422 Chapter 8: Microwave Filters

Impedance inverters Admittance inverters

K ZL J YL
± 90° ± 90°

Zin = K 2/ZL Yin = J 2/YL

(a)

/4 /4

Z0 = K Y0 = J

(b)

/2 /2 /2 /2

jB
Z0 jX Z0 Y0 Y0

K = Z0 tan ⎜/2⎜ J = Y0 tan ⎜/2⎜

K J
X= B=
1 – (K/Z0)2 1 – (J/Y0)2

 = –tan–1 2X  = –tan–1 2B
Z0 Y0
(c)
⫺C ⫺C C

C ⫺C ⫺C

K = 1/C J = C
(d)

FIGURE 8.38 Impedance and admittance inverters. (a) Operation of impedance and admittance
inverters. (b) Implementation as quarter-wave transformers. (c) Implementation
using transmission lines and reactive elements. (d) Implementation using capacitor
networks.

8.6 STEPPED-IMPEDANCE LOW-PASS FILTERS

A relatively easy way to implement low-pass filters in microstrip or stripline is to use alter-
nating sections of very high and very low characteristic impedance lines. Such filters are
usually referred to as stepped-impedance, or hi-Z , low-Z filters, and are popular because
they are easier to design and take up less space than a similar low-pass filter using stubs.
Because of the approximations involved, however, their electrical performance is not as
good, so the use of such filters is usually limited to applications where a sharp cutoff is not
required (for instance, in rejecting out-of-band mixer products).

Approximate Equivalent Circuits for Short Transmission Line Sections

We begin by finding the approximate equivalent circuits for a short length of transmis-
sion line having either a very large or a very small characteristic impedance. The ABCD
 8.6 Stepped-Impedance Low-Pass Filters 423

jX jX
2 2

X = Z 0
l
jB

(a) (b)

B = Y0
l

(c)

FIGURE 8.39 Approximate equivalent circuits for short sections of transmission lines. (a) T-
equivalent circuit for a transmission line section having β  π/2. (b) Equivalent
circuit for small β and large Z 0 . (c) Equivalent circuit for small β and small Z 0 .

parameters of a length of line having characteristic impedance Z 0 are given in Table 4.1;
the conversion in Table 4.2 can then be used to find the impedance parameters as
A
Z 11 = Z 22 = = − j Z 0 cot β , (8.81a)
C
1
Z 12 = Z 21 = = − j Z 0 csc β . (8.81b)
C
The series elements of the T-equivalent circuit are
cos β − 1 β
Z 11 − Z 12 = − j Z 0 = j Z 0 tan , (8.82)
sin β 2
while the shunt element of the T-equivalent is Z 12 . If β < π/2, the series elements have a
positive reactance (inductors), while the shunt element has a negative reactance (capacitor).
We thus have the equivalent circuit shown in Figure 8.39a, where
X β
= Z 0 tan , (8.83a)
2 2
1
B= sin β . (8.83b)
Z0
Now assume a short length of line (say β < π/4) and a large characteristic impedance.
Then (8.83) approximately reduces to
X Z0β , (8.84a)
B 0, (8.84b)
which implies the equivalent circuit of Figure 8.39b (a series inductor). For a short length
of line and a small characteristic impedance, (8.83) approximately reduces to
X 0, (8.85a)
B Y0 β , (8.85b)
which implies the equivalent circuit of Figure 8.39c (a shunt capacitor). So the series induc-
tors of a low-pass prototype can be replaced with high-impedance line sections (Z 0 = Z h ),
and the shunt capacitors can be replaced with low-impedance line sections (Z 0 = Z ). The
424 Chapter 8: Microwave Filters

ratio Z h /Z should be as large as possible, so the actual values of Z h and Z are usually set
to the highest and lowest characteristic impedance that can be practically fabricated. The
lengths of the lines can then be determined from (8.84) and (8.85); to get the best response
near cutoff, these lengths should be evaluated at ω = ωc . Combining the results of (8.84)
and (8.85) with the scaling equations of (8.67) allows the electrical lengths of the inductor
sections to be calculated as
L R0
β = (inductor) (8.86a)
Zh
and the electrical length of the capacitor sections as
CZ
β = (capacitor), (8.86b)
R0
where R0 is the filter impedance and L and C are the normalized element values (the gk )
of the low-pass prototype.

