Order this document

MOTOROLA by AN721/D

SEMICONDUCTOR
APPLICATION NOTE
AN721

IMPEDANCE MATCHING NETWORKS

APPLIED TO RF POWER TRANSISTORS
Prepared by: B. Becciolini

1. INTRODUCTION LS RBB′

Some graphic and numerical methods of impedance

matching will be reviewed here. The examples given will refer ZIN CC CDE CTE RE
to high frequency power amplifiers.
Although matching networks normally take the form of
filters and therefore are also useful to provide frequency Where:
discrimination, this aspect will only be considered as a RE = emitter diffusion resistance
corollary of the matching circuit. CDE,CTE = diffusion and transition capacitances of the
Matching is necessary for the best possible energy emitter junction
transfer from stage to stage. In RF-power transistors the RBB′ = base spreading resistance
input impedance is of low value, decreasing as the power CC = package capacitance
LS = base lead inductance
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 Figure 2. Equivalent Circuit for the Input Impedance of
stage. Impedance transformation ratios of 10 or even 20 are RF-Power Transistors
not rare. Interstage matching has to be made between two
Most of the VHF high band transistors will have the series
complex impedances, which makes the design still more
resonant frequency within their operating range, i.e. be
difficult, especially if matching must be accomplished over
purely resistive at one single frequency fs, while the parallel
a wide frequency band.
resonant frequency fp will be outside.
Parameters for one or two gigahertz transistors will be
2. DEVICE PARAMETERS beyond fs and approach fp. They show a high value of Rp
and Xp with inductive character.
2.1 INPUT IMPEDANCE A parameter that is very often used to judge on the
broadband capabilities of a device is the input Q or Q lN
The general shape of the input impedance of RF-power
defined simply as the ratio Rp/Xp. Practically QlN ranges
transistors is as shown in Figure 1. It is a large signal
around 1 or less for VHF devices and around 5 or more for
parameter, expressed here by the parallel combination of
microwave transistors.
a resistance Rp and a reactance Xp (Ref. (1)).
Q IN is an important parameter to consider for broadband
The equivalent circuit shown in Figure 2 accounts for the
matching. Matching networks normally are low-pass or
behavior illustrated in Figure 1.
pseudo low-pass filters. If Q lN is high, it can be necessary
With the presently used stripline or flange packaging,
to use band-pass filter type matching networks and to allow
most of the power devices for VHF low band will have their
insertion losses. But broadband matching is still possible.
Rp and Xp values below the series resonant point fs. The
This will be discussed later.
input impedance will be essentially capacitive.
2.2 OUTPUT IMPEDANCE
Rp XP The output impedance of the RF-power transistors, as
given by all manufacturers’ data sheets, generally consists
of only a capacitance C OUT. The internal resistance of the
IND. transistor is supposed to be much higher than the load and
f is normally neglected. In the case of a relatively low internal
f resistance, the efficiency of the device would decrease by
fS fP CAP. fS fP
the factor:
1 + RL/RT
where RL is the load resistance, seen at the collector-emitter
terminals, and RT the internal transistor resistance equal to:
Figure 1. Input Impedance of RF-Power Transistors as a 1
,
Function of Frequency ωT, (CTC + CDC)

RF Application
Motorola, Inc. 1993 Reports 1
 AN721
defined as a small signal parameter, where: The different reactance values will be calculated and
ωT = transit angular frequency determined graphically. Increasing the number of reactances
CTC + CDC = transition and diffusion capacitances broadens the bandwidth. However, networks consisting of
at the collector junction more than four reactances are rare. Above four reactances,
The output capacitance C OUT, which is a large signal the improvement is small.
parameter, is related to the small signal parameter CCB, the
4.1 NUMERICAL DESIGN
collector-base transition capacitance.
Since a junction capacitance varies with the applied 4.1.1 Two-Resistance Networks
voltage, C OUT differs from C CB in that it has to be averaged
over the total voltage swing. For an abrupt junction and Resistance terminations will first be considered. Figure 4
assuming certain simplifications, C OUT = 2 C CB. shows the reactive L-section and the terminations to be
Figure 3 shows the variation of C OUT with frequency. matched.
C OUT decreases partly due to the presence of the collector
X2
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. R
R1 X1 R2 = 1
n
COUT n>1

Figure 4. Two-Reactance Matching Network

Matching or exact transformation from R2 into R1 occurs

at a single frequency fo.
At fo, X1 and X2 are equal to:
f
X1 = " R1 R2
R1 – R 2
= R1
1

Figure 3. Output Capacitance COUT as a Function n–1

of Frequency
X2 = # R2 (R1 – R2) = R1
n–1
1
3. OUTPUT LOAD
At fo: X1 ⋅ X2 = R1 ⋅ R2
In the absence of a more precise indication, the output
load RL is taken equal to: X1 and X2 must be of opposite sign. The shunt reactance is
in parallel with the larger resistance.
[VCC – VCE(sat)]2 The frequency response of the L-section is shown in
RL =
2POUT Figure 5, where the normalized current is plotted as a
function of the normalized frequency.
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 I2 1.0 2
related to the device, except for V CE(sat). The load value Io 4
5 2 Qo = 1
is primarily dictated by the required output power and the 0.8 6
peak voltage; it is not matched to the output impedance of 8
the device. 10 4 Qo = 2
0.6 5
At higher frequencies this approximation becomes less 6
exact and for microwave devices the load that must be 8
0.4 10 Qo = 3
presented to the device is indicated on the data sheet. This
parameter will be measured on all Motorola RF-power 0.5 0.7 0.9 1.1 1.3 1.5
devices in the future. f/fo LOW-PASS — fo/f HIGH-PASS
Strictly speaking, impedance matching is accomplished
R1 X2
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
E X1 R2
reactive component) that depends essentially on the power
I2
demanded and the supply voltage.

