You are on page 1of 91

ME1000 RF CIRCUIT DESIGN

This courseware product contains scholarly and technical information and


is protected by copyright laws and international treaties. No part of this
publication may be reproduced by any means, be it transmitted,
transcribed, photocopied, stored in a retrieval system, or translated into
any language in any form, without the prior written permission of Acehub
Vista Sdn. Bhd. or their respective copyright owners.
The use of the courseware product and all other products developed
and/or distributed by Acehub Vista Sdn. Bhd. are subject to the applicable
License Agreement.
For further information, see the Courseware Product License Agreement.

http://dreamcatcher.asia/cw
11

3B. RF Microwave Filters

1.0 Basic Filter Theory

Introduction
An ideal filter is a linear 2-port network that provides
perfect transmission of signal for frequencies in a
certain passband region, infinite attenuation for
frequencies in the stopband region, and a linear
phase response in the passband (to reduce signal
distortion).
The goal of filter design is to approximate the ideal
requirements within acceptable tolerance with circuits
or systems consisting of real components.

Categorization of Filters

Low pass filter (LPF), high pass filter (HPF), bandpass filter (BPF),
bandstop filter (BSF), arbitrary type, etc.

In each category, the filter can be further divided into active and passive
types.

In an active filter, there can be amplification of the signal power in the


passband region; a passive filter do not provide power amplification in
the passband.

Filters used in electronics can be constructed from resistors, inductors,


capacitors, transmission line sections, and resonating structures (e.g.,
piezoelectric crystal, Surface Acoustic Wave (SAW) devices,
mechanical resonators, etc.).
Filter
An active filter may contain a transistor, FET,
and an op-amp.
LPF
HPF
BPF

Active

Passive

Active

Passive
5

Filter Frequency Response


Frequency response implies the behavior of the filter
with respect to steady-state sinusoidal excitation
(e.g., energizing the filter with a sine voltage or
current source and observing its output).
There are various approaches to displaying the
frequency response:

Transfer function H() (the traditional approach)


Attenuation factor A()
S-parameters, e.g., s21()
Others, such as ABCD parameters, etc.

Filter Frequency Response (contd)

|H()|

Low pass filter (passive)

Transfer
function

1
V1()

Filter
H()

V2()

ZL

V2
H
(1.1a)
V1
Complex value

Arg(H())

A()/dB
50
40
30
20
10
3
0

Real value

V2
Attenuation A 20 Log10
V1
c

(1.1b)

Filter Frequency Response (contd)

Low pass filter (passive) continued...


For the impedance matched system, using s21 to observe the filter
response is more convenient, as this can be easily measured using a
vector network analyzer (VNA).
Zc
Zc
a1
b2
Vs
Zc
Zc

20log|s21()|

Zc

Filter

Zc

Arg(s21())

Transmission line
is optional

0 dB

b
b
s11 1
s21 2
a1 a 0
a1 a 0
2
2

Complex value
8

Filter Frequency Response (contd)

Low pass filter (passive) continued...


A()/dB
50
40
30
20
10
3
0

Passband

Transition band

Stopband

Cut-off frequency (3 dB)


V1()

Filter
H()

V2()

ZL

Filter Frequency Response (contd)

High pass filter (passive)

|H()|

A()/dB

Transfer
function

Passband

50
40

30
20
10
3
0

Stopband

10

Filter Frequency Response (contd)


Bandpass filter (passive)

Bandstop filter

A()/dB

A()/dB

40

40

30

30

20

20

10
3
0

10
3
0

1 o 2

|H()|
1

1
|H()|

Transfer
function

Transfer
function

1 o 2

11

Basic Filter Synthesis Approaches

Image Parameter Method.

Filter

Zo

Zo

Zo

Zo

Zo

H1()

H2()

Consider a filter to be a
cascade of linear 2-port
networks.
Synthesize or realize each
2-port network, so that
the combine effect gives the
required frequency response.
The image impedance seen
at the input and output of
each network is maintained.

Zo

Response of
a single
network

Zo

Hn()

Zo

The combined
response
12

Basic Filter Synthesis Approaches (contd)

Insertion Loss Method.


|H()|

Filter

Zo

Approximate ideal filter response


with polynomial function:
Ideal
Approximate with rational polynomial
function
sn a
s n 1 a s a

H s K

n 1
1
o
n
n

1
s bn 1s
b1s bo

We can also use Attenuation Factor or |


s21| for this.
Use the RCLM circuit synthesis theorem to
come up with a resistive terminated
LC network that can produce the
Z
approximate response.
o

Zo

13

Our Scope
Only concentrate on passive LC and stripline filters.
Filter synthesis using the Insertion Loss Method
(ILM). The Image Parameter Method (IPM) is more
efficient and suitable for simple filter designs, but has
the disadvantage that arbitrary frequency response
cannot be incorporated into the design.

14

2.0 Passive LC Filter Synthesis Using


the Insertion Loss Method

15

Insertion Loss Method (ILM)

The insertion loss method (ILM) enables a systematic way to


design and synthesize a filter with various frequency responses.

The ILM method also enables a filter performance to be


improved in a straightforward manner, at the expense of a
higher order filter.

A rational polynomial function is used to approximate the ideal |


H()|, A(), or |s21()|.

Phase information is totally ignored.

Ignoring phase simplifies the actual synthesis method. An LC


network is then derived which will produce this approximated
response.

The attenuation A() can be cast into power attenuation ratio,


called the Power Loss Ratio, PLR, which is related to A()2.

