Freescale Semiconductor

Application Note

AN721
Rev. 1, 10/2005

NOTE: The theory in this application note is still applicable,

but some of the products referenced may be discontinued.

Impedance Matching Networks

Applied to RF Power Transistors
By: B. Becciolini

1. INTRODUCTION
Some graphic and numerical methods of impedance
matching will be reviewed here. The examples given will refer
to high frequency power amplifiers.
Although matching networks normally take the form of
filters and therefore are also useful to provide frequency
discrimination, this aspect will only be considered as a
corollary of the matching circuit.
Matching is necessary for the best possible energy
transfer from stage to stage. In RF-power transistors the
input impedance is of low value, decreasing as the power
increases, or as the chip size becomes larger. This
impedance must be matched either to a generator of
generally 50 ohms internal impedance or to a preceding
stage. Impedance transformation ratios of 10 or even 20 are
not rare. Interstage matching has to be made between two
complex impedances, which makes the design still more
difficult, especially if matching must be accomplished over
a wide frequency band.

Rp

IND.
f
f
fS

RF Application Information
Freescale Semiconductor

CAP.

LS

ZIN

fP

CC

CDE

RBB

CTE

RE

Where:
RE = emitter diffusion resistance
CDE,CTE = diffusion and transition capacitances of the
emitter junction
RBB = base spreading resistance
CC = package capacitance
LS = base lead inductance

2.1 INPUT IMPEDANCE

Freescale Semiconductor, Inc., 2005. All rights reserved.

fP

fS

Figure 1. Input Impedance of RF-Power Transistors as

a Function of Frequency

2. DEVICE PARAMETERS

The general shape of the input impedance of RF-power

transistors is as shown in Figure 1. It is a large signal
parameter, expressed here by the parallel combination of
a resistance Rp and a reactance Xp (Ref. (1)).
The equivalent circuit shown in Figure 2 accounts for the
behavior illustrated in Figure 1.
With the presently used stripline or flange packaging,
most of the power devices for VHF low band will have their
Rp and Xp values below the series resonant point fs. The
input impedance will be essentially capacitive.
Most of the VHF high band transistors will have the series
resonant frequency within their operating range, i.e. be
purely resistive at one single frequency fs, while the parallel
resonant frequency fp will be outside.
Parameters for one or two gigahertz transistors will be
beyond fs and approach fp. They show a high value of Rp
and Xp with inductive character.
A parameter that is very often used to judge on the
broadband capabilities of a device is the input Q or QlN
defined simply as the ratio Rp/Xp. Practically QlN ranges
around 1 or less for VHF devices and around 5 or more for
microwave transistors.

XP

Figure 2. Equivalent Circuit for the Input Impedance of

RF-Power Transistors
QIN is an important parameter to consider for broadband
matching. Matching networks normally are low-pass or
pseudo low-pass filters. If QlN is high, it can be necessary
to use band-pass filter type matching networks and to allow
insertion losses. But broadband matching is still possible.
This will be discussed later.
2.2 OUTPUT IMPEDANCE
The output impedance of the RF-power transistors, as
given by all manufacturers data sheets, generally consists
of only a capacitance COUT. The internal resistance of the
transistor is supposed to be much higher than the load and
is normally neglected. In the case of a relatively low internal
resistance, the efficiency of the device would decrease by
the factor:
1 + RL/RT

AN721
1

where RL is the load resistance, seen at the collector-emitter

terminals, and RT the internal transistor resistance equal to:
1
T, (CTC + CDC)

defined as a small signal parameter, where:

T
CTC + CDC

= transit angular frequency

= transition and diffusion capacitances
at the collector junction

The output capacitance COUT, which is a large signal

parameter, is related to the small signal parameter CCB, the
collector-base transition capacitance.
Since a junction capacitance varies with the applied
voltage, COUT differs from CCB in that it has to be averaged
over the total voltage swing. For an abrupt junction and
assuming certain simplifications, COUT = 2 CCB.
Figure 3 shows the variation of COUT with frequency.
COUT decreases partly due to the presence of the collector
lead inductance, but mainly because of the fact that the
base-emitter diode does not shut off anymore when the
operating frequency approaches the transit frequency fT.

4. MATCHING NETWORKS
In the following matching networks will be described by
order of complexity. These are ladder type reactance
networks.
The different reactance values will be calculated and
determined graphically. Increasing the number of reactances
broadens the bandwidth. However, networks consisting of
more than four reactances are rare. Above four reactances,
the improvement is small.
4.1 NUMERICAL DESIGN
4.1.1 Two-Resistance Networks
Resistance terminations will first be considered. Figure 4
shows the reactive L-section and the terminations to be
matched.
X2

R1

X1

R1
n

R2 =
n>1

Figure 4. Two-Reactance Matching Network

COUT

Matching or exact transformation from R2 into R1 occurs

at a single frequency fo.
At fo, X1 and X2 are equal to:
X1 =
R1

R2
R1 - R2

= R1

n-1

Figure 3. Output Capacitance COUT as a Function

of Frequency

3. OUTPUT LOAD
In the absence of a more precise indication, the output
load RL is taken equal to:
RL =

[VCC - VCE(sat)]2

X2 = 

R2 (R1 - R2) = R1

n-1
1

At fo: X1 X2 = R1 R2
X1 and X2 must be of opposite sign. The shunt reactance is
in parallel with the larger resistance.
The frequency response of the L-section is shown in
Figure 5, where the normalized current is plotted as a
function of the normalized frequency.

2POUT

I2

with VCE(sat) equal to 2 or 3 volts, increasing with frequency.

The above equation just expresses a well-known relation,
but also shows that the load, in first approximation, is not
related to the device, except for VCE(sat). The load value
is primarily dictated by the required output power and the
peak voltage; it is not matched to the output impedance of
the device.
At higher frequencies this approximation becomes less
exact and for microwave devices the load that must be
presented to the device is indicated on the data sheet. This
parameter will be measured on all Freescale RF-power
devices in the future.
Strictly speaking, impedance matching is accomplished
only at the input. Interstage and load matching are more
impedance transformations of the device input impedance
and of the load into a value RL (sometimes with additional
reactive component) that depends essentially on the power
demanded and the supply voltage.

