Exercise notes

Microwave Semiconductor Devices and Circuits I

Winter Term 2008 / 2009
Contents
1 Planar microwave circuits 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Waves on transmission lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Waveguide geometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3.1 Stripline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.2 Coplanar waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.3 Slotline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.4 Finline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.5 Microstripline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Substrate materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Transmission line theory and discontinuities 6
2.1 Transmission lines and terminations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 Matching with /4 transmission line . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Microstrip open circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.2 Step in width of a microstripline . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Rectangular junction and microstrip bends . . . . . . . . . . . . . . . . . . . . . 12
2.3 CAD-Exercise: Matching Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.1 Design of

4
matching networks using ideal transmission lines . . . . . . . . . . 12
2.3.2 Design of

4
matching network using microstrip transmission lines . . . . . . . . 13
3 Microwave circuits using coupled transmission lines 14
3.1 Open-circuited two-conductor line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Bandpass element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Systematic lter design 18
4.1 Insertion-loss method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Bandpass lter design equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3 CAD-Exercise: Design of Chebyshev bandpass lter . . . . . . . . . . . . . . . . . . . 22
4.3.1 Filter specication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3.2 Computation of even- and odd mode impedances . . . . . . . . . . . . . . . . . 23
4.3.3 Simulation using ideal transmission lines and couplers . . . . . . . . . . . . . . 23
4.3.4 Simulation using microstrip transmission lines and couplers . . . . . . . . . . . 23
4.3.5 Optimization of the lter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5 Signal ow graphs 25
References 30
1 Planar microwave circuits 1
1 Planar microwave circuits
1.1 Introduction
The use of waveguide components for microwave circuits leads in general to large circuit sizes and inte-
gration of different components (like transistors, diodes etc.) is difcult. Integrated circuits at microwave
frequencies can be easily built up in a planar technology like low-frequency multilayer PCBs (Printed
Circuit Boards) so integration of discrete components is simple. These circuits are called MICs (Mi-
crowave Integrated Circuit) if the integration is realized in a hybrid way (transmission lines, passive and
active components on the same carrier substrate, see Fig. 1). A monolithic integration is called MMIC
(Monolithic Microwave Integrated Circuit, see Fig. 2). Here, all functions are realized on a single waver.
Figure 1: Oscillator MIC using dielectric resonators [6].
Figure 2: Single-sideband modulator MMIC with GaAs FET [6].
1.2 Waves on transmission lines
This section introduces the properties of waves on transmission lines. A detailed description of the
underlying theory can be found in [1]. A short piece of a TEM transmission line can be described by the
equivalent circuit shown in Fig. 3 with the distributed circuit elements L

, C

, R

, und G

. In this way a
generalized voltage and current can be dened:
V = V
+
e
jz
+V

e
jz
, (1)
I = I
+
e
jz
I

e
jz
. (2)
V
+
, I
+
are waves in positve and V

, I

are waves in negative direction. The propagation constant can

be computed from L

, C

, R

, und G

:
1 Planar microwave circuits 2
Z ,
0
g,l
z
Z
L
Z
g
V
g
L'dz R'dz
C'dz
G'dz
V(z)
I(z) I(z+dz)
V(z+dz)
Figure 3: Equivalent circuit of TEM transmission line (length dz).
=

(R

+ jL

)(G

+ jC

) = + j. (3)
is the attenuation of the transmission line and the phase constant at angular frequency . From those
the following parameters can be dened:
k
0
=

0
=
2

0
=

c
0
, (4)
and

e f f
=

k
0

2
=

wg

2
=

c
0
v
p

2
, (5)

e f f
is the effective dielectic constant. (
0
) and (c
0
) are the free-space wavelength and the speed of
light, respectively.
wg
and v
p
are the wavelength and the phase velocity on the transmission line. In the
TEM-case, the characteristic impedance can be computed from (zero losses):
Z
0
=

e f f
c
0
C

(6)
TEM-waves are not supported by some waveguides since a TEM-wave requires at least two conductors
and a homogeneous and loss-free medium of propagation. Practical geometries do not satisfy this con-
dition, so the axial components of the electric and magnetic eld do not vanish. The waves are TM, TE
or even hybrid. However, at microwave frequencies the transversal eld components are dominant, so
the waves can be treated as quasi-TEM. The following denitions of the characteristic impedance are
common:
Z
0
=
P
2I
2
, (7)
Z
0
=
V
I
. (8)
P is the power of the wave, which can be computed from the poynting-vector.
1.3 Waveguide geometries
The most common planar waveguide geometries are described in this section.
1 Planar microwave circuits 3
Figure 4: Stripline
1.3.1 Stripline
This structure consists of a metal stripe between two ground metallizations. The space is lled with a
dielectric substrate (Fig. 4). In most applications the metal stripe is in the center and the substrate is
homogeneous. The guided wave is a TEM-wave.
Advantages: Good shielding due to ground metallizations. The effective dielectic constant is equal to
the dielectic constant of the substrate. Therefore, the dielectric constant is in general relatively high,
which leads to circuit miniaturization. Dispersion is low.
Drawbacks: An additional cover substrate is necessary. Higher losses compared to microstrip lines
since all energy is concentrated within the (lossy) substrate. It is more difcult to integrate discrete
circuit elements.
Effective dielectric constant and characteristic impedance (static case) [2]:

e f f
=
r
(9)
Z
0
=

0
K(k)
4K(k

)
(10)
k = sech

w
2h

, k

= tanh

w
2h

and values for K, computed from the complete elliptic integrals. The
computation can also carried out by the method of conformal mapping.
1.3.2 Coplanar waveguide
Figure 5: Coplanar waveguide with additional ground plane (top) and without (bottom).
This structure (Fig. 5) consists of a metal stripe between two lateral ground metallizations on a dielectric
substrate. The classical coplanar waveguide comes without a ground metallization. However, to improve
1 Planar microwave circuits 4
electromagnetic shielding an additional ground plane can be used. The lateral dimensions of the grounds
metallizations should be at least 5b.
Advantages: Low dispersion since the ratio of the power, which is transported within the air and the
substrate is independent of the frequency of operation. Therefore, the characteristic impedance can
be adjusted simply by the ratio of
a
b
. Close packed circuit geometries are possible with this kind of
waveguide. Short circuits can easily be realized.
Drawbacks: The additional ground plane may cause problems due to the excitation of parallel plate
modes. This increases the losses on the transmission line.
Effective dielectric constant and characteristic impedance for coplanar waveguide (substrate extends to
innity at lower z-direction) [3]:

e f f
=

r
+1
2
, (11)
Z
0
=

0
K(k

)
4

e f f
K(k)
. (12)
k =
a
b
and k

1k
2
. The computation can be carried out again by the method of conformal mapping.
1.3.3 Slotline
Figure 6: Slotline.
This structure (Fig. 6) is built up from two metallizations separated by a single slot with the width w.
This waveguide is used in the millimeter-wave range. The guided wave is quasi TE. The expressions of
the transmission line parameters are complicated and will not be repeated here.
1.3.4 Finline
A slotline (Fig. 7) inserted into a waveguide is called nline. This transmission line is used for tran-
sitions between waveguides and planar circuits. There are serveral different congurations of a nline
(unilateral, bilateral and antipodal).
1.3.5 Microstripline
This structure (Fig. 8) consists simply of a metal stripe of width w and thickness t above a ground
metallization with a substrate of dielectric constant
r
and height h in between. The guided waves are
hybrid TE/TM-waves. This structure is dispersive, so a quasistatic calculation can be carried out only
approximately.
Advantages: Simple design. Junctions and open circuits can be easily realized. Discrete components
can be inserted with no difculty.
1 Planar microwave circuits 5
Figure 7: Finline.
Figure 8: Microstripline.
Drawbacks: Short circuits are more difcult to realize (via-hole required). Especially at higher frequen-
cies parastic elements of this vertical interconnect may cause deviations from an "ideal" short circuit.
The mircostripline is the most common transmission line for MICs. Therefore, this exercise has a special
focus on it. However, the design procedures presented in the following can also be used for different
types of transmission lines.
The effective dielectic constant and characteristic impedance in the quasistatic case (TEM) can be
computed by the following formulas [3]:

e f f
=

r
+1
2
+

r
1
2
F (13)
F =
1

1
10h
w
(14)
and
Z
L
(
w
h
,
r
) =
Z
L0

e f f
. (15)
2 Transmission line theory and discontinuities 6
Z
L0
= Z
L
(
r
= 1) = 60ln

8h
w
+
w
4h

(16)
The thickness of the metal stripe has been neglected in the formulas given above. However, all common
CAD tools account for the thickness, so the more complicated formulas will not be repeated here.
1.4 Substrate materials
Substrates should have low losses at RF frequencies. Materials with high dielectric constant can be
used for circuit miniaturization since the guided wavelength is reduced. The drawback of these sub-
strates when used in combination with microstriplines is a eld concentration within the substrate and
subsequently higher losses. Antenna substrates should exhibit a low dielectric constant to reduce eld
concentration within the substrate. Antenna efciency is increased by the way. Good mechanical stabil-
ity and easy fabrication of the material are other important factors. If active components are used a good
thermal conductivity of the substrates is important, too.
Standard FR4 material, which is used for most low-frequency PCBs, can not be used for high-frequency
applications since the dielectic constant and the thickness are subject to large deviations (up to 20 %)
and the losses are relatively high. Resonant applications like lters are very sensitive to these parameters
and can not be realized on such substrates. Some suitable materials are listed in [5].
A common material used for microwave substrates is PTFE (Teon). Substrates with PTFE as base
material exhibit dielectric constants of 2 to 10. The high dielectric constant materials are lled with
ceramic material.
Another standard material is alumina (Al
2
O
3
) with good thermal conductivity and very low losses. But
this substrate is expensive and is difcult to machine.
The substrates are plated on both sides with a conductor (Cu or Ag).
1.5 Fabrication
The following is a description of the fabrication process at the Arbeitsbereich. A mask with the mi-
crowave circuit structure is drawn with a CAD tool on a PC. The mask data are sent to a photo mask
manufacturer. The substrates are coated with a photoresist. The mask is attached to the substrate and is
exposed by a UV-light source. The exposed photoresist is developed and so the structure is transferred
onto the substrate. Subsequently, the substrate is etched and the remaining photoresist removed. This
completes the fabrication.
2 Transmission line theory and discontinuities
2.1 Transmission lines and terminations
Fig. 9 shows a transmission line of propagation constant , characteristic impedance Z
0
and length l
terminated with the impedance Z
L
. The input impedance of the transmission line is:
Z
1
= Z
0
Z
L
+Z
0
tanh(l)
Z
0
+Z
L
tanh(l)
. (17)
Zero losses:
2 Transmission line theory and discontinuities 7
Z ,
0
g,l
z
Z
L
Z
g
V
g
Figure 9: General termination at the end of a transmission line.
Z
1
= Z
0
Z
L
+ jZ
0
tan(l)
Z
0
+ jZ
L
tan(l)
. (18)
Two important cases will be considered in the following: transmission lines with a length of /4 and
/2. These correspond to resonant circuits from Fig. 10.
Z
g
V
g
L
C
Z
g
V
g
L
C
Figure 10: Series- and parallel resonant circuits for comparison with /4 transmission lines.
/4 transmission line:
l =

