Download as pdf or txt
Download as pdf or txt
You are on page 1of 77

EKT 356

MICROWAVE COMMUNICATIONS

CHAPTER 4:
MICROWAVE FILTERS

1
INTRODUCTION
What is a Microwave filter ?
 linear 2-port network
 controls the frequency response at a certain point in
a microwave system
 provides perfect transmission of signal for
frequencies in a certain passband region
 infinite attenuation for frequencies in the stopband
region
 a linear phase response in the passband (to reduce
signal distortion).

2
INTRODUCTION

 The goal of filter design is to approximate the ideal


requirements within acceptable tolerance with
circuits or systems consisting of real components.

f1
f2

f3
Commonly used block Diagram of a Filter

3
INTRODUCTION
Why Use Filters?
 RF signals consist of:
1. Desired signals – at desired frequencies
2. Unwanted Signals (Noise) – at unwanted
frequencies
 That is why filters have two very important
bands/regions:
1. Pass Band – frequency range of filter where it
passes all signals
2. Stop Band – frequency range of filter where it
rejects all signals

4
INTRODUCTION

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 active filter, there can be amplification of the of the signal
power in the passband region, passive filter do not provide
power amplification in the passband.
 Filter 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, and also mechanical resonators etc.).
 Active filter may contain transistor, FET and Op-amp.
Filter

LPF HPF BPF

Active Passive Active Passive


5
INTRODUCTION

Types of Filters
1. Low-pass Filter 2. High-pass Filter
f1 f1
f1 f2

f2 f2

 Passes low freq  Passes high freq


 Rejects high freq  Rejects low freq

6
INTRODUCTION

3. Band-pass Filter 4.Band-stop Filter


f1 f1 f1
f2
f2 f2

f3 f3 f3

Rejects a small range of


 Passes a small range
freq
of freq
Passes all other freq
 Rejects all other freq

7
INTRODUCTION

Filter Parameters
 Pass bandwidth; BW(3dB) = fu(3dB) – fl(3dB)
 Stop band attenuation and frequencies,
 Ripple difference between max and min of
amplitude response in passband
 Input and output impedances
 Return loss
 Insertion loss
 Group Delay, quality factor

8
INTRODUCTION
|H(ω)|
 Low-pass filter (passive).
Transfer
1 V2 (ω )
function
( )
Hω = (1.1a)
V1(ω) A Filter V2(ω)
V1 (ω )
H(ω) ZL

ω
ωc
Arg(H(ω))

A(ω)/dB

50
40 ω

30
 V2 (ω ) 
Attenuation A = −20 Log10  
20
 (1.1b)
 V1 (ω ) 
10
3
0 ω
ωc

9
INTRODUCTION

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

Zc Zc
Filter
Zc
20log|s21(ω)| Arg(s21(ω))
Transmission line
0dB is optional
b b
s11 = 1 s21 = 2
a1 a =0 a1 a =0
2 2
ω ω
ωc
Complex value

10
INTRODUCTION

Low pass filter response (cont)


A(ω)/dB
Transition band
Passband
50
40
30
20
10
3 Stopband
0 ω
ωc

Cut-off frequency (3dB)

V1(ω) A Filter V2(ω)


H(ω) ZL

11
INTRODUCTION

High Pass filter

A(ω)/dB
Transfer
|H(ω)| function
50 Passband
1 40

30
20
10
3
ω ω
0
ωc ωc

Stopband

12
INTRODUCTION
Band-pass filter (passive). Band-stop filter.
A(ω)/dB A(ω)/dB

40 40

30 30
20 20
10 10
3 3
ω ω
0 ω1 ωo ω2 0
ω1 ωo ω2

|H(ω)| |H(ω)| Transfer


function
1 Transfer 1
function

ω ω
ω1 ωo ω2 ω1 ωo ω2

13
INTRODUCTION
Insertion Loss
Pass BW (3dB)
Filter R esp on se
0

12 .1 24 G H z
Q factor
-1 0 7 .9 0 24 G H z -3 .0 0 3 8 d B
-3 .0 0 5 7 dB

-2 0

-3 0

Inp ut R eturn L oss


-4 0
Ins ertio n Lo ss

-5 0
6 8 10 12 14
F re q u e n cy (G H z)

Figure 4.1: A 10 GHz Parallel Coupled Filter Response


Stop band frequencies and attenuation 14
FILTER DESIGN METHODS

Filter Design Methods

Two types of commonly used design methods:


1. Image Parameter Method
2. Insertion Loss Method

• Image parameter method yields a usable filter


• However, no clear-cut way to improve the design i.e to control the
filter response

15
FILTER DESIGN METHODS

Filter Design Methods


•The insertion loss method (ILM) allows a systematic way
to design and synthesize a filter with various frequency
response.