Io

1.0

2
4
5
0.8 6
8
10
0.6
0.4
0.5

Qo = 1

4
5
6
8
10

Qo = 2

Qo = 3

0.7
0.9
1.1
1.3
1.5
f/fo LOWPASS
fo/f HIGHPASS
X2

R1

X1

I2

R2

n = R1/R2
Qo = n 1

Figure 5. Normalized Frequency Response for the

L-Section in Low-Pass or High-Pass Form
AN721

RF Application Information
Freescale Semiconductor

If X1 is capacitive and consequently X2 inductive, then:

fo

X1 = -

and

X2 =

R2

R1

f
f

R1 - R2

R2 (R1 - R2) =

fo

fo

=-

I2

where Io =

=
(n - 1)2

 f 4
f

Xext

Xext = X2 Xint

The normalized current absolute value is equal to:

Io

Xint

R2

n-1

n-1

R1

fo

R1

R1

Xint

Xext =

Xext

X1 Xint
Xin X1

n
-2

 f 2 + (n + 1)2
f
o

nE
, and is plotted in Figure 5 (Ref. (2)).
2 R1

If X1 is inductive and consequently X2 capacitive, the only

change required is a replacement of f by fo and vice-versa.
The L-section has low pass form in the first case and
high-pass form in the second case.

Figure 6. Termination Reactance Compensation

4.1.1.1 Use of transmission lines and inductors
In the preceding section, the inductance was expected
to be realized by a lumped element. A transmission line can
be used instead (Fig 7).
R1

R1

Zo(,L)

The Q of the circuit at fo is equal to:

Qo =

X2
R2

R1
X1

n-1

For a given transformation ratio n, there is only one

possible value of Q. On the other hand, there are two
symmetrical solutions for the network, that can be either a
low-pass filter or a high-pass filter.
The frequency fo does not need to be the center
frequency, (f1 + f2)/2, of the desired band limited by f1
and f2.
In fact, as can be seen from the low-pass configuration
of Figure 5, it may be interesting to shift fo toward the high
band edge frequency f2 to obtain a larger bandwidth w, where
w=

2 (f1 + f2)
f2 - f1

This will, however, be at the expense of poorer harmonic

rejection.
Example:
For a transformation ratio n = 4, it can be determined from
the above relations:
Bandwidth w
Max insertion losses
X1/R1

0.1
0.025
1.730

0.3
0.2
1.712

If the terminations R1 and R2 have a reactive component

X, the latter may be taken as part of the external reactance
as shown in Figure 6.
This compensation is applicable as long as
QINT =

XINT
R2

or

R1
XINT

R2

(a)

R2

(b)

Figure 7. Use of a Transmission Line in the L-Section

As can be seen from the computed selectivity curves
(Fig. 8) for the two configurations, transmission lines result
in a larger bandwidth. The gain is important for a
transmission line having a length L = /4 ( = 90) and a
characteristic impedance Zo= R1 R2. It is not significant
for lines short with respect to /4. One will notice that there
is an infinity of solutions, one for each value of C, when using
transmission lines.
4.1.2 Three-Reactance Matching Networks
The networks which will be investigated are shown in
Figure 9. They are made of three reactances alternatively
connected in series and shunt.
A three-reactances configuration allows to make the
quality factor Q of the circuit and the transformation ratio n
= R2/R1 independent of each other and consequently to
choose the selectivity between certain limits.
For narrow band designs, one can use the following
formulas (Ref. (5) AN267, where tables are given):
Network (a):
XC1 = R1/Q

<n - 1

Q must be first selected

XC2 = R2

R1/R2
(Q2 + 1) -

Tables giving reactance values can be found in Ref. (3)

and (4).

XL =

R1
R2

QR1 + (R1R2/XC2)
Q2 + 1

AN721
RF Application Information
Freescale Semiconductor

POUT rel.
1.0
.9
.8
(b)
.7
.6
.5
.4

(a)

.3
.2
.1
0.2

0.4

0.6

0.8

1.0

RG

1.2

1.4

f/fo

Z()

RG

1.6

RL

RL

(a)

(b)
TRANSFORMATION RATIO n = 10

Figure 8. Bandwidth of the L-Section for n = 10

(a) with Lumped Constants
(b) with a Transmission Line (/4)
(a)

(b)

XL

XL1

Network (c):

XL2

XL1 = Q R1
R1

(c)

XC1

XC2

R2

R1

XC1

R2

XC2 = A R2
XC1 =

XL

XC2

R1

XC1

R2 > R1

R2

Figure 9. Three-Reactance Matching Networks

Network (b):
XL1 = R1Q
XL2 = R2 B
XC1 =

A
Q+B

Q must be first selected

A = R1 (1 + Q2)
B=

A
R2

B
Q-A

Q must be first selected

A=

R1 (1 + Q2)
R2

-1

B = R1 (1 + Q2)

The network which yields the most practical component

values, should be selected for a given application.
The three-reactance networks can be thought of as being
formed of a L-section (two reactances) and of a
compensation reactance. The L-section essentially performs
the impedance transformation, while the additional reactance
compensates for the reactive part of the transformed
impedance over a certain frequency band.
Figure 10 shows a representation in the Z-plane of the
circuit of Figure 9 (a) split into two parts R1 - C1 - L1 and
C2 - R2.

-1

AN721
4

RF Application Information
Freescale Semiconductor

Again, if one of the terminations has a reactive

component, the latter can be taken as a part of the matching
network, provided it is not too large (see Fig. 6).