16

tra
x
E

More on ILM

There is a historical reason why phase information is


ignored. Original filter synthesis methods are
developed in the 1920s60s, for voice
communication. The human ear is insensitive to
phase distortion, thus only the magnitude response
(e.g., |H()|, A()) is considered.
Modern filter synthesis can optimize a circuit to meet
both magnitude and phase requirements. This is
usually done using computer optimization procedures
with goal functions.

17

Power Loss Ratio (PLR)


Zs

Vs

Lossless
2-port network
PA

Pin

ZL
PL

PLR Power available from source network

Power delivered to Load


P
PA
1
inc

PLoad
2
2
1 1
PA 1 1

(2.1a)

PLR large,
high attenuation
P
LR large, high attenuation
P
closetoto1,1,low
lowattenuation
attenuation
PLRLRclose
For
Forexample,
example,aalow
lowpass
pass
filter
filterresponse
responseisisshown
shown
below:
below:
PLR(f)
High
attenuation

Low pass filter PLR

Low
attenuation

fc

f
18

PLR and s21

In terms of incident and reflected waves, assuming ZL = Zs = ZC.


b1
a1
Zc
Vs

PA

b2
Lossless
2-port network
Pin

Zc
PL

1a 2
2
1
PA
a1
2
PLR

PL
b2
1b 2
2
2
PLR 1
(2.1b)
2
s 21

19

PLR for the Low Pass Filter (LPF)

Since |1()|2 is an even function of , it can be written in terms of 2 as:

(2.2)

PLR can be expressed as:


1

M 2

1
N 2

1
2
1 1
M 2

1
M 2 N 2

Various types of polynomial

PLR

M 2
M 2 N 2

This is also known


as Characteristic Polynomial

PLR 1 P 2

M 2
N 2

(2.3a)
(2.3b)

function in P can
used for P(). The
P be
requirement is that P() must either be an odd or even function. Among the
classical polynomial functions are:
Maximally flat or Butterworth functions
Equal ripple or Chebyshev functions
Elliptic function
The
Thecharacteristics
characteristicswe
weneed
need
2
Many, many more
from [P()] 2for LPF:

from [P()] for LPF:


2
[P()]
[P()]2
00for
for<<c c
2
[P()]
[P()]2>>
>>11for
for>>
>>c c

20

Characteristic Polynomial Functions

Maximally flat or Butterworth:

Equal ripple or Chebyshev:

P
c

(2.4a)

NN==order
orderofofthe
the
Characteristic
Characteristic
Polynomial
PolynomialP()
P()

P C N , ripple factor

C0 1

C N C1
C 2C C , n 2
n 1
n 2
n

Bessel or linear phase:

(2.4b)

P 2 B j B j 1

B0 s 1

(2.4c)

BN s B1 s s 1

Bn s 2 s 1 Bn 1 s s 2 Bn 2 s , n 2
21

Examples of PLR for the Low Pass Filter

PLR of the low pass filter using 4th order polynomial functions (N
= 4) Butterworth, Chebyshev (ripple factor =1), and Bessel.
Normalized to c = 1 rad/s, k = 1.

2
2

PLR (chebyshev) 1 k 2 8 4 1
c
c

Ideal

1 10

1 10

k=1

Chebyshev

bt ( )

4
2
PLR ( Butterworth) 1 k
c

cb( )

PLR100

bs ( )

Butterworth
10

Bessel
1

If we convert into dB,


this ripple is equal to
3 dB

0.5

1.5

PLR ( Bessel ) 1 k 2 B j B j 1

B s 1 s
105 c

10 s
c

45 s
c

105 s 105
c

22

Examples of PLR for the Low Pass Filter (contd)

PLR of the low pass filter using the Butterworth characteristic


2
polynomial, normalized to c = 1 rad/s, k = 1.
N

PLR ( Butterworth) 1 k 2
c

5
1 10

PLR( 2) 4
1 10

N=7

PLR( 3)

PLR( 4)1 10
PLR( 5)
PLR( 6)
PLR( 7)

N=6

N=5
N=4

100

N=3
10

N=2

0.5

1.5

Conclusion:
Conclusion:
The
Thetype
typeofof
polynomial
polynomial
function
functionand
and
the
theorder
order
determine
determinethe
the
attenuation
attenuationrate
rate
ininthe
thestopband.
stopband.

23

Characteristics of Low Pass Filters Using Various


Polynomial Functions
Butterworth: Moderately linear phase response,
slow cutoff, smooth attenuation in the passband.
Chebyshev: Bad phase response, rapid cutoff for a
similar order, contains ripple in the passband. May
have impedance mismatch for N even.
Bessel: Good phase response, linear. Very slow
cutoff. Smooth amplitude response in the passband.

24

Low Pass Prototype Design

A lossless linear, passive, reciprocal network that can produce the


insertion loss profile for the low pass filter is the LC ladder network.

Many researchers have tabulated the values for the L and C for the low
pass filter with cut-off frequency c = 1 rad/s, that works with the source
and load impedance Zs = ZL = 1 .

This low pass filter is known as the Low Pass Prototype (LPP).

As the order N of the polynomial P increases, the required element also


increases. The no. of elements = N.
1

L1=g2

C1=g1

L1=g1

g0= 1

L2=g4

C2=g3

RL= gN+1

L2=g3

C1=g2

C2=g4

RL= gN+1

Dual of each
other

25

Low Pass Prototype Design (contd)


The LPP is the building block from which real filters
may be constructed.
Various transformations may be used to convert it
into a high pass, bandpass, or other filter of arbitrary
center frequency and bandwidth.
The following slides show some sample tables for
designing LPP for Butterworth and Chebyshev
amplitude response of PLR.