•ILM method also allows 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(ω)|, A(ω) or |s21(ω)|.

16
Filter Design Methods

 Phase information is totally ignored.Ignoring


phase simplified the actual synthesis method.
An LC network is then derived that will
produce this approximated response.

 Here we will use A(ω) following [2]. The


attenuation A(ω) can be cast into power
attenuation ratio, called the Power Loss Ratio,
PLR, which is related to A(ω).
FILTER DESIGN METHODS
Zs

Lossless
Vs 2-port network ZL
PA PL
Pin
PPLR large, high attenuation
LR large, high attenuation
PPLR close to 1, low attenuation
Γ1(ω) LR close to 1, low attenuation
For
Forexample,
example,aalow-pass
low-pass
filter
filterresponse
responseisisshown
shown
PLR = Power available from source network below:
below:
Power delivered to Load
P PA PLR(f)
= inc = = 1
(2.1a)
PA 1− Γ1 (ω )  1− Γ1 (ω )
PLoad 2 2 High
  attenuation

Low
1 attenuation
Low-Pass filter PLR
0 f
fc
18
PLR and s21
In terms of incident and reflected waves, assuming ZL=Zs = ZC.
b1
a1 b2

Zc
Lossless
Vs 2-port network Zc
PA PL
Pin

1a 2 2
PA 2 1 a1
PLR = = =
PL 1b 2 b2
2 2

PLR = 1 (2.1b)
2
s21

19
FILTER RESPONSES

Filter Responses

Several types filter responses:


- Maximally flat (Butterworth)
- Equal Ripple (Chebyshev)
- Elliptic Function
- Linear Phase

20
THE INSERTION LOSS METHOD

Practical filter response:


Maximally flat:
- also called the binomial or Butterworth response,
- is optimum in the sense that it provides the flattest possible
passband response for a given filter complexity.
- no ripple is permitted in its attenuation profile
N
ω 
PLR = 1 + k  
2
[4.1]
 ωc 
ω – frequency of filter
ωc – cutoff frequency of filter
N – order of filter

21
THE INSERTION LOSS METHOD

Equal ripple
- also known as Chebyshev.
- sharper cutoff
- the passband response will have ripples of amplitude 1 +k2

ω
PLR = 1 + k T  2 2
N 4.2]

 ωc 
ω – frequency of filter
ωc – cutoff frequency of filter
N – order of filter

22
THE INSERTION LOSS METHOD

Figure 4.2: Maximally flat and equal-ripple low pass filter response.

23
THE INSERTION LOSS METHOD

Elliptic function:
- have equal ripple responses in the passband and
stopband.
- maximum attenuation in the passband.
- minimum attenuation in the stopband.

Linear phase:
- linear phase characteristic in the passband
- to avoid signal distortion
- maximally flat function for the group delay.

24
THE INSERTION LOSS METHOD

Figure 4.3: Elliptic function low-pass filter response


25
THE INSERTION LOSS METHOD

Low-pass
Filter
Prototype
Specification
Design

Scaling &
Conversion
Normally done using
simulators

Optimization Filter
& Tuning Implementation

Figure 4.4: The process of the filter design by the insertion


loss method.
26
THE INSERTION LOSS METHOD

Low Pass Filter Prototype

Figure 4.5: Low pass filter prototype, N = 2

27
THE INSERTION LOSS METHOD

Low Pass Filter Prototype – Ladder Circuit

Figure 4.6: Ladder circuit for low pass filter prototypes and their
element definitions. (a) begin with shunt element. (b) begin with 28
series element.
THE INSERTION LOSS METHOD

g0 = generator resistance, generator conductance.


gk = inductance for series inductors, capacitance for shunt
capacitors.
(k=1 to N)

gN+1 = load resistance if gN is a shunt capacitor, load


conductance if gN is a series inductor.

As a matter of practical design procedure, it will be


necessary to determine the size, or order of the filter. This is
usually dictated by a specification on the insertion loss at
some frequency in the stopband of the filter.