L1

R1

C1

R2

C2

C1

L2
Z
Z =

Z =

1+

R1
2 R12 C12

+ j

1+

R2
2 C22 R22

L1 -

1+

j2 C2 R22

x =

L1
C1 R12

R2(G2)

= R + jX

R
R1

-1

R22

jC1

1 + 2 C12 R12

+ jC2 -

2L22

G/R1 - G2

B = -

1 + 2 C12 R12
R2

Y = G + jB =

R R2 - R2

2C12R1

Y = G + jB =
R1

R1(G1)

C2

= R + jX

1 + 2 C22 R22

x =
x, x

Z
R12C1
2 R12 C12

jL2
R22

B = (1/kR2 - G)

+ 2L22

1/GR2 - 1

k =L2/C2R22

B
N

R
R2

R1

Exact transformation from R1 into R2 occurs at the points

of intersection M and N. Impedances are then conjugate or
Z = R + jX and Z = R + jX with R = R and X = - X.
The only possible solution is obtained when X and - X
are tangential to each other. For the dashed curve,
representing another value of L1 or C1, a wider frequency
band could be expected at the expense of some ripple inside
the band. However, this can only be reached with four
reactances as will be shown in section 4.1.3.
With a three-reactance configuration, there are not
enough degrees of freedom to permit X = - X and
simultaneously obtain the same variation of frequency on
both curves from M to point N.
Exact transformation can, therefore, only be obtained at
one frequency.
The values of the three reactances can be calculated by
making
dX
dX
=.
dR
dR

The general solution of these equations leads to

complicated calculations. Therefore, computed tables should
be used.
One will note on Figure 10 that the compensation
reactance contributes somewhat to impedance
transformation, i.e. R varies when going from M to R2.
The circuit of Figure 9 (b) is dual with respect to the first
one and gives exactly the same results in a Y-plane
representation.
Circuit of Figure 9 (c) is somewhat different since only
one intersection M exists as shown in Figure 11. Narrower
frequency bands must be expected from this configuration.
The widest band is obtained for C1 = .

M
G or G

Figure 10. Z-Plane Representation of the Circuit of

Figure 9(a)

X = - X, R = R and

(f1)

G1

G2

B . B

Figure 11. Y-Plane Representation of the Circuit

of Figure 9(c)
4.1.3 Four-Reactance Networks
Four-reactance networks are used essentially for
broadband matching. The networks which will be considered
in the following consist of two two-reactance sections in
cascade. Some networks have pseudo low-pass filter
character, others band-pass filter character. In principle, the
former show narrower bandwidth since they extend the
impedance transformation to very low frequencies
unnecessarily, while the latter insure good matching over a
wide frequency band around the center frequency only (see
Fig. 14).
X2

a)

R1

c)

R1

X1

X3

X4

Z1

X1

R2

Z2

X2

R1

d)

X3

X4

X2

b)

R2

X1

R1

X4

X1

X3

X2

R2

X4

M
X3

R2

R1 = n.R2

Figure 12. Four-Reactance Networks

AN721

RF Application Information
Freescale Semiconductor

the selectivity curve will be different from the curves given

in Figure 14. Figure 15 shows the selectivity for networks
(a) and (b) when the source resistance R1 is infinite.
From Figure 15 it can be seen that network (a) is more
sensitive to R1 changes than network (b).
0 + j1

n-1

, X2 = R1

Q2
R1

X3 =
n(

, X4 =

n - 1)

R1

2
4
0.65 1

8
10
1.35 1.46

16
1.73

(b)

0.316 j0.465
(1 + j1.47)

n-1

For network (a) normally, at point (M), Z1 and Z2 are

complex. This pseudo low-pass filter has been computed
elsewhere (Ref. (3)). Many tables can be found in the
literature for networks of four and more reactances having
Tchebyscheff character or maximally-flat response (Ref. (3),
(4) and (6)).
Figure 13 shows the transformation path from R1 to R2
for networks (a) and (b) on a Smith-Chart (refer also to
section 4.2, Graphic Design).
Case (a) has been calculated using tables mentioned in
Ref. (4).
Case (b) has been obtained from the relationship given
above for X1 . . . X4. Both apply to a transformation ratio
equal to 10 and for R1 = 1.
There is no simple relationship for X1 . . . X4 of network
(b) if f1 is made different from f2 for larger bandwidth.
Figure 14 shows the respective bandwidths of network
(a) and (b) for the circuits shown in Figure 13.
If the terminations contain a reactive component, the
computed values for X1 or X4 may be adjusted to
compensate for this.
For configuration (a), it can be seen from Figure 13, that
in the considered case the Qs are equal to 1.6.
For configuration (b) Q1, which is equal to Q2, is fixed
for each transformation ratio.
n
Q1 = Q2

b1

0.28 j0.14
(2.8 + j1.4)

X1 X4 = R1 R2 = X2 X3

1 + j0
x2

n-1

0.316 j0

x4

0.1

NORMALIZED IMP.
NORMALIZED ADM.

b1

x2

R1

0.28 + j0.45
(1 j1.6)
(a)

b3
b3

0.1 + j0.147
Q1 (0.316
j4.65)
0

X1 =

Q2

0.1 + j0.16
Q1
(2.8 j4.5)

x4

The two-reactance sections used in above networks have

either transformation properties or compensation properties.
Impedance transformation is obtained with one series
reactance and one shunt reactance. Compensation is made
with both reactances in series or in shunt.
If two cascaded transformation networks are used,
transformation is accomplished partly by each one.
With four-reactance networks there are two frequencies,
f1 and f2, at which the transformation from R1 into R2 is exact.
These frequencies may also coincide.
For network (b) for instance, at point M, R1 or R2 is
transformed into R1R2 when both frequencies fall together.
At all points (M), Z1 and Z2 are conjugate if the transformation
is exact.
In the case of Figure 12 (b) the reactances are easily,
calculated for equal frequencies:

Q =

R1 = 1
R2 = 0.1
X1 = 0.68
(
B1 = 1.47
X2 = 0.465

X3 = 0.215
(
B3 = 4.65
X4 = 0.147

Q1 = 1.6 = Q2
X2

(a)
R1

Q1 = Q2 = 1.47
X4

X1

X3

X2

(b)
R2

R1

X4

X1

X3

R2

Figure 13. Transformation Paths for Networks

(a) and (b)
ATTEN.
0
1
2
3
4
5
6
7
8
9
10

n-1

The maximum value of reactance that the terminations

may have for use in this configuration can be determined
from the above values of Q.
If R1 is the load resistance of a transistor, the internal
transistor resistance may not be equal to R1. In this case