4. MATCHING NETWORKS n = R1/R2

Qo = √n – 1
In the following matching networks will be described by
order of complexity. These are ladder type reactance Figure 5. Normalized Frequency Response for the
networks. L-Section in Low-Pass or High-Pass Form

2 RF Application Reports
 AN721
If X1 is capacitive and consequently X2 inductive, then:
fo R2 fo 1 R2 Xint Xext Xext = X2 – Xint
X1 = – R1 =– R1
f R1 – R 2 f n–1

f f n–1
and X2 = R2 (R1 – R2) = R1
fo fo 1
X1 ⋅ Xint
The normalized current absolute value is equal to: R1 Xint Xext Xext =
Xin - X1

I2 2 n
=
Io f 4 f 2
(n – 1)2 ⋅ ǒ Ǔ – 2 ǒ Ǔ + (n + 1)2
fo fo Figure 6. Termination Reactance Compensation

4.1.1.1 Use of transmission lines and inductors

nE
where Io = , and is plotted in Figure 5 (Ref. (2)).
2 ⋅ R1 In the preceding section, the inductance was expected
to be realized by a lumped element. A transmission line can
If X1 is inductive and consequently X2 capacitive, the only be used instead (Fig 7).
change required is a replacement of f by fo and vice-versa.
The L-section has low pass form in the first case and R1 L R1 Zo(Θ,L)
high-pass form in the second case.
The Q of the circuit at fo is equal to:
C R2 C R2
X2 R1
Qo = = = n–1
R2 X1
(a) (b)
For a given transformation ratio n, there is only one
possible value of Q. On the other hand, there are two Figure 7. Use of a Transmission Line in the L-Section
symmetrical solutions for the network, that can be either a
low-pass filter or a high-pass filter. As can be seen from the computed selectivity curves (Fig.
The frequency fo does not need to be the center 8) for the two configurations, transmission lines result in a
frequency, (f1 + f2)/2, of the desired band limited by f1 larger bandwidth. The gain is important for a transmission
and f2. line havlng a length L = λ/4 (Θ = 90°) and a characteristic
In fact, as can be seen from the low-pass configuration impedance Zo=√ R1 ⋅ R2. It is not significant for lines short
of Figure 5, it may be interesting to shift fo toward the high with respect to λ/4. One will notice that there is an infinity
band edge frequency f2 to obtain a larger bandwidth w, where of solutions, one for each value of C, when using
2 (f1 + f2)
transmission lines.
w=
f2 – f1 4.1.2 Three-reactance matching networks
This will, however, be at the expense of poorer harmonic The networks which will be investigated are shown in
rejection. Figure 9. They are made of three reactances alternatively
Example: connected in series and shunt.
A three-reactances configuration allows to make the
For a transformation ratio n = 4, it can be determined from quality factor Q of the circuit and the transformation ratio n
the above relations: = R2/R1 independent of each other and consequently to
Bandwidth w 0.1 0.3 choose the selectivity between certain limits.
Max insertion losses 0.025 0.2 For narrow band designs, one can use the following
X1/R1 1.730 1.712 formulas (Ref. (5) AN-267, where tables are given):
If the terminations R1 and R2 have a reactive component Network (a):
X, the latter may be taken as part of the external reactance XC1 = R1/Q Q must be first selected
as shown in Figure 6.
This compensation is applicable as long as R 1 R2
XC2 = R2
R1
XINT R1 (Q2 + 1) –
QINT = or <n–1 R2
R2 XINT
QR1 + (R1R2/XC2)
Tables giving reactance values can be found in Ref. (3) XL =
Q2 + 1
and (4).

RF Application Reports 3
 AN721
POUT rel.

1.0
.9

.8
(b)
.7

.6
.5

.4
(a)
.3
.2

.1

f/fo
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

RG L RG Z(Θ)

C 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 (λ/4)

(a) XL (b) XL1 XL2 Network (b):

XL1 = R1Q Q must be first selected
XL2 = R2 ⋅ B A = R1 (1 + Q2)
R1 XC1 XC2 R2 R1 XC1 R2
A A
XC1 = B= –1
Q+B R2

(c)
XL XC2 Network (c):
XL1 = Q ⋅ R1 Q must be first selected
R2 > R1
R1 XC1 R2 R1 (1 + Q2)
XC2 = A ⋅ R2 A= –1
R2

B
XC1 = B = R1 ⋅ (1 + Q2)
Figure 9. Three-Reactance Matching Networks Q–A
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.

4 RF Application Reports
 AN721
L1 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).
R1 C1 C2 R2
L2 C1

Z′ Z″
R1 2
R1 C1 R2(G2) C2 R1(G1)
Z′ = + jω L1 – = R′ + jX′
1 + ω2 R12 C12 1 + ω2 R12 C12

R2 jω2 C2 R22 Y″ Y′
Z″ = – = R″ + jX″
1 + ω C2 R2
2 2 2 1 + ω2 C22 R22
ω2C12R1 jωC1
Y′ = G′ + jB′ = –
1 + ω2 C12 R12 1 + ω2 C12 R12
L1 R′ R1
 x′ =ǒ – Ǔ –1 R2 jωL2
C1 R12 R1 R′ Y″ = G″ + jB″ = + jωC2 –
R2 + ω2L22
2 R22 + ω2L22
x′, – x″
 x″ = R′ R2 – R′2
B′ = – G′/R′1 – G′2 B″ = (1/kR2 – G″) 1/G″R2 – 1
x′

ω k =L2/C2R22
–x″ M
ω M′ B′
(f1)
N′ ω B″
R′ M
R2 R1 G′ or G″
G1 G2
ω
Figure 10. Z-Plane Representation of the Circuit of
Figure 9(a)
B′ .– B″