26

Table for the Butterworth LPP Design


N

g1

g2

g3

g4

g5

g6

g7

g8

1
2
3
4
5
6

2.0000
1.4142
1.0000
0.7654
0.6180
0.5176

1.0000
1.4142
2.0000
1.8478
1.6180
1.4142

1.0000
1.0000
1.8478
2.0000
1.9318

1.0000
0.7654
1.6180
1.9318

1.0000
0.6180
1.4142

1.0000
0.5176

1.0000

0.4450

1.2470

1.8019

2.0000

1.8019

1.2470

0.4450

1.0000

0.3902

1.1111

1.6629

1.9615

1.9615

1.6629

1.1111

0.3902

g9

1.0000

See Example 2.1 in the following slides on how the constant values g1, g2, g3,
etc., are obtained.

27

Table for the Chebyshev LPP Design

N
1
2
3
4
5
6

N
1
2
3
4
5
6

Ripple factor 20log10 = 0.5 dB


g1
0.6986
1.4029
1.5963
1.6703
1.7058
1.7254

g2
1.0000
0.7071
1.0967
1.1926
1.2296
1.2479

g3

g4

g5

g6

g7

1.9841
1.5963
2.3661
2.5408
2.6064

1.0000
0.8419
1.2296
1.3137

1.9841
1.7058
2.4578

1.0000
0.8696

1.9841

Ripple factor 20log10 = 3.0 dB


g1
1.9953
3.1013
3.3487
3.4389
3.4817
3.5045

g2
1.0000
0.5339
0.7117
0.7483
0.7618
0.7685

g3

g4

g5

g6

g7

5.8095
3.3487
4.3471
4.5381
4.6061

1.0000
0.5920
0.7618
0.7929

5.8095
3.4817
4.4641

1.0000
0.6033

5.8095
28

Table for the Maximally-Flat Time Delay LPP Design

29

Example 2.1: Finding the Constants for the LPP


Design
a
xtr

Consider a simple case of a 2nd order low pass filter:


R

Vs

V1

R V
1 jRC s
R jL 1 jRRC

Thus

Vs

jL
R V1

1/jC

RV

RV

s
R R jL s1 jRC
2
2 R RLC j L R 2C

2
PL 21R V1

Vs R

and

2 2 2 LC R 2 2 L R 2C

PA 81R Vs

Therefore we can compute the power loss ratio as:


2

PLR P A
L
P

Vs
8R

Vs R

2
2

2 2 2 LC R 2 2 L R 2C

2
2 4

1 1 2 L R 2C LC 2 LC

2

4R

2
2

1 2 2 R 2 2 2 LC 2 L R 2C 2
8R

[P()]2
30

Example 2.1: Finding the Constants for the LPP


Design (contd)
a
xtr

PLR can be written in terms of polynomial of 2:

2
2 4

PLR 1 1 2 L R 2C LC 2 LC
1 a1 2 a2 4
2

4R

(E1.1)

For Butterworth response with k = 1, c = 1:

2 1 4 1 0 2 1 4

PLR ( Butterworth) 1 2

Comparing equations (E1.1) and (E1.2):


a2 1 LC
1 LC 2
2

(E1.3)

(E1.2)

L R 2C

2
a1 0 1 2 L R 2C LC 0

LC

4R
21R

(E1.4)

Setting R = 1 for the Low Pass Prototype (LPP):

R 1 Thus from equation (E1.4):

LC 2 C 2 2

LC 12 L C 2 L2 C 2 2 LC 0

C 2 1.4142

L C2 0
LC

Using (E1.3)

L C 1.4142

Compare
Comparethis
thisresult
resultwith
with
N=2
in
the
table
for
the
N=2 in the table for the
LPP
LPPButterworth
Butterworthresponse.
response.
This
direct
brute
force
This direct brute force
approach
approachcan
canbe
be
extended
to
N=3,
extended to N=3,4,4,5
5
31

tra
x
E

Example 2.1: Verification


AC
AC
AC1
Start=0.01 Hz
Stop=2.0 Hz
Step=0.01 Hz
Vin
R
R2
R=1 Ohm

V_AC
SRC1
Vac=polar(1,0) V
Freq=freq

Vout
L
L1
L=1.4142 H
R=

C
C1
C=1.4142 F

R
R1
R=1 Ohm

32

Example 2.1: Verification (contd)

tra
x
E

Eqn PL=0.5*mag(Vout)*mag(Vout)

Eqn PA=1/8

Eqn PLR=PA/PL

m1

m1
freq=160.0mHz
m1=-3.056

dB(Vout/0.5)

-10

2.5E4

PLR

2.0E4

Power loss ratio


versus frequency

1.5E4

-20
-30
-40
-50

1.0E4

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

freq, Hz

5.0E3

3 dB at 160 mHz (milliHertz!!),


which is equivalent to 1 rad/s

0.0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

freq, Hz

33

Impedance Denormalization and Frequency


Transformation of LPP

Once the LPP filter is designed, the cut-off frequency c can be


transformed to other frequencies.

Furthermore the LPP can be mapped to other filter types such


as high pass, bandpass, and bandstop.

This frequency scaling and transformation entails changing the


value and configuration of the elements of the LPP.

Finally the impedance presented by the filter at the operating


frequency can also be scaled, from unity to other values; this is
called impedance denormalization.

Let Zo be the new system impedance value. The following slide


summarizes the various transformation from the LPP filter.