X3 = 0.169
(
B3 = 5.9
X4 = 0.160

(a) X1 = 0.624
(
B1 = 1.6
X2 = 0.59

(a)
(b)
(b)
(a)

f/fo
.3 .4 .5 .6 .7 .8 .9 1 1.1

1.3

1.5

1.7

1.9

Figure 14. Selectivity Curves for Networks (a) and (b)

of Figure 13

AN721
6

RF Application Information
Freescale Semiconductor

L2

REL. ATTEN.
2
1
0
1
2
3
4
5
6
7
8
9
10

L1

R2
(G2)

(a)

(b)

(a)

Y = G + jB =
f/fo
.4 .5 .6 .7 .8 .9 1 1.1

1.3

1.5

1.7

B =

Figure 15. Selectivity Curves for Networks (a) and (b)

of Figure 13 with Infinite R1

R2
(G1)

C2

Y = G + jB =

(b)

C1

Y
R1

R12 + (L1 1/c1)2

R2
R 22 + 2L 22

j(L1 1/c1)
R12 + (L1 1/c1)2

+ j C2 -

L2
R 22 + 2L 22

G
G2
R1
1
1
GR2

B = (1/kR2 G)

As mentioned earlier, the four-reactance network can also

be thought of as two cascaded two-reactance sections; one
used for transformation, the other for compensation.
Figure 16 shows commonly used compensation networks,
together with the associated L-section.
The circuit of Figure 16 (a) can be compared to the
three-reactance network shown in Figure 9(c). The difference
is that capacitor C2 of that circuit has been replaced by a
L-C circuit. The resulting improvement may be seen by
comparing Figure 17 with Figure 11.

k =

L2
C 2 R 22

B
G2

G1

B
COMPENSATION NETWORK

(a)
R2

R1

(b)

C.N.

R2

R1

(c)

C.N.

R2

C.N.

R1 = n.R2

R1

Figure 16. Compensation Networks Used with

a L-Section

By adding one reactance, exact impedance

transformation is achieved at two frequencies. It is now
possible to choose component values such that the point
of intersection M occurs at the same frequency f1 on both
curves and simultaneously that N occurs at the same
frequency f2 on both curves. Among the infinite number of
possible intersections, only one allows to achieve this.

Figure 17. Y-Plane Representation of the Circuit

of Figure 16(a)
When M and N coincide in, M, the new dX/df = dX/df
condition can be added to the condition X = - X (for
three-networks) and similarly R = R and dR/df = dR/df.
If f1 is made different from f2, a larger bandwidth can be
achieved at the expense of some ripple inside the band.
Again, a general solution of the above equations leads
to still more complicated calculations than in the case of
three-reactance networks. Therefore, tables are preferable
(Ref. (3), (4) and (6)).
The circuit of Figure 16(b) is dual of the circuit of
Figure 14(a) and does not need to be treated separately. It
gives exactly the same results in the Z-plane. Figure 16(c)
shows a higher order compensation requiring six reactive
elements.
The above discussed matching networks employing
compensation circuits result in narrower bandwidths than the
former solutions (see paragraph 4.1.3) using two
transformation sections. A matching with higher order
compensation such as in Figure 16(c) is not recommended.
Better use can be made of the large number of reactive
elements using them all for transformation.
When the above configurations are realized using short
portions of transmission lines, the equations or the usual
tables no longer apply. The calculations must be carried out
on a computer, due to the complexity. However, a graphic
method can be used (see next section) which will consist
essentially in tracing a transformation path on the Z-Y-chart
using the computed lumped element values and replacing
it by the closest path obtained with distributed constants. The

AN721
RF Application Information
Freescale Semiconductor

bandwidth change is not significant as long as short portions

of lines are used (Ref. (13)).

1.0
n=4
0.9

4.1.4 Matching Networks Using Quarter-ave

Transformers

0.8

At sufficiently high frequencies, where /4-long lines of

practical size can be realized, broadband transformation can
easily be accomplished by the use of one or more
/4-sections.
Figure 18 summariz es the main relations for
(a) one-section and (b) two-section transformation.
A compensation network can be realized using a /2-long
transmission line.

0.7

0.6

0.8

1.0

1.2

1.4

o/

Figure 20. Selectivity Curves for One, Two and Three

/4-Sections
4.1.5 Broadband Matching Using Band-Pass Filter
Type Networks. High Q Case.

(a)
R1 = nR2
R2

R1

Z = R1 R2 = R2 n =

(b)
R2

R1
n

4
Z 1 = R 13 R 2
Z2

Z1

4
Z 2 = R1 R 2 3

R1

at A, R = R1 R2
A

Figure 18. Transformation Networks Using /4-Long

Transmission Lines
Figures 19 and 20 show the selectivity curves for different
transformation ratios and section numbers.

The above circuits are applicable to devices having low

input or output Q, if broadband matching is required.
Generally, if the impedances to be matched can be
represented for instance by a resistor R in series with an
inductor L (sometimes a capacitor C) within the band of
interest and if L is sufficiently low, the latter can be
incorporated into the first inductor of the matching network.
This is also valid if the representation consists of a shunt
combination of a resistor and a reactance.
Practically this is feasible for Qs around one or two. For
higher Qs or for input impedances consisting of a series or
parallel resonant circuit (see Fig. 2), as it appears to be for
large bandwidths, a different treatment must be followed.
Let us first recall that, as shown by Bode and Fano (Ref.
(7) and (8)), limitations exist on the impedance matching of
a complex load. In the example of Figure 21, the load to be
matched consists of a capacitor C and a resistor R in shunt.
RG

Exponential lines
Exponential lines have largely frequency independent
transformation properties.
The characteristic impedance of such lines varies
exponentially with their length I:

Z = Zo ekl

ZIN

MATCHING
NETWORK
(LOSSLESS)

Figure 21. General Matching Conditions

where k is a constant, but these properties are preserved

only if k is small.