29
THE INSERTION LOSS METHOD
Low Pass Filter Prototype – Maximally Flat

Figure 4.7: Attenuation versus normalized frequency for maximally flat


filter prototypes.
30
THE INSERTION LOSS METHOD

Figure 4.8: Element values for maximally flat LPF prototypes

31
THE INSERTION LOSS METHOD
Low Pass Filter Prototype – Equal Ripple
For an equal ripple low pass filter with a cutoff frequency ωc =
1, The power loss ratio is:
PLR = 1 + k T (ω )
2 2
N
[4.3]

Where 1 + k2 is the ripple level in the passband. Since the


Chebyshev polynomials have the property that
0
TN (ω ) = 
1
[4.3] shows that the filter will have a unity power loss ratio at ω
= 0 for N odd, but the power loss ratio of 1 + k2 at ω = 0 for N
even : two cases to consider depending on N

32
THE INSERTION LOSS METHOD

Figure 4.9: Attenuation versus normalized frequency for equal-ripple filter


33
prototypes. (0.5 dB ripple level)
THE INSERTION LOSS METHOD

Figure 4.10: Element values for equal ripple LPF prototypes (0.5 dB ripple
level)
34
THE INSERTION LOSS METHOD

Figure 4.11: Attenuation versus normalized frequency for equal-ripple filter


prototypes (3.0 dB ripple level)
35
THE INSERTION LOSS METHOD

Figure 4.12: Element values for equal ripple LPF prototypes (3.0 dB ripple
level).
36
EXAMPLE 4.1

Design a maximally flat low pass filter with a cutoff


freq of 2 GHz, impedance of 50 Ω, and at least 15 dB
insertion loss at 3 GHz. Compute and compare with an
equal-ripple (3.0 dB ripple) having the same order.

37
FILTER TRANSFORMATIONS

 Impedance scaling: In the prototype design, the


source and load resistance are unity (except for equal
ripple filters with even N, which have non unity load
resistance).

38
FILTER TRANSFORMATIONS

Low Pass Filter Prototype – Impedance Scaling

'
L = R0L [4.4a]

C '
C = [4.4b]
R0
'
R = R0 s
[4.4c]

'
R L = R0RL [4.4d]

39
FILTER TRANSFORMATIONS

Low Pass Filter Prototype – Frequency Scaling

Frequency scaling: To change the cut-off


frequency of a LP prototype from unity to ωc
requires to scale the frequency dependence of
the filter by the factor 1/ ωc which is
accomplished by replacing ω by ω/ωc

Frequency scaling for the low pass filter: ω


ω ←
ωc

40
FILTER TRANSFORMATIONS

The new Power Loss Ratio, P’LR

P’LR (ω) = PLR (ω/ωc) [4.5]

Cut off occurs when ω/ωc = 1 or ω = ωc

The new element values of the prototype filter:

ω
jX k = j L k = j ω L 'k [4.6]
ωc
ω
jB k = j C k = j ω C k' [4.7]
ωc

41
FILTER TRANSFORMATIONS

The new element values are given by:

' Lk R0 Lk
L =
k = [4.8a]

ω ωc
' CkCk
C =k = [4.8b]
ω R0ωc

42
FILTER TRANSFORMATIONS

Low pass to high pass transformation

The frequency substitution: ωc


ω←− [4.9]
ω
The new component values are given by:

' 1
C =k
[4.10a]
R 0ω c L k
' R0 [4.10b]
L =
k
ω cC k

43
BANDPASS & BANDSTOP TRANSFORMATIONS

Low pass to Bandpass transformation

 ω ω 0  1  ω ω0 
ω0
ω←  −  =  −  [4.11]
ω2 − ω1  ω0 ω  ∆  ω0 ω 
Centre freq.
ω2 − ω1 [4.12]
Where, ∆=
ω0 Edges of
passband
The center frequency is:

ω0 = ω1ω2 [4.13]

44
BANDPASS & BANDSTOP TRANSFORMATIONS

The series inductor, Lk, is transformed to a series LC circuit with


element values: L
L 'k = k
[4.14a]
∆ω0
' ∆
C k = [4.14b]
ω 0Lk
The shunt capacitor, Ck, is transformed to a shunt LC circuit with
element values: ∆
L 'k = [4.15a]
ω 0C k