Exact transformation from R1 into R2 occurs at the points Figure 11. Y-Plane Representation of the Circuit
of intersection M and N. Impedances are then conjugate or of Figure 9(c)
Z′ = R′ + jX′ and Z″ = R″ + jX″ with R′ = R″ and X′ = – X″.
The only possible solution is obtained when X′ and – X″ 4.1.3 Four-reactance networks
are tangential to each other. For the dashed curve, Four-reactance networks are used essentially for
representing another value of L1 or C1, a wider frequency broadband matching. The networks which will be considered
band could be expected at the expense of some ripple inside in the following consist of two two-reactance sections in
the band. However, this can only be reached with four cascade. Some networks have pseudo low-pass filter
reactances as will be shown in section 4.1.3. character, others band-pass filter character. In principle, the
With a three-reactance configuration, there are not former show narrower bandwidth since they extend the
enough degrees of freedom to permit X′ = – X″ and impedance transformation to very low frequencies unneces-
simultaneously obtain the same variation of frequency on sarily, while the latter insure good matching over a wide
both curves from M′ to point N′. frequency band around the center frequency only (see
Exact transformation can, therefore, only be obtained at Fig. 14).
one frequency.
The values of the three reactances can be calculated by a) X2 M X3 b) X2 M X4
making
dX′ dX″ X1 X4 X1 X3
X′ = – X″, R′ = R″ and =– . R1 R2 R1 R2
dR′ dR″

The general solution of these equations leads to

Z1 Z2
complicated calculations. Therefore, computed tables should c) d)
X2 M X1 X2 M X4
be used.
One will note on Figure 10 that the compensation
reactance contributes somewhat to impedance transforma- X1 X3
R1 R R1 R2
tion, i.e. R′ varies when going from M to R2. X3 X4 2
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. R1 = n.R2
Circuit of Figure 9 (c) is somewhat different since only Figure 12. Four-Reactance Networks
one intersection M exists as shown in Figure 11. Narrower
frequency bands must be expected from this configuration. The two-reactance sections used in above networks have
The widest band is obtained for C1 = ∞. either transformation properties or compensation properties.

RF Application Reports 5
 AN721
Impedance transformation is obtained with one series in Figure 14. Figure 15 shows the selectivity for networks
reactance and one shunt reactance. Compensation is made (a) and (b) when the source resistance R1 is infinite.
with both reactances in series or in shunt. From Figure 15 it can be seen that network (a) is more
If two cascaded transformation networks are used, sensitive to R1 changes than network (b).
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. 0 + j1
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. Q2
0.28 + j0.45 NORMALIZED IMP.
At all points (M), Z1 and Z2 are conjugate if the transformation 0.1 + j0.16 (1 – j1.6) NORMALIZED ADM.
Q1
is exact. (2.8 – j4.5)
(a)
In the case of Figure 12 (b) the reactances are easily, b′3 b1
calculated for equal frequencies: 0.1 + j0.147
Q′1 (0.316 b3
– j4.65) 0.316 – j0
∞

x4
x’4
1 + j0
R1 n–1
X1 = , X2 = R1 0 0.1

x2
n
n–1 x′2 b′1

X1 ⋅ X4 = R1 ⋅ R2 = X2 ⋅ X3 0.28 – j0.14
(2.8 + j1.4) (b)

R1 R1 0.316 – j0.465
X3 = , X4 = n–1 Q′2
(1 + j1.47)
n
n( n – 1)
R1 = 1
For network (a) normally, at point (M), Z1 and Z2 are
R2 = 0.1
complex. This pseudo low-pass filter has been computed
elsewhere (Ref. (3)). Many tables can be found in the
(a) X1 = 0.624 X3 = 0.169 X′1 = 0.68 X′3 = 0.215
literature for networks of four and more reactances having ( ( ( (
Tchebyscheff character or maximally-flat response (Ref. (3), B1 = 1.6 B3 = 5.9 B′1 = 1.47 B′3 = 4.65
(4) and (6)). X2 = 0.59 X4 = 0.160 X′2 = 0.465 X′4 = 0.147
Figure 13 shows the transformation path from R1 to R2
for networks (a) and (b) on a Smith-Chart (refer also to Q1 = 1.6 = Q2 Q′1 = Q′2 = 1.47
section 4.2, Graphic Design).
(a) X2 X4 (b) X′2 X′4
Case (a) has been calculated using tables mentioned in
Ref. (4).
Case (b) has been obtained from the relationship given R1 X1 X3 R2 R1 X′1 X′3 R2
above for X1 . . . X4. Both apply to a transformation ratio
equal to 10 and for R1 = 1.
There is no simple relationship for X′1 . . . X′4 of network
(b) if f1 is made different from f2 for larger bandwidth.
Figure 13. Transformation Paths for Networks
Figure 14 shows the respective bandwidths of network
(a) and (b)
(a) and (b) for the circuits shown in Figure 13.
If the terminations contain a reactive component, the
computed values for X1 or X 4 may be adjusted to
ATTEN.
compensate for this.
–0
For configuration (a), it can be seen from Figure 13, that –1
(a)
in the considered case the Q’s are equal to 1.6. –2
For configuration (b) Q′1, which is equal to Q′2, is fixed –3
(b)
for each transformation ratio. –4
–5
(b)
n 2 4 8 10 16 –6 (a)
Q′ = n–1
Q′1 = Q′2 0.65 1 1.35 1.46 1.73 –7
–8
–9
The maximum value of reactance that the terminations
–10 f/fo
may have for use in this configuration can be determined
from the above values of Q′. .3 .4 .5 .6 .7 .8 .9 1 1.1 1.3 1.5 1.7 1.9
If R1 is the load resistance of a transistor, the internal
transistor resistance may not be equal to R1. In this case Figure 14. Selectivity Curves for Networks (a) and (b)
the selectivity curve will be different from the curves given of Figure 13