The reflection coefficient between transformed load and

generator is equal to:
=

1.0

ZIN - Rg
ZIN + Rg

= 0, perfect matching,
= 1, total reflection.
The ratio of reflected to incident power is:

0.9
0.8

Pr
n=6

0.7

Pi

2
o/

0.6

0.8

1.0

1.2

The fundamental limitation on the matching takes the

form:

= oln  1  d  RC

1.4

Figure 19. Selectivity Curves for Two /4-Section

Networks at Different Transformation Ratios

= 2

Bode equation

and is represented in Figure 22.

AN721
8

RF Application Information
Freescale Semiconductor

ln

S=

RC
c

Figure 22. Representation of Bode Equation

The meaning of Bode equation is that the area S under the
curve cannot be greater than /RC and therefore, if matching
is required over a certain bandwidth, this can only be done at
the expense of less power transfer within the band. Thus,
power transfer and bandwidth appear as interchangeable
quantities.
It is evident that the best utilization of the area S is
obtained when is kept constant over the desired band
c and made equal to 1 over the rest of the spectrum. Then
= e

c RC

within the band and no power transfer happens outside.

A network fulfilling this requirement cannot be obtained
in practice as an infinite number of reactive elements would
be necessary.
If the attenuation a is plotted versus the frequency for
practical cases, one may expect to have curves like the ones
shown in Figure 23 for a low-pass filter having Tchebyscheff
character.

Thus, devices having relatively high input Qs are usable

for broadband operation, provided the consequent higher
attenuation or reflection introduced is acceptable.
The general shape of the average insertion losses or
attenuation a (neglecting the ripple) of a low-pass impedance
matching network is represented in Figure 24 as a function
of 1/Q for different numbers of network elements n (ref. (3)).
For a given Q and given ripple, the attenuation decreases
if the number n of the network element increases. But above
n = 4, the improvement is small.
For a given attenuation a and bandwidth, the larger n the
smaller the ripple.
For a given attenuation and ripple, the larger n the larger
the bandwidth.
Computations show that for Q < 1 and n 3 the
attenuation is below 0.1 db approximately. The impedance
transformation ratio is not free here. The network is a true
low-pass filter. For a given load, the optimum generator
impedance will result from the computation.
Before impedance transformation is introduced, a
conversion of the low-pass prototype into a band-pass filter
type network must be made. Figure 25 summarizes the main
relations for this conversion.
r is the conversion factor.
For the band-pass filter, QlN max or the maximum possible
input Q of a device to be matched, has been increased by
the factor r (from Figure 25, QIN max = r QIN max).
Impedance inverters will be used for impedance
transformation. These networks are suitable for insertion into
a band-pass filter without affecting the transmission
characteristics.
a

LOWPASS

BANDPASS

a max2
a min2
a max1
wc =

a min1
1

r=

Figure 23. Attenuation versus Angular Frequency for

Different Bandwidths with Same Load

L1
Ro

C2

QIN max =
dB
4

a 2
1

C 1

For a given complex load, an extension of the bandwidth

from 1 to 2 is possible only with a simultaneous increase
of the attenuation a. This is especially noticeable for Qs
exceeding one or two (see Figure 24).

QIN max =

L3
C4

cL1
Ro
oL 1
Ro

R5

L 1

R5

Ro C
2

L 2

L1 = r.L1

C 1 =

C2 = r.C2

L 2 =
etc.

1
o2 L1
1
o 2 C2

n=2

Figure 25. Conversion from Low-Pass into

Band-Pass Filter

0
0.1

1/Q

Figure 24. Insertion Losses as a Function of 1/Q

Figure 26 shows four impedance inverters. It will be

noticed that one of the reactances is negative and must be
combined in the band-pass network with a reactance of at
AN721

RF Application Information
Freescale Semiconductor

least equal positive value. Insertion of the inverter can be

made at any convenient place (Ref. (3) and (9)).
INVERTER

EQUIVALENT TO:
C
1/n2 1/n

C
1 1/n

a)

nC

n:1
nC

b)

(4) Convert the element values found by step (3) into

series or parallel resonant circuit parameters,
(5) Insert the impedance inverter in any convenient
place.

(1 n)C

(n2 n)C

In the above discussions, the gain roll-off has not been

taken into account. This is of normal use for moderate
bandwidths (30% for ex.). However, several methods can
be employed to obtain a constant gain within the band
despite the intrinsic gain decrease of a transistor with
frequency.
Tables have been computed elsewhere (Ref. (10)) for
matching networks approximating 6 db/octave attenuation
versus frequency.
Another method consists in using the above mentioned
network and then to add a compensation circuit as shown
for example in Figure 27.
L2

L(1/n2 1/n)

L(1 1/n)

C2

n:1
R = ZIN = CONSTANT vs. f.

R1
c)

L/n

L1C1 = L2C2
R

L
L1

R1 = R
b2 = 1/L1C1 = 1/L2C2

C1

(HIGH BAND EDGE)

n:1

L/n

Q2 = bL2/R = Q1 = bC1R

L
d)

L
1n

L
n2 n

Figure 27. Roll-Off Compensation Network

n:1

Figure 26. Impedance Inverters

When using the band-pass filter for matching the input
impedance of a transistor, reactances L1 C1 should be
made to resonate at o by addition of a convenient series
reactance.
As stated above, the series combination of Ro, L1 and
C1 normally constitutes the equivalent input network of a
transistor when considered over a large bandwidth. This is
a good approximation up to about 500 MHz.
In practice the normal procedure for using a bandpass
filter type matching network will be the following:
(1) For a given bandwidth center frequency and input
impedance of a device to be matched e.g. to 50 ohms,
first determine QlN from the data sheet as
oL1
Ro

after having eventually added a series reactor for

centering,
(2) Convert the equivalent circuit R o L 1 C 1 into a
low-pass prototype Ro L1 and calculate QlN using the
formulas of Figure 25,
(3) Determine the other reactance values from tables
(Ref. (3)) for the desired bandwidth,

Resonance b is placed at the high edge of the frequency

band. Choosing Q correctly, roll-off can be made 6 db/octave.
The response of the circuit shown in Figure 27 is
expressed by:
1

b 2
1 + Q2 -
b

, where < b

This must be equal to /b for 6 db/octave compensation.