' Ck
C k = [4.15b]
∆ω0

45
BANDPASS & BANDSTOP TRANSFORMATIONS

Low pass to Bandstop transformation


−1
 ω ω0 
ω ← −∆ −  [4.16]
 ω0 ω 
ω2 − ω1
Where, ∆=
ω0
The center frequency is:

ω0 = ω1ω2
46
BANDPASS & BANDSTOP TRANSFORMATIONS

The series inductor, Lk, is transformed to a parallel LC circuit with


element values: ∆L
L'k = k
[4.17a]
ω0
' 1
C =
k [4.17b]
ω0 ∆Lk
The shunt capacitor, Ck, is transformed to a series LC circuit with
element values: 1
L'k = [4.18a]
ω 0 ∆C k
' ∆C k
C = k [4.18b]
ω0

47
BANDPASS & BANDSTOP TRANSFORMATIONS

48
THE INSERTION LOSS METHOD

Low-pass
Filter
Prototype
Specification
Design

Scaling &
Conversion
Normally done using
simulators

Optimization Filter
& Tuning Implementation

49
SUMMARY OF STEPS IN FILTER
DESIGN

A. Filter Specification
1. Max Flat/Equal Ripple,
2. If equal ripple, how much pass band ripple allowed? If max
flat filter is to be designed, cont to next step
3. Low Pass/High Pass/Band Pass/Band Stop
4. Desired freq of operation
5. Pass band & stop band range
6. Max allowed attenuation (for Equal Ripple)

50
SUMMARY OF STEPS IN FILTER
DESIGN (cont)

B. Low Pass Prototype Design


1. Min Insertion Loss level, Number of Filter
Order/Elements by using IL values
2. Determine whether shunt capacitance model or
series inductance model to use
3. Draw the low-pass prototype ladder diagram
4. Determine elements’ values from Prototype Table

51
SUMMARY OF STEPS IN FILTER
DESIGN (cont)

C. Scaling and Conversion


1. Determine whether if any modification to the
prototype table is required (for high pass, band
pass and band stop)
2. Scale elements to obtain the real element values

52
SUMMARY OF STEPS IN FILTER
DESIGN (cont)

D. Filter Implementation
1. Put in the elements and values calculated from
the previous step
2. Implement the lumped element filter onto a
simulator to get the attenuation vs frequency
response

53
EXAMPLE 4.2

Design a band pass filter having a 0.5 dB


equal-ripple response, with N = 3. The center
frequency is 1 GHz, the bandwidth is 10%, and
the impedance is 50 Ω.

54
EXAMPLE 4.3

Design a five-section high pass lumped element


filter with 3 dB equal-ripple response, a cutoff
frequency of 1 GHz, and an impedance of 50 Ω.
What is the resulting attenuation at 0.6 GHz?

55
Filter Realization Using Distributed Circuit
Elements (1)
 Lumped-element filter realization using surface
mounted inductors and capacitors generally works
well at lower frequency (at UHF, say < 3 GHz).
 At higher frequencies, the practical inductors and
capacitors loses their intrinsic characteristics.
 A limited range of component values : available from
manufacturer – difficult design at microwave freq.
 Therefore for microwave frequencies (> 3 GHz),
passive filter is usually realized using distributed
circuit elements such as transmission line sections.

56
Cont’d…
 Distributed elements : open cct TL stubs or short cct TL stubs.
 At microwave freq, the distance between filter components is not
negligible.
 Richard’s transformation:
 Can be used to convert lumped elements to TL sections
 Kuroda’s identities:
 Can be used to physically separate the filter elements by using TL
sections.
 The four kuroda’s identities operations:

 Physically separate transmission line stubs

 Transform series stubs into shunt stubs or vice versa

 Change impractical characteristic impedances into more


realizable values.