34

Impedance Denormalization and Frequency


Transformation of LPP (contd)
LPP to
High Pass

LPP to
Low Pass
L

Zo L
c

LZ o
o

1
c LZ o

o LZ o

Zo
cC

C
Z o c

C
o Z o

Center frequency

o 1 2 2 or

1 2

1
o LZ o

Z o
oC

Fractional bandwidth

(2.5a)

LPP to
Bandstop

LPP to
Bandpass


2 1 (2.5b)
o

LZ o
o

Zo
o C
C
o Zo
Note
Notethat
thatthe
theinductor
inductoralways
always
multiplies
with
Z
while
the
o
multiplies with Zo while the
capacitor
capacitordivides
divideswith
withZZo
o

35

Summary of Passive LC Filter Design Flow Using the


ILM Method
Step 1: From the requirements, determine the order
and type of approximation functions to use.

Insertion loss (dB) in the passband ?


Attenuation (dB) in the stopband ?
Cut-off rate (dB/decade) in the transition band ?
Tolerable ripple?
Linearity of phase?

Step 2: Design a normalized low pass prototype


(LPP) using the L and C elements.
|H()|

L1=g2
C1=g1

L2=g4

C2=g3

1
RL= gN+1

36

Summary of Passive Filter Design Flow Using the


ILM Method (contd)

Step 3: Perform frequency scaling, and denormalize the


impedance.
|H()|
50

Vs

15.916pF

79.58nH

0.7072nH

0.1414pF

15.916pF

1
0.7072nH

RL
50

Step 4: Choose suitable lumped components, or transform the


lumped circuit design into distributed realization.

All
Alluses
usesthe
themicrostrip
microstripstripline
striplinecircuit
circuit
37

Filter vs. Impedance Transformation Network


tra
x
E

If we ponder carefully, the sharp observer will notice


that the filter can be considered as a class of
impedance transformation network.
In the passband, the load is matched to the source
network, much like a filter.
In the stopband, the load impedance is highly
mismatched from the source impedance.
However, the procedure described here only applies
to the case when both load and source impedance
are equal and real.
38

Example 2.2A: LPF Design Butterworth Response

Design a 4th order Butterworth low pass filter, Rs = RL= 50 ,


fc = 1.5 GHz.

Steps 1 & 2: LPP

L1=0.7654H

g 0= 1

Step 3: Frequency scaling


and impedance denormalization

L2=1.8478H

C1=1.8478F

L1=4.061 nH

g0=1/50

c 2 1.5GHz 9.4248 109 rad/s


Z o 50

C2=0.7654F

RL= 1

L
L Zo n
c
Cn
C
Z o c

L2=9.803 nH

C1=3.921 pF

C2=1.624 pF

R Z o Rn

RL= 50

39

Example 2.2B: LPF Design Chebyshev Response

Design a 4th order Chebyshev low pass filter, 0.5 dB ripple


factor, Rs = 50 , fc = 1.5 GHz.

Steps 1 & 2: LPP

L1=1.6703H

g 0= 1

L2=2.3661H

C1=1.1926F

Step 3: Frequency scaling


and impedance denormalization L =8.861 nH
1

g0=1/50

c 2 1.5GHz 9.4248 109 rad/s


Z o 50

C2=0.8419F

RL=
1.9841

L
L Zo n
c
Cn
C
Z o c

L2=12.55 nH

C1=2.531 pF

C2=1.787 pF

R Z o Rn

RL=
99.2

40

Example 2.2 (contd)


Ripple is roughly
0.5 dB

dB(LPF_butterworth..S(2,1))
dB(S(2,1))

5
0

Butterworth

Computer simulation result


using AC analysis (ADS2003C)

-10

|s21|
-20

Chebyshev

-30
0.5

1.0

1.5

2.0

2.5

3.0

freq, GHz

Note: Equation used in Data Display of ADS2003C


to obtain a continuous phase display with the built-in
function phase( ).
Eqn Phase_chebyshev = if (phase(S(2,1))<0) then phase(S(2,1)) else (phase(S(2,1))-360)

-50

P hase_butterworth
P hase_chebyshev

0.0

Better phase
linearity for Butterworth
LPF in the passband

-100
-150

Butterworth

Arg(s21)
-200
-250

Chebyshev

-300
-350
0.0

0.5

1.0

1.5

2.0

2.5

3.0

freq, GHz

41

Example 2.3: BPF Design


Design a bandpass filter with Butterworth (maximally
flat) response.
N=3
Center frequency fo = 1.5 GHz
3 dB Bandwidth = 200 MHz or f1 = 1.4 GHz, f2 =
1.6 GHz
Impedance = 50

42

Example 2.3 (contd)

From table, design the low pass prototype (LPP) for 3rd order
Butterworth response, c = 1.

Steps 1 & 2: LPP

Zo=1

2<0o

g2
2.000H

g1
1.000F

g3
1.000F

g4
1

Simulated result
using PSpice

c 2f c 1
f c 21 0.1592 Hz
Voltage across g4

43

Example 2.3 (contd)

LPP to bandpass transformation

Impedance denormalization

1 2 1.4GHz
2 2 1.6GHz
fo

Step 3: Frequency scaling


and impedance denormalization
LZ o
o

o LZ o

50

Vs

f1 f 2 1.497GHz

2
o
C
o Z o

79.58 nH

0.133

Z o
oC

0.1414 pF

RL
15.916 pF

0.7072 nH

15.916 pF

0.7072 nH

50

44

Example 2.3 (contd)


Simulated result using PSpice:

Voltage across RL

45

All Pass Filter

tra
x
E

There is also another class of filter known as the All Pass Filter (APF).

This type of filter does not produce any attenuation in the magnitude
response, but provides phase response in the band of interest.