4
l =

2
. (19)
For small deviations of frequency =
0
+: l =

2
+ with =

2

0
. The input
impedance can be calculated from eq. (18):
Z
1
Z
0

jZ
0
Z
L

jZ
L
Z
0

. (20)
A /4 transmission line terminated with an open circuit (Z
L
=) has the input impedance:
Z
1
Z
0
= j. (21)
The impedance of a series resonant circuit is:
Z
s
= jL+
1
jC
(22)
2 Transmission line theory and discontinuities 8
=
1
2
LC
jC
(23)
=
1

1+

2
j(
0
+)C
(24)
j2L, (25)
Higher order terms have been neglected.
0
=
1

LC
. Comparison with eq. (21) yields:
C =
4

0
Z
0
, (26)
L =

4
Z
0

0
. (27)
The short circuited transmission line can be modelled by an equivalent parallel resonant circuit:
C =

4
1
Z
0

0
, (28)
L =
4

Z
0

0
. (29)
/2 transmission line:
The calculation is essentially the same; the series- and parallel resonant circuit are interchanged in
this case.
impedance- and admittance inverter
A /4 transmission line has following input impedance:
Z
1
=
Z
2
0
Z
L
. (30)
The termination Z
L
is "inverted". Therefore, a /4 transmission line can be used as an impedance
inverter. These elements are required in lter theory and can also be used for transformation of a
series circuit into a parallel circuit.
2.1.1 Matching with /4 transmission line
The most simple matching network is a single /4 transmission line (Fig. 11) [1]. The impedance
transfomation function with tan =t is:
Z
in
= Z
1
Z
L
+ jZ
1
t
Z
1
+ jZ
L
t
. (31)
The reection coefcient vanishes if Z
1
=

Z
0
Z
L
is chosen for a particular frequency. For the reection
coefcient follows:
2 Transmission line theory and discontinuities 9
Z
0
Z , l
1
b = p/2
Z
L
Z
0
V
g
Z
in
Figure 11: Single stage matching network with /4 transmission line.
=
Z
in
Z
0
Z
in
+Z
0
(32)
=
Z
1
Z
L
+ jZ
2
1
t Z
0
Z
1
Z
0
Z
L
t
Z
1
Z
L
+ jZ
2
1
t +Z
0
Z
1
+Z
0
Z
L
t
(33)
=
Z
1
(Z
L
Z
0
) + jt(Z
2
1
Z
0
Z
L
)
Z
1
(Z
L
+Z
0
) + jt(Z
2
1
+Z
0
Z
L
)
(34)
=
Z
L
Z
0
Z
L
+Z
0
+2 jt

Z
0
Z
L
(35)
For broadband applications the input bandwidth is an important gure of merit. For the calculation the
magnitude of the reection coefcient is evaluated:
=
|Z
L
Z
0
|

(Z
L
+Z
0
)
2
+4t
2
Z
0
Z
L
(36)
=

(Z
L
Z
0
)
2
Z
2
l
+Z
2
0
+(4
s
2
c
2
+2)Z
0
Z
L
(37)
=

(Z
L
Z
0
)
2
(Z
L
Z
0
)
2
+Z
1
Z
L
1
c
2
(38)
=
1

1+

Z
0
Z
L
Z
L
Z
0

2
1
c
2
. (39)
s =sinand c =cos. All impedances are considered as real quantities. For a maximuminput reection

m
follows the maximum / minimum electrical length of the matching transmission line
m
:

m
= cos
1

2
m

Z
0
Z
L
(Z
L
Z
0
)

1
2
m

. (40)
The bandwidth for TEM transmission lines with =l =
f
m
f
0

2
and f
0
as frequency of operation is:
f = 2( f
0
f
m
) = 2( f
0

2 f
0

m
). (41)
2 Transmission line theory and discontinuities 10
The usable bandwidth of a matching network is relatively small. In order to increase the usable bandwidth
a multi stage matching network can be used. An exemplary network is calculated in the following. Fig.
12 shows a network of cascaded /4 transmission lines.
Z
0 Z , l
1
b = Q Z , l
2
b = Q Z , l
N
b = Q Z
L
Z
g
V
g
G
0
G
1 G
2
G
N
Figure 12: Multi stage matching network of cascaded /4 transmision lines.
For the reection coefcients at each transmission line follows:

n
=
Z
n+1
Z
n
Z
n+1
+Z
n
. (42)
Z
n
are the characteristic impedances of each transmission line (Fig. 12).
n
is only the reection coef-
fection at the respective boundary and not the total reection coefcient. For the following calculation
it is assumed that Z
L
> Z
1
and all characteristic impedances are real quantities (for Z
L
< Z
1
the signs of
the reection coefcients must be inverted). A maximally at reection coefcient in the passband is
desired. The function of the reection coefcient vs. frequency shall:
vanish at f
0
like
all N1 derivatives.
Such a function is given by:
= A(1+e
j2
)
N
; (43)
the magitude of the reection coefcient is = A2
N
cos
N
.
The impedances of each single stage and the constant A must be calculated:
= A(1+e
j2
)
N
= A
N

n=0

N
n

e
2 jn
(44)
with the binomial coefcients:

N
n

=
N!
(Nn)!n!
. (45)
If the difference of the impedances is not too large, multiple reections can be neglected (rst order
reections only). This assumption leads to:
=
N

n=0

n
e
j2n
. (46)
Comparison with eq. (44) yields:
2 Transmission line theory and discontinuities 11

n
= A

N
n

= A

N
Nn

=
Nn
. (47)
The equation is symmetrical:
n
=
Nn
.
The missing expressions for the characteristic impedances and the coefcient A can be calculated from
the reection at f = 0 (= 0). In this case the reection is simply:
(= 0) = 2
N
A =
Z
L
Z
0
Z
L
+Z
0