57
Filter Realization Using Distributed Circuit
Elements (2)
 Recall in the study of Terminated Transmission Line Circuit that a
length of terminated Tline can be used to approximate an inductor and
capacitor.
 This concept forms the basis of transforming the LC passive filter into
distributed circuit elements.
l

≅ L ≅
Zc , β
l

Zc , β
≅ C ≅ Zo

Zo
Zo
Zo ≅

58
Filter Realization Using Distributed Circuit
Elements (3)
 This approach is only approximate. There will be deviation between the
actual LC filter response and those implemented with terminated Tline.
 Also the frequency response of distributed circuit filter is periodic.
 Other issues are shown below.
Connection physical
length cannot be
How do we implement series Tline ignored at
connection ? (only practical for microwave region,
Zo
certain Tline configuration) comparable to λ

Zo
Thus
Thussome
sometheorems
theoremsareareused
usedtoto
facilitate the transformation of LC
facilitate the transformation of LC
circuit
circuitinto
intostripline
striplinemicrowave
microwavecircuits.
circuits.
Chief
Chiefamong
amongthese
theseare
arethe
theKuroda’s
Kuroda’s
Identities
Identities(See
(SeeAppendix)
Appendix)

59
More on Approximating L and C with
Terminated Tline: Richard’s Transformation
l
Z in = jZ c tan (βl ) = jωL = jLω
Zin ≅ L
Zc , β tan (βl ) = ω
(3.1.1a)
Zc = L
l

Zin ≅ C Yin = jYc tan (βl ) = jωC = jCω


Zc , β
tan (βl ) = ω
(3.1.1b)
Yc = 1 = C
Zc Wavelength at
For LPP design, a further requirment is cut-off frequency
that:
⇒ tan  2π 
tan (βl ) = ωc = 1 λc (3.1.1c)
l  =1⇒ l =
 λc  8
60
More on Approximating L and C with
Terminated Tline: Richard’s Transformation

61
Kuroda’s Identities

 As taken from [2]. Note: The inductor represents


2 Z2 shorted Tline while the capacitor
n =1+
Z1 l
represents open-circuit Tline.
l

1
Z1 β Z2/n2 β Z1
Z2
n

l l

1
Z1 Z2 β n2Z1 β 2
n Z2

l l 1: n2
Z1
Z1 Z2 β Z2 /n2 β
n2

l l n2: 1
1 1
β n2Z1 2
Z2 Z1 β n Z2

62
Example – LPF Design Using Stripline

 Design a 3rd order Butterworth Low-Pass Filter. Rs = RL= 50Ohm, fc =


1.5GHz.
Step 1 & 2: LPP
g1 g3
1.000H 1.000H
Zo=1

g4
g2 1
2.000F

Length = λc/8
1 = 0.500 for all Tlines
2.000 at ω = 1 rad/s

63
Example – LPF Design Using Stripline

 Design a 3rd order Butterworth Low-Pass Filter. Rs = RL= 50Ohm, fc =


1.5GHz.
Step 3: Convert to Tlines using Richard’s Transformation Length = λc/8
for all Tlines
at ω = 1 rad/s

1 = 0.500
2.000

64
Example Cont…
Step 4: Add extra Tline on the series
connection Extra T-lines

Length = λc/8
for all Tlines
at ω = 1 rad/s

65
Example Cont…
Step 5: apply Kuroda’s Step 6: apply Kuroda’s
1st Identity. 2nd Identity.

Similar operation is
performed here

66
Example Cont…
After applying Kuroda’s Identity.

Length = λc/8
Since
Sinceall
allTlines
Tlineshave
havesimilar
similarphysical
physical
for all Tlines
length,
length,this
thisapproach
approachto
tostripline
striplinefilter
filter
at ω = 1 rad/s
implementation
implementationisisalso
alsoknown
knownas as
Commensurate
CommensurateLine LineApproach.
Approach.

67
Example Cont…
Step 5: Impedance and frequency denormalization.

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

Zc/Ω λ/8 @ 1.5GHz /mm W /mm Length = λc/8


50 13.45 2.85 for all Tlines at
25 12.77 8.00 f = fc = 1.5GHz
100 14.23 0.61
68
Example Cont…

Step 6: The layout (top view)

69
Example 2

Design a low pass filter for fabrication using microstrip


lines. The specifications are: cutoff freq of 4 GHz, third
order, impedance of 50 ohms and a 3dB equal ripple
characteristics

g1 = 3.3487 = L1
g2 = 0.7117 = C2
g3 = 3.3487 = L3
g4 = 1.0000 = RL

70
Example 2: Richard’s Transformation

71
Example 2: Cont’d…

Z1 =3.3487 Z1 =3.3487
UE =1 (Z2 =1) UE =1 (Z2 =1)

72
Cont’d…

UE UE

73
Example 5.8 (cont) Kuroda’s Identity

74
Example 5.8 (cont)

75
Kuroda Identities

76

You might also like