6 RF Application Reports
 AN721
REL. ATTEN. L2 L1 C1

2
1 R2 C2 R2
0 (G2) (G1)
–1
–2
–3 Y″ Y′
–4
R1 j(ωL1 – 1/ωc1)
–5 (a) (b) (a) (b) Y′ = G′ + jB′ = –
–6 R12 + (ωL1 – 1/ωc1)2 R12 + (ωL1 – 1/ωc1)2
–7
–8 R2 L2
Y″ = G″ + jB″ = + jω ǒC2 – Ǔ
–9 R22 + ω2L22 R22 + ω2L22
–10 f/fo
.4 .5 .6 .7 .8 .9 1 1.1 1.3 1.5 1.7 G′
B′ = ± – G′2
R1
1 L2
Figure 15. Selectivity curves for networks (a) and (b) B″ = (1/kR2 – G″) –1 k =
G″R2 C2 R22
of Figure 13 with Infinite R1
B′
As mentioned earlier, the four-reactance network can also G1 G2
be thought of as two cascaded two-reactance sections; one G
M′
used for transformation, the other for compensation. ω B″
N′
Figure 16 shows commonly used compensation networks, M
together with the associated L-section.
ω
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. B

COMPENSATION NETWORK Figure 17. Y-Plane Representation of the Circuit

(a)
of Figure 16(a)

When M′ and N′ coincide in, M, the new dX′/df = dX″/df

R2 R1 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.
(b) C.N.
Again, a general solution of the above equations leads
to still more complicated calculations than in the case of
R2 three-reactance networks. Therefore, tables are preferable
R1
(Ref. (3), (4) and (6)).
The circuit of Figure 16(b) is dual of the circuit of
R1 = n.R2 Figure 14(a) and does not need to be treated separately. It
(c) C.N. C.N. gives exactly the same results in the Z-plane. Figure 16(c)
shows a higher order compensation requiring six reactive
elements.
R2 R1 The above discussed matching networks employing
compensation circuits result in narrower bandwidths than the
former solutions (see paragraph 4.1.3) using two trans-
formation sections. A matching with higher order com-
pensation such as in Figure 16(c) is not recommended.
Figure 16. Compensation Networks Used with Better use can be made of the large number of reactive
a L-Section elements using them all for transformation.
When the above configurations are realized using short
By adding one reactance, exact impedance transforma- portions of transmission lines, the equations or the usual
tion is achieved at two frequencies. It is now possible to tables no longer apply. The calculations must be carried out
choose component values such that the point of intersection on a computer, due to the complexity. However, a graphic
M′ occurs at the same frequency f1 on both curves and method can be used (see next section) which will consist
simultaneously that N′ occurs at the same frequency f2 on essentially in tracing a transformation path on the Z-Y-chart
both curves. Among the infinite number of possible using the computed lumped element values and replacing
intersections, only one allows to achieve this. it by the closest path obtained with distributed constants. The

RF Application Reports 7
 AN721
bandwidth change is not significant as long as short portions 1.0
of lines are used (Ref. (13)). n=4
0.9
4.1.4 Matching networks using quarter-wave
transformers 0.8

At sufficiently high frequencies, where λ/4-long lines of 0.7 1 2 3

practical size can be realized, broadband transformation can
easily be accomplished by the use of one or more
λo/λ
λ/4-sections. 0.6 0.8 1.0 1.2 1.4
Figure 18 summarizes the main relations for (a) one-
section and (b) two-section transformation. Figure 20. Selectivity Curves for One, Two and Three
A compensation network can be realized using a λ/2-long λ/4-Sections
transmission line.
4.1.5 Broadband matching using band-pass filter type
networks. High Q case.
(a)
The above circuits are applicable to devices having low
R1 = nR2
R2 input or output Q, if broadband matching is required.
Z R1
R1 Generally, if the impedances to be matched can be
Z = R1 R2 = R2 √n =
√n 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
(b) 4 incorporated into the first inductor of the matching network.
Z1 = R13 R2 This is also valid if the representation consists of a shunt
R2 R1 4 combination of a resistor and a reactance.
Z2 Z1 Z2 = R1 R23 Practically this is feasible for Q’s around one or two. For
higher Q’s or for input impedances consisting of a series or
at A, R = √R1 R2 parallel resonant circuit (see Fig. 2), as it appears to be for
A large bandwidths, a different treatment must be followed.
Let us first recall that, as shown by Bode and Fano (Ref.
Figure 18. Transformation Networks Using λ/4-Long (7) and (8)), limitations exist on the impedance matching of
Transmission Lines 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.
Figures 19 and 20 show the selectivity curves for different RG
transformation ratios and section numbers.
MATCHING
Exponential lines
V ZIN NETWORK C R
Exponential lines have largely frequency independent (LOSSLESS)
transformation properties.
The characteristic impedance of such lines varies
exponentially with their length I: Figure 21. General Matching Conditions
Z = Zo ⋅ ekl
The reflection coefficient between transformed load and
where k is a constant, but these properties are preserved generator is equal to:
only if k is small. ZIN – Rg
Γ=
1.0 ZIN + Rg
Γ = 0, perfect matching,
0.9 Γ = 1, total reflection.
The ratio of reflected to incident power is:
0.8
Pr
=  Γ 2
0.7 n=6 4 2 Pi
The fundamental limitation on the matching takes the

∫ω = olnǒ 1Γ Ǔ dω v RCπ

λo/λ form:
0.6 0.8 1.0 1.2 1.4 ∞
Bode equation
Figure 19. Selectivity Curves for Two λ/4-Section
Networks at Different Transformation Ratios and is represented in Figure 22.