1
2
ln
Z
L
Z
0
. (48)
The approximation is based on the power series of the ln-function. The reections at the single transmis-
sion lines can be calculated by the same approximation:

n
=
Z
n+1
Z
n
Z
n+1
+Z
n

1
2
ln
Z
n+1
Z
n
. (49)
with eq. (47) - (49) follows:
ln
Z
n+1
Z
n
=
1
2
N

N
n

ln
Z
L
Z
0
. (50)
The impedances can be calculated with this equation (starting with Z
1
).
For the maximum specied input reection
m
the bandwidth is (cf. eq. (43)):
= || = |A(1+e
j2
)
N
| = (51)
= |A|2
N
| cos
N
| (52)

m
= cos
1

2
m
ln
Z
L
Z
0

1
N
(53)
f
f
0
= 2
4

cos
1

2
m
ln
Z
L
Z
0

1
N
. (54)
This bandwidth is considerably larger than that of the single stage network.
2.2 Discontinuities
Discontinuities occur when one transmission line parameter changes abruptly. If these changes are small
compared with the wavelength discontinuities can be described by an equivalent circuit model of discrete
components. Some important discontinuities and their equivalent circuit models will be described in the
following. The additional parasitic effects which occur are:
distortion of the electric and magnetic eld and of the current distribution on the transmission line.
Higher order modes are excited.
a fraction of the power may be radiated or surwave waves may be excited.
Most CAD-Tools include accurate models of the most common discontinuities.
2 Transmission line theory and discontinuities 12
2.2.1 Microstrip open circuit
An abrupt ending of the microstripline. This should result in an impedance of Z
L
=. However, there is
an electrical eld around edge of the ending which results in a capacitive load C
L
of the microstripline.
This effect can also be described by an additional piece of transmission line l
L
. The equivalent circuit
elements are given by [3]:
l
L
=
C
L
C

=C
L
Z
0
c
0

e f f
. (55)
l
L
= 0, 412h
(
e f f
+0.3)(
w
h
+0.26)
(
e f f
0.258)(
w
h
+0.813)
. (56)
The error is smaller than 5%.
2.2.2 Step in width of a microstripline
A step in width of the microstripline from w
1
to w
2
changes not only the characteristic impedance but
also introduces parasitic effects. There is a distortion of the current densitiy which has an inductive
effect (L
s
). The stray electric eld has a capacitive effect C
p
or can be modelled by an additional piece
of transmission line l
s1
(cf. microstrip open circuit). The equivalent circuit elements are given by [3]:
L
s
h
=
2
0

ln

1/sin(

2
Z
L1,0
Z
L2,0
)

, (57)
C
p
= (

e f f 1
Z
L1
c
0

r
w
1
h
)(w
1
w
2
)/2, (58)
l
S1
=

1.35

r
+0.44

(1
w
1
w
2
)h, (59)
Z
Li,0
is the impedance of width i without dielectric (air). The above formulas are valid for w
1
> w
2
.
2.2.3 Rectangular junction and microstrip bends
A rectangular junction of three tranmission lines has the following parasitics: distortion of the current
density and distortion of the electric and magnetic elds. Again, the current distortion can be modelled
by an inductance and the distortion of the electric eld can be modelled by an capacitance. A commonly
used bend is the optimal right-angle mitered bend. This bend compensates for the parasitics:
L
w
=

2(1.04+1.3e
1.35
w
h
), (60)
b =

2w
L
2
. (61)
2.3 CAD-Exercise: Matching Networks
2.3.1 Design of

4
matching networks using ideal transmission lines
Task: Design and comparison of single and multiple stage matching networks (up to fourth order) at
f = 2 GHz and Z
0
= 50 matched to Z
1
= 100 .
2 Transmission line theory and discontinuities 13
computation of characteristic impedances (eq. 50)
here: 50 matched to 100
N=1:
n = 0 Z
1
= Z
0
EXP(
1
2

1
0

ln(
100
50
)) =
N=2:
n = 1 Z
1
= Z
0
EXP(
1
4

2
0

ln(
100
50
)) =
n = 2 Z
2
= Z
1
EXP(
1
4

2
1

ln(
100
50
)) =
N=3:
n = 1 Z
1
= Z
0
EXP(
1
8

3
0

ln(
100
50
)) =
n = 2 Z
2
= Z
1
EXP(
1
8

3
1

ln(
100
50
)) =
n = 3 Z
3
= Z
2
EXP(
1
8

3
2

ln(
100
50
)) =
N=4:
n = 1 Z
1
= Z
0
EXP(
1
16

4
0

ln(
100
50
)) =
n = 2 Z
2
= Z
1
EXP(
1
16

4
1

ln(
100
50
)) =
n = 3 Z
3
= Z
2
EXP(
1
16

4
2

ln(
100
50
)) =
n = 4 Z
4
= Z
3
EXP(
1
16

4
3

ln(
100
50
)) =
drawing of the schematic and simulation of the network Draw the networks in ADS schematic
window and start the simulation. The bandwidth can be obtained from the simulated scattering
parameters.
B
1
= MHz B
4
= MHz
2.3.2 Design of

4
matching network using microstrip transmission lines
Task: Design and comparison of single and multiple stage matching networks (up to fourth order) at
f = 2 GHz and Z
0
= 50 matched to Z
1
= 100 .
computation of microstrip line widths and lengths (ADS-Tool LineCalc) A substrate with fol-
lowing parameters is used:
h = 1 mm
r
= 3 tan = 0.001 h
cu
= 35 m
w
1
= mm l
1
= mm
w
2
= mm l
2
= mm
w
3
= mm l
3
= mm
w
4
= mm l
4
= mm
drawing of the schematic and simulation of the network Draw the networks in ADS schematic
window and start the simulation. The bandwidth can be obtained from the simulated scattering
parameters.
B
1
= MHz B
4
= MHz
Due to the different microstrip line widths additional parasitic effects are introduced. Insert an
appropriate simulation model (step in width) between each transmission line and compare the
results with the ideal case.
3 Microwave circuits using coupled transmission lines 14
3 Microwave circuits using coupled transmission lines
Beside simple transmission line circuits like

