Test & Measurement websi t e l i st ed bel ow. Chal l enge
yoursel f t o i mpedance mat chi ng games based on pri nci pl es
and exampl es di scussed i n t hi s appl i cat i on not e!
Cl i ck on t he URL bel ow or t ype t he address i n 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
NarrowBand
Amplifier Design
Figure 6
Smi t h Chart f or 300MHz
ampl i f i er desi gn exampl e.
In this case, the circle to be used for
finding series C is the one passing
through the center of the chart, as
shown by the solid line in Fig. 6.
Increasing shunt inductive
susceptance moves impedance
points clockwise along constant
conductance circles. These
circles are like the constant
resistance circles, but they
are on another Smith Chart,
one that is just the reverse
of the chart shown in Fig. 6.
The constantconductance
circles for shunt L all pass
through the leftmost point of
the chart rather than the rightmost
point. The circles to be used are
those passing through s*
11
and s*
22
,
as shown by the dashed lines in Fig. 6.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
42
Once t hese circles have been locat ed,
t he normalized values of L and C
needed for t he mat ching net works are
calculat ed from readings t aken from t he
react ance and suscept ance scales of t he
Smit h Chart s.
Each element ’s react ance or
suscept ance is t he difference bet ween
t he scale readings at t he t wo end point s
of a circular arc. Which arc corresponds
t o which element is indicat ed in Fig. 6.
The final net work and t he element
values, normalized and unnormalized,
are shown in Fig. 7.
7
Figure 7
A 300MHz ampl i f i er wi t h
mat chi ng net works f or
maxi mum power gai n.
NarrowBand
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
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 26 nH
1
x
nH
156
2 0 3 10
83
9
= =
( . )
x
pF
1
2 0 3 10 3 5 50
3
9
= =
( . )( . )( )
pF
1
25 = =
x 2 0 3 10
9
( . )(0.42)(50)
50
= =
x 2 0 3 10
9
( . )(1.01)
Calculat ions:
The animat ion t o t he right demonst rat es how t o use a
Smit h Chart t o design a mat ching net work bet ween a t ransist or
out put and a resist ive load. As previously described, t o
maximize t he power delivered t o t he load, t he s*
22
paramet er of
t he t ransist or must be mat ched t o t he load impedance, 50 Ω in
t his 300MHz amplifier example. This mat ching is achieved
using t he LC circuit shown at t he right .
St art ing from t he 50 Ω load, t he series capacit ance is varied t o
move t he impedance point along t he circle of const ant 50 Ω
resist ance on t he Smit h Chart . The capacit ance is adjust ed unt il
it int ersect s t he const ant conduct ance circle on which s*
22
is
sit t ing. Varying t he shunt induct ance t hen moves t he impedance
point along t his const ant
conduct ance circle as
indicat ed by t he admit t ance
Smit h Chart . To reach s*
22
,
t he shunt induct ance is
adjust ed unt il t he impedance
point reaches s*
22
.
43
7
NarrowBand
Amplifier Design
50Ω L
C
2N3478
Animation 2
Impedance mat chi ng
usi ng t he Smi t h Chart
f or t he mat chi ng
net work shown at t he
l ef t . Cl i ck over t he chart
t o st art ani mat i on.
http://www.hp.com/go/tminteractive
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
C = 1000 pF
L = 10000 nH
S
22
*
44
Designing a broadband amplifier, t hat is, one
which has nearly const ant gain over a
prescribed frequency range, is a mat t er of
surrounding a t ransist or wit h ext ernal
element s in order t o compensat e for t he
variat ion of forward gain,

s
21

wit h
frequency.
This can be done in eit her of t wo ways—
first , negat ive feedback, or second, select ive
mismat ching of t he input and out put
circuit ry. We will use t he second met hod.
When feedback is used, it is usually
convenient t o convert t o y or zparamet ers
( for shunt or series feedback, respect ively)
using t he conversion equat ions given in
Appendix B and a digit al comput er.
Equat ion 24 for t he unilat eral t ransducer
power gain can be fact ored int o t hree part s,
as shown t o t he right :
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
·
·
·
−
−
·
−
−
Γ
Γ
Γ
Γ
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
45
When a broadband amplifier is designed by selective
mismatching, the gain contributions of G
1
and G
2
are varied to
compensate for the variations of G
0
= s
21

2
with frequency.
Suppose that the 2N3478 transistor whose sparameters are
given in Fig. 4 is to be used in a broadband amplifier that will
operate from 300 MHz to 700 MHz. The amplifier is to be
driven from a 50 Ω source and is to drive a 50 Ω load.
According to Figure 4f,
To realize an amplifier with a constant gain of 10 dB, source
and load matching networks must be found that will decrease
the gain by 3 dB at 300 MHz, leave the gain the same at
450 MHz, and increase the gain 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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
46
Alt hough in t he general case bot h a source
mat ching net work and a load mat ching
net work would be designed, G
1max
( i.e.,
G
1
for Γ
s
= s*
11
) for t his t ransist or is
less t han 1 dB over t he frequencies of
int erest , which means t here is lit t le t o
be gained by mat ching t he source.
Consequent ly, for t his example, only
a loadmat ching net work will be
designed. Procedures for designing
sourcemat ching net works are
ident ical t o t hose used for designing
loadmat ching net works.
The first st ep in t he design of t he
loadmat ching net work is t o plot s*
22
over t he required frequency range on t he
Smit h Chart , Fig. 8a.
8
Broadband
Amplifier Design
s
22
*
700
450
300
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Figure 8a
A pl ot of s*
22
over t he f requency
range f rom 300 MHz t o 700 MHz.
47
8
Broadband
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Next , a set of const ant gain circles is drawn. As shown in Fig. 8b,
each circle is drawn for a single frequency; it s cent er is on a line
bet ween t he cent er of t he Smit h Chart and t he point
represent ing s*
22
at t hat frequency. The dist ance from
t he cent er of t he Smit h Chart t o t he cent er of t he
const ant gain circle is given by t he following
equat ions, which also appear in Appendix B:
where
The radius of t he const ant gain circle is:
ρ
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
·
−
− − ( )
ρ
2
2
r
ρ
2
ρ
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
Const ant gai n ci rcl es.
48
For t his example, t hree circles will be drawn, one for G
2
= 3 dB
at 300 MHz, one for G
2
= 0 dB at 450 MHz, and one for G
2
=
+4 dB at 700 MHz. Since s
22
 for t his t ransist or is const ant at
0.85 over t he frequency range [see Figure 4( b) ], G
2max
for all
t hree circles is ( 0.278)
1
, or 5.6 dB. The t hree const ant gain
circles are indicat ed in Fig. 8b.
The required mat ching net work must t ransform t he cent er of t he
Smit h Chart , represent ing 50 , t o some point on t he 3 dB
circle at 300 MHz, t o some point on t he 0 dB circle at 450 MHz,
and t o some point on t he +4 dB circle at 700 MHz. There are
undoubt edly many net works t hat
will do t his. One sat isfact ory
solut ion is a combinat ion of t wo
induct ors, one in shunt and one in
series, as shown in Fig. 9.
Z
L
= 50
50
L
series
L
shunt
2N3478
8
Broadband
Amplifier Design
Figure 9
A combi nat i on of shunt
and seri es i nduct ances
i s a sui t abl e mat chi ng
net work f or t he
broadband ampl i f i er.
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
49
8
Broadband
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Shunt and series element s move impedance point s
on t he Smit h Chart along const ant conduct ance
and const ant resist ance circles, as explained in
t he narrowband design example. As shown in
Fig. 10a, t he shunt induct ance t ransforms t he
50 Ω load alng a circle of const ant conduct ance
and varying ( wit h frequency) induct ive
suscept ance. The series induct or t ransforms
t he combinat ion of t he 50 Ω load and t he
shunt induct ance along circles of const ant
resist ance and varying induct ive react ance.
Opt imizing t he values of shunt and series L
is an it erat ive process wit h t wo goals:
– t he t ransformed load reflect ion
t erminat es on t he right gain circle at
each frequency, and
– t he suscept ance component decreases
wit h frequency and t he react ance component
increases wit h frequency. ( This rule applies t o
induct ors; capacit ors would behave in t he
opposit e way.)
3
0
0
M
H
z
4
5
0
M
H
z
7
0
0
M
H
z
Constant
conductance
circleTransformation
due to L
shunt
Constant resistance
circlesTransformation
due to L
series
Γ
L
locus
Figure 10a
Matching paths
for shunt and
series L.
50
Once appropriat e const ant 
conduct ance and const ant 
resist ance circles have been found,
t he react ances and suscept ances
of t he element s can be read
direct ly from t he Smit h Chart .
Then t he element values are
calculat ed, t he same as t hey were
for t he narrowband design.
Figure 10b is a schemat ic diagram
of t he complet ed broadband
amplifier, wit h unnormalized
element values.
Figure 10b
Broadband amplif ier wit h const ant gain
of 10 dB f rom 300 MHz t o 700 MHz.
50Ω
Z
S
= 50Ω
20.4 nH
36.4 nH
2N3478
Inductance Calculations:
j L
Z
j j
Z
j L
j
series
shunt
ω
ω
0
0
3 64 0 44 3 2
1 3
=  =
= 
( . . ) .
L nH nH
series
π
3 2 50
2 0 7
36 4 = =
( . ) ( )
( . )
.
.
L nH
shunt
π
50
1 3 2 0 3
20 4 = =
( . ) ( ) ( . )
.
From 700 MHz data,
From 300 MHz data,
8
Broadband
Amplifier Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
51
Desi gn of Refl ecti on Ampl i fi ers
and Osci l l ators
When t he real part of t he input impedance of a
net work is negat ive, t he corresponding input
reflect ion coefficient ( Equat ion 17) is great er
t han one, and t he net work can be used as t he
basis for t wo import ant t ypes of circuit s,
reflect ion amplifiers and oscillat ors. A reflect ion
amplifier ( Fig. 11) can be realized wit h a
circulat or—a nonreciprocal t hreeport device—
and a negat iveresist ance device.
The circulat or is used t o separat e t he incident
( input ) wave from t he larger wave reflect ed by
t he negat iveresist ance device. Theoret ically, if
t he circulat or is perfect and has a posit ive real
charact erist ic impedance Z
0
, an amplifier wit h
infinit e gain can be built by select ing a negat ive
resist ance device whose input impedance has a
real part equal t o Z
0
and an imaginary part
equal t o zero ( t he imaginary part can be set
equal t o zero by t uning if necessary) .
Figure 11
Ref l ect i on ampl i f i er
consi st s of ci rcul at or and
t ransi st or wi t h negat i ve
i nput resi st ance.
9
Stability Considerations
Circulator
Input Output
Two port with
s'
11
> 1
(Real part of input
impedance is
negative)
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
H
52
Amplifiers, of course, are not supposed to oscillate, whether they
Figure 3, the ratio
of the reflected voltage wave b
1
to the input voltage wave b
s
is
where s′
11
is the input reflection coefficient with Γ
s
= 0 ( Z
2
= Z
0
)
and an arbitrary load impedance Z
L
, as defined in Equation 19.
If at some frequency
( 25)
the circuit is unstable and it will oscillate at that frequency.
On the other hand, if
the device is unconditionally stable and will not oscillate, what ever
t he phase angle of Γ
s
might be.
′ < s
s
11
1
Γ
Γ
s
s′ =
11
1
b
b
s
s
s s
1
11
11
1
=
′
− ′ Γ
9
Stability Considerations
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Computer Aided
Engineering tools (CAE)
CAE sof t ware t ool s are
used i n t he desi gn process
t o si mul at e act ual devi ce
and ci rcui t behavi or so
desi gns can be eval uat ed
bef ore t hey' re bui l t . The
CAE approach i s f ast er,
produces accurat e resul t s,
and i s easi er t o f ol l ow
t han manual met hods
usi ng graphi cal desi gn ai ds.
CAE t ool s are part of t he
t ot al engi neeri ng sol ut i on.
are reflection amplifiers or some other kind. There is a convenient
criterion based upon scattering parameters for determining whether
a device is stable or potentially unstable with given source and load
impedances. Referring again to the flow graph of
53
To see how t hese principles of
st abilit y are applied in design
problems, consider t he
t ransist or oscillat or design
illust rat ed in Fig. 12. In t his
case t he input reflect ion
coefficient s′
11
is t he reflect ion
coefficient looking int o t he
collect or circuit , and t he
‘source’ reflect ion coefficient Γ
s
is one of t he t wo t ankcircuit
refect ion coefficient s, Γ
T1
or
Γ
T2
. From equat ion 19,
′ · +
−
s s
s s
s
L
L
11 11
12 21
22
1
Γ
Γ
9
Stability Considerations
150 Ω
Γ
T
1
ω
0
, Q
Γ
T
2
200 Ω Z
L
, Γ
L
s'
11
15 Ω
ω
0
, Q
Parallel
Tank
Circuit
Series
Tank
Circuit
Gain
Element
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Figure 12 The t ransi st or
osci l l at or i s desi gned by
choosi ng a t ank ci rcui t
such t hat
Γ
s
s′ ·
11
1
54
To make the transistor oscillate, s′
11
and Γ
s
must be adjusted so
that they satisfy equation 25. There are four steps in the design
procedure:
– Measure the four scattering parameters of the transistor as
functions of frequency.
– Choose a load reflection coefficient Γ
L
that makes s′
11
greater
than unity. In general, it may also take an external feedback
element that increases s
12
s
21
to make s′
11
greater than one.
– Plot 1/s′
11
on a Smith Chart. ( If the network analyzer is
being used to measured the sparameters of the transistor,
1/s′
11
can be measured directly by reversing the reference and
test channel connections between the reflection test unit and
the harmonic frequency converter. The polar display with a
Smith Chart overlay then gives the desired plot immediately.)
– Connect either the series or the parallel tank circuit to the
collector circuit and tune it so that Γ
T1
or Γ
T2
is large enough
to satisfy equation 25. ( The tank circuit reflection coefficient
plays the role of Γ
s
in this equation.)
9
Stability Considerations
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
55
Figure 13 shows a Smit h Chart plot of
1/s
11
for a high frequency t ransist or
in t he commonbase configurat ion.
Load impedance Z
L
is 200 ,
which means t hat
L
referred t o
50 is 0.6. Reflect ion
coefficient s
T1
and
T2
are
also plot t ed as funct ions of t he
resonant frequencies of t he
t wo t ank circuit s. Oscillat ions
occur when t he locus of
T1
or
T2
passes t hrough t he shaded
region. Thus, t his t ransist or
would oscillat e from 1.5 t o
2.5 GHz wit h a series t uned circuit ,
and from 2.0 t o 2.7 GHz wit h a
parallel t uned circuit .
— Di ck An derson , 1967 an d 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 t ransi st or wi l l osci l l at e i n t he shaded area bet ween 1.5 and 2.5 GHz
wi t h a seri est uned ci rcui t and bet ween 2.0 and 2.7 GHz wi t h a paral l el t uned ci rcui t .
Stability Considerations
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
56
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
In addit ion t o previous references list ed earlier and repeat ed
again here, t he following papers and books were list ed in t he
1967 HP Jou rn al art icle as sources for informat ion on
sparamet er design procedures and flow graphs. Current
references are also ment ioned.
– J. K. Hunt on, ‘Analysis of Microwave Measurement
Techniques by Means of Signal Flow Graphs,’
IRE Transact ions on Microwave Theory and Techniques,
Vol. MTT 8, No. 2, March, 1960.
– D.C Youla, ‘On Scat t ering Mat rices Normalized t o Complex
Port Numbers,’ Proc. IRE, Vol. 49, No. 7, July, 1961.
– J.G. Linvill and J.F. Gibbons, ‘Transist ors and Act ive
Circuit s,’ McGrawHill, 1961. ( No sparamet ers, but good
t reat ment of Smit h Chart design met hods.)
– N. Kuhn, ‘Simplified Signal Flow Graph Analysis,’
Microwave Journal, Vol. 6, No, 11, November, 1963.
– K. Kurokawa, ‘Power Waves and t he Scat t ering Mat rix,’
IEEE Transact ions on Microwave Theory and Techniques,
Vol. MTT13, No. 2, March, 1965.
A
Additional Reading
on SParameters
57
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
– F. Weinert , ‘Scat t ering Paramet ers Speed Design of
HighFrequency Transist or Circuit s,’
Elect ronics, Vol. 39, No. 18, Sept . 5, 1966.
– G. Fredricks, ‘How t o Use SParamet ers for Transist or
Circuit Design,’ EEE Vol. 14, No. 12, Dec., 1966.
Among many modern reference sources on t he subject , t he
following book, first published in 1969, is definit ely a classic:
– Smit h, Phillip H., ‘Elect ronic Applicat ions of t he Smit h
Chart in Waveguide, Circuit and Component Analysis,’
Noble Publishing Classic Series, Tucker, Georgia, 1995,
ISBN1884932398, 237 pp.
We also ment ion a useful t ext book cont aining 5 chapt ers,
2 appendices, and problem set s. This t ext present s a unified
t reat ment of t he analysis and design of microwave t ransist or
amplifiers using scat t ering paramet ers t echniques:
– G. Gonzalez, ‘Microwave Transist or Amplifiers:
Analysis and Design,’ Prent ice Hall, 1984,
ISBN 0135816467, 240 pp.
A
Additional Reading
on SParameters
Keep up to date
Thi s book by P. H. Smi th
referenced here must be
consi dered the ul ti mate
source on Smi th Charts.
Many excel l ent arti cl es on
the use of Smi th charts
have appeared i n trade
publ i cati ons such as the
Mi crowave Journal , and
educati onal Smi th Chart
software i s sol d over the
i nternet.
A
V
V
s
s s
V
L
L
· ·
+
− + ′
2
1
21
22 11
1
1 1
( )
( )( )
Γ
Γ
′ · +
−
s s
s s
s
S
S
22 22
12 21
11
1
Γ
Γ
′ · +
−
s s
s s
s
L
L
11 11
12 21
22
1
Γ
Γ
58
I nput refl ecti on coeffi ci ent w i th
arbi trary Z
L
Output refl ecti on coeffi ci ent w i th
arbi trary Z
s
Vol tage gai n w i th
arbi trary 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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
59
B
Scattering Parameter
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
Γ
L
Γ
L
Γ 2Re( N)
S
Γ 2Re( M)
Available Power Gain
G
A
Power available from network
Power available from source
= =
−
−
+ −
−
s
s s D
S
S
21
2 2
22
2 2
11
2 2
1
1
Γ
Γ
Transducer Power Gain
G
T
Power delivered to load
Power available from source
= =
−
−
−
( )
−
( )
−
s
s s s s
S L
S L L S
21
2 2
2
11 22 12 21
2
1 1
1 1
Γ Γ
Γ Γ Γ Γ )
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
60
B
Scattering Parameter
Relationships
Unilateral Transducer Power Gain
Maximum Unilateral Transducer Power Gain when s
and s Maximum obtained 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
·
( )
·
−

.
`
,
−

.
`
,
− −
·
<
< · ·
·
−

.
`
,
−

.
`
,
Γ Γ
Γ Γ
Γ Γ .
* *
·· 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
·
·
−
−
·
−
−
·
−
·
Γ
Γ
Γ
Γ
max
,
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
61
Constant Gai n Ci rcl es (Uni l ateral case: s
12
= 0)
B
Scattering Parameter
Relationships
ρ
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
ρ
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
62
B
Scattering Parameter
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Unilateral figure of merit
All twoport 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’s
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
Scattering Parameter
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
64
B s s D
B s s D
1 11
2
22
2 2
2 22
2
11
2 2
1
1
· + − −
· + − −
B
Scattering Parameter
Relationships
(Use minus sign
when is positive,
plus sign when
is negative.)
B
B
1
1
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Source and load for Simultaneous Match
Maximum Available Power Gain
If ,
where
Γ
Γ
mS
mL
A
M
B B M
M
N
B B N
N
K G
s
s
K K
K C
·
t −
]
]
]
]
]
·
t −
]
]
]
]
]
> · t −

.
`
,
·
−
*
*
max
1 1
2
2
2
2 2
2
2
2
21
12
2
1
4
2
4
2
1 1
65
sparameters
in terms of zparameters
zparameters
in terms of sparameters
B
Scattering Parameter
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
66
sparameters
in terms of yparameters
yparameters
in terms of sparameters
B
Scattering Parameter
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
67
sparameters
in terms of hparameters
hparameters
in terms of sparameters
B
Scattering Parameter
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
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
68
The h–, y–, and z–parameters listed in
previous tables are all normalized to Z
0
.
If h′, y′, z′ are the actual parameters, then:
B
Scattering Parameter
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 vari ous scat t eri ng
paramet ers are al l
normal i zed by t he
ref erence i mpedance, Z
0
.
Thi s i mpedance i s usual l y
t he charact eri st i c
i mpedance of t he
t ransmi ssi on l i ne i n whi ch
t he net work of i nt erest i s
embedded. Normal i zi ng
t he scat t eri ng paramet ers
makes t he Smi t h Chart
readi l y appl i cabl e t o
t ransmi ssi on l i nes of any
i mpedance. In addi t i on,
i mpedance and
admi t t ance val ues can be
pl ot t ed on t he same
chart .
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
69
El ectroni c Desi gn Automati on ( EDA)
The Sof t w a r e Re vol ut i on
In t he 30 years t hat have elapsed since t he
publicat ion of Dick Anderson' s art icle,
comput er aided engineering ( CAE) t ools have
been developed for t he highfrequency design
met hods t hat were t radit ionally implement ed
using pencil and paper. These comput er
soft ware programs run on UNIX workst at ions
and PCs, and do much more t han merely
assist in comput at ionint ensive design t asks.
Modern CAE t ools for highfrequency
design eliminat e t he need for simplifying
assumpt ions ( such as, s
12
= 0) and can
accurat ely simulat e act ual device, circuit , or
syst em behavior. They enable broadband
solut ions, offer opt imizat ion and yieldanalysis capabilit ies, and
provide answers t o “What if?” quest ions. CAE t ools also speed t he
analysis of a wide range of RF and microwave devices, circuit s,
and syst ems, for a short er t ime t o market , while lowering cost s.
C
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
CAE tools for High
Frequency Design
C
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
CAE tools for High
Frequency Design
70
The CAE t ools of int erest t o RF and
microwave designers include t hose
summarized below:
Smalls ignal ( s parame t e r) s imulat ion —
Smallsignal analysis CAE t ools simulat e
response over a range of frequencies, so act ual implement at ions
perform more closely t o design paramet ers. Mat ching circuit s are
easily det ermined, and can be readily opt imized, saving t ime.
A yieldanalysis feat ure allows t he select ion of component s in
mat ching net works for t he best product ion yield, saving cost s.
Large s ignal s imulat ion — This powerful analysis t ool
includes t he harmonic balance implement at ion, useful for
oscillat or design and many ot her problems.
Circuit Enve lope s imulat ion — Efficient ly analyzes circuit s
and feedback loops in t he presence of modulat ed or t ransient
highfrequency signals.
Circuit Envelope Simulation
The wavef orm above
t ypi f i es t he modul at ed
and t ransi ent si gnal s t hat
can be ef f i ci ent l y anal yzed
usi ng ci rcui t envel ope
si mul at or sof t ware. Ci rcui t
envel ope si mul at i on i s
orders of magni t ude f ast er
t han t radi t i onal SPICE
si mul at i on sof t ware i f t he
envel ope bandwi dt h of t he
RF carri er f requency i s
much smal l er t han t he
carri er f requency i t sel f .
Thi s i s t he case i n many
communi cat i ons, and radar
ci rcui t s and subsyst ems.
71
C
CAE tools for High
Frequency Design
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Time domain analys is — A CAE t ool t hat is especially
useful for simulat ing t he response of digit al syst ems at
high clock rat es.
Planar e le ct romagne t ic analys is — This simulat or
accurat ely comput es t he s, y, or zparamet ers of arbit rarily
shaped, mult ilayer planar st ruct ures such as st riplines.
3D e le ct romagne t ic analys is — This CAE t ool accurat ely
comput es t he sparamet ers for passive, t hreedimensional,
mult iport st ruct ures.
Sys t e m analys is — Syst em and boardlevel simulat ors offer
discret et ime and frequencydomain capabilit ies; can analyze
and opt imize complicat ed syst em t opologies; handle complex
waveforms; and perform physical layout design.
Mode ling s ys t e ms — Hardware and soft ware are combined t o
ext ract paramet ers needed for accurat e act ive device modeling.
http://www.hp.com/go/hpeesof
Visit t he HP EEsof websit e for t he lat est in
CAE news, product s, soft ware, and solut ions.
High Flying Software
The pat t ern of a horn
ant enna, di spl ayed usi ng
HP Hi ghFrequency
St ruct ure Si mul at or
3D el ect romagnet i c
vi sual i zat i on sof t ware,
shows beam shapes
i n bot h azi mut h and
el evat i on i n a si ngl e pl ot .
Relevant Products
Features of HP 8720D VNA’s
• Bui l t i n synt hesi zed source
wi t h 1 Hz resol ut i on
• Al l ows measuri ng al l f our
sparamet ers wi t h a
si ngl e connect i on
• Cont i nuous updat es f or
t woport error correct i on
http:/ / w w w.hp.com/ go/ tmdatasheets
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Vector netw ork anal yzers w i th bui l t i n synthesi zed
sources cover 50 M Hz to up to 40 GHz
The HP 8720D vect or net work analyzer family
charact erizes RF and microwave component s
from 50 MHz up t o 40 GHz. They combine a
fast synt hesized source, t uned receiver, and
Sparamet er t est set in a single inst rument .
The devices have t he performance and
flexibilit y t o solve difficult measurement
problems and cut t est t imes, all at an
at t ract ive cost . Use t hem t o quickly and
accurat ely measure magnit ude and phase of
all four sparamet ers, as well as group delay,
plus t he absolut e out put power of microwave
component s.
Product ivit y is enhanced wit h pass/fail t est ing, direct print er/
plot t er out put of result s, advanced marker funct ions, save/
recall of t est configurat ions t o int ernal memory or a built in
floppy disk drive, and t est sequencing for aut omat ion. Opt ions
allow highpower t est s, frequency offset mixer t est ing, and
highaccuracy noncoaxial and onwafer measurement s.
HP 8720D Seri es Vector Netw ork Anal yzers
M i crow ave netw ork anal yzers
w i th coverage up to 110 GHz
The HP 8510C microwave vect or net work analyzer family
provide complet e solut ions for charact erizing t he linear
behavior of eit her act ive or passive net works over t he
45 MHz t o 110 GHz range. A complet e syst em consist s of
t he HP 8510C net work analyzer, an sparamet er t est set ,
and a compat ible RF source. Also available are fully
int egrat ed syst ems, t est ed and verified prior t o shipment .
The HP 8510C displays measurement result s in log/linear
magnit ude, phase, or group delay format on a large, color CRT
wit h t wo independent , yet ident ical, channels. The impact of
syst emat ic errors is removed by virt ually “realt ime” error
correct ion, so a t est device can be adjust ed while it ’s being
measured. Effect ive direct ivit y and source mat ch can be improved
t o as much as 60 dB. The HP 85161B soft ware leads t he operat or
one st ep at a t ime, from set up and calibrat ion t o hardcopy result s.
Visit t he HP Test & Measurement websit e and find more t han
1,000 up t o dat e product dat asheet s. ( Adobe Acrobat users wit h
t he Weblink plugin may click direct ly on t he URL below.)
Relevant Products
HP 8510C Microwave Network Analyzers
Features of the HP 8510C
• 45 MHz to 110 GHz
frequency range
• Real t i me errorcorrect ed
measurement s
• 60 dB ef f ect i ve di rect i vi t y
and source mat ch
• Up t o 100 dB dynami c
range
• 0.001 dB, 0.01 degree,
0.01 ns resol ut i on
• Opt i onal t i me domai n
and pul sed RF
measurement capabi l i t y
http://www.hp.com/go/tmdatasheets
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Relevant Products
http:/ / w w w.hp.com/ go/ tmdatasheets
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
RF netw ork anal yzer w i th i ntegrated s parameter test set
performs characteri zati ons from 30 kHz to 6 GHz
The HP 8753D RF vect or net work analyzer will
simplify and speed your device, component , or
net work measurement s in t he 30 kHz t o 6 GHz
range. A 1Hz resolut ion swept synt hesized source,
sparamet er t est set , and sensit ive receiver are
int egrat ed int o t his compact inst rument , which is
simple t o set up and use in t he lab or on t he
product ion line. The HP 8573D provides magnit ude
and phase informat ion, offers up t o 110 dB dynamic
range, makes group delay and t ime domain
measurement s, and uses vect or accuracy
enhancement t o minimize measurement uncert aint y.
To increase your t hroughput in product ion, t he HP 8753D offers
feat ures such as t he t est sequence funct ion, which allows you t o
make a measurement once from t he front panel and aut omat ically
save t he keyst rokes wit hout an ext ernal comput er. The analyzer’s
fast CPU clock rat e, LIF and DOS format s for out put t o t he
built in disk drive or an ext ernal disk drive, and a 512 KB
nonvolat ile memory also help improve your product ivit y.
HP 8753D RF Netw ork Anal yzer
Features of the HP 8753D
•Bui l ti n sparameter test
set, synthesi zed source
•Opti onal ti medomai n
and sweptharmoni c
measurements
•Up to 110 dB dynami c range
•Superb accuracy, wi th
comprehensi ve cal i brati on
•Save/recal l to bui l ti n
di sk dri ve
Time t o market is crit ical t o product success in t oday’s compet it ive
environment . CAE t ools from Hewlet t Packard’s EEsof Division
give companies a compet it ive edge by simplifying and expedit ing
t he development of RF and microwave circuit s and syst ems. For
cost savings and
manufact uring
yield improve
ment s, designs
can be accurat ely
modeled, simu
lat ed, and opt i
mized before
t hey are act ually
produced.
A comprehensive
set of CAE t ools
from HP EEsof includes highfrequency circuit , elect romagnet ic,
and syst em simulat ors, layout t ools, device modeling syst ems and
libraries, and links t o inst rument at ion and t hirdpart y design
soft ware. Visit t he HP EEsof websit e for a cont inuously updat ed
list of news, product s, soft ware, and solut ions.
Relevant Products
CAE sol uti ons for RF and mi crow ave desi gn
http://www.hp.com/go/hpeesof
ELECTRONIC DESIGN AUTOMATION (EDA)
SOLUTIONS FROM HP EEsof
HIGHFREQUENCY
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
http:/ / w w w.hp.com/ go/ hpeesof
Fast, accurate el ectromagneti c si mul ati on of 3D
hi gh frequency structures saves desi gn ti me
Wit h t oday’s comput er t echnology, t he best met hod
for designing high frequency circuit s and machined st ruct ures
is t o comput e t heir behavior wit h 3D elect romagnet ic ( EM)
simulat or soft ware, making improvement s t hrough successive
simulat ions, t hen building designs t o t he refined specificat ions.
This approach is more cost effect ive and t akes less t ime t han
building and t est ing prot ot ype aft er prot ot ype in t he lab.
The HP High Frequency St ruct ure Simulat or ( HP HFSS) ,
Release 5.0, models arbit rarilyshaped, passive 3D st ruct ures
such as ant ennas, machined component s, and RF and digit al
circuit s. Using accuracydriven adapt ive solut ion refinement ,
t he soft ware produces accurat e result s t en t imes fast er, using
half t he memory of previous releases. It s drawing environment
and part s library simplify specifying complex st ruct ures.
For comprehensive evaluat ions, animat ed EM fields, surface
current s, vect or plot s, ant enna polar pat t erns and t abular,
generalized sparamet ers and Smit h chart s can be viewed.
Relevant Products
951
Features of the HP 85180A
•Reduces "cut and try"
prototypi ng, for l ower
devel opment costs and a
qui cker ti me to market
•Yi el ds accurate resul ts ten
ti mes faster, usi ng hal f the
memory of previ ous rel eases
•Requi res mi ni mal user
knowl edge of EM fi el d theory
•For PC and UNIX pl atforms
H
HP 85180A Hi gh Frequency Structure Si mul ator
Test & Measurement
Applicat ion Not e
SPar amet er Techniques
HP offers an ext ensive curriculum of educat ion services at
locat ions worldwide. To help highfrequency design
engineers learn how t o use CAE t ools quickly, we provide
t raining courses t aught by t he expert s from HP’s EEsof
Division. Advanced RF/microwave CAE courses are also
conduct ed t o help ext end t he capabilit ies and hone t he
skills of experienced CAE users.
HP educat ion classes are scheduled
regularly. The curriculum can be t ailored
t o a company’s special needs, and t raining
sessions can be conduct ed at it s sit e.
Visit HP’s World Wide Websit e for t he complet e and
cont inuously updat ed list of Test & Measurement class
schedules and locat ions around t he world. Click
on t he URL below or t ype t he address in your
browser. Regist er direct ly online!
http://www.hp.com/go/tmeducation
Relevant Services
RF and mi crow ave educati on and trai ni ng
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Worldwide Contacts
HewlettPackard Test & Measurement offices
As iaPacific
HewlettPackard Asia Pacific Ltd.
HewlettPackard Hong Kong Ltd.
1721/F Shell Tower,Times Square
1 Matheson Street, Causeway Bay
Hong Kong
85 2 599 7777
For Japan: 0120421345
Aus t ralia/Ne w Ze aland
HewlettPackard Australia Ltd.
3141 Joseph Street
Blackburn, Victoria 3130
Australia
1 800 627 485
Canada
HewlettPackard Ltd.
5150 Spectrum Way
Mississauga, Ontario L4W 5G1
905 206 4725
Europe /Middle Eas t /Africa
HewlettPackard S.A.
150, Route du Nantd’Avril
1217 Meyrin 1
Geneva, Switzerland
For Europe:
41 22 780 81 11
For Middle East/Africa:
41 22 780 41 11
Lat in Ame rica
Latin American Region
Headquarters
HewlettPackard Company
Waterford Building, Suite 950
5200 Blue Lagoon
Miami, FL 33126
3052674245
Unit e d St at e s of Ame rica
HewlettPackard Company
8004524844
Visit HP’s Wor ld Wide Websit e for t he complet e and up
t o dat e list ing of Test & Measur ement sales office
addr esses and call cent er phone number s. Click on t he
URL below or t ype t he addr ess in your br owser.
http://www.hp.com/go/tmdir
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
©
Copyrights and Credits
Authors, contri butors and producers
H
Test & Measurement
Applicat ion Not e 951
SPar amet er Techniques
Aut hors
Dick Anderson – original author
of this application note.
Lee Smith – Sparameter guru,
mathematical programming,
technical illustrations, updated &
revised content, and animations.
Jeff Gruszynski – expert domain
consultant, content perspectives,
sidebars text, and source images.
Cont ribut ors
Walt Patstone – technical
content and sidebar editing.
Chuck McGuire, Mike
C’debaca – new sparameter
measurements & device output.
Graphic &
Int e ract ion De s ign
Ev Shafrir – original concept,
art direction, interaction design,
digital publishing & production,
content supervision, overall
direction, & project management.
Christina Bangle – freestyle
illustrations.and graphic reviews.
Kathy Cunningham – page
layout design and Quark guru.
Leann Scully – source images.
Bus ine s s Manage r
Nyna Casey – funding, support,
excitement, and encouragement.
© Copyright HewlettPackard Company 19961997 All Rights Reserved.
Adaptation, reproduction, translation, extraction, dissemination or dis
assembly without prior written permission is prohibited except as allowed
under the copyright laws. SParameter Techniques for Faster, More
Accurate Network Design is electronically published as part of the HP
Test & Measurement Digital Application Note Library for the World Wide
Web, November 1996. Original printed publication Number 59521130.
Richard W. Anderson
Hol ds a 1959 BSEE degree
f rom Ut ah St at e Uni versi t y
and a 1963 MSEE degree
f rom St anf ord Uni versi t y.
Duri ng hi s 37 years wi t h HP
Di ck has cont ri but ed t o t he
devel opment of numerous
mi crowave and ot her T&M
i nst rument s. Current l y, he
i s HP Vi ce Presi dent , and
General Manager of t he
T&M Mi crowave and
Communi cat i ons Group.
H
Test & Measurement Application Note 951 SParameter Techniques
Contents
1. Foreword and Introduction 2. TwoPort Network Theory 3. Using SParameters 4. Network Calculations with Scattering Parameters 5. Amplifier Design using Scattering Parameters 6. Measurement of SParameters 7. NarrowBand Amplifier Design 8. Broadband Amplifier Design 9. Stability Considerations and the Design of Reflection Amplifiers and Oscillators Appendix A. Additional Reading on SParameters Appendix B. Scattering Parameter Relationships Appendix C. The Software Revolution Relevant Products, Education and Information Contacting HewlettPackard
© Copyright HewlettPackard Company, 1997. 3000 Hanover Street, Palo Alto California, USA.
H
Test & Measurement Application Note 951 SParameter Techniques
Foreword
This application note is based on an article written for the February 1967 issue of the HewlettPackard Journal, yet its content remains important today. Sparameters are an essential part of highfrequency design, though much else has changed during the past 30 years. During that time, HP has continuously forged ahead to help create today's leading test and measurement environment. We continuously apply our capabilities in measurement, communication, and computation to produce innovations that help you to improve your business results. In wireless communications, for example, we estimate that 85 percent of the world’s GSM (Groupe Speciale Mobile) telephones are tested with HP instruments. Our accomplishments 30 years hence may exceed our boldest conjectures. This interactive application note revises and updates the1967 article for online electronic media. It reflects the changes in our industry, while reminding us of the underlying scientific basis for the technology, and takes advantage of a potent new information dissemination capability, the World Wide Web. We hope you find this tutorial useful. Richard Anderson, HP Vice President and General Manager, Microwave and Communications Group
HEWLETTPACKARD JOURNAL
Cover: A NEW MICROWAVE INSTRUMENT SWEEP MEASURES GAIN, PHASE IMPEDANCE WITH SCOPE OR METER READOUT; page 2 See Also: THE MICROWAVE ANALYZER IN THE FUTURE; page 11 SPARAMETERS THEORY AND APPLICATIONS; page 13
FEBRUARY 1967
3
February 1967 HP Journal Cover of issue in which the original “SParameters Theory and Application,” written during Christmas holiday 1966, first appeared. HP Journal is now online at: www.hp.com/go/journal
again without regard to the contents of the network. Maxwell’s equations All electromagnetic behaviors can ultimately be explained by Maxwell’s four basic equations: ∇⋅D=ρ ∇⋅B= 0 ∂t ∂D ∇ × H = j+ ∂t ∇×E= − ∂B However. can be completely characterized by parameters measured at the network terminals (ports) without regard to the contents of the networks.1 4 H Test & Measurement Application Note 951 SParameter Techniques Introduction Linear networks. To show how sparameters ease microwave design. Once the parameters of a network have been determined. Data obtained with a network analyzer is used to illustrate amplifier design. Efficient design requires the use of approximations such as lumped and distributed models. and how you can best take advantage of their abilities. and relates them to more familiar concepts such as transducer power gain and voltage gain. and capable of providing a great insight into a measurement or design problem. analytically convenient. . or nonlinear networks operating with signals sufficiently small to cause the networks to respond in a linear manner. this application note describes sparameters and flow graphs. its behavior in any external environment can be predicted. Sparameters are important in microwave design because they are easier to measure and work with at high frequencies than other kinds of parameters. They are conceptually simple. Solving them can be quite difficult. it isn’t always possible or convenient to use these equations directly.
Distributed models are needed at RF frequencies and higher to account for the increased behavioral impact of those physical effects. 1 is excited by voltage sources V1 and V2. the circuits. . an input port – NETWORK – 2 and an output port. To characterize the performance of such a network. variables represent the excitation of the network (independent Why are models needed? variables). each of which has certain advantages. If the behavior of components. and systems. like the network shown in Figure 1. and the remaining two represent the response of Models help us predict the the network to the excitation (dependent variables).PORT + V1 V considering a network with only two ports. where some physical effects can be ignored because they are so small. Each parameter set is related to a set of Figure 1 four variables associated with the twoport model.2 5 H Test & Measurement Application Note 951 SParameter Techniques TwoPort Network Theory I1 I2 Although a network may have any number of ports. Two of these General twoport network. any Port 1 Port 2 of several parameter sets can be used. network parameters can be explained most easily by + TWO . network currents I1 and I2 will be related by the following Lumped models are useful at equations (assuming the network behaves linearly): I1 = y 11 V1 + y 12 V2 I 2 = y 21 V1 + y 22 V2 (1) (2) lower frequencies. network of Fig.
y21. threeport. is the ratio of the current at port 2 to the voltage at port 1 with port 2 short circuited. and nport models simplify the input / output response of active and passive devices and circuits into “black boxes” described by a set of four linear parameters. Z (resistances). For example. as shown in equation 3. four measurements are required to determine the four parameters y11. Lumped models use representations such as Y (conductances). or yparameters. Distributed models use sparameters (transmission and reflection coefficients). Each measurement is made with one port of the network excited by a voltage source while the other port is short circuited. with port voltages selected as independent variables and port currents taken as dependent variables. y21. the relating parameters are called shortcircuit admittance parameters.2 6 H Test & Measurement Application Note 951 SParameter Techniques TwoPort Network Theory In this case. the forward transadmittance. . I y 21 = 2 V1 V = 0 (output short circuited) 2 (3) Twoport models Twoport. y22. y12. In the absence of additional information. and h (a mixture of conductances and resistances).
except that the variables and the parameters describing their relationships would be different. and a mixture of conductances and resistances parameters (h). are a parameter set that relates to the traveling waves that are scattered or reflected when an nport network is inserted into a transmission line. “Scattering parameters. . by two linear equations similar to equations 1 and 2. the network would have been described. However.” which are commonly referred to as sparameters. all parameter sets contain the same information about a network. Appendix B “Scattering Parameter Relationships” contains tables converting scattering parameters to and from conductance parameters (y). as before. resistance parameters (z).2 7 H Test & Measurement Application Note 951 SParameter Techniques TwoPort Network Theory If other independent and dependent variables had been chosen. and it is always possible to calculate any set in terms of any other set.
separately adjusted at each measurement frequency. At higher frequencies these measurements typically require tuning stubs. to reflect short or open circuit conditions to the device terminals. but a tuning stub shunting the input or output may cause a transistor to oscillate. Sparameters. on the other hand. and there is very little chance for oscillations to occur. are usually measured with the device imbedded between a 50 Ω load and source.3 8 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques The ease with which scattering parameters can be measured makes them especially well suited for describing transistors and other active devices. Measuring most other parameters calls for the input and output of the device to be successively opened and short circuited. especially at RF frequencies where lead inductance and capacitance make short and open circuits difficult to obtain. Not only is this inconvenient and tedious. This can be hard to do. making the measurement invalid. .
This means that scattering parameters can be measured on a device located at some distance from the measurement transducers. scattering parameters are measures of reflection and transmission of voltage waves through a twoport electrical network. Kurokawa [Appendix A]. . do not vary in magnitude at points along a lossless transmission line. as in this photograph. the terminal current I i . The variables ai and bi are normalized complex voltage waves incident on and reflected from the ith port of the network. They are defined in terms of the terminal voltage Vi .3 9 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques Another important advantage of sparameters stems from the fact that traveling waves. unlike terminal voltages and currents. The amounts reflected and transmitted are characterized by optical reflection and transmission coefficients. provided that the measuring device and the transducers are connected by lowloss transmission lines. Similarly. These parameters describe the interrelationships of a new set of variables (ai . where the asterisk denotes the complex conjugate: V + Zi Ii ai = i 2 Re Z i (4) * Vi − Z i I i bi = 2 Re Z i (5) Transmission and Reflection When light interacts with a lens. bi). Derivation Generalized scattering parameters have been defined by K. and an arbitrary reference impedance Z i . part of the light incident on the woman's eyeglasses is reflected while the rest is transmitted.
Z0. A material’s characteristic impedance. N: ZL Z0 ε 377 = 1 N Figure 2 Twoport network showing incident waves (a 1 . . The wave functions used to define sparameters for a twoport network are shown in Fig. s 12 and s 21 are the same as optical transmission coefficients. The sparameters s 11 and s 22 are the same as optical reflection coefficients. For the remainder of this article. ZS a1 b1 TWO . a 2 ) and reflected waves (b 1 .3 VS 10 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques For most measurements and calculations it is convenient to assume that the reference impedance Zi is positive and real. all variables and parameters will be referenced to a single positive real impedance. is inversely related to the index of refraction.PORT NETWORK a2 b2 Scattering parameters relationship to optics Impedance mismatches between successive elements in an RF circuit relate closely to optics. The flow graph for this network appears in Figure 3. then. b 2 ) used in sparameter definitions. Z0. where there are successive differences in the index of refraction. 2.
3 a 2= 11 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques The independent variables a1 and a2 are normalized incident voltages. and b2. as follows: V i1 V1 + I1 Z 0 voltage wave incident on port 1 = = a1 = 2 Z0 Z0 Z0 V2 + I 2 Z 0 V voltage wave incident on port 2 = = i2 2 Z0 Z0 Z0 (6) (7) Dependent variables b1. are normalized reflected voltages: V −I Z V voltage wave reflected from port 1 = r1 b1 = 1 1 0 = 2 Z0 Z0 Z0 V2 − I 2 Z 0 voltage wave reflected from port 2 Vr 2 = = b2 = 2 Z0 Z0 Z0 (8) (9) .
as circuit element size approaches the wavelengths of the operating frequencies.3 s 11= s 22 = s 21 = s 12 = 12 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques The linear equations describing the twoport network are then: b 1 = s 11 a 1 + s 12 a 2 (10) b 2 = s 21 a 1+ s 22 a 2 (11) The sparameters s11. s21. We have to use a distributed circuit element model and sparameters. b1 = Reverse transmission (insertion) gain with the input port a2 a 1 =0 terminated in a matched load. and s12 are: b1 = Input reflection coefficient with a 1 a = 0 the output port terminated by a 2 matched load (Z =Z sets a =0) L 0 2 (12) b2 = Output reflection coefficient a 2 a = 0 with the input terminated by a 1 matched load (ZS=Z0 sets Vs=0) b2 = Forward transmission (insertion) a 1 a = 0 gain with the output port 2 terminated in a matched load. (13) (14) (15) Limitations of lumped models At low frequencies most circuits behave in a predictable manner and can be described by a group of replaceable. s22. . The physical arrangements of the circuit components can no longer be treated as black boxes. such a simplified type of model becomes inaccurate. lumpedequivalent black boxes. At microwave frequencies.
the reflection coefficients s11 and s22 can be plotted on Smith charts. I1 This relationship between reflection coefficient and impedance is the basis of the Smith Chart transmissionline calculator. Sparameters Sparameters and distributed models provide a means of measuring. and easily manipulated to determine matching networks for optimizing a circuit design. describing. They are used for the design of many products. .3 and 13 H Using SParameters (16) Test & Measurement Application Note 951 SParameter Techniques Notice that V1 − Z0 b1 Z − Z0 I1 s 11 = = = 1 Z 1 + Z0 V1 a1 + Z0 I1 Z1 = Z 0 (1 + s11 ) (1 − s11 ) (17) where Z 1 = V1 is the input impedance at port 1. Consequently. such as cellular telephones. and characterizing circuit elements when traditional lumpedequivalent circuit models cannot predict circuit behavior to the desired level of accuracy. converted directly to impedance.
Click over image to animate. Ferrell and J. Bell Lab’s Phillip H.B. but Smith suspected that a grid with orthogonal circles might be more practical. One version. using a transformation developed by coworkers E. McRae that accommodates all data values from zero to infinity. causing normalized resistance and reactance values greater than unity to become compressed towards the right side of the Smith Chart. The Smith Chart is used to plot reflectances. but the circular grid lines allow easy reading of the corresponding impedance. the rectangular grid lines of the impedance plane are transformed to circles and arcs on the Smith Chart. In 1937 he constructed the basic Smith Chart still used today. The transformation between the impedance plane and the Smith Chart is nonlinear. a polar coordinate form. Vertical lines of constant resistance on the impedance plane are transformed into circles on the Smith Chart.W. Showing transformations graphically To ease his RF design work. worked for all values of impedance components. As the animation shows. . Animation 1 Transformation between the impedance plane and the Smith Chart.3 14 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques Smith Chart Transformation The movie at the right animates the mapping between the complex impedance plane and the Smith Chart. Smith developed increasingly accurate and powerful graphical design aids. Horizontal lines of constant reactance on the impedance plane are transformed into arcs on the Smith Chart.
insertion gain gives by far the greater insight into the operation of the network. By comparison. both familiar quantities to engineers. namely that they are simply gains and reflection coefficients.3 15 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques Advantages of SParameters The previous equations show one of the important advantages of sparameters. For example. . the yparameter corresponding to insertion gain s21 is the ‘forward transadmittance’ y21 given by equation 3. some elements in modern highspeed digital circuits require characterization with distributed models and sparameters for accurate performance prediction. Clearly. A short pulse with steep edges has a signal spectrum with relatively high power levels at very high frequencies. some of the yparameters described earlier in this article are not so familiar. As a result. Digital pulses Digital pulses are comprised of highorder harmonic frequencies that determine the shape of the pulse.
a 2 2 = Power incident on the output of the network. b1 2 = Power reflected from the input port of the network. . = Power available from a Z0 source minus the power delivered to the input of the network. b1. b2 2 = Power reflected from the output port of the network. = Power reflected from the load.3 a1 2 H Using SParameters Test & Measurement Application Note 951 SParameter Techniques Another advantage of sparameters springs from the simple relationship between the variables a1. The design methods employed at that time combined distributed measurements and lumped circuit design. and b2. provided a powerful incentive for improved high frequency design methods during World War II. The Smith Chart met that need. = Power available from a source impedance Z 0 . = Power incident on the load. 16 Radar The development of radar. There was an urgent need for an efficient tool that could integrate measurement and design. and various power waves: = Power incident on the input of the network. a2 . which uses powerful signals at short wavelengths to detect small objects at long distances. = Power that would be delivered to a Z0 load.
3 s11 2 H Using SParameters Power reflected from the network input Power incident on the network input Power reflected from the network output Power incident on the network output Power delivered to a Z0 load Power available from Z0 source Test & Measurement Application Note 951 SParameter Techniques The previous four equations show that sparameters are simply related to power gain and mismatch loss. quantities which are often of more interest than the corresponding voltage functions: = = = s 22 2 s 21 2 = Transducer power gain with Z 0 load and source s12 2 = Reverse transducer power gain with Z0 load and source 17 .
The transfer parameters s12 and s21 are a measure of the complex insertion gain. As dimensionless expressions of gain and reflection. References J. but flow graphs make sparameter calculations extremely simple. 1960. November.4 18 H Network Calculations with Scattering Parameters Test & Measurement Application Note 951 SParameter Techniques Signal Flow Graphs Scattering parameters turn out to be particularly convenient in many network calculations. Of course. Vol. . Therefore. ‘Simplified Signal Flow Graph Analysis. Flow graphs will be used in the examples that follow. and the driving point parameters s11 and s22 are a measure of the input and output mismatch loss. 2. March. No. they are strongly recommended.’ IRE Transactions on Microwave Theory and Techniques. Kuhn. 11. the sparameters not only give a clear and meaningful physical interpretation of the network performance. Hunton.’ Microwave Journal. ‘Analysis of Microwave Measurement Techniques by Means of Signal Flow Graphs. it is not necessary to use signal flow graphs in order to use sparameters. MTT8. N. K. Vol. This is especially true for power and power gain calculations. but also form a natural set of parameters for use with signal flow graphs [See references here and also in Appendix A ]. 1963. 6. No.
s11 = b1 a1 a = Γ b = 0 2 L 2 . 2. Node an bS = represents the wave coming into the device from another device at port n. is the flow graph representation of the system of Fig. right. 3. The complex scattering coefficients are then represented as multipliers on ΓS = ZS ZS branches connecting the nodes within the network and in adjacent networks. This confirms the definition of s11: Figure 3 Flow graph for twoport network appearing in Figure 2. each port is represented by two nodes.Z0 + Z0 a1 s21 s11 b1 s12 Test & Measurement Application Note 951 SParameter Techniques In a signal flow graph. and node bn represents the wave leaving the device at port n.4 19 H Network Calculations with Scattering Parameters VS Z 0 ZS + Z 0 1 . Fig. b2 s22 ΓL = ZL  Z0 ZL + Z 0 a2 Figure 3 shows that if the load reflection coefficient ΓL is zero (ZL = Z0) there is only one path connecting b1 to a1 (flow graph rules prohibit signal flow against the forward direction of a branch arrow).
curves of constant standing wave ratio. The Nontouching Loop Rule The nontouching loop rule provides a simple method for writing the solution of any flow graph by inspection. .4 H Network Calculations with Scattering Parameters Test & Measurement Application Note 951 SParameter Techniques The simplification of network analysis by flow graphs results from the application of the “nontouching loop rule. In an article published in 1944. Smith described an improved version and showed how to use it with either impedance or admittance coordinates. for example.” This rule applies a generalized formula to determine the transfer function between any two nodes within a complex system. constant attenuation. The solution T (the ratio of the output variable to the input variable) is defined. T= ∑ Tk ∆ k k ∆ 20 ∆k = The value of ∆ not touching the kth forward path. . where: Tk = path gain of the kth forward path ∆ = 1 – Σ ( all individual loop gains) + Σ ( loop gain products of all possible combinations of 2 nontouching loops) – Σ ( loop gain products of all possible combinations of 3 nontouching loops) + . and constant reflection coefficient are all circles coaxial with the center of the diagram. Better Smith Charts On the copyrighted Smith Chart. More recent improvements include double Smith Charts for impedance matching and a scale for calculating phase distance. Refinements to the original form have enhanced its usefulness. The nontouching loop rule is explained below. .
and no combinations of three or more nontouching loops. only one combination of two nontouching loops. no node being encountered more than once. respectively. For example. and its gain is s21. There are two paths from bs to b1. The Smith Chart performs a highly useful translation between the distributed and lumped models and is used to predict circuit and system behavior. There are three individual loops. are similar to lumped models in many respects because they are related to the input impedance and output impedance. Therefore. the two sparameters typically represented using Smith Charts. Loop gain is the product of the branch multipliers around the loop. A loop is a path that originates and terminates on the same node. s 11 and s 22 . and a forward path is a path connecting the input node to the output node. Path gain is the product of all the branch multipliers along the path. the value of ∆ for this network is ( ) s11 s12 s21 s22 Sparameters & Smith Charts Test & Measurement Application Note 951 SParameter Techniques ∆ = 1 − (s11 ΓS + s 21 s12 ΓL ΓS + s 22 ΓL ) + s11 s 22 ΓL ΓS The transfer function from bs to b2 is therefore s 21 b2 = bs ∆ Invented in the 1960’s. where no node is encountered more than once.4 21 H Network Calculations with Scattering Parameters A path is a continuous succession of branches. . their path gains are s21s12ΓL and s11 respectively. Sparameters are a way to combine distributed design and distributed measurement. in Figure 3 there is only one forward path from bs to b2.
In the following equations. respectively. Transducer power gain: P Power delivered to the load = L Power available from the source PavS P L = P( incident on load) − P( reflected from load) = b 2 (1 − ΓL ) bS 2 2 2 ( 1 − ΓS 2 2 ) 2 2 b2 bS (1 − ΓS )(1 − ΓL ) . the load and source are described by their reflection coefficients ΓL and ΓS.4 GT = PavS = GT = 22 H Network Calculations with Scattering Parameters Test & Measurement Application Note 951 SParameter Techniques Transducer Power Gain Using scattering parameter flowgraphs and the nontouching loop rule. referenced to the real characteristic impedance Z0. it is easy to calculate the transducer power gain with an arbitrary load and source.
s 21 bS 1 − s 11ΓS − s 22 ΓL − s 21 s12 ΓL ΓS + s11ΓS s 22 ΓL s 21 = (1 − s 11ΓS )(1 − s 22 ΓL ) − s 21 s12 ΓL ΓS s 21 (1 − ΓS )(1 − ΓL ) (1 − s11ΓS )(1 − s 22 ΓL ) − s 21 s12 ΓL ΓS 2 2 2 2 (18) Two other parameters of interest are: 1) Input reflection coefficient with the output termination arbitrary and Zs = Z0.4 b2 = GT = 23 H Network Calculations with Scattering Parameters Test & Measurement Application Note 951 SParameter Techniques Using the nontouching loop rule. Because sparameters allow the interactions between such components to be simply predicted and calculated. including amplifiers. . they make it possible to maximize performance in areas such as power transfer. s11 = ′ b1 s (1 − s 22 ΓL ) + s 21 s12 ΓL = 11 1 − s 22 ΓL a1 s s Γ = s11 + 21 12 L 1 − s 22 ΓL (19) Obtaining maximum performance Sparameters are used to characterize RF and microwave components that must operate together. and antennas (and free space). and frequency response. transmission lines. directivity.
cabling.4 AV = 24 H Network Calculations with Scattering Parameters V1 = ( a 1 + b 1 ) Z 0 = Vi 1 + Vr 1 V2 = ( a 2 + b 2 ) Z 0 = Vi 2 + Vr 2 a 2 = ΓL b 2 b 1 = s11a 1 ′ Test & Measurement Application Note 951 SParameter Techniques 2) Voltage gain with arbitrary source and load impedances V2 V1 Waveguides A radar system delivers a large amount of energy from a microwave (µW) source to the transmitting antenna. which are other parameter sets used very often for specifying transistors at lower frequencies. y. driving point characteristics. etc. These hollow metal tube constructions conduct µW energy much like a plumbing system. and zparameters. In the design of waveguides. we can test for signal reflections and transmission quality with sparameters. The high field strengths cause short circuits in standard wires. . Also included are conversion formulas between sparameters and h. b (1 + ΓL ) s 21(1 + ΓL ) AV = 2 = a 1(1 + s11 ) (1 − s 22 ΓL )(1 + s11 ) ′ ′ (20) Appendix B contains formulas for calculating many oftenused network functions (power gains. and coax. so waveguides are used.) in terms of scattering parameters.
The HP 83017A microwave system amplifier achieves 0. To keep the discussion from becoming bogged down in extraneous details. the emphasis in these examples will be on sparameter design methods. The frequencydependent impedances (or dispersion) in this integrated circuit can not be modeled by lumpedequivalent circuit elements. but sparameters can accurately characterize the amplifier's response.5 GHz bandwidth by incorporating the HP TC702 GaAs MESFET TWA IC.5 25 H Test & Measurement Application Note 951 SParameter Techniques Amplifier Design Traveling wave amplifier Sparameters are extensively used for designing RF/µW circuits such as the HP TC702 distributed traveling wave amplifier (TWA) enlarged in the photograph below. and mathematical manipulations will be omitted wherever possible. .5–26. Using Scattering Parameters The remainder of this application note will show with several examples how sparameters are used in the design of transistor amplifiers and oscillators.
7 GHz for a 2N3478 transistor in the commonemitter configuration. These graphs are the results of sweptfrequency measurements made with the classic HP 8410A microwave network analyzer. Outermost circle on Smith Chart overlay corresponds to  s 11  = 1. the basic measurement techniques have not.6 26 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques Most design problems will begin with a tentative selection of a device and the measurement of its sparameters. Measurements made with a modern network analyzer are presented at the end of this section. While the measurement tools have changed over the past 30 years. The movement of s 11 with frequency is approximately along circles of constant resistance. . are a set of oscillograms showing complete sparameter data btween 100 MHz and 1. Figures 4a – 4e. which appear to the right and on the next two pages. indicative of series capacitance and inductance. They were originally published as part of the 1967 HP Journal article. 1700 MHz S11 100 MHz Figure 4a s 11 of a 2N3478 transistor measured with the classic HP 8410A network analyzer.
While the phase of s12 is relatively insensitive to the frequency.6 Figure 4b H Measurement of SParameters 100 MHz Test & Measurement Application Note 951 SParameter Techniques S12 10 dB/cm –30 dB S22 1700 MHz Displayed on the same scale as Figure 4a. the magnitude of s 12 increases about 6dB/octave. S12 50°/cm 100 MHz Figure 4c –110° 1700 MHz Magnitude and phase of s 12. s22 moves between the indicated frequencies roughly along circles of constant conductance. 27 . characteristic of a shunt RC equivalent circuit.
6 28 H Measurement of SParameters S21 10 dB/cm 0 dB Test & Measurement Application Note 951 SParameter Techniques S21 10 dB/cm S21 50°/cm 0 dB S21 50°/ cm + 20° + 90° 100 MHz 1700 MHz 100 MHz 1700 MHz Figure 4d – Magnitude and phase of s 21 . while the phase decreases linearly above 500 MHz. 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. The magnitude of s 21 decays with a slope of about 6 dB/octave.
4d to that of Fig.35 cm of line. The phase of s21. the unilateral figure of merit. The bottom curve u. a compensating linear phase shift was introduced.1 u 100 MHz 1 GHz . and then it rolls off at a slope of 6 dB/octave. The magnitude of s21 is essentially constant to 125 MHz. Figure 4f Top curve:  s 21  from Fig. the magnitude of s21 from Fig. By adjusting a calibrated line stretcher in the network analyzer. 4e resulted.4e required 3.162 1 . measured with a vector voltmeter. 4 is replotted on a logarithmic frequency scale.3162 . 4d is replotted on a logarithmic frequency scale. 4f. as seen in Fig. Magnitude (dB) 20 10 0 – 10 – 20 – 30 10 MHz s 21  10 3. calculated from sparameters. Data below 100 MHz was measured with an HP 8405A vector voltmeter. that is equivalent to a pure time delay of 112 picoseconds.03162 10 GHz Frequency Magnitude . along with additional data on s21 below 100 MHz. 4d. To go from the phase curve of Fig. varies linearly with frequency above about 500 MHz.6 29 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques In Fig. and the phase curve of Fig.
6 where 30 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques After removal of the constantdelay. The intuitive sense that designers gain from developing an understanding of these approximations can eliminate much frustration. the phase angle of s21 for this transistor (Fig. . s21 behaves like a single pole in the frequency domain. passing through +135° at 125 MHz. the 3 dB point of the magnitude curve. while also decreasing development costs.2 = 21 dB (21) Importance of simple approximations Using firstorder approximations such as equation 21 is an important step in circuit design.w j T0 s 21 = w 1+ j w 0 T0 = 112 ps w =p2 f w 0 = p 2 ¥ 125 MHz s 210 = 11. 4e) varies from 180° at dc to +90° at high frequencies. The acquired insight can save hours of time that otherwise might be wasted generating designs that cannot possibly be realized in the lab. component.s 210e . This expression is . The time delay T0 = 112 ps is due primarily to the transit time of minority carriers (electrons) across the base of this npn transistor. In other words. or linearphase. and it is possible to write a closed expression for it.
00 GHz. This plot shows s 11 and s 22 on a Smith Chart.6 31 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques The sparameters of an 2N3478 transistor shown in Figures 4a through 4f were measured with the classic HP 8410A network analyzer. The frequency index of this point is referenced in the other plots. The marker set at 47 MHz represents the 3 dB gain roll off point of s 21 . . Figures 5a through 5e represent the actual sparameters of this transistor between 0. the sparameters of an 2N3478 transistor are shown remeasured with a modern HP 8753 network analyzer. measured by an HP 8753 network analyzer. In Figures 5a through 5e. S11 S 22 Figure 5a Sparameters of 2N3478 transistor in commonemitter configuration.300 MHz and 1.
Both the phase angle and magnitude decrease dramatically as the frequency is increased. the magnitude increases rapidly at low frequencies.6 Magnitude (dB) Figure 5b H Measurement of SParameters S12 Phase (deg) Test & Measurement Application Note 951 SParameter Techniques S 21 Frequency A plot of the magnitude and phase of s 12 . 32 Figure 5c A polar plot of s 21 . The frequency marker shown is at the 3 dB point. While the phase of s 12 depends only weakly on the frequency. .
6 Magnitude (dB) 33 H Measurement of SParameters S 21 Test & Measurement Application Note 951 SParameter Techniques S 21 Frequency (Hz) Frequency (GHz) Figure 5e The phase angle. Removing this time delay allows the phase distortion to be viewed with much greater resolution. Phase (deg) . Figure 5d The magnitude of s 21 plotted on a log scale showing the 6 dB/octave rolloff above 75 MHz. in degrees. A time delay of 167 ps was deembedded from the measured data using the analyzer's electricaldelay feature to get a response with a singlepole transfer characteristic. of s 21 .
a compensating linear phase shift was introduced electronically in the network analyzer. RF through lightwave. or impedance characteristic of linear networks across a broad range of frequencies.6 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques In Fig. Vector Network Analyzers Most modern design projects. Measurements are made with a VNA.0 cm of line using a calibrated line stretcher. After removal of the constantdelay. transfer function. passing through +135° at 75 MHz. 5c is replotted on a logarithmic frequency scale. The phase angle of s21 as seen in Fig. component. and then rolls off at a slope of 6 dB/octave. 5e to the second phase curve required removing a pure time delay of 167 picoseconds. Thirty years ago this operation was accomplished by deembedding 5. s21 behaves like a single 34 pole in the frequency domain. the magnitude of s21 from Fig. 5e varies linearly with frequency above 500 MHz. In other words. an instrument that accurately measures the sparameters. the phase angle of s21 for this transistor (Fig. Today it’s performed by software in the network analyzer. . This established an accurate calibration for measuring the device. or linearphase. 5e) varies from 180° at dc to +90° at high frequencies. The magnitude of s21 is essentially constant to 75 MHz. To better characterize phase distortion. 5d. the 3 dB point of the magnitude curve. Vector Network Analyzer. To go from the first phase curve of Fig. These programs require complete sparameter data on each component. use sophisticated simulation software to model system performance from components through subsystems. resulting in the second phase curve of Figure 5e.
Today. computeraided engineering (CAE) tools now can be used iteratively to simulate and refine the design. and simulate and optimize the performance of the complete system implementation being developed.4 = 18. however.6 H Measurement of SParameters − s 210 e − jωT0 1+ j ω ω0 T0 = 167 ps where ω = 2π f ω 0 = 2 π × 75 MHz s 210 = 8.5 dB Test & Measurement Application Note 951 SParameter Techniques Since s21 behaves like a single pole in the frequency domain. Complete network characterization Vector Network Analyzers (VNA) are ideal for applications requiring complete network characterization. They use narrowband detection to achieve wide dynamic range and provide noisefree data. through the process of electronic design automation (EDA). repeated here. . VNAs are often combined with powerful computerbased electronic design automation (EDA) systems to both measure data. Subsequently. This expression is the same as equation 21. it is possible to write a closed expression for it. we have design technology undreamed of in 1967. singlepole approximation for s21 is an important step in circuit design. These tools combine accurate models 35 with performanceoptimization and yieldanalysis capabilities. s 21 = The time delay T0= 167 ps is due primarily to the transit time of minority carriers (electrons) across the base of this npn transistor. Using the firstorder. Removing this time delay using the electricaldelay feature of the vector network analyzer allows the phase distortion to be viewed with much greater resolution.
extract precision numerical data. and sometimes unavoidable. Device differences – Semiconductor manufacturing processes evolve over time. especially in characteristics not guaranteed on the datasheet.6 36 H Measurement of SParameters Test & Measurement Application Note 951 SParameter Techniques Explanation of Measurement Discrepancies You may have noticed a difference between the measured and caclulated data from the 1967 HP Journal article and the data obtained for this updated application note. Modern network anaylzers contain sophisticated automated techniques that enhance measurement processing capabilities and reduce operator errors. or display electronic markers. or highresolution graphics to perform calibration. Calibration techniques used in 1967 were procedurally and mathematically simpler than those used today. Two major sources account for these differences: Measurement techniques – Early network analyzers did not have onboard computers. an HPIB standard. Nevertheless. performance differences. successive generations of parts like the 2N3478 can exhibit unintentional. Device engineers attempt to produce identical transistors with different processes. . Both sets of data are fundamentally correct.
4—there is a possibility that it can be neglected.7 37 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Suppose now that this 2N3478 transistor is to be used in a simple amplifier. Since reverse gain s12 for this transistor is quite small—50 dB smaller than forward gain s21. That is. and optimized for power gain at 300 MHz by means of lossless input and output matching networks. the design problem will be much simpler. using the formula given in Appendix B. according to Fig. u = s11 s12 s 21 s 22 (1 − s11 )(1 − s 22 ) 2 2 (22) . because setting s12 equal to zero will make the design equations much less complicated. In determining how much error will be introduced by assuming s12 = 0. If this is so. the first step is to calculate the unilateral figure of merit u. If the signal flow in the reverse direction is much smaller than the flow in the forward direction. Unilateral figure of merit All twoport models are bilateral. operating between a 50 Ω source and a 50 Ω load. it is possible to make the simplification that the reverse flow is zero. so both the forward and reverse signal flow must be considered. 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.
often exceptional. for reasons other than it simplifies amplifier design. performance characteristics. Incidentally. 4f.03. Discrete transistors remain the best choice for many applications. A small feedback factor means that the input characteristics of the completed amplifier will be independent of the load. calculated from the measured parameters. is an important and desirable property for a transistor to have. This is small enough to justify the assumption that s12 = 0. isolation of source and load is an important consideration. the maximum value of u is about 0. discrete devices are still available. manufactured and selected for specific. Highfrequency transistors Discrete transistors were the mainstay of highfrequency system design in 1967 when Dick Anderson wrote the article on which this application note is based. Now if GTu is the transducer power gain with s12 = 0 and GT is the actual transducer power gain.25 dB at 100 MHz. a small reverse gain. such as sensitive firststage amplifiers in satellite TV receivers. Thirty years later. so the maximum error in this case turns out to be about +0.7 38 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques A plot of u as a function of frequency. 4f. appears in Fig. s12. In most amplifiers. or feedback factor. and the output will be independent of the source impedance. . the maximum error introduced by using GTu instead of GT is given by the following relationship: 1 (1 + u )2 < GT 1 < G Tu (1 − u )2 (23) From Fig.
7 39 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Returning now to the 300MHz amplifier design. Simple seriescapacitor. is G Tu = s 21 (1 − ΓS )(1 − ΓL ) 1 − s11ΓS 2 2 2 2 1 − s 22 ΓL 2 (24) When  s11 and  s22 are both less than one. The next step in the design is to synthesize matching networks that will transform the 50 Ω load and source impedances to the impedances corresponding to reflection coefficients of s*11 and s*22. . called Traveling Wave Tubes (TWTs). as they are in this case. Since this is to be a singlefrequency amplifier. but will also provide a handy means of biasing the transistor—via the inductor—and of isolating the dc bias from the load and the source. respectively. the unilateral expression for transducer power gain. which are best characterized by sparameters. the matching networks need not be complicated. obtained either by setting s12 = 0 in equation 18 or looking in Appendix B. Satellite Broadcast Signals Satellites provide broad geographical signal coverage over a wide band of frequencies by using high power vacuum tubes. shuntinductor networks will not only do the job. maximum GTu occurs for ΓS = s*11 and ΓL = s*22 (Appendix B).
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. which is shown on the next page. http://www. You will find an interactive Impedance Matching Model at the HP Test & Measurement website listed below. Another matching network transforms the 50 Ω source impedance to s*22. To eliminate reflections in an amplifier. a combination of constantresistance and constantconductance circles is found. 6. s22. leading from the center of the chart. The circles on the Smith Chart are constantresistance circles. its length equal to the magnitude of the reflection coefficient being plotted. 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. Next. representing 50 Ω.7 40 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Values of L and C to be used in the matching networks for the 300MHz amplifier are determined using the Smith Chart of Fig. points corresponding to s11. increasing series capacitive reactance moves an impedance point counterclockwise along these circles. First. to s*11 and s*22.com/go/tminteractive .hp. s*11. one matching network is carefully designed to transform the 50 Ω load impedance to s*11. Each point represents the tip of a vector leading away from the center of the chart. and s*22 at 300 MHz are plotted. and its angle equal to the phase of the coefficient.
6. but they are on another Smith Chart. s* 11 L1 s* 22 C1 C2 s11 L2 s22 . 6. as shown by the solid line in Fig. The circles to be used are those passing through s*11 and s*22. These circles are like the constantresistance circles. one that is just the reverse of the chart shown in Fig. The constantconductance circles for shunt L all pass through the leftmost point of the chart rather than the rightmost point. as shown by the dashed lines in Fig. Figure 6 Smith Chart for 300MHz amplifier design example. 6. Increasing shunt inductive susceptance moves impedance points clockwise along constantconductance circles.7 41 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques In this case. the circle to be used for finding series C is the one passing through the center of the chart.
42)(50) 50 ( 0.7 42 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Once these circles have been located. normalized and unnormalized.01) = 25 pF = 26 nH .3 x 109 ) 1 C2 = = 3 pF 9 )( 3.3 x 10 C1 = L1 = 1 2p 2p ( 0. C1 2N3478 L1 L2 C2 50W Calculations: 50 = 156 W X = L 2 0.3 x 109 )(0. 6.5 )( 50 ) 2 p ( 0. Which arc corresponds to which element is indicated in Fig. are shown in Fig. Each element’s reactance or susceptance is the difference between the scale readings at the two end points of a circular arc.3 x 109 )(1. the normalized values of L and C needed for the matching networks are calculated from readings taken from the reactance and susceptance scales of the Smith Charts. 7.32 156 L2 = = 83 nH 2 p ( 0. The final network and the element values. 50W Figure 7 A 300MHz amplifier with matching networks for maximum power gain.
S* 22 C = 1000 pF L = 10000 nH Animation 2 Impedance matching using the Smith Chart for the matching network shown at the left. To reach s*22. 2N3478 L 50 Ω the shunt inductance is adjusted until the impedance point reaches s*22. This matching is achieved using the LC circuit shown at the right. http://www. As previously described. Click over the chart to start animation.hp. the s*22 parameter of the transistor must be matched to the load impedance.7 43 H NarrowBand Amplifier Design Test & Measurement Application Note 951 SParameter Techniques The animation to the right demonstrates how to use a Smith Chart to design a matching network between a transistor output and a resistive load. The capacitance is adjusted until it intersects the constant conductance circle on which s*22 is sitting. to maximize the power delivered to the load. Starting from the 50 Ω load. the series capacitance is varied to move the impedance point along the circle of constant 50 Ω resistance on the Smith Chart.com/go/tminteractive . 50 Ω in this 300MHz amplifier example. Varying the shunt inductance then moves the impedance point along this constant C conductance circle as indicated by the admittance Smith Chart.
or zparameters (for shunt or series feedback. negative feedback. it is usually convenient to convert to y. We will use the second method. GTu = G0 G1G2 G0 = s21 G1 = 2 2 2 1 − Γs 1 − s11Γs 1 − ΓL 2 G2 = Equation 24 for the unilateral transducer power gain can be factored into three parts. that is. is a matter of surrounding a transistor with external elements in order to compensate for the variation of forward gain. When feedback is used.  s21 with frequency. This can be done in either of two ways— first.8 44 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Designing a broadband amplifier. as shown to the right: 1 − s22ΓL 2 . one which has nearly constant gain over a prescribed frequency range. selective mismatching of the input and output circuitry. respectively) using the conversion equations given in Appendix B and a digital computer. or second.
Suppose that the 2N3478 transistor whose sparameters are given in Fig. leave the gain the same at 450 MHz. the gain contributions of G1 and G2 are varied to compensate for the variations of G0= s212 with frequency. . According to Figure 4f. and increase the gain by 4 dB at 700 MHz. 4 is to be used in a broadband amplifier that will operate from 300 MHz to 700 MHz. The amplifier is to be driven from a 50 Ω source and is to drive a 50 Ω load. source and load matching networks must be found that will decrease the gain by 3 dB at 300 MHz.8 s 21 45 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques When a broadband amplifier is designed by selective mismatching. 2 = 13 dB at 300 MHz = 10 dB at 450 MHz = 6 dB at 700 MHz To realize an amplifier with a constant gain of 10 dB.
Fig.. only a loadmatching network will be designed.8 46 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Although in the general case both a source matching network and a load matching network would be designed. 8a. G1 for Γs = s*11) for this transistor is less than 1 dB over the frequencies of interest. . for this example. Consequently. G1max (i. Procedures for designing sourcematching networks are identical to those used for designing loadmatching networks. s* 22 700 450 300 Figure 8a A plot of s* 22 over the frequency range from 300 MHz to 700 MHz.e. The first step in the design of the loadmatching network is to plot s*22 over the required frequency range on the Smith Chart. which means there is little to be gained by matching the source.
8 where 47 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Next. its center is on a line between the center of the Smith Chart and the point representing s*22 at that frequency. which also appear in Appendix B: Figure 8 Constantgain circles. r2 = 1 − g 2 s 22 1 − s 22 (1 − g 2 ) G2 G 2 max = G 2 (1 − s 22 ) G2 = 0 dB at 450 MHz 2 2 G2 = +4 dB at 700 MHz r2 ρ 2 g2 = r2 r2 ρ ρ 2 2 The radius of the constantgain circle is: ρ2 = 1 − g 2 (1 − s 22 ) 1 − s 22 (1 − g 2 ) 2 2 G2 = – 3 dB at 300 MHz . 8b. a set of constantgain circles is drawn. As shown in Fig. The distance from the center of the Smith Chart to the center of the constant gain circle is given by the following equations. each circle is drawn for a single frequency.
278)1. as shown in Fig. and to some point on the +4 dB circle at 700 MHz. Lshunt ZL = 50W . G2max for all three circles is (0. 2N3478 Figure 9 A combination of shunt and series inductances is a suitable matching network for the broadband amplifier. The required matching network must transform the center of the Smith Chart. 8b.85 over the frequency range [see Figure 4(b)]. one for G2 = 0 dB at 450 MHz. representing 50 W . The three constantgain circles are indicated in Fig. Since s22 for this transistor is constant at 0. three circles will be drawn. to some point on the 0 dB circle at 450 MHz. or 5. one in shunt and one in 50W series.8 48 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques For this example. One satisfactory Lseries solution is a combination of two inductors. and one for G2 = +4 dB at 700 MHz. 9. one for G2 = 3 dB at 300 MHz. There are undoubtedly many networks that will do this. to some point on the 3 dB circle at 300 MHz.6 dB.
(This rule applies to inductors.8 49 H Broadband Amplifier Design Test & Measurement Application Note 951 SParameter Techniques Shunt and series elements move impedance points Constant resistance Matching paths on the Smith Chart along constantconductance circlesTransformation for shunt and and constantresistance circles. the narrowband design example. as explained in due to Lseries series L. capacitors would behave in the opposite way. The series inductor transforms Hz 00 M the combination of the 50 Ω load and the Hz 50 M shunt inductance along circles of constant 00 resistance and varying inductive reactance. As shown in Fig.) Constant conductance circleTransformation due to Lshunt 7 . and – the susceptance component decreases with frequency and the reactance component increases with frequency. Figure 10a 3 4 Optimizing the values of shunt and series L is an iterative process with two goals: – the transformed load reflection terminates on the right gain circle at each frequency. 10a. the shunt inductance transforms the ΓL locus 50 Ω load alng a circle of constant conductance Hz and varying (with frequency) inductive M susceptance.
0.2 Z0 ( 3. Then the element values are calculated.4 nH 50Ω Inductance Calculations: From 700 MHz data. with unnormalized element values. the same as they were for the narrowband design. 20.8 50 H Broadband Amplifier Design 36.64 .3 ) Figure 10b Broadband amplifier with constant gain of 10 dB from 300 MHz to 700 MHz. . Figure 10b is a schematic diagram of the completed broadband amplifier. j ω L series = j ( 3.4 nH ZS = 50Ω 2N3478 Test & Measurement Application Note 951 SParameter Techniques Once appropriate constantconductance and constantresistance circles have been found. the reactances and susceptances of the elements can be read directly from the Smith Chart. Z0 = .44 ) = j 3.4 nH (1.j 1.3 ) (2 π ) ( 0.4 nH 2 π ( 0.3 j ω L shunt L shunt = 50 = 20.7 ) From 300 MHz data.2 ) ( 50 ) L series = nH = 36.
11) can be realized with a circulator—a nonreciprocal threeport device— and a negativeresistance device. reflection amplifiers and oscillators. The circulator is used to separate the incident (input) wave from the larger wave reflected by the negativeresistance device. an amplifier with infinite gain can be built by selecting a negativeresistance device whose input impedance has a real part equal to Z0 and an imaginary part equal to zero (the imaginary part can be set equal to zero by tuning if necessary). Input Two port with s'11 > 1 (Real part of input impedance is negative) Circulator Output Figure 11 Reflection amplifier consists of circulator and transistor with negative input resistance. if the circulator is perfect and has a positive real characteristic impedance Z0. . the corresponding input reflection coefficient (Equation 17) is greater than one. A reflection amplifier (Fig. Theoretically.9 51 H Test & Measurement Application Note 951 SParameter Techniques Stability Considerations Design of Reflection Amplifiers and Oscillators When the real part of the input impedance of a network is negative. and the network can be used as the basis for two important types of circuits.
as defined in Equation 19. On the other hand. if 1 s ′11 < Γs the device is unconditionally stable and will not oscillate. are not supposed to oscillate. CAE tools are part of the total engineering solution. whatever 52 the phase angle of Γs might be. Referring again to the flow graph of Figure 3. of course. the ratio of the reflected voltage wave b1 to the input voltage wave bs is s ′11 b1 = bs 1 − Γs s11 ′ where s′11 is the input reflection coefficient with Γs= 0 ( Z2= Z0 ) and an arbitrary load impedance ZL . There is a convenient criterion based upon scattering parameters for determining whether a device is stable or potentially unstable with given source and load impedances. and is easier to follow than manual methods using graphical design aids.9 H Test & Measurement Application Note 951 SParameter Techniques Stability Considerations Amplifiers. Γs s′11 = 1 (25) the circuit is unstable and it will oscillate at that frequency. . If at some frequency 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. produces accurate results. whether they are reflection amplifiers or some other kind. The CAE approach is faster.
12. Q 150 Ω To see how these principles of stability are applied in design problems. Figure 12 The transistor oscillator is designed by choosing a tank circuit such that Γs s′11 = 1 ΓT1 Parallel Tank Circuit 200 Ω ZL. ΓL s s Γ s ′11 = s 11 + 12 21 L 1 − s 22 ΓL Gain s' 11 Element 15 Ω ω0. ΓT1 or ΓT2. and the ‘source’ reflection coefficient Γs is one of the two tankcircuit refection coefficients. From equation 19.9 53 H Test & Measurement Application Note 951 SParameter Techniques Stability Considerations ω0. consider the transistor oscillator design illustrated in Fig. In this case the input reflection coefficient s′11 is the reflection coefficient looking into the collector circuit. Q ΓT2 Series Tank Circuit .
1/s′11 can be measured directly by reversing the reference and test channel connections between the reflection test unit and the harmonic frequency converter. – Choose a load reflection coefficient ΓL that makes s′11 greater than unity. s′11 and Γs must be adjusted so that they satisfy equation 25.) . it may also take an external feedback element that increases s12 s21 to make s′11 greater than one. (If the network analyzer is being used to measured the sparameters of the transistor. There are four steps in the design procedure: – Measure the four scattering parameters of the transistor as functions of frequency.9 54 H Test & Measurement Application Note 951 SParameter Techniques Stability Considerations To make the transistor oscillate.) – Connect either the series or the parallel tank circuit to the collector circuit and tune it so that ΓT1 or ΓT2 is large enough to satisfy equation 25. (The tank circuit reflection coefficient plays the role of Γs in this equation. The polar display with a Smith Chart overlay then gives the desired plot immediately. In general. – Plot 1/s′11 on a Smith Chart.
5 to 2. this transistor would oscillate from 1. Reflection coefficients G T1 and G T2 are also plotted as functions of the resonant frequencies of the two tank circuits.5 and 2. Load impedance ZL is 200 W .0 GHz s'11 w 0 w G T2 0 G T1 Figure 13 The transistor will oscillate in the shaded area between 1.5 GHz 2.7 GHz with a paralleltuned circuit. Thus. Oscillations occur when the locus of G T1 or G T2 passes through the shaded region. which means that G L referred to 50 W is 0. 1967 and 1997 1.9 55 H Test & Measurement Application Note 951 SParameter Techniques Stability Considerations 2.0 to 2.7 GHz Figure 13 shows a Smith Chart plot of 1/s¢ 11 for a high frequency transistor in the commonbase configuration. . and from 2.5 GHz with a seriestuned circuit and between 2.7 GHz with a parallel tuned circuit.5 GHz with a series tuned circuit.6.5 GHz 2.0 and 2. — Dick Anderson.
– K. Vol. March. No. ‘Transistors and Active Circuits. No.G. Vol. (No sparameters. 1960.’ Proc. No. Kuhn. No.F. 1963. ‘Analysis of Microwave Measurement Techniques by Means of Signal Flow Graphs.’ Microwave Journal. Hunton. the following papers and books were listed in the 1967 HP Journal article as sources for information on sparameter design procedures and flow graphs. 6. – J. . Vol. Kurokawa. Current references are also mentioned. 1965. 1961. 7. 2. K. – D. 1961.8. MTT13. Linvill and J. November.’ IRE Transactions on Microwave Theory and Techniques. – J.’ IEEE Transactions on Microwave Theory and Techniques.) – N. IRE. 2. Vol. March. MTT. 49. ‘Simplified Signal Flow Graph Analysis.C Youla.A 56 H Additional Reading on SParameters Test & Measurement Application Note 951 SParameter Techniques In addition to previous references listed earlier and repeated again here. but good treatment of Smith Chart design methods. ‘On Scattering Matrices Normalized to Complex Port Numbers. Gibbons. 11.’ McGrawHill. July. ‘Power Waves and the Scattering Matrix.
– G. Circuit and Component Analysis. This text presents a unified treatment of the analysis and design of microwave transistor amplifiers using scattering parameters techniques: – G. Dec. 2 appendices. 5. 39. 1966. ISBN1884932398. ‘How to Use SParameters for Transistor Circuit Design. the following book. Sept. 1966. and problem sets. ‘Microwave Transistor Amplifiers: Analysis and Design.. No. ISBN 0135816467. No. Among many modern reference sources on the subject.’ Electronics. and educational Smith Chart software is sold over the internet. 240 pp. is definitely a classic: – Smith.’ Noble Publishing Classic Series. H. 237 pp. Keep up to date This book by P. Weinert. Phillip H. We also mention a useful textbook containing 5 chapters. Tucker.’ Prentice Hall. Georgia. 1984. 18. 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 Journal. Fredricks. ‘Electronic Applications of the Smith Chart in Waveguide.’ EEE Vol. 1995..A 57 H Additional Reading on SParameters Test & Measurement Application Note 951 SParameter Techniques – F. 14. ‘Scattering Parameters Speed Design of HighFrequency Transistor Circuits. . Gonzalez. 12. Vol. first published in 1969.
B 58 H Scattering Parameter Relationships + V1 a1 b1 TWO .PORT NETWORK Test & Measurement Application Note 951 SParameter Techniques a2 b2 + V2 b1 = s11 a1 + s12 a 2 b2 = s 21 a1 + s 22 a 2 – – Input reflection coefficient with arbitrary ZL s s Γ s11 = s11 + 12 21 L ′ 1 − s 22 ΓL Output reflection coefficient with arbitrary Zs s s Γ s22 = s22 + 12 21 S ′ 1 − s11 ΓS V2 s21(1 + ΓL ) = V1 (1 − s22 ΓL )(1 + s11) ′ Voltage gain with arbitrary ZL and ZS AV = .
B Power Gain G= GA = GT = 59 H Scattering Parameter Relationships = 2 2 s 21 1 − ΓL Test & Measurement Application Note 951 SParameter Techniques Power delivered to load Power input to network 2 2 2 2 1 − s11 + ΓL s 22 − D − 2Re( ΓL N) Available Power Gain Power available from network Power available from source = 2 2 s 21 1 − ΓS 2 2 2 2 1 − s 22 + ΓS s11 − D − 2Re( ΓS M) 2 s 21 2 1 − ΓS 2 1 − ΓL Transducer Power Gain Power delivered to load Power available from source = (1 − s11ΓS ) (1 − s 22 ΓL ) − s12 s 21 ΓL ΓS ) 2 .
2 .B G Tu = Gu = 60 H Scattering Parameter Relationships ( ) Test & Measurement Application Note 951 SParameter Techniques Unilateral Transducer Power Gain s12 = 0 2 2 2 s 21 1 − ΓS 1 − ΓL G 0 = s 21 G1 = 2 2 2 1 − s 11 ΓS 2 1 − s 22 ΓL 2 = G 0 G1 G 2 1 − ΓS 1 − s 11 ΓS 1 − ΓL 2 Maximum Unilateral Transducer Power Gain when s 11 < 1 * and s 22 < 1. Maximum obtained for ΓS = s11 and ΓL = s * 22 G2 = G i max = 1 − s 22 ΓL 1 1 − s ii 2 2 s 21 2 2 2 1 − s11 1 − s 22 = G 0 G1 max G 2 max i = 1.
B ri = ρi = gi = 61 H Scattering Parameter Relationships Test & Measurement Application Note 951 SParameter Techniques Constant Gain Circles (Unilateral case: s12 = 0) • 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: Constant Gain Circle g i sii 1 − sii (1 − g i ) 2 • radius of circle: 1 − g i (1 − sii ) 1 − s ii (1 − g i ) 2 2 ri S* ii ρ where i = 1. 2. and i Gi Gi max = Gi (1 − sii ) 2 .
so both the forward and reverse signal flow must be considered. (1 + u 2 ) 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. it’s possible to make the simplification that the reverse flow is zero.B u= 1 62 H Scattering Parameter Relationships s11 s 22 s12 s 21 Test & Measurement Application Note 951 SParameter Techniques Unilateral Figure of Merit 2 ( 1 − s11 )( 1 − s 22 2 ) Error Limits on Unilateral Gain Calculations < GT G Tu < 1 (1 − u 2 ) Unilateral figure of merit All twoport models are bilateral. . If the signal flow in the reverse direction is much smaller than the flow in the forward direction.
B C = K −1 K= 63 H Scattering Parameter Relationships Test & Measurement Application Note 951 SParameter Techniques Conditions for Absolute Stability : No passive source or load will cause a network to oscillate.port network can be simultaneously matched with a real source and load : K > 1 or C < 1 where C = Linvill C Factor and c. s 22 < 1 s12 s 21 − M * b. s 22 s12 s 21 − N* 2 − D 2 >1 1 + D − s11 2 2 − s 22 2 D = s11 s 22 − s12 s 21 M = s11 − Ds * 22 * N = s 22 − Ds11 2 s12 s 21 . s11 2 −D 2 >1 Condition that a two. if conditions a. and c are all satisfied. s11 < 1. b. a .
plus sign when B1 is negative. G A max = where K = C −1 s 21 2 K ± K − 1 s12 .B 64 H Scattering Parameter Relationships Test & Measurement Application Note 951 SParameter Techniques Source and load for Simultaneous Match 2 B1 ± B1 − 4 M 2 ΓmS = M * 2 2M 2 B2 ± B2 − 4 N 2 ΓmL = N* 2 2N (Use minus sign 2 2 2 B1 = 1 + s11 − s 22 − D when B1 is positive.) B 2 = 1 + s 22 2 − s11 2 − D 2 Maximum Available Power Gain If K > 1.
B
65
H
Scattering Parameter Relationships
Test & Measurement Application Note 951 SParameter Techniques
sparameters in terms of zparameters
( z 11 − 1)( z 22 + 1) − z 12 z 21 ( z 11 + 1)( z 22 + 1) − z 12 z 21 2z 12 ( z 11 + 1)( z 22 + 1) − z 12 z 21 2z 21 ( z 11 + 1)( z 22 + 1) − z 12 z 21 ( z 11 + 1)( z 22 − 1) − z 12 z 21 ( z 11 + 1)( z 22 + 1) − z 12 z 21
zparameters in terms of sparameters
z 11 = (1 + s 11 )(1 − s 22 ) + s 12 s 21 (1 − s 11 )(1 − s 22 ) − s 12 s 21 2s 12 (1 − s 11 )(1 − s 22 ) − s 12 s 21 2s 21 (1 − s 11 )(1 − s 22 ) − s 12 s 21 (1 + s 22 )(1 − s 11 ) + s 12 s 21 (1 − s 11 )(1 − s 22 ) − s 12 s 21
s 11 =
s 12 =
z 12 =
s 21 =
z 21 =
s 22 =
z 22 =
B
s 11 = s 12 = s 21 = s 22 = 66
H
Scattering Parameter Relationships
Test & Measurement Application Note 951 SParameter Techniques
sparameters in terms of yparameters
(1 − y 11 )(1 + y 22 ) + y 12 y 21 (1 + y 11 )(1 + y 22 ) − y 12 y 21 −2y 12 (1 + y 11 )(1 + y 22 ) − y 12 y 21 −2y 21 (1 + y 11 )(1 + y 22 ) − y 12 y 21 (1 + y 11 )(1 − y 22 ) + y 12 y 21 (1 + y 11 )(1 + y 22 ) − y 12 y 21
yparameters in terms of sparameters
y 11 = (1 + s 22 )(1 − s 11 ) + s 12 s 21 (1 + s 11 )(1 + s 22 ) − s 12 s 21 −2s 12 (1 + s 11 )(1 + s 22 ) − s 12 s 21 −2s 21 (1 + s 11 )(1 + s 22 ) − s 12 s 21 (1 + s 11 )(1 − s 22 ) + s 12 s 21 (1 + s 11 )(1 + s 22 ) − s 12 s 21
y 12 =
y 21 =
y 22 =
B
67
H
Scattering Parameter Relationships
Test & Measurement Application Note 951 SParameter Techniques
sparameters in terms of hparameters
( h 11 − 1)( h 22 + 1) − h 12 h 21 ( h 11 + 1)( h 22 + 1) − h 12 h 21 2h 12 ( h 11 + 1)( h 22 + 1) − h 12 h 21 −2h 21 ( h 11 + 1)( h 22 + 1) − h 12 h 21 (1 + h 11 )(1 − h 22 ) + h 12 h 21 ( h 11 + 1)( h 22 + 1) − h 12 h 21
hparameters in terms of sparameters
h 11 = (1 + s 11 )(1 + s 22 ) − s 12 s 21 (1 − s 11 )(1 + s 22 ) + s 12 s 21 2s 12 (1 − s 11 )(1 + s 22 ) + s 12 s 21 −2s 21 (1 − s 11 )(1 + s 22 ) + s 12 s 21 (1 − s 22 )(1 − s 11 ) − s 12 s 21 (1 − s 11 )(1 + s 22 ) + s 12 s 21
s 11 =
s 12 =
h 12 =
s 21 =
h 21 =
s 22 =
h 22 =
z′ are the actual parameters. If h′. y′. In addition. Z 0 . and z–parameters listed in previous tables are all normalized to Z0. impedance and admittance values can be plotted on the same chart. y–.B 68 H Scattering Parameter Relationships Test & Measurement Application Note 951 SParameter Techniques The h–. Normalizing the scattering parameters makes the Smith Chart readily applicable to transmission lines of any impedance. . then: y 11 = y 11 / Z 0 ′ y 12 = y 12 / Z 0 ′ y ′ = y 21 / Z 0 21 y ′ = y 22 / Z 0 22 h 11 = h 11Z 0 ′ h 12 = h 12 ′ h ′ = h 21 21 h ′ = h 22 / Z 0 22 z 11 = z 11Z 0 ′ z 12 = z 12 Z 0 ′ z ′ = z 21Z 0 21 z ′ = z 22 Z 0 22 Parameter Normalization The various scattering parameters are all normalized by the reference impedance. This impedance is usually the characteristic impedance of the transmission line in which the network of interest is embedded.
They enable broadband solutions. and provide answers to “What if?” questions.C 69 H CAE tools for HighFrequency Design Test & Measurement Application Note 951 SParameter Techniques Electronic Design Automation (EDA) The Software Revolution In the 30 years that have elapsed since the publication of Dick Anderson's article. These computer software programs run on UNIX workstations and PCs. computer aided engineering (CAE) tools have been developed for the highfrequency design methods that were traditionally implemented using pencil and paper. CAE tools also speed the analysis of a wide range of RF and microwave devices. and do much more than merely assist in computationintensive design tasks. while lowering costs. or system behavior. circuits. . s12 = 0) and can accurately simulate actual device. circuit. Modern CAE tools for highfrequency design eliminate the need for simplifying assumptions (such as. offer optimization and yieldanalysis capabilities. and systems. for a shorter time to market.
. Circuit Envelope Simulation The waveform above typifies the modulated and transient signals that can be efficiently analyzed using circuit envelope simulator software. and can be readily optimized. A yieldanalysis feature allows the selection of components in matching networks for the best production yield. saving costs. Matching circuits are easily determined. 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. Largesignal simulation — This powerful analysis tool includes the harmonic balance implementation. and radar circuits and subsystems.C 70 H CAE tools for HighFrequency Design Test & Measurement Application Note 951 SParameter Techniques The CAE tools of interest to RF and microwave designers include those summarized below: Smallsignal (sparameter) simulation — Smallsignal analysis CAE tools simulate response over a range of frequencies. useful for oscillator design and many other problems. Circuit Envelope simulation — Efficiently analyzes circuits and feedback loops in the presence of modulated or transient highfrequency signals. This is the case in many communications. saving time. so actual implementations perform more closely to design parameters.
Visit the HP EEsof website for the latest in CAE news. can analyze and optimize complicated system topologies. or zparameters of arbitrarily shaped. multiport structures. displayed using HP HighFrequency Structure Simulator 3D electromagnetic visualization software. Modeling systems — Hardware and software are combined to extract parameters needed for accurate active device modeling.hp. handle complex waveforms. multilayer planar structures such as striplines. shows beam shapes in both azimuth and elevation in a single plot. threedimensional. and solutions. High Flying Software The pattern of a horn antenna. y.C 71 H CAE tools for HighFrequency Design Test & Measurement Application Note 951 SParameter Techniques Timedomain analysis — A CAE tool that is especially useful for simulating the response of digital systems at high clock rates. System analysis — System and boardlevel simulators offer discretetime and frequencydomain capabilities. products.com/go/hpeesof . 3D electromagnetic analysis — This CAE tool accurately computes the sparameters for passive. software. http://www. Planar electromagnetic analysis — This simulator accurately computes the s. and perform physical layout design.
Productivity is enhanced with pass/fail testing. save/ recall of test configurations to internal memory or a builtin floppy disk drive. tuned receiver. as well as group delay. and highaccuracy noncoaxial and onwafer measurements. plus the absolute output power of microwave components.com/go/tmdatasheets . and test sequencing for automation. Test & Measurement Application Note 951 SParameter Techniques Features of HP 8720D VNA’s • Builtin synthesized source with 1 Hz resolution • Allows measuring all four sparameters with a single connection • Continuous updates for twoport error correction http://www. direct printer/ plotter output of results. all at an attractive cost. frequency offset mixer testing. and Sparameter test set in a single instrument. Use them to quickly and accurately measure magnitude and phase of all four sparameters. Options allow highpower tests.H Relevant Products HP 8720D Series Vector Network Analyzers Vector network analyzers with builtin synthesized sources cover 50 MHz to up to 40 GHz The HP 8720D vector network analyzer family characterizes RF and microwave components from 50 MHz up to 40 GHz. advanced marker functions.hp. The devices have the performance and flexibility to solve difficult measurement problems and cut test times. They combine a fast synthesized source.
hp. Effective directivity and source match can be improved to as much as 60 dB. (Adobe Acrobat users with the Weblink plugin may click directly on the URL below. channels. The HP 85161B software leads the operator one step at a time. so a test device can be adjusted while it’s being measured. Also available are fully integrated systems. phase.01 degree.01 ns resolution • Optional time domain and pulsed RF measurement capability http://www. or group delay format on a large. 0.com/go/tmdatasheets . Visit the HP Test & Measurement website and find more than 1.H Relevant Products HP 8510C Microwave Network Analyzers Microwave network analyzers with coverage up to 110 GHz The HP 8510C microwave vector network analyzer family provide complete solutions for characterizing the linear behavior of either active or passive networks over the 45 MHz to 110 GHz range.000 up to date product datasheets. tested and verified prior to shipment. an sparameter test set. yet identical. The HP 8510C displays measurement results in log/linear magnitude. The impact of systematic errors is removed by virtually “realtime” error correction. color CRT with two independent. and a compatible RF source.) Test & Measurement Application Note 951 SParameter Techniques Features of the HP 8510C • 45 MHz to 110 GHz frequency range • Realtime errorcorrected measurements • 60 dB effective directivity and source match • Up to 100 dB dynamic range • 0. from setup and calibration to hardcopy results.001 dB. 0. A complete system consists of the HP 8510C network analyzer.
and sensitive receiver are integrated into this compact instrument. synthesized source • Optional timedomain and sweptharmonic measurements • Up to 110 dB dynamic range • Superb accuracy. Test & Measurement Application Note 951 SParameter Techniques Features of the HP 8753D • Builtin sparameter test set. component. offers up to 110 dB dynamic range. The analyzer’s fast CPU clock rate. sparameter test set. A 1Hz resolution swept synthesized source.H Relevant Products HP 8753D RF Network Analyzer RF network analyzer with integrated sparameter test set performs characterizations from 30 kHz to 6 GHz The HP 8753D RF vector network analyzer will simplify and speed your device.com/go/tmdatasheets . which is simple to set up and use in the lab or on the production line. or network measurements in the 30 kHz to 6 GHz range. which allows you to make a measurement once from the front panel and automatically save the keystrokes without an external computer. To increase your throughput in production. with comprehensive calibration • Save/recall to builtin disk drive http://www. the HP 8753D offers features such as the test sequence function. and a 512 KB nonvolatile memory also help improve your productivity. LIF and DOS formats for output to the builtin disk drive or an external disk drive. makes group delay and time domain measurements. The HP 8573D provides magnitude and phase information. and uses vector accuracy enhancement to minimize measurement uncertainty.hp.
and links to instrumentation and thirdparty design software. products.com/go/hpeesof . and system simulators.H Relevant Products CAE solutions for RF and microwave design Test & Measurement Application Note 951 SParameter Techniques Time to market is critical to product success in today’s competitive environment. simulated. layout tools. http://www. For cost savings and SOLUTIONS FROM HP EEsof manufacturing yield improvements. and solutions. Visit the HP EEsof website for a continuously updated list of news. CAE tools from HewlettPackard’s EEsof Division give companies a competitive edge by simplifying and expediting the development of RF and microwave circuits and systems. A comprehensive HIGHFREQUENCY ELECTRONIC DESIGN AUTOMATION (EDA) set of CAE tools from HP EEsof includes highfrequency circuit. designs can be accurately modeled. software. device modeling systems and libraries. electromagnetic.hp. and optimized before they are actually produced.
using half the memory of previous releases. the software produces accurate results ten times faster. vector plots. Release 5. Using accuracydriven adaptive solution refinement. The HP High Frequency Structure Simulator (HP HFSS). machined components.0. passive 3D structures such as antennas. For comprehensive evaluations.H Relevant Products HP 85180A High Frequency Structure Simulator Fast. This approach is more cost effective and takes less time than building and testing prototype after prototype in the lab. Its drawing environment and parts library simplify specifying complex structures. making improvements through successive simulations. using half the memory of previous releases • Requires minimal user knowledge of EM field theory • For PC and UNIX platforms http://www. models arbitrarilyshaped. animated EM fields. the best method for designing high frequency circuits and machined structures is to compute their behavior with 3D electromagnetic (EM) simulator software. antenna polar patterns and tabular. Test & Measurement Application Note 951 SParameter Techniques Features of the HP 85180A • Reduces "cut and try" prototyping.hp.com/go/hpeesof . surface currents. then building designs to the refined specifications. for lower development costs and a quicker time to market • Yields accurate results ten times faster. accurate electromagnetic simulation of 3D highfrequency structures saves design time With today’s computer technology. generalized sparameters and Smith charts can be viewed. and RF and digital circuits.
com/go/tmeducation . Register directly online! Test & Measurement Application Note 951 SParameter Techniques http://www. we provide training courses taught by the experts from HP’s EEsof Division.hp.H Relevant Services RF and microwave education and training HP offers an extensive curriculum of education services at locations worldwide. Visit HP’s World Wide Website for the complete and continuously updated list of Test & Measurement class schedules and locations around the world. and training sessions can be conducted at its site. Click on the URL below or type the address in your browser. Advanced RF/microwave CAE courses are also conducted to help extend the capabilities and hone the skills of experienced CAE users. The curriculum can be tailored to a company’s special needs. HP education classes are scheduled regularly. To help highfrequency design engineers learn how to use CAE tools quickly.
Suite 950 5200 Blue Lagoon Miami.hp. http://www. FL 33126 3052674245 United States of America HewlettPackard Company 8004524844 Visit HP’s World Wide Website for the complete and up to date listing of Test & Measurement sales office addresses and call center phone numbers. 5150 Spectrum Way Mississauga. HewlettPackard Hong Kong Ltd. Switzerland For Europe: 41 22 780 81 11 For Middle East/Africa: 41 22 780 41 11 Test & Measurement Application Note 951 SParameter Techniques Latin America Latin American Region Headquarters HewlettPackard Company Waterford Building. Route du Nantd’Avril 1217 Meyrin 1 Geneva. 1721/F Shell Tower.A.H Worldwide Contacts HewlettPackard Test & Measurement offices AsiaPacific HewlettPackard Asia Pacific Ltd. Causeway Bay Hong Kong 85 2 599 7777 For Japan: 0120421345 Australia/New Zealand HewlettPackard Australia Ltd. Click on the URL below or type the address in your browser. Ontario L4W 5G1 905 206 4725 Europe/Middle East/Africa HewlettPackard S. Victoria 3130 Australia 1 800 627 485 Canada HewlettPackard Ltd.Times Square 1 Matheson Street. 3141 Joseph Street Blackburn. 150.com/go/tmdir .
Christina Bangle – freestyle illustrations. More Accurate Network Design is electronically published as part of the HP Test & Measurement Digital Application Note Library for the World Wide Web. content supervision. and animations. Holds a 1959 BSEE degree from Utah State University and a 1963 MSEE degree from Stanford University. updated & revised content. and source images. SParameter Techniques for Faster. dissemination or disassembly without prior written permission is prohibited except as allowed under the copyright laws. Richard W. content perspectives. Adaptation. & project management. Lee Smith – Sparameter guru. Contributors Walt Patstone – technical content and sidebar editing. Chuck McGuire. mathematical programming.and graphic reviews. sidebars text. support. excitement. technical illustrations. Test & Measurement Application Note 951 SParameter Techniques © Copyright HewlettPackard Company 19961997 All Rights Reserved. During his 37 years with HP Dick has contributed to the development of numerous microwave and other T&M instruments. Kathy Cunningham – page layout design and Quark guru. Original printed publication Number 59521130. © Graphic & Interaction Design Ev Shafrir – original concept. Business Manager Nyna Casey – funding. . Jeff Gruszynski – expert domain consultant.H Copyrights and Credits Authors. and encouragement. Mike C’debaca – new sparameter measurements & device output. art direction. he is HP Vice President. contributors and producers Authors Dick Anderson – original author of this application note. digital publishing & production. Currently. overall direction. November 1996. Anderson Leann Scully – source images. extraction. interaction design. translation. reproduction. and General Manager of the T&M Microwave and Communications Group.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.