EXAMPLE 8.6 STEPPED-IMPEDANCE FILTER DESIGN

Design a stepped-impedance low-pass filter having a maximally flat response and

a cutoff frequency of 2.5 GHz. It is desired to have more than 20 dB insertion
loss at 4 GHz. The filter impedance is 50 ; the highest practical line impedance
is 120 , and the lowest is 20 . Consider the effect of losses when this filter is
implemented with a microstrip substrate having d = 0.158 cm, r = 4.2, tan δ =
0.02, and copper conductors of 0.5 mil thickness.
Solution
To use Figure 8.26 we calculate
ω 4.0
−1= − 1 = 0.6;
ωc 2.5

L2 L4 L6

C1 C3 C5

(a)

l1 l2 l3 l4 l5 l6

Z0 Zl Zh Zl Zh Zl Zh Z0

(b)

(c)

FIGURE 8.40 Filter design for Example 8.6. (a) Low-pass filter prototype circuit. (b) Stepped-
impedance implementation. (c) Microstrip layout of final filter.
 8.6 Stepped-Impedance Low-Pass Filters 425

then the figure indicates N = 6 should give the required attenuation at 4.0 GHz.
Table 8.3 gives the low-pass prototype values as
g1 = 0.517 = C1 ,
g2 = 1.414 = L 2 ,
g3 = 1.932 = C3 ,
g4 = 1.932 = L 4 ,
g5 = 1.414 = C5 ,
g6 = 0.517 = L 6 .
The low-pass prototype filter is shown in Figure 8.40a.
Next, (8.86a) and (8.86b) are used to replace the series inductors and shunt
capacitors with sections of low-impedance and high-impedance lines. The re-
quired electrical line lengths, β i , along with the physical microstrip line widths,
Wi , and lengths, i , are given in the table below.

Section Z i = Z or Z h ( ) β i (deg) Wi (mm) i (mm)

1 20 11.8 11.3 2.05

2 120 33.8 0.428 6.63
3 20 44.3 11.3 7.69
4 120 46.1 0.428 9.04
5 20 32.4 11.3 5.63
6 120 12.3 0.428 2.41

The final filter circuit is shown in Figure 8.40b, with Z = 20 and Z h =

120 . Note that β < 45◦ for all but one section. The microstrip layout of the
filter is shown in Figure 8.40c.
Figure 8.41 shows the calculated amplitude response of the filter, with and
without losses. The effect of loss is to increase the passband attenuation to about

10
Attenuation (dB)

Hi-Z, Lo-Z
20

Lumped element

30
0 1.0 2.0 3.0 4.0 5.0
Frequency (GHz)

FIGURE 8.41 Amplitude response of the stepped-impedance low-pass filter of Example 8.6, with
(dotted line) and without (solid line) losses. The response of the corresponding
lumped-element filter is also shown.
426 Chapter 8: Microwave Filters

1 dB at 2 GHz. The response of the corresponding lumped-element filter is also

shown in Figure 8.41. The passband characteristic is similar to that of the stepped
impedance filter, but the lumped-element filter gives more attenuation at higher
frequencies. This is because the stepped-impedance filter elements depart sig-
nificantly from the lumped-element values at higher frequencies. The stepped-
impedance filter may have other passbands at higher frequencies, but the response
will not be perfectly periodic because the lines are not commensurate. ■

8.7 COUPLED LINE FILTERS

The parallel coupled transmission lines discussed in Section 7.6 (for directional couplers)
can be used to construct many types of filters. Fabrication of multisection bandpass or
bandstop coupled line filters is particularly easy in microstrip or stripline form for band-
widths less than about 20%. Wider bandwidth filters generally require very tightly coupled
lines, which are difficult to fabricate. We will first study the filter characteristics of a single
quarter-wave coupled line section, and then show how these sections can be used to design
a bandpass filter [7]. Other filter designs using coupled lines can be found in reference [1].

Filter Properties of a Coupled Line Section

A parallel coupled line section is shown in Figure 8.42a, with port voltage and current
definitions. We will derive the open-circuit impedance matrix for this four-port network by
considering the superposition of even- and odd-mode excitations [8], which are shown in
Figure 8.42b. Thus, the current sources i 1 and i 3 drive the line in the even mode, while i 2
and i 4 drive the line in the odd mode. By superposition, we see that the total port currents,
Ii , can be expressed in terms of the even- and odd-mode currents as
I1 = i 1 + i 2 , (8.87a)
I2 = i 1 − i 2 , (8.87b)
I3 = i 3 − i 4 , (8.87c)
I4 = i 3 + i 4 . (8.87d)

First consider the line as being driven in the even mode by the i 1 current sources. If
the other ports are open-circuited, the impedance seen at port 1 or 2 is
e
Z in = − j Z 0e cot β . (8.88)

The voltage on either conductor can be expressed as

va1 (z) = vb1 (z) = Ve+ [e− jβ(z− ) + e jβ(z− ) ]

= 2Ve+ cos β( − z), (8.89)

so the voltage at port 1 or 2 is

va1 (0) = vb1 (0) = 2Ve+ cos β = i 1 Z in

e
.

This result and (8.88) can be used to rewrite (8.89) in terms of i 1 as

cos β( − z)
va1 (z) = vb1 (z) = − j Z 0e i1 . (8.90)
sin β