8 RF Application Reports
 AN721
1 Thus, devices having relatively high input Q’s are usable
ln
Γ for broadband operation, provided the consequent higher
attenuation or reflection introduced is acceptable.
The general shape of the average insertion losses or
π attenuation a (neglecting the ripple) of a low-pass impedance
S=
RC matching network is represented in Figure 24 as a function
of 1/Q for different numbers of network elements n (ref. (3)).
ωc ω For a given Q and given ripple, the attenuation decreases
if the number n of the network element increases. But above
Figure 22. Representation of Bode Equation n = 4, the improvement is small.
For a given attenuation a and bandwidth, the larger n the
The meaning of Bode equation is that the area S under the smaller the ripple.
curve cannot be greater than π/RC and therefore, if matching For a given attenuation and ripple, the larger n the larger
is required over a certain bandwidth, this can only be done at the bandwidth.
the expense of less power transfer within the band. Thus, Computations show that for Q < 1 and n ≤ 3 the
power transfer and bandwidth appear as interchangeable attenuation is below 0.1 db approximately. The impedance
quantities. transformation ratio is not free here. The network is a true
It is evident that the best utilization of the area S is low-pass filter. For a given load, the optimum generator
obtained when Γ is kept constant over the desired band impedance will result from the computation.
ωc and made equal to 1 over the rest of the spectrum. Then Before impedance transformation is introduced, a
conversion of the low-pass prototype into a band-pass filter
π
– type network must be made. Figure 25 summarizes the main
Γ = e ωc RC
relations for this conversion.
within the band and no power transfer happens outside. r is the conversion factor.
A network fulfilling this requirement cannot be obtained For the band-pass filter, QlN max or the maximum possible
in practice as an infinite number of reactive elements would input Q of a device to be matched, has been increased by
be necessary. the factor r (from Figure 25, Q′IN max = r ⋅ QIN max).
If the attenuation a is plotted versus the frequency for Impedance inverters will be used for impedance
practical cases, one may expect to have curves like the ones transformation. These networks are suitable for insertion into
shown in Figure 23 for a low-pass filter having Tchebyscheff a band-pass filter without affecting the transmission
character. characteristics.

2 a a
a 1
LOW-PASS BAND-PASS
a max2

a min2

a max1
ωa ωo ωb
a min1 ∆ω
wc =
ω 2 ∆ω
ω1 ω2 ωo
r=
∆ω
Figure 23. Attenuation versus Angular Frequency for C1′
Different Bandwidths with Same Load
L1 L3 L1′
R5 R5
For a given complex load, an extension of the bandwidth Ro C2 C4 Ro C ′
from ω1 to ω2 is possible only with a simultaneous increase 2 L2′
of the attenuation a. This is especially noticeable for Q’s
exceeding one or two (see Figure 24).
ωcL1 1
QIN max = L1′ = r.L1 C1′ =
dB Ro ωo2 L1′
4 ωoL1′ 1
QIN max = C2′ = r.C2 L2′ =
3 Ro ωo2 C2′
n=2 etc.
a 2
3
Figure 25. Conversion from Low-Pass into
1 ∞ Band-Pass Filter
0
0.1 1 1/Q Figure 26 shows four impedance inverters. It will be
noticed that one of the reactances is negative and must be
Figure 24. Insertion Losses as a Function of 1/Q combined in the band-pass network with a reactance of at

RF Application Reports 9
 AN721
least equal positive value. Insertion of the inverter can be (4) Convert the element values found by step (3) into
made at any convenient place (Ref. (3) and (9)). series or parallel resonant circuit parameters,
(5) Insert the impedance inverter in any convenient
INVERTER EQUIVALENT TO: place.
C C In the above discussions, the gain roll-off has not been
1 – 1/n 1/n2 – 1/n 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
a) nC C despite the intrinsic gain decrease of a transistor with
frequency.
Tables have been computed elsewhere (Ref. (10)) for
n:1
matching networks approximating 6 db/octave attenuation
nC C
versus frequency.
Another method consists in using the above mentioned
b) (1 – n)C (n2 – n)C network and then to add a compensation circuit as shown
for example in Figure 27.

L2 C2
n:1
L(1 – 1/n) L(1/n2 – 1/n)
R = ZIN = CONSTANT vs. f.
R1
L1C1 = L2C2
c)
L/n L R R1 = R
L1 C1 ωb2 = 1/L1C1 = 1/L2C2

n:1 (HIGH BAND EDGE)

L/n Q2 = ωbL2/R = Q1 = ωbC1R
L
d) L L
1–n n2 – n
Figure 27. Roll-Off Compensation Network