4
-transformators coupled transmission lines are widely
used. These are simple multiconductor ciruits. However, for the operating frequency range the coupled
line sections can be transformed to noncoupled lines. If the operating frequency range is sufciently
narrow, an equivalanet circuit model of lumped elements exists. This is illustrated in the following by
two coupled open-circuited transmission lines.
3.1 Open-circuited two-conductor line
The following calculation is carried out for a homogeneous and symmetrical two-conductor transmission
line (Fig. 13, 14).
w
w
s
(a) (b)
Evenmode
Oddmode
I
1
I
1
I
1
-I
1
Conductor(a)
Conductor(b)
Figure 13: Two coupled microstrip lines (a) conductors, (b) eld-plot of even- and odd mode.
The generalized currents and voltages on the transmission line can be described by forward- and back-
ward travelling waves:
V
a
(z) = V
+
e
e
jz
+V

e
e
jz
+V
+
o
e
jz
+V

o
e
jz
, (62)
I
a
(z) =
V
+
e
Z
e
e
jz

e
Z
e
e
jz
+
V
+
o
Z
o
e
jz

o
Z
o
e
jz
, (63)
V
b
(z) = V
+
e
e
jz
+V

e
e
jz
V
+
o
e
jz
V

o
e
jz
, (64)
I
b
(z) =
V
+
e
Z
e
e
jz

e
Z
e
e
jz

V
+
o
Z
o
e
jz
+
V

o
Z
o
e
jz
(65)
e: even mode, o: odd mode, +: forward-travelling wave, backward-travelling wave, 1: conductor 1,
2: conductor 2, Z
e
: even mode impedance, Z
o
: odd mode impedance.
The coupled transmission lines can be calculated with the above formulas.
For circuit synthesis it is convenient to have an equivalent circuit model without coupled transmission
lines (see Fig. 15).
For comparison of the two circuits the results for short-circuit and open-circuit terminations are com-
puted. For unsymmetrical networks this has to be done on both sides (input and output). Since the
network is symmetrical one can immediately set Z
1
= Z
3
for the equivalent circuit model in Fig 15.
The following boundary conditions are obtained:
port 1 (z = 0) - xed voltage V
1
:
3 Microwave circuits using coupled transmission lines 15
w
w
s
Conductor(a)
Conductor(b)
Port1
Port2
Figure 14: Two coupled open-circuited tranmission lines.
Figure 15: Equivalent tranmission line circuit model for two coupled open-circuited transmission lines.
V
a
(0) =V
1
=V
+
e
+V

e
+V
+
o
+V

o
. (66)
open circuit at the end of conductor a (z = l):
I
a
(l) = 0 =Y
e
V
+
e
e
jl
Y
e
V

e
e
jl
+Y
o
V
+
o
e
jl
Y
o
V

o
e
jl
. (67)
open circuit at the end of conductor b (z = 0):
I
b
(0) = 0 =Y
e
V
+
e
Y
e
V

e
Y
o
V
+
o
+Y
o
V

o
. (68)
open-circuit termination of port 2 (z = l):
I
b
(l) = 0 =Y
e
V
+
e
e
jl
Y
e
V

e
e
jl
Y
o
V
+
o
e
jl
+Y
o
V

o
e
jl
. (69)
short-circuit termination of port 2:
V
b
(l) = 0 =V
+
e
e
jl
+V

e
e
jl
V
+
o
e
jl
V

o
e
jl
. (70)
This is a 4 4 system of equations, which has to be solved for the unknown forward- and backward
travelling waves. The impedance of port 1 for the open-circuit termination of port 2 is:
3 Microwave circuits using coupled transmission lines 16
Z
oc
= j
Z
e
+Z
o
2
cot l (71)
For the short-circuit termination the impedance is:
Z
sc
= j
(Z
e
Z
o
)
2
2(Z
e
+Z
o
)
tanl j
Z
e
Z
o
Z
e
+Z
o
cot l (72)
The impedances (open/short-circuit) for the equivalent circuit model (Fig. 15) are:
Z

oc
= j(Z
1
+Z
2
)cot (73)
and
Z

sc
= j
Z
2
1
Z
1
+Z
2
tan j
Z
2
2
+2Z
1
Z
2
Z
1
+Z
2
cot (74)
Comparison leads to:
Z
1
+Z
2
=
Z
e
+Z
o
2
, (75)
Z
2
1
Z
1
+Z
2
=
(Z
e
Z
o
)
2
2(Z
e
+Z
o
)
(76)
The impedances of the equivalent cirfcuit model can now be calculated:
Z
1
= Z
o
= Z
3
, (77)
Z
2
=
Z
e
Z
o
2
(78)
This demonstrates the equivalence of both circuits (coupled transmission lines and transmission line with
stubs). The equivalent circuit model can be analyzed by conventional transmission line theory.
3.2 Bandpass element
A bandpass can be realized with multiple coupled transmission lines. We use two coupled transmission
lines in series connection (Fig. 16). For a small frequency range this circuit can be further simnplied.
We assume an electrical length of the tranmission lines of =

2
(

4
-transmission lines). The open-
circuited stub lines can be described by a LC-serial resonant circuit. The inductance is L =
Z
4
0
and the
capacitance C =
4

