UNIT-V

IIR DIGITAL FILTERS

Contents
Introduction

Filtering of signals
Classification of filters

Analog and digital

Based on frequency response

Practical analog filter specifications

LPF. HPF, BPF and BSF.

Analog filters approximation

Butterworth and Chebyshev

Finding Order and normalized Stable filter
Design examples

Analog to analog transformations

Design of IIR Digital filters from analog filters
Impulse invariance, step invariance and bilinear
transformations
Design examples

Analog to digital transformations

Jan 6, 2016

Introduction

Filtering