n:1 Resonance ωb is placed at the high edge of the frequency

band. Choosing Q correctly, roll-off can be made 6 db/octave.
Figure 26. Impedance Inverters The response of the circuit shown in Figure 27 is
expressed by:
When using the band-pass filter for matching the input 1
, where ω < ωb
impedance of a transistor, reactances L′1 C′1 should be ω ωb 2
1 + Q2 ǒ ω – ω Ǔ
made to resonate at ωo by addition of a convenient series b
reactance.
This must be equal to ω/ωb for 6 db/octave compensation.
As stated above, the series combination of Ro, L′1 and
At the other band edge a, exact compensation can be
C′1 normally constitutes the equivalent input network of a
obtained if:
transistor when considered over a large bandwidth. This is
a good approximation up to about 500 MHz. ωb 2
ǒ Ǔ –1
In practice the normal procedure for using a bandpass ωa
Q=
filter type matching network will be the following: ωa ωb 2
– ǒ Ǔ
ωb ωa
(1) For a given bandwidth center frequency and input im-
pedance of a device to be matched e.g. to 50 ohms,
first determine QlN′ from the data sheet as 4.1.6 Line Transformers
ωoL′1 The broadband properties of line transformers make them
Ro very useful in the design of broadband impedance matching
networks (Ref. (11) and (12)).
after having eventually added a series reactor for cen-
A very common form is shown by Figure 28. This is a
tering,
4:1 impedance transformer. Other transformation ratios like
(2) Convert the equivalent circuit RoL1′ C1′ into a low- 9:1 or 16:1 are also often used but will not be considered
pass prototype Ro L1 and calculate QlN using the for- here.
mulas of Figure 25, The high frequency cut-off is determined by the length
(3) Determine the other reactance values from tables of line which is usually chosen smaller than λ min/8. Short
(Ref. (3)) for the desired bandwidth, lines extend the high frequency performance.

10 RF Application Reports
 AN721

Z0 = Rg/2
Rg
Rg
RL =
4

Figure 28. 4:1 Line Transformer Line Transformer on P.C. Board (Figure 29)

35 µ
The low frequency cut-off is determined first by the length Cu 66 µ
of line, long lines extending the low frequency performance Ribbon thickness
Electrical tape
of the transformer. Low frequency cut-off is also improved
EE3990
by a high even mode impedance, which can be achieved Permacel New Brunswick
by the use of ferrite material. With matched ends, no power
ISOLATION
is coupled through the ferrite which cannot saturate. Cu
ISOLATION
For matched impedances, the high frequency attenuation
a of the 4:1 transformer is given by:

(1 + 3 cos 2πI/λ)2 + 4 sin2 2πI/λ FERRITE

a=
4 (1 + cos 2πI/λ)2

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

The characteristic impedance of the line transformer must
be equal to:
Zo = √Rg ⋅ R .
25 mm
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 – Stick one ribbon (1.5 mm wide) against the other (2.5)
or two ribbons stuck together and connected as shown in Total length per ribbon – 9 cm.
Figures 29 and 30. Turns 3.5

Figure 30. 4:1 Copper Ribbon Line Transformer

EPOXY 0.2 mm THICK
with Ferrite
35 µ

ÉÉÉÉÉÉÉ
ÉÉÉÉÉÉÉ 15 22
in mm

ÉÉÉÉÉÉÉ
Cu

10
61.5
68.5
Line Transformer on Ferrite (Figure 30)

ÉÉÉÉÉÉÉÉÉÉÉÉÉ
STICK ONE AGAINST THE OTHER

ÉÉÉÉÉÉÉÉÉÉÉÉÉ
ÉÉÉÉÉÉÉÉÉÉÉÉÉ
4.2 GRAPHIC DESIGN

ÉÉÉÉÉÉÉÉÉÉÉÉÉ
The common method of graphic design makes use of the

ÉÉÉÉÉÉÉÉÉÉÉÉÉ
Impedance-Admittance Chart (Smith Chart).
It is applicable to all ladder-type networks as encountered

ÉÉÉÉÉÉÉÉÉÉÉÉÉ
in matching circuits.

ÉÉ É
ÉÉÉÉÉÉÉÉÉÉÉÉÉ
Matching is supposed to be realized by the successive
algebraic addition of reactances (or susceptances) to a given

ÉÉ É ÉÉ
ÉÉÉÉÉÉÉÉÉÉÉÉÉ start impedance (or admittance) until another end impedance

ÉÉ ÉÉ
ÉÉÉÉÉÉÉÉÉÉÉÉÉ
(or admittance) is reached.
Impedance chart and admittance chart can be super-
imposed and used alternatively due to the fact that an
immittance point, defined by its reflection coefficient Γ with
respect to a reference, is common to the Z-chart and the
Figure 29. 4:1 Line Transformer on P.C. Board Y-chart, both being representations in the Γ-plane.

RF Application Reports 11
 AN721

Z – Rs γi
Γ= γr + jγ i +1
Z + Rs b = – 0.5
g = 0.5 INDUCTIVE
Gs – Y
Γ= b<0
Gs + Y g=1
b = –1 γr
b=0
1
Rs = = Characteristic impedance of the line –1 +1
Gs SHORT b=1
OPEN
More precisely, the Z-chart is a plot in the Γ-plane, while CAPACITIVE
the Y-chart is a plot in the – Γ-plane. The change from the g=0 b>0
Γ to – Γ-plane is accounted for in the construction rules given –1 b = 0.5
below. g>0
Figure 31 and 32 show the representation of normalized – Γ-PLANE
Z and Y respectively, in the Γ-plane.
The Z-chart is used for the algebraic addition of series Y
= y = g + jb
reactances. The Y-chart is used for the algebraic addition Gs
of shunt reactances. 1 – γr2 – γ i2
g=
For the practical use of the charts, it is convenient to make
(1 + γr)2 + γ i2
the design on transparent paper and then place it on a usual INDUCTIVE
Smith-chart of impedance type (for example). For the – 2γ i
b= g
addition of a series reactor, the chart will be placed with (1 + γr)2 + γ i2
“short” to the left. For the addition of a shunt reactor, it will 1 – g2 – b2 CAPACITIVE
be rotated by 180° with “short” (always in terms of γr =
impedance) to the right. (1 + g)2 + b2
– 2b b
γi γi =
+1 (1 + g)2 + b2 RECTANGULAR COORDINATES
x = 0.5