APF is often used in conjunction with LPF, BPF, HPF, etc., to


compensate for phase distortion.

Example of the APF response

|H(f)|

Arg(H(f))

1
0

Nonlinear
phase in
passband
f

|H(f)|

Arg(H(f))

1
0

|H(f)|

BPF

APF

Zo

Arg(H(f))

1
0

Linear
phase in
passband
f
46

Example 2.4: Practical RF BPF Design Using SMD


Discrete Components
CPWSub

C
Ct3
C=Ct_value2 pF

L
Lt1
L=Lt_value nH
R=

Term
Term1
Num=1
Z=50 Ohm

C
Ct1
C=Ct_value pF

CP WSUB
CP WSub1
H=62.0 mil
Er=4.6
Mur=1
Cond=5.8E+7
T=1.38 mil
TanD=0.02
Rough=0.0 mil

S-PARAMETERS

Var
E qn

S_P aram
SP 1
Start=0.1 GHz
Stop=3.0 GHz
Step=1.0 MHz

INDQ
1_0pF_NP O_0603
CP WG
L4
C1
CP W1
L=15.0 nH
Q=90.0
Subst="CP WSub1"
b82496c3229j000
4_7pF_NPF=
O_0603
800.0 MHz
W=50.0 mil
L2
C2
G=10.0 mil
Mode=proportional to freq
param=
SIMID
0603-C
(2.2
nH
+
-5%)
L=28.0 mm
Rdc=0.1 Ohm

VAR
VAR1
Lt_value=4.8
Ct_value=3.5
Ct_value2=2.9

C
Ct2
C=Ct_value pF

CP WG
CP W2
b82496c3229j000
Subst="CP WSub1"
L3
W=50.0
mil
4_7pF_NP
O_0603
param=SIMID 0603-C (2.2
nH +-5%)
G=10.0 mil
C3
L=28.0 mm

L
Lt2
L=Lt_value nH
R=

C
Ct45
C=Ct_value2 pF

T erm
T erm2
Num=2
Z=50 Ohm

47

Example 2.4 (contd)


BPF synthesis
using synthesis
tool E-syn
of ADS2003C

48

Example 2.4 (contd)


|s21|/dB

Measured
Simulated

-20
-20

Measurement is performed with the


Agilent 8753ES Vector Network
Analyzer, using Full OSL calibration

-40
-40

Arg(s21)/degree

-60
-60
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

freq, GHz
freq, GHz

200
200

phase(RF_BPF_measured..S(2,1))
phase(RF_BPF_m
easure
d))
..S(2,1))
phase(S(2
,1
phase(S(2,1))

dB(RF_BPF_measured..S(2,1))
dB(RF_BPF_m
eB(S(2
asure,1
d..S(2
d
)) ,1))
dB(S(2,1))

100
100

-100
-100

-200
-200
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

freq, GHz
freq, GHz

49

3.0 Microwave Filter Realization Using


Stripline Structures

50

3.1 Basic Approach

51

Filter Realization Using Distributed Circuit Elements


Lumped-element filter realization using surface
mounted inductors and capacitors generally works well
at lower frequencies (at UHF, say < 3 GHz).
At higher frequencies, the practical inductors and
capacitors loses their intrinsic characteristics.
Also, a limited range of component values is available
from the manufacturer.
Therefore, for microwave frequencies (> 3 GHz), the
passive filter is usually realized using distributed circuit
elements such as transmission line sections.
Here we will focus on stripline microwave circuits.

52

Filter Realization Using Distributed Circuit Elements


(contd)

Recall in the study of Terminated Transmission Line Circuit that a length


of terminated Tline can be used to approximate an inductor and a
capacitor.
This concept forms the basis of transforming the LC passive filter into
distributed circuit elements.
l
Zc ,
l
Zc ,

L
Zo

Zc ,

Zc ,

Zo

Zc ,

Zo

Zo

53

Filter Realization Using Distributed Circuit Elements


(contd)

Zc ,

This approach is only an approximation. There will be deviation


between the actual LC filter response and those implemented with
terminated Tline.
Also, the frequency response of the distributed circuit filter is periodic.
Other issues are shown below.
Connection of physical
How do we implement a series Tline
length cannot be
connection? (only practical for
ignored at the
certain Tline configuration)
microwave region,
Zo
comparable to
Zo

Zc ,

Zc ,

Thus
Thussome
sometheorems
theoremsare
areused
usedtoto
facilitate
facilitatethe
thetransformation
transformationof
ofthe
theLC
LC
circuit
circuitinto
intostripline
striplinemicrowave
microwavecircuits.
circuits.
Chief
Chiefamong
amongthese
theseare
arethe
theKurodas
Kurodas
Identities
Identities(See
(SeeAppendix
Appendix1)
1)

54

More on Approximating L and C with Terminated


Tline: Richards Transformation
Z in jZ c tan l jL jL
tan l

Zin

Zc L

Zc ,

(3.1.1a)

Here, instead of fixing Zc and tuning l to approach an L or C,


we allow Zc to be a variable too.
l

Zin

Zc ,

Yin jYc tan l jC jC


tan l
(3.1.1b)
Yc 1 C

For LPP design, a further requirement is


that:

tan l c 1

Zc

Wavelength at
cut-off frequency

tan
l 1 l
(3.1.1c)

8
c

55

Example 3.1: LPF Design Using Stripline

Design a 3rd order Butterworth low pass filter, Rs = RL= 50 ,


fc = 1.5 GHz.
g3
1.000H

g4
1

g2
2.000F

Zc=1.000

Zo=1

g1
1.000H

Step 3: Convert to Tlines


Length = c/8
for all Tlines
at = 1 rad/s

Zc=0.500

1 0.500
2.000

Zc=1.000

Steps 1 & 2: LPP

56

Example 3.1 (contd)


Step 4: Add an extra Tline on the series
connection and apply Kurodas
2nd Identity.