0
Z
.
0
is the angular frequency (for =

2
) and Z the chracteristic impedance of the
transmission line. The resulting equivalent circuit model is shown in Fig.16.
This model uses impedance inverters which are described by the length of the connecting transmission
line (

4
) and the parameter K =
Z
e
Z
o
2
.
The impedances in the reference planes (Fig. 18) are:
3 Microwave circuits using coupled transmission lines 17
Figure 16: (a) Bandpass element, (b) equivalent circuit model.
Figure 17: Simpied equivalent circuit model for the bandpass element (impedance inverter K).
Z
A
=
K
2
Z
L
, (79)
Z
B
= Z
S
+
K
2
Z
L
, (80)
Z
C
=
K
2
Z
S
+
K
2
Z
L
. (81)
The input admittance is Y
C
=
Z
S
K
2
+
1
Z
L
. If a serial resonant circuit is chosen for Z
S
= jL
1
jC
this
circuit is transformed to a parallel resonant circuit (C

=
L
K
2
, L

=
1
K
2
C
). The resulting equivalent circuit
model is shown in Fig. 19.
This is a standard bandpass circuit element with following parameters:
L =
Z
o
4
0
, (82)
C =
4

1
Z
o
, (83)
4 Systematic lter design 18
Figure 18: Impedance inverter reference planes.
Figure 19: Further simplied circuit: standard bandpass.
L

=
(Z
e
Z
o
)
2
2
0
Z
o
, (84)
C

=
Z
o
(Z
e
Z
o
)
2

0
. (85)
It is now possible to determine the impedances of the coupled transmission lines from the element values.
This is necessary for a systematic lter design, which is described in the next section.
4 Systematic lter design
Filters are among the few circuit components which can be synthesized analytically. Starting from the
specications and the lter topology a lter prototype can be designed with the procedure described in the
following. The practical realization needs serveral optimzing steps because of some assumptions which
must be made in the theory (e.g. losses and dispersion are neglected). Even if the CAD-Tools provide
more sophisticated models a design has to be made with care. Numerical optimization of the lter
structure is possible but sometimes the optimization process yields unphysical results. Therefore, any
optimization has to be closely monitored. Filter design can be carried without complete understanding of
the theory by use of lter tables. However, for specialized applications a good knowledge of lter theory
is necessary.
After a brief introduction of the insertion-loss method common lter prototypes are presented. An equiv-
alent circuit model for edge-coupled nicrostriplines has been developed in the last section. Design for-
mulas are given for translation from circuit theory lter prototypes to the edge-coupled lters. The nal
design is carried out in the second CAD-exercise.
4 Systematic lter design 19
4.1 Insertion-loss method
The lter is treated as a two-port device and described by scattering parameters. The quanity of interest
is the reection coefcient from which the power loss ratio
P
LR
=
1
1
2
=
1
1

(86)
(input power / power delivered to load) can be computed ( = ||). For lter synthesis P
LR
() or ()
are given as a function of frequency. Some boundary conditions must be observed:
passive circuits: 1
the normalized impedance Z =
Z
Z
c
can be described by
Z() = R() + jX() (87)
Z
c
is the reference impedance (design specication). R() is an even function, X() an odd
function of angular frequency. This requirement follows from a theorem of circuit theory that the
response of a circuit to a real-time-dependent driving function must also be real. The equivalent
condition for the reection coefcient is (() =

()). The reection coefcient can be

written as:
() =
R() 1+ jX()
R() +1+ jX()
. (88)
Therefore, the function
2
() has as argument
2
.
any lowpass-imepdance function can be described by the ratio of two polynomials M, N:

2
() =
M(
2
)
M(
2
) +N(
2
)
=
(R() 1)
2
+X
2
())
(R() +1)
2
+X
2
())
. (89)
The power loss ratio can now be expresed as:
P
LR
= 1+
M(
2
)
N(
2
)
. (90)
So far any polynomials can be chosen for lter synthesis. However, there are some standard-polynomials,
which are frequently used in microwave engineering. The two most inmportant will be described in the
following.
Maximally Flat lter (Butterworth lter)
The Butterworth lter is also known as the maximally at lter due to the fact that it has the most at
passband response. This kind of lter is useful if signal distortions in the passband must be kept at
minimum. The lter response is less steep compared with other lters. Therefore, this lter is less useful
for ltering signals which are located closely to each other in the spectrum. The lter polynomials are:
4 Systematic lter design 20
M(
2
) = k
2

2N
,
N(
2
) = 1,
P
LR
= 1+k
2

2N
. (91)
N is the lter order and
c
is the cutoff frequency.
Chebyshev lter
The Chebyshev lter is also known is the equal-ripple lter due to the occurence of ripples in the
passband. This lter has a much sharper cutoff region separating the passband and stopband compared
with the Butterworth lter. The power loss ratio of this lter is:
P
LR
= 1+k
2
T
2
N
(

c
). (92)

c
is the cutoff frequency.
The rst Chebyshev polynomials are:
T
N
(

c
) = cos

Ncos
1

T
0
(x) = 1
T
1
(x) = x
T
2
(x) = 2x
2
1
T
3
(x) = 4x
3
3x
T
4
(x) = 8x
4
8x
2
+1 (93)
The maximum reection coefcient in the passband
max
is specied by the factor k:

max
=
k

1+k
2
. (94)
Therefore, k
2
is called passband tolerance. It is now possible to design a lowpass-prototype. For re-
alization of highpass, bandpass or bandstop lters the lowpass prototype can be used as design basis.
Afterwards a frequency transformtation is used:

= fct(). (95)
Thus, the power loss ratio becomes:
P
LR
= 1+P(
2
) = 1+P( f
2
()). (96)
P can be obtained from (eq. 92). For lowpass to bandpass transformation following formula is used:

= f () =

0

. (97)
4 Systematic lter design 21
The bandwidth is =
2

1
and the center frequency:

0
=

2
1. (98)
Systematic design of Chebyshev bandpass lters:
1. specication of cutoff frequencies, ripple attenuation and lter order. The lter order can be ob-
tained by specication of x dB stopband attenuation at y GHz distance from center frequency.
This can be done graphically.
2. lookup of lter coefcients (table). For specialized lter designs the lter coefcients may not be
found in a table. For these, the lter coefcients must be calculated analytically.
3. the lter coefcients g
n
are normalized values (impedance: 1, cutoff frequency: 1Hz) for lumped
element lter prototypes and represent either capacitors or inductors.
4. the lter prototype is now complete.
4.2 Bandpass lter design equations
the Chebyshev bandpass lter prototype shall be realized using:
microstrip transmission lines
edge-coupled, open-circuited

2
-resonators
In the last section an equivalent circuit model of edge-coupled microstriplines was presented. Now
equations must be found for the inverse problem: synthesis of edge-coupled microstriplines from the
values of the equivalent circuit model. The design equations are given in the following:

1
=

1
2
0
,
Psin
1
=
K

10

1
2
tan
1
+K

10
2
,
s =
1
1
2
tan
1
+K

10
2
. (99)
The impedance inverters can be calculated (impedances normalized to Z
c
):
K

10
= K

N,N+1
=
1

g
0
g
1
=
1

g
N
g
N+1
,
K

k,k+1
=
1

g
k
g
k+1
. (100)

N
k+1,k
=

2
k,k+1
+
1
4
tan
2

1
(101)
With the above equations the even- and odd mode impedances can be computed:
4 Systematic lter design 22
Z
1
e
= Z
N+1
e
= Z
c
(1+Psin
1
)
Z
1
o
= Z
N+1
o
= Z
c
(1Psin
1
)
Z
k+1
e
= Z
Nk+1
e
= Z
c
s(

N
k+1,k
+K

k,k+1
)
Z
k+1
o
= Z
Nk+1
o
= Z
c
s(

N
k+1,k
K

k,k+1
). (102)
The electrical parameters of the lter prototype using coupled microstriplines are now determined. The
physical dimensions can be obtained by use of CAD-Tools (e.g. ADS LineCalc).
4.3 CAD-Exercise: Design of Chebyshev bandpass lter
The design is carried out in ve steps:
1. specication of the lter and lookup of lter coefcients for the lter prototype
2. computation of even- and odd mode impedances using the design equations
3. simulation of the lter using ideal transmission lines and couplers
4. simulation of the lter using microstrip transmission lines and couplers
5. optimization of the lter
4.3.1 Filter specication
The lter shall have:
a center frequency of f
0
=

0
2
= 5.0 GHz
a bandwidth of 0.5 GHz, i.e. the cutoff frequencies are f
d
=

d
2
= 4.75 and 5.25 GHz, respectively
a stopband attenuation of at least 15 dB at f
s
=

s
2
= 4.5 resp. 5.5 GHz
a ripple attenuation of no more than 0.5 dB
For lookup in lter tables the above values have to be normalized.
normalized angular frequency:

=

0

0


0

s
) =
required lter order (table 4.03-7): N=
lter coefcients (table 4.05-2(a)): g
0
=
g
1
=
g
2
=
g
3
=
g
4
=
4 Systematic lter design 23
4.3.2 Computation of even- and odd mode impedances
The even and odd mode impedances can be calculated using the design equations (note the symmetry of
the lter: Z
e/o1
= Z
e/o4
and Z
e/o2
= Z
e/o3
):

1
=
K
1,0
=
1

g
0
g
1
=
K
2,1
=
1

g
1
g
2
=
K
3,2
=
1

g
2
g
3
=
K
4,3
=
1

g
3
g
4
=
P sin
1
=
P =

N
k+1,k
=

K
2
k+1,k
+
1
4
tan
2

1
=

N
2,1
=
s =
1
0.5tan
1
+K
2
10
=
Z
e1
= Z
c
(1+P sin
1
) =
Z
e2
= Z
c
s(

N
21
+K
21
) =
Z
o1
= Z
c
(1P sin
1
) =
Z
o2
= Z
c
s(

N
21
K
21
) =
4.3.3 Simulation using ideal transmission lines and couplers
Draw a circuit of ideal transmission lines and couplers in ADS (scrollbar: TLINES->IDEAL). Add S-
Parameter ports and a simulation control block and simulate the lter.
4.3.4 Simulation using microstrip transmission lines and couplers
The lter shall be realized on a microwave substrate with following parameters:

r
= 10.8
h = 0.635 mm
tan = 0.0025
metal thickness: T = 35 m
Add a substrate parameter block in ADS and enter the values. The physical parameters of the microstrip
transmission lines and couplers must be calculated before a simulation can be carried out. Start the
LineCalc-Tool (Menu->Tools->LineCalc). Choose the waveguide structure and enter the substrate para-
meters. Now enter the electrical parameters of the coupled lines and calculate the physical dimensions
of the coupler.
4 Systematic lter design 24
Z o. Z
e
Z
o
E [] w [mm] s [mm] P [mm] K o. K
odd
K
even
MS(in/out) X

2
X X
CPL1
CPL2
OST1 X X X
OST2 X X X
Draw a new circuit using microstrip elements (TLINES->MICROSTRIP). Simulate the lter (including
open-end effect of open-circuited microstrip lines).
4.3.5 Optimization of the lter
The center frequency has shifted due to the capacitance at the end of the open-circuited microstrip lines.
The effective length can be computed by the formula of Kammestad:
l
e f f
= 0.412 h

e f f
+0.3

e f f
0.258

w
h
+0.262
w
h
+0.813
(103)
Compute the effective transmission line lengths.
w [mm] Z K bzw.
e f f
l
e f f
[mm]
w
h
MS
MS
Draw a new circuit and compensate for the open-end capacitance by shortening the transmission lines.
Simulate the optimized lter and compare the results with the previous simulations!
5 Signal ow graphs 25
5 Signal ow graphs
microwave n-ports can be described by scattering parameters:

b
1
b
2

b
n

= S

a
1
a
2

a
n

S is the scattering matrix, S

i j
are the scattering parameters, a
1
a
n
: incident waves, b
1
b
n
:
reected waves
the signal ow graph consists of nodes and directed line segments (see two-port Fig. 20):
S
a
2
b
2
a
1
b
1
Figure 20: Microwave two-port.
b
1
= S
11
a
1
+S
12
a
2
b
2
= S
21
a
1
+S
22
a
2
a
2
b
2
a
1
b
1
S
21
S
12
S
11
S
22
Rules for signal ow graphs
a pair of linear equations [x
2
=C
21
x1 ; x
3
=C
32
x2] has the graphical representation:
x
2
x
1
x
3
c
21
c
32
x
3
x
1
c
21
c
32
Figure 21: Illustration of rule one.
If there are no other inputs at node 2 the node can be omitted. The transmission factor is
equal to the product of the separate transmission factors.
5 Signal ow graphs 26
parallel paths can be combined:
x
2
= A
21
x
1
+B
21
x
1
+C
21
x
1
= (A
21
+B
21
+C
21
)x
1
The graphical representation is:
x
2
x
1
C
21
B
21
A
21 x
2
x
1
A +B
21 21 21
+C
Figure 22: Illustration of rule two.
combination when a node has one output and several inputs:
x
4
=C
41
x
1
+C
42
x
2
+C
43
x
3
und x
5
=C
54
x
4
(104)
The graphical representation is:
x
2
x
3
C
43
C
42
C
41
x
1
x
4
x
5
C
54
x '
4
x
2
C
43
C
42
C
41
x
1
x
3
x
5
C
54
x '
4
x '
4
C
54
C
54
Figure 23: Illustration of rule three.
For the output signals follows:
x
5
=C
54
x
4
=C
54
C
41
x
1
+C
54
C
42
x
2
+C
54
C
43
x
3
combination when a node has one input and several outputs:
x
1
x
3
C
41
C
31
C
21
x
2
x
4
x
0
C
10
x '
1
x
3
C
41
C
31
C
21
x
2
x
4
x
0
C
10
x '
1
x '
1
C
10
C
10
Figure 24: Illustration of rule four.
5 Signal ow graphs 27
feedback loops should be eliminated to obtain more simple graphs. A feedback loop is de-
scribed by the equations:
x
2
= c
21
x
1
+C
23
x
3
x
3
= C
32
x
2
x
4
= C
43
x
3
First node 3 is duplicated:
x
3
x
2
C
32
C
23
C
43
C
21
x
1
x
4
x
2
C
32
C
23
C
43
C
21
x
1
x
4
C
32
x '
3
Figure 25: Illustration of rule ve.
The new equations are:
x
2
= C
21
x
1
+C
23
C
32
x
2
x
2
=
C
21
1C
23
C
32
x
1
x

3
= C
32
x
2
; x
4
=C
43
x

3
The equations are the same except for node x

3
. The last reduction yields:
x
4
x
1
C
21
x
2
C
32 43
C
1-C C
23 32
Figure 26: Illustration of rule ve.
Signal ow graphs of common circuit elements:
5 Signal ow graphs 28
a
1
b
1
1
1
r
L
Figure 27: Termination.
Figure 28: Matched transmission line.
a) termination: b
1
=
L
a
1
b) matched transmission line (no losses): b
1
= e
jl
a
2
; b
2
= e
jl
a
1
c) generator:

g
=
a
1
b
1
=
Z
g
Z
0
Z
g
+Z
0
Z
g
V
a
1
b
1
G
g
V
1
1
Figure 29: Generator.
d) detector:

M
=
Z
M
Z
0
Z
M
+Z
0
5 Signal ow graphs 29
Z
M
a
1
b
1
G
M
1
1
M
M
Figure 30: Detector.
References 30
References
[1] H.G. Unger, Elektromagnetische Wellen auf Leitungen, 3. Auage, Huethig Buch Verlag, Heidel-
berg 1991.
[2] B.C. Wadell, Transmission Line Design Handbook, Artech House, Boston, 1991.
[3] R.K. Hofmann, Integrierte Mikrowellenschaltungen, Springer Verlag, Berlin, 1983.
[4] J. Helszajn, Microwave Planar Passive Circuits and Filters, John Wiley & Sons, Chichester, 1994.
[5] R.E. Collin, Foundations for Microwave Engineering, 2nd Edition, McGraw Hill, 1992, 1966.
[6] G.D. Vendelin, A.M. Pavio, U.L. Rohde, Microwave Circuit Design Using Linear and Nonlinear
Techniques, John Wiley & Sons, New York, 1990.
[7] I.Bahl, P.Bhartia, Microwave Solid State Circuit Design, John Wiley & Sons, 1988.
[8] G.L. Matthaei, L. Young, E.M.T. Jones, Microwave lters, impedance-matching networks, and
coupling structures, McGraw Hill, 1964.
[9] M.W. Medley, Microwave and RF Circuits: Analysis, Synthesis and Design, Artech House, 1993.
[10] K. C. Gupta, R. Garg, I. Bahl, P. Bhartia, Microstrip Lines and Slotlines, Artech House, 1996.
[11] W. Baechtold, Mikrowellentechnik: Kompakte Grundlagen fuer das Studium, Vieweg Verlag, 1999.