INDUCTIVE Radius: 1/(1 + g)

r=1 g circles
r=0 x>0 Center at: γr = – g/(1 + g); γ i = 0
x=1
r = 0.5 γr
–1 0 +1
SHORT x=0 x=–1 Radius: 1/b
OPEN b circles
Center at: γr = – 1, γ i = – 1/b
CAPACITIVE
x = – 0.5 x<0
–1 Figure 32. Representation of the Normalized Y-Values in
the Γ-Plane
r>0
Γ-PLANE The following design rules apply. They can very easily
be found by thinking of the more familiar Z and Y
Z representation in rectangular coordinates.
= z = r + jx
Rs x
For joining two impedance points, there are an infinity of
1 – γr2 – γ i2 solutions. Therefore, one must first decide on the number
r=
(1 – γr)2 + γ i2 of reactances that will constitute the matching network. This
INDUCTIVE
number is related essentially to the desired bandwidth and
2γ i r
x= the transformation ratio.
(1 – γr)2 + γ i2
CAPACITIVE Chart Using
r2 + x2 – 1
γr = to be Curve of
(r + 1)2 + x2 Addition of Used Direction Constant
2x RECTANGULAR COORDINATES
γi = Series R Z Open (in terms of x
(r + 1)2 + x2 admittance)
Series G Y Short (in terms of b
Radius: 1/(1 + r) admittance)
r circles r
Center at: γr = , γi = 0 Series C (+ 1/jωC) Z ccw r
1+r
Shunt C (+ jωC) Y cw g
Radius: 1/x
x circles Series L (+ jωL) Z cw r
Center at: γr =1 , γ i = 1/x Shunt L (+ 1/jωL) Y ccw g

Figure 31. Representation of the Normalized Z Values in Secondly, one must choose the operating Q of the circuit,
the Γ-Plane which is also related to the bandwidth. Q can be defined at

12 RF Application Reports
 AN721
each circuit node as the ratio of the reactive part to the real [Vcc – Vce (sat)]2 (28 – 3)2
part of the impedance at that node. The Q of the circuit, which or = 15.6 ohms .
2 x Pout 40
is normally referred to, is the highest value found along the
path.
The collector capacitance given by the data sheet is
Constant Q curves can be superimposed to the charts
40 pF, corresponding to a capacitive reactance of 22.7 ohms.
and used in conjunction with them. In the Γ-plane Q-curves
The output impedance seen by the collector to insure the
are circles with a radius equal to
required output power and cancel out the collector
1 capacitance must be equal to a resistance of 15.6 ohms in
1+ parallel with an inductance of 22.7 ohms. This is equivalent
Q2
to a resistance of 10.6 ohms in series with an inductance
of 7.3 ohms.
and a center at the point ± 1/Q on the imaginary axis, which
The input Q is equal to, 1.1/1.94 or 0.57 while the output

ǒ Ǔ
is expressed by:
Q is 7.3/10.6 or 0.69.
x 2γ i 1 2 1 It is seen that around this frequency, the device has good
Q= = γ r2 + γ i + =1+ .
r 1 – γ r2 – γ i2 Q Q2 broadband capabilities. Nevertheless, the matching circuit
will be designed here for a narrow band application and the
The use of the charts will be illustrated with the help of effective Q will be determined by the circuit itself not by the
an example. device.
The following series shunt conversion rules also apply: Figure 34 shows the normalized impedances (to
50 ohms).

X 0.022 0.146
R ≅ R′ X′

0.039 0.212

R′ 1 R
R= G′ = =
R′2 R′ R2 + X2
1+ NORMALIZED INPUT IMPEDANCE LOAD IMPEDANCE
X′2
Figure 34. Normalized Input and Output Impedances for
X′ 1 X the 2N5642
X= – B′ = =
X′2 X′ R2 + X2
1+ Figure 35 shows the diagram used for the graphic design
R′2
of the input matching circuit. The circuit Q must be larger
Figure 33 shows the schematic of an amplifier using the than about 5 in this case and has been chosen equal to 10.
2N5642 RF power transistor. Matching has to be achieved At Q = 5, C1 would be infinite. The addition of a finite value
at 175 MHz, on a narrow band basis. of C1 increases the circuit Q and therefore the selectivity.
The normalized values between brackets in the Figure are
VCC = 28 V admittances (g + jb).

0 + j1
RFC
L4 PARALLEL CAPACITANCE C2

Q– (b2)
C1 L3 0.039 + j0.39
CIRCLE (0.25 – j2.5) 1 + j1.75
(10)
(0.25 – j0.42)
RFC SERIES
50 Ω 50 Ω INDUCTANCE L3 SERIES
C2 C0 C5
(x3) CAPACITANCE C1
(x1)
0 0.039 + j0.022 1 + j0 ∞
2N5642
Figure 35. Input Circuit Design
Figure 33. Narrow-Band VHF Power Amplifier
At f = 175 MHz, the following results are obtained:
ωL3 = 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 N
L3 = 16.8 nH
at these conditions is equal to 2.6 ohms in parallel with – 1 1
ωC2 = b2 = (2.5 – 0.42) = 0.0416 mhos
200 pF (see data sheet). This converts to a resistance of 50 50
1.94 ohms in series with a reactance of 1.1 ohm.
The collector load must be equal to: N C2 = 37.8 pF

RF Application Reports 13
 AN721

1 Rated output power:

= 50x1 = 50 ⋅ 1.75 = 87.5 ohms
ωC1 30 W for 8 W input at 175 MHz. From the data sheet it
N C1 = 10.4 pF
appears that at 125 MHz, 30 W output will be achieved with
about 4 W input.
Figure 36 shows the diagram for the output circuit,
Output impedance:
designed in a similar way.

0 + j1 [Vcc – Vce (sat)]2 100

= = 1.67 ohms
2 x Pout 60

Cout = 180 pF at 125 MHz

0.212 + j0.146
5.2 CIRCUIT SCHEMATIC
1 + j0
0 ∞ 12.5 V
C7

SERIES INDUCTANCE L4
(x4) PARALLEL CAPACITANCE C5
0.212 – j0.4 (b5)
(1 + j1.9) RFC2
C6
Figure 36. Output Circuit Design

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

ωL4 = 50 ⋅ x4 = 50 ⋅ (0.4 + 0.146) = 27.3 ohms RFC1
N L4 = 24.8 nH L3 C5

1 1 L2
ωC5 = b5 = ⋅ 1.9 = 0.038 mhos
50 50

N C5 = 34.5 pF T1 C1
C4 T2

The circuit Q at the output is equal to 1.9. 50 Ω 50 Ω

The selectivity of a matching circuit can also be
determined graphically by changing the x or b values
C1 = 300 pF (chip) T1, T2 see Figure 30
according to a chosen frequency change. The diagram will L2 = 4.2 nH (adjust) RFC1 = 7 turns
give the VSWR and the attenuation can be computed. L3 = 8 nH (adjust) 6.3 mm coil diameter
The graphic method is also useful for conversion from C4 = 130 pF (chip) 0.8 mm wire diameter
a lumped circuit design into a stripline design. The C5 = 750 pF (chip) RFC2 = 3 turns on ferrite bead
immittance circles will now have their centers on the 1 + jo C6 = 2.2 µF C7 = 0.68 µF
point.
At low impedance levels (large circles), the difference Figure 37. Circuit Schematic
between lumped and distributed elements is small.

5. PRACTICAL EXAMPLE 5.3 TEST RESULTS

The example shown refers to a broadband amplifier stage W

using a 2N6083 for operation in the VHF band 118 – 30
PIN = 2 W
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
20
associated with a feedback system for distortion compensa-
tion. 1W
POUT
Line transformers will be used at the input and output.
Therefore the matching circuits will reduce to two-reactance 10
networks, due to the relatively low impedance transformation 0.5 W
ratio required.
MHz
5.1 DEVICE CHARACTERISTICS 0
118 124 128 132 136
Input impedance of the 2N6083 at 125 MHz: f
Rp = 0.9 ohms
Cp = – 390 pF Figure 38. POUT vs Frequency

14 RF Application Reports
 AN721
% Acknowledgments:
PIN = 2 W The author is indebted to Mr. T. O’Neal for the fruitful
70
discussions held with him. Mr. O’Neal designed the circuit
60 shown in Figure 37; Mr. J. Hennet constructed and tested
1W
the lab model.
50
η 40 0.5 W 6. LITERATURE
30 1. Motorola Application Note AN-282A “Systemizing RF
20 Power Amplifier Design”
2. W. E. Everitt and G. E. Anner “Communication Engi-
10
neering” McGraw-Hill Book Company, Inc.
MHz 3. G. L. Matthaei, L. Young, E. M. T. Jones “Microwave
118 124 128 132 136
Filters, Impedance-Matching Networks and Coupling
f
Structures” McGraw-Hill Book Company, Inc.
Figure 39. η vs Frequency 4. G. L. Matthaei “Tables of Chebyshew Impedance-
Transforming Networks of Low-Pass Filter Form” Proc.
db.
IEEE, August 1964
0
5. Motorola Application Note AN-267 “Matching Network
Prefl.
Designs with Computer Solutions”
Pin – 5 PIN = 2 W 6. E. G. Cristal “Tables of Maximally Flat Impedance-
– 10
Transforming Networks of Low-Pass-Filter Form” IEEE
Transactions on Microwave Theory and Techniques
1W
– 15 Vol. MTT 13, No 5, September 1965 Correspondence
0.5 W 7. H. W. Bode “Network Analysis and Feedback Amplifier
– 20 Design” D. Van Nostrand Co., N.Y.
8. R. M. Fano “Theoretical Limitations on the Broadband
MHz Matching of Arbitrary Impedances” Journal of Franklin
118 124 128 132 136 Institute, January – February 1950
f 9. J. H. Horwitz “Design Wideband UHF-Power Amplifi-
Figure 40. PREFL./PIN vs Frequency ers” 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 Mi-
crowave 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” München,
Oldenburg 1951

118 – 136 MHz Amplifier (see Figure 37)

Before Coil Adjustment

RF Application Reports 15
 AN721

Motorola reserves the right to make changes without further notice to any products herein. Motorola makes no warranty, representation or guarantee regarding
the suitability of its products for any particular purpose, nor does Motorola 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 can and do vary in different
applications. All operating parameters, including “Typicals” must be validated for each customer application by customer’s technical experts. Motorola does
not convey any license under its patent rights nor the rights of others. Motorola 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 Motorola product could create a situation where personal injury or death may occur. Should Buyer purchase or use Motorola products for any such
unintended or unauthorized application, Buyer shall indemnify and hold Motorola 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 Motorola was negligent regarding the design or manufacture of the part.
Motorola and are registered trademarks of Motorola, Inc. Motorola, Inc. is an Equal Opportunity/Affirmative Action Employer.

Literature Distribution Centers:

USA: Motorola Literature Distribution; P.O. Box 20912; Phoenix, Arizona 85036.
EUROPE: Motorola Ltd.; European Literature Centre; 88 Tanners Drive, Blakelands, Milton Keynes, MK14 5BP, England.
JAPAN: Nippon Motorola Ltd.; 4-32-1, Nishi-Gotanda, Shinagawa-ku, Tokyo 141, Japan.
ASIA PACIFIC: Motorola Semiconductors H.K. Ltd.; Silicon Harbour Center, No. 2 Dai King Street, Tai Po Industrial Estate, Tai Po, N.T., Hong Kong.

16 AN721/D
RF Application Reports
◊