Z1 1.0

Z2=1

Yc
1
n2Z

0.5
2

n2Z1=2

Zc=1.000

Extra Tline

Zc=1.000

Extra Tline

Z
n2 1 2
Z1

1 1 2
1

Zc=1.0
Zc=1.0

Zc=0.500

Similar operation is
performed here

Length = c/8
for all Tlines
at = 1 rad/s
57

Example 3.1 (contd)


After applying Kurodas 2nd Identity
1
Zc=2.0

Zc=2.0

Zc=2.000

Zc=0.500

Zc=2.000

Length = c/8
for all Tlines
at = 1 rad/s

Since
Sinceall
allTlines
Tlineshave
havesimilar
similarphysical
physical
length,
length,this
thisapproach
approachto
tostripline
striplinefilter
filter
implementation
implementationisisalso
alsoknown
knownas
as
Commensurate
CommensurateLine
LineApproach.
Approach.
58

Example 3.1 (contd)


Step 5: Impedance and frequency denormalization
50
Zc=100

Here we multiply all


impedance with Zo = 50

Zc=100

Zc=100

Zc=25

Zc=100

Length = c/8
for all Tlines at
f = fc = 1.5 GHz

50

Microstrip line using double-sided FR4 PCB (r = 4.6, H=1.57 mm)


Zc/
50
25
100

/8 @ 1.5 GHz/mm
13.45
12.77
14.23

W/mm
2.85
8.00
0.61

We
Wecan
canwork
workout
outthe
thecorrect
correctwidth
widthW
Wgiven
giventhe
the
impedance,
dielectric
constant,
and
thickness.
impedance, dielectric constant, and thickness.
From
FromW/H
W/Hratio,
ratio,the
theeffective
effectivedielectric
dielectricconstant
constant
eff can
be
determined.
Use
this
together
with
eff can be determined. Use this together with
frequency
frequencyatat1.5
1.5GHz
GHztotofind
findthe
thewavelength.
wavelength.
59

Example 3.1 (contd)


Step 6: The layout (top view)

60

Example 3.1 (contd)


Simulated results
S-PARAMETERS

MSub

MTEE
Tee1
Subst="MSub1"
W1=2.85 mm
W2=0.61 mm
W3=0.61 mm

MLIN
TL1
Subst="MSub1"
W=2.85 mm
L=25.0 mm

Term
Term1
Num=1
Z=50 Ohm

MTEE
Tee3
Subst="MSub1"
W1=0.61 mm
W2=0.61 mm
W3=8.00 mm

MLIN
TL3
Subst="MSub1"
W=0.61 mm
L=14.23 mm

MLOC
TL6
Subst="MSub1"
W=0.61 mm
L=14.23 mm

MTEE
Tee2
Subst="MSub1"
W1=0.61 mm
W2=2.85 mm
W3=0.61 mm

MLIN
TL4
Subst="MSub1"
W=0.61 mm
L=14.23 mm

MLOC
TL5
Subst="MSub1"
W=8.0 mm
L=12.77 mm

MLIN
TL2
Subst="MSub1"
W=2.85 mm
L=25.0 mm

MLOC
TL7
Subst="MSub1"
W=0.61 mm
L=14.23 mm

L
L2
C
L=5.305 nH
C1
R=
C=4.244 pF

Term
Term2
Num=2
Z=50 Ohm

m1
freq=1.500GHz
m1=-6.092
Term
Term2
Num=2
Z=50 Ohm

dB(Butter_LPF_LC..S(2,1))
dB(S(2,1))

MSUB
MSub1
H=1.57 mm
Er=4.6
Mur=1
Cond=1.0E+50
Hu=3.9e+034 mil
T=0.036 mm
TanD=0.02
Rough=0 mil

Term
Term1
Num=1
Z=50 Ohm

S_Param
SP1
Start=0.2 GHz
Stop=4.0 GHz
Step=5 MHz

L
L1
L=5.305 nH
R=

m1
-10

-20

-30

-40
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

freq, GHz

61

Conclusions for Section 3.1


Further tuning is needed to optimize the frequency
response.
The method illustrated is good for the low pass and
bandstop filter implementation.
For high pass and bandpass, other approaches are
needed.

62

3.2 Further Implementations

63

Realization of LPF Using the Step-Impedance


Approach
A relatively easy way to implement LPF using
stripline components.
Using alternating sections of high and low
characteristic impedance Tlines to approximate the
alternating L and C elements in an LPF.
Performance of this approach is marginal as it is an
approximation, where a sharp cutoff is not required.
As usual, beware of parasitic passbands!!!

64

Equivalent Circuit of a Transmission Line Section


T-network equivalent circuit
Z11 Z12

Ideal lossless Tline

Z11 Z12

Z12

Positive
susceptance
Z11 Z12 jZ c
jZ c

1
sin l


jZ c
sin 2
l
2
l
2

1 cos 2

jZ c tan

l
cos
sin l


2 sin 2 cos 2

l
2

Positive reactance

2 sin 2 2
l
2

l
2

l
2

Zc

Z11 Z 22 jZ c cot l

(3.2.1a)

Z12 Z 21 jZ c cosec l (3.2.1b)


o e o e k o

(3.2.1c)

65

Approximation for High and Low ZC

When l < /2, the series element can be thought of as an inductor and
the shunt element can be considered a capacitor.