At the other band edge a, exact compensation can be
obtained if:

 b  2 - 1
Q=

a
b

 b  2
a

4.1.6 Line Transformers

The broadband properties of line transformers make them
very useful in the design of broadband impedance matching
networks (Ref. (11) and (12)).
A very common form is shown by Figure 28. This is a
4:1 impedance transformer. Other transformation ratios like
9:1 or 16:1 are also often used but will not be considered
here.
The high frequency cut-off is determined by the length
of line which is usually chosen smaller than min/8. Short
lines extend the high frequency performance.

AN721
10

RF Application Information
Freescale Semiconductor

Z0 = Rg/2

Rg

RL =

Rg
4

Figure 28. 4:1 Line Transformer

Line Transformer on P.C. Board (Figure 29)

The low frequency cut-off is determined first by the length

of line, long lines extending the low frequency performance
of the transformer. Low frequency cut-off is also improved
by a high even mode impedance, which can be achieved
by the use of ferrite material. With matched ends, no power
is coupled through the ferrite which cannot saturate.
For matched impedances, the high frequency attenuation
a of the 4:1 transformer is given by:
a=

Cu

ISOLATION

35

66

Ribbon thickness
Electrical tape
EE3990
Permacel New
Brunswick
Cu
ISOLATION

(1 + 3 cos 2I/)2 + 4 sin2 2I/

FERRITE

4 (1 + cos 2I/)2

For I = /4, a = 1.25 or 1 db; for I = /2, a = .

The characteristic impedance of the line transformer must
be equal to:
Zo = Rg R .

Figures 29 and 30 show two different realizations of 4:1

transformers for a 50 to 12.5 ohm-transformation designed
for the band 118 - 136 MHz.
The transformers are made of two printed circuit boards
or two ribbons stuck together and connected as shown in
Figures 29 and 30.

25 mm

- Stick one ribbon (1.5 mm wide) against the other (2.5)

Total length per ribbon - 9 cm.
Turns 3.5

Figure 30. 4:1 Copper Ribbon Line Transformer

with Ferrite

EPOXY 0.2 mm THICK

35

Cu

in mm
15 22

10
61.5
68.5

Line Transformer on Ferrite (Figure 30)

STICK ONE AGAINST THE OTHER

Figure 29. 4:1 Line Transformer on P.C. Board

4.2 GRAPHIC DESIGN

The common method of graphic design makes use of the
Impedance-Admittance Chart (Smith Chart).
It is applicable to all ladder-type networks as encountered
in matching circuits.
Matching is supposed to be realized by the successive
algebraic addition of reactances (or susceptances) to a given
start impedance (or admittance) until another end impedance
(or admittance) is reached.
Impedance chart and admittance chart can be
superimposed and used alternatively due to the fact that an
immittance point, defined by its reflection coefficient with
respect to a reference, is common to the Z-chart and the
Y-chart, both being representations in the -plane.

AN721
RF Application Information
Freescale Semiconductor

11

Z - Rs

Z + Rs

= Characteristic impedance of the line

Gs

1
SHORT

x=0

x= 1

+1
OPEN

x = 0.5

r =
i =

r =

INDUCTIVE

2 i

1 - g2 - b2

- 2b
(1 +

g)2

b2

Center at: r = - g/(1 + g); i = 0

Radius: 1/b

Figure 32. Representation of the Normalized Y-Values

in the -Plane
The following design rules apply. They can very easily
be found by thinking of the more familiar Z and Y
representation in rectangular coordinates.
For joining two impedance points, there are an infinity of
solutions. Therefore, one must first decide on the number
of reactances that will constitute the matching network. This
number is related essentially to the desired bandwidth and
the transformation ratio.

Addition of

Chart
to be
Used

Series R

Open (in terms of

admittance)

Series G

Short (in terms of

admittance)

Series C (+ 1/jC)

ccw

Shunt C (+ jC)

cw

Series L (+ jL)

cw

Shunt L (+ 1/jL)

ccw

(r + 1)2 + x2
RECTANGULAR COORDINATES

(r + 1)2 + x2

r circles

x circles

Radius: 1/(1 + r)
Center at: r =

r
1+r

RECTANGULAR COORDINATES

Radius: 1/(1 + g)

CAPACITIVE

2x

CAPACITIVE

(1 + g)2 + b2

(1 - r)2 + i2
r2 + x2 - 1

(1 + r)2 + i2

Center at: r = - 1, i = - 1/b

r>0

x=

b=

INDUCTIVE

- 2 i

b circles

PLANE
Z
= z = r + jx
Rs
1 - r2 - i2
r=
(1 - r)2 + i2

b = 0.5

Y
= y = g + jb
Gs
1 - r2 - i2
g=
(1 + r)2 + i2

g circles

CAPACITIVE
x<0

1
PLANE

INDUCTIVE
x>0

x=1

r = 0.5

CAPACITIVE
b>0
g>0

i =

r=1

+1
OPEN

b=1

g=0

i
+1

r=0

b=0

1
SHORT

More precisely, the Z-chart is a plot in the -plane, while

the Y-chart is a plot in the - -plane. The change from the
to - -plane is accounted for in the construction rules given
below.
Figure 31 and 32 show the representation of normalized
Z and Y respectively, in the -plane.
The Z-chart is used for the algebraic addition of series
reactances. The Y-chart is used for the algebraic addition
of shunt reactances.
For the practical use of the charts, it is convenient to make
the design on transparent paper and then place it on a usual
Smith-chart of impedance type (for example). For the
addition of a series reactor, the chart will be placed with
short to the left. For the addition of a shunt reactor, it will
be rotated by 180 with short (always in terms of
impedance) to the right.

x = 0.5

INDUCTIVE
b<0

g=1
b = 1

Gs + Y

b = 0.5

g = 0.5

Gs - Y

Rs =

i
+1

r + j i

, i = 0

Radius: 1/x
Center at: r =1 , i = 1/x

Direction

Using
Curve of
Constant

Figure 31. Representation of the Normalized Z Values

in the -Plane
AN721
12

RF Application Information
Freescale Semiconductor

Secondly, one must choose the operating Q of the circuit,

which is also related to the bandwidth. Q can be defined at
each circuit node as the ratio of the reactive part to the real
part of the impedance at that node. The Q of the circuit, which
is normally referred to, is the highest value found along the
path.
Constant Q curves can be superimposed to the charts
and used in conjunction with them. In the -plane Q-curves
are circles with a radius equal to
1+

1
Q2

and a center at the point 1/Q on the imaginary axis, which

is expressed by:
Q=

2 i
x
=
r
1 - r2 - i2

r2 + i +

1
1 2
=1+
Q
Q2

The use of the charts will be illustrated with the help of

an example.
The following series shunt conversion rules also apply:

The collector load must be equal to:

[Vcc - Vce (sat)]2

or

2 x Pout

(28 - 3)2
40

= 15.6 ohms .

The collector capacitance given by the data sheet is

40 pF, corresponding to a capacitive reactance of 22.7 ohms.
The output impedance seen by the collector to insure the
required output power and cancel out the collector
capacitance must be equal to a resistance of 15.6 ohms in
parallel with an inductance of 22.7 ohms. This is equivalent
to a resistance of 10.6 ohms in series with an inductance
of 7.3 ohms.
The input Q is equal to, 1.1/1.94 or 0.57 while the output
Q is 7.3/10.6 or 0.69.
It is seen that around this frequency, the device has good
broadband capabilities. Nevertheless, the matching circuit
will be designed here for a narrow band application and the
effective Q will be determined by the circuit itself not by the
device.
Figure 34 shows the normalized impedances (to
50 ohms).
0.022

X
R

0.146

X
0.039

R=

1+

G =

R2

1+

R
R2 + X2

NORMALIZED INPUT IMPEDANCE

X2

X=

1
R

- B =

X2

1
X

0.212

LOAD IMPEDANCE

Figure 34. Normalized Input and Output Impedances

for the 2N5642

X
R2 + X2

R2

Figure 33 shows the schematic of an amplifier using the

2N5642 RF power transistor. Matching has to be achieved
at 175 MHz, on a narrow band basis.
VCC = 28 V

Figure 35 shows the diagram used for the graphic design

of the input matching circuit. The circuit Q must be larger
than about 5 in this case and has been chosen equal to 10.
At Q = 5, C1 would be infinite. The addition of a finite value
of C1 increases the circuit Q and therefore the selectivity.
The normalized values between brackets in the Figure are
admittances (g + jb).
0 + j1

RFC
PARALLEL CAPACITANCE C2

L4
C1

Q
CIRCLE
(10)

L3

RFC

50

C0

1 + j1.75
(0.25 j0.42)

SERIES
INDUCTANCE L3
(x3)

50

C2

(b2)

0.039 + j0.39
(0.25 j2.5)

C5

SERIES
CAPACITANCE C1
(x1)
1 + j0

0.039 + j0.022

2N5642

Figure 35. Input Circuit Design

Figure 33. Narrow-Band VHF Power Amplifier

At f = 175 MHz, the following results are obtained:

L3 = 50x3 = 50 (0.39 - 0.022) = 18.5 ohms


The rated output power for the device in question is 20 W

at 175 MHz and 28 V collector supply. The input impedance
at these conditions is equal to 2.6 ohms in parallel with 200 pF (see data sheet). This converts to a resistance of
1.94 ohms in series with a reactance of 1.1 ohm.

C2 =

L3 = 16.8 nH

1
1
b2 =
(2.5 - 0.42) = 0.0416 mhos
50
50


C2 = 37.8 pF

AN721
RF Application Information
Freescale Semiconductor

13

Rated output power:

1
= 50x1 = 50 1.75 = 87.5 ohms
C1


C1 = 10.4 pF

Figure 36 shows the diagram for the output circuit,

designed in a similar way.

30 W for 8 W input at 175 MHz. From the data sheet it

appears that at 125 MHz, 30 W output will be achieved with
about 4 W input.
Output impedance:
[Vcc - Vce (sat)]2

0 + j1

2 x Pout

100
60

= 1.67 ohms

Cout = 180 pF at 125 MHz

5.2 CIRCUIT SCHEMATIC

0.212 + j0.146

12.5 V

1 + j0

C7

SERIES INDUCTANCE L4
(x4)
0.212 j0.4
(1 + j1.9)

PARALLEL CAPACITANCE C5
(b5)

RFC2
C6

Figure 36. Output Circuit Design

Here, the results are (f = 175 MHz):

RFC1

L4 = 24.8 nH

L2

1
1
C5 =
b5 =
1.9 = 0.038 mhos
50
50


C5 = 34.5 pF

The circuit Q at the output is equal to 1.9.

The selectivity of a matching circuit can also be
determined graphically by changing the x or b values
according to a chosen frequency change. The diagram will
give the VSWR and the attenuation can be computed.
The graphic method is also useful for conversion from
a lumped circuit design into a stripline design. The
immittance circles will now have their centers on the 1 + jo
point.
At low impedance levels (large circles), the difference
between lumped and distributed elements is small.

C5

L3

L4 = 50 x4 = 50 (0.4 + 0.146) = 27.3 ohms

T1

C4

C1

T2

50

50

C1 = 300 pF (chip)
L2 = 4.2 nH (adjust)
L3 = 8 nH (adjust)
C4 = 130 pF (chip)
C5 = 750 pF (chip)
C6 = 2.2 F

T1, T2 see Figure 30

RFC1 = 7 turns
6.3 mm coil diameter
0.8 mm wire diameter
RFC2 = 3 turns on ferrite bead
C7 = 0.68 F

Figure 37. Circuit Schematic

5.3 TEST RESULTS
W

5. PRACTICAL EXAMPLE

30

The example shown refers to a broadband amplifier stage

using a 2N6083 for operation in the VHF band 118 136 MHz. The 2N6083 is a 12.5 V-device and, since
amplitude modulation is used at these transmission
frequencies, that choice supposes low level modulation
associated with a feedback system for distortion
compensation.
Line transformers will be used at the input and output.
Therefore the matching circuits will reduce to two-reactance
networks, due to the relatively low impedance transformation
ratio required.

PIN = 2 W

20
1W

POUT
10

0.5 W
MHz

0
118

124

128

132

136

5.1 DEVICE CHARACTERISTICS

Input impedance of the 2N6083 at 125 MHz:

Figure 38. POUT vs Frequency

Rp = 0.9 ohms
Cp = - 390 pF

AN721
14

RF Application Information
Freescale Semiconductor

Acknowledgments

The author is indebted to Mr. T. ONeal for the fruitful

discussions held with him. Mr. ONeal designed the circuit
shown in Figure 37; Mr. J. Hennet constructed and tested
the lab model.

PIN = 2 W

70
60

1W

50
40

6. LITERATURE

0.5 W

1. Freescale Application Note AN282A Systemizing RF

Power Amplifier Design
2. W. E. Everitt and G. E. Anner Communication
Engineering McGraw-Hill Book Company, Inc.
3. G. L. Matthaei, L. Young, E. M. T. Jones Microwave
Filters, Impedance-Matching Networks and Coupling
Structures McGraw-Hill Book Company, Inc.
4. G.
L.
Matthaei
Tables
of
Chebyshew
Impedance-Transforming Networks of Low-Pass Filter
Form Proc. IEEE, August 1964
5. Freescale Application Note AN267 Matching Network
Designs with Computer Solutions
6. E. G. Cristal Tables of Maximally Flat
Impedance-Transforming Networks of Low-Pass-Filter
Form IEEE Transactions on Microwave Theory and
Techniques Vol. MTT 13, No 5, September 1965
Correspondence
7. H. W. Bode Network Analysis and Feedback Amplifier
Design D. Van Nostrand Co., N.Y.
8. R. M. Fano Theoretical Limitations on the Broadband
Matching of Arbitrary Impedances Journal of Franklin
Institute, January - February 1950
9. J. H. Horwitz Design Wideband UHF-Power
Amplifiers Electronic Design 11, May 24, 1969
10. O. Pitzalis, R. A. Gilson Tables of Impedance Matching
Networks which Approximate Prescribed Attenuation
Versus Frequency Slopes IEEE Transactions on
Microwave Theory and Techniques Vol. MTT-19, No 4,
April 1971
11. C. L. Ruthroff Some Broadband Transformers Proc.
IRE, August 1959
12. H. H. Meinke Theorie der H. F. Schaltungen Mnchen,
Oldenburg 1951

30
20
10
MHz
118

124

128

132

136

Figure 39. vs Frequency

db.
0
Prefl.
Pin 5

PIN = 2 W

10
1W
15
0.5 W
20

118

124

128
f

132

MHz
136

Figure 40. PREFL./PIN vs Frequency

118 - 136 MHz Amplifier (see Figure 37)

Before Coil Adjustment

AN721
RF Application Information
Freescale Semiconductor

15

How to Reach Us:

Home Page:
www.freescale.com
E - mail:
[email protected]
USA/Europe or Locations Not Listed:
Freescale Semiconductor
Technical Information Center, CH370
1300 N. Alma School Road
Chandler, Arizona 85224
+1 - 800 - 521 - 6274 or +1 - 480 - 768 - 2130
[email protected]
Europe, Middle East, and Africa:
Freescale Halbleiter Deutschland GmbH
Technical Information Center
Schatzbogen 7
81829 Muenchen, Germany
+44 1296 380 456 (English)
+46 8 52200080 (English)
+49 89 92103 559 (German)
+33 1 69 35 48 48 (French)
[email protected]
Japan:
Freescale Semiconductor Japan Ltd.
Headquarters
ARCO Tower 15F
1 - 8 - 1, Shimo - Meguro, Meguro - ku,
Tokyo 153 - 0064
Japan
0120 191014 or +81 3 5437 9125
[email protected]
Asia/Pacific:
Freescale Semiconductor Hong Kong Ltd.
Technical Information Center
2 Dai King Street
Tai Po Industrial Estate
Tai Po, N.T., Hong Kong
+800 2666 8080
[email protected]
For Literature Requests Only:
Freescale Semiconductor Literature Distribution Center
P.O. Box 5405
Denver, Colorado 80217
1 - 800 - 441 - 2447 or 303 - 675 - 2140
Fax: 303 - 675 - 2150
[email protected]

Document Number: AN721

Rev. 1, 10/2005

16

Information in this document is provided solely to enable system and software

implementers to use Freescale Semiconductor products. There are no express or
implied copyright licenses granted hereunder to design or fabricate any integrated
circuits or integrated circuits based on the information in this document.
Freescale Semiconductor reserves the right to make changes without further notice to
any products herein. Freescale Semiconductor makes no warranty, representation or
guarantee regarding the suitability of its products for any particular purpose, nor does
Freescale Semiconductor assume any liability arising out of the application or use of
any product or circuit, and specifically disclaims any and all liability, including without
limitation consequential or incidental damages. Typical parameters that may be
provided in Freescale Semiconductor data sheets and/or specifications can and do
vary in different applications and actual performance may vary over time. All operating
parameters, including Typicals, must be validated for each customer application by
customers technical experts. Freescale Semiconductor does not convey any license
under its patent rights nor the rights of others. Freescale Semiconductor products are
not designed, intended, or authorized for use as components in systems intended for
surgical implant into the body, or other applications intended to support or sustain life,
or for any other application in which the failure of the Freescale Semiconductor product
could create a situation where personal injury or death may occur. Should Buyer
purchase or use Freescale Semiconductor products for any such unintended or
unauthorized application, Buyer shall indemnify and hold Freescale Semiconductor
and its officers, employees, subsidiaries, affiliates, and distributors harmless against all
claims, costs, damages, and expenses, and reasonable attorney fees arising out of,
directly or indirectly, any claim of personal injury or death associated with such
unintended or unauthorized use, even if such claim alleges that Freescale
Semiconductor was negligent regarding the design or manufacture of the part.
Freescalet and the Freescale logo are trademarks of Freescale Semiconductor, Inc.
All other product or service names are the property of their respective owners.
Freescale Semiconductor, Inc. 2005. All rights reserved.

AN721
RF Application Information
Freescale Semiconductor