X
l
Z11 Z12 Z c tan

2
2

For l < /4 and Zc = ZH >> 1:

For l < /4 and Zc = ZL 1:


Z11 - Z12

1
1
B
sin l
Z12
Zc

X ZH l
B0
1
X 0
B
l
ZL

Z11 - Z12

When Zc >> 1
l < /4

Z12

jX/2

X ZH l

jX/2

jB

When Zc 1
l < /4

B YLl
66

Approximation for High and Low ZC (contd)

Note that l < /2 implies a physically short Tline. Thus a short


Tline with high Zc (e.g., ZH) approximates an inductor.

lL

c L
ZH

A short Tline with low Zc (e.g., ZL) approximates a capacitor.

c CZ L
lC

(3.2.2a)

(3.2.2b)

The ratio of ZH/ZL should be as high as possible. Typical values:


ZH = 100 to 150 , ZL = 10 to 15 .

67

Example 3.2: Mapping an LPF Circuit into a Step


Impedance Tline Network

For instance, consider the LPF Design Example 2.2A


(Butterworth).

Let us use the microstrip line. Since a microstrip Tline with low
Zc is wide and a Tline with high Zc is narrow, the transformation
from circuit to physical layout would be as follows:
L1=4.061 nH

g0=1/50

L2=9.803 nH

C1=3.921 pF

C2=1.624 pF

RL= 50

68

Example 3.2: Physical Realization of LPF

Using the microstrip line, with r = 4.2, d = 1.5 mm:

Zc = 15
Zc = 50
Zc = 110

W/d
10.0
2.0
0.36

d/mm
1.5
1.5
1.5

W/mm
15.0
3.0
0.6

e
3.68
3.21
2.83

L eL ko eL 2f c 3.3356 10 9 60.307 s 1
H eH ko eH 2f c 3.3356 10 9 53.258s 1

L1 = 4.061 nH, L2 = 9.083 nH, C1 = 3.921 pF, C2 = 1.624 pF

69

Example 3.2: Physical Realization of LPF (contd)


l1

c L1
6.5mm
ZH H

Verification:

H l1 0.392 4 0.7854
L l2 0.490 4 0.7854

CZ
l2 c 1 L 9.2mm
L

H l3 0.905 4 0.7854

l3 15.0mm
l4 3.8mm

3.0 mm

L l4 0.202 4 0.7854
l1

l2

l3

Nevertheless we still
proceed with the implementation. It will be seen
that this will affect the
accuracy of the 3 dB cut-off
point of the filter.

l4
50 line

50 line

To 50
Load
15.0 mm

0.6 mm

70

Example 3.2: Step Impedance LPF Simulation with


ADS Software

Transferring the microstrip line design to ADS:

Microstrip line substrate model

Microstrip line model


Microstrip step junction
model

71

Example 3.2: Step Impedance LPF Simulation with


ADS Software (contd)
m1
freq=1.410GHz
dB(S(2,1))=-3.051
0

m1

dB(S(2,1))

-5

-10

-15

-20

-25
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

freq, GHz

72

Example 3.2: Step Impedance LPF Simulation with


ADS Software (contd)
However if we extent the stop frequency for the S-parameter
simulation to 9 GHz...
Parasitic passbands,
artifacts due to using
m1
freq=1.410GHz
transmission lines
dB(S(2,1))=-3.051
0

m1

-5

dB(S(2,1))

-15

-25
0

freq, GHz
73

Example 3.2: Verification with Measurement


The 3 dB point is around 1.417 GHz!

The actual LPF constructed in year


2000. The Agilent 8720D Vector
Network Analyzer is used to perform
the S-parameter measurement.

74

Example 3.3: Realization of BPF Using a Coupled


Microstrip Line
a
xtr

Based on the BPF design of Example 2.3:


50

Vs

79.58 nH

15.916 pF

To source
network

o
4

J1
90o

0.7072 nH

Admittance
inverter
Tline
J2
90o

0.1414 pF

15.916 pF

See appendix (using Richards transformation


and Kurodas identities)
J3
90o

J4
90o

To RL

An array of coupled
microstrip line

o
4

Section 1

0.7072 nH

RL
50

An equivalent circuit
model for coupled Tlines
with open circuit at
two ends.

Section 2

Section 3

Section 4

o ==wavelength
wavelengthatatoo
o
75

Example 3.3: Realization of BPF Using the Coupled


Microstrip Line (contd)
a
xtr
E

Each section of the coupled stripline contains three parameters: S, W,


d. These parameters can be determined from the values of the odd and
even mode impedances (Zoo & Zoe) of each coupled line.
W

Zoo and Zoe are in turn depend on the gain of the corresponding
admittance inverter J.
From Example 2.3

And each Jn is given by:

Z oe Z o 1 JZ o JZ o 2

fo

Z oo

f1 f 2 1.497GHz

2
o

2 g1

J1 Z1

Z 1 JZ

1 2 1.4GHz
2 2 1.6GHz

0.133

JZ
o
o

J n 2 Z1
for n 2,3,4 N
g n 1g n
o

J N 1 Z1

2 g N g N 1
76

Example 3.3: Realization of BPF Using the Coupled


Microstrip Line (contd)
a
xtr
E

Section 1:

J1

0.009163
2 g1

1
Zo

Section 2:

J2

1
2Zo

Z oo1 Z o 1 J1Z o J1Z o 2 37.588


Z oe1 Z o 1 J1Z o J1Z o 2 83.403

Z oo 2 Z o 1 J 2 Z o J 2 Z o 2 43.680

0.002969
g1g 2

Z oe 2 Z o 1 J 2 Z o J 2 Z o 2 58.523

0.002969
g 2 g3

Z oe3 83.403

0.009163
2 g3 g 4

Z oe 4 58.523

Section 3:

J 3 2 Z1

Z oo3 37.588

Section 4:

J 4 Z1
o

Z oo 4 43.680

Note:
Note:
gg1=1.0000
1=1.0000
gg2=2.0000
2=2.0000
gg3=1.0000
3=1.0000
gg4=1.0000
=1.0000
4

77

Example 3.3: Realization of BPF Using the Coupled


Microstrip Line (contd)
a
xtr
E

In this example, an edge-coupled microstrip line is used to implement


the coupled transmission line structures needed in the BPF. Stripline
does not suffer from dispersion and its propagation mode is pure TEM
mode, however it is more difficult to implement physically due to the
fact that the trace is buried within the dielectric.
Design equations for coupled microstrip line implemented are widely
tabulated.
Here we will use FR4 (r = 4.6, r = 1.0) substrate with 1.0mm dielectric
thickness, and 1 ounce copper (about 36m thick) copper laminate.
The conductivity of copper is around 5.8107 Siemens/meter.
Furthermore we will use the LineCal tool in Advanced Design System
to work out the dimensions needed for the coupled microstrip line.

78

Example 3.3: Realization of BPF Using


tra
x
E
the Coupled Microstrip Line (contd)
Using the LineCal tool to work out the dimensions for sections
1 and 3.
Strategy:
1) We tune the W and
S for the specified
Zoo and Zoe.
2) Then we tune the length L
to meet the electrical length of
/2 (quarter wavelength) at
1.5GHz.
Zoe
Zoo
Zo
Fix the frequency at Voltage coupling
Electrical length (l ), 90o for quarter
1.5GHz, the center factor in dB
wavelength.
of passband

79

Example 3.3: Realization of BPF Using


tra
x
E
the Coupled Microstrip Line (contd)
Using the LineCal tool to work out the dimensions for
sections 2 and 4.

80

Example 3.3: Realization of BPF Using


tra
x
E
the Coupled Microstrip Line (contd)
Alternatively we can implement our own design tool, as shown below implemented on
Microsoft Excel .

Based on design equations from Garg R., Bahl I.


J.,Characteristics of coupled microstriplines, IEEE
Transaction on Microwave Theory and Techniques,
MTT-27, No.7, pp. 700-705,July 1979.

Strategy:
1) We tune the W and
S for the specified Zoo and Zoe.
2) Based on the width W of
a single trace, we work out
the effective permittivity, and
use this to calculate the phase
velocity.
3) From this we find the
wavelength at 1.5GHz and
work out the required quarter
wavelength.

81

Example 3.3: Coupled-Line BPF Simulation with ADS


Software
a
xtr
E

Using ideal transmission line elements:


Ideal open circuit
Ideal coupled tline

82

Example 3.3: Coupled-Line BPF Simulation with ADS


Software (contd)
a
xtr
E

Parasitic passbands. Artifacts due to using distributed


elements, these are not present if lumped components
are used.
1.0

mag(S(2,1))

0.8

0.6

0.4

0.2

0.0
1.0

1.5

2.0

2.5

2fo

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

10.0

freq, GHz

83

Example 3.3: Coupled-Line BPF Simulation With


ADS Software (contd)
a
xtr

Using a practical stripline model:

Stripline substrate model

Coupled stripline model


Open circuit
model

84

Example 3.3: Coupled-Line BPF Simulation with ADS


Software (contd)
a
xtr
E

Attenuation due to losses in the conductor and dielectric


1.0
0.9
0.8

mag(S(2,1))

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

10.0

freq, GHz

85

Items for Self-Study


Network analysis and realizability theory
Synthesis of terminated RLCM 1-port circuits
Ideal impedance and admittance inverters and
practical implementation
Periodic structures theory
Filter design by the Image Parameter Method (IPM).

86

Other Types of Stripline Filter

LPF

HPF:

BPF:
SMD capacitor

87

Other Types of Stripline Filter (contd)

More BPFs:

BSF:

88

Appendix 1 Kurodas Identities

89

Kurodas Identities

As extracted from
Ref. [2]
l
1
Z2

Z1

Z2

Z1

Z1
n

n2Z1

1
2
n Z2

Z2/n

Note:
Note:The
The
length
lengthofofall
all
transmission
transmission
lines
linesisis
l l==/8
/8

1: n2
2

Z1
n2

l
1
Z2

l
Z1

Z2

Z1

Z2/n

l
Z1

Z
n2 1 2

Note: The inductor represents the


shorted Tline while the capacitor
represents the open-circuit Tline.

n2Z1

n2: 1

1
n2Z

90

References
[1] R. E. Collin, Foundations for Microwave Engineering, 2nd Edition 1992,
McGraw-Hill.
[2] D. M. Pozar, Microwave Engineering, 2nd Edition 1998, John-Wiley &
Sons.* (3rd Edition 2005, John-Wiley & Sons is now available)
Other more advanced references:
[3] W. Chen (Editor), The Circuits and Filters Handbook, 1995, CRC
Press.*
[4] I. Hunter, Theory and Design of Microwave Filters, 2001, The Institution
of Electrical Engineers.*
[5] G. Matthaei, L. Young, E.M.T. Jones, Microwave Filters, ImpedanceMatching Networks, and Coupling Structures, 1980, Artech House.*
[6] F. F. Kuo, Network Analysis and Synthesis, 2nd Edition 1966, JohnWiley & Sons.
**Recommended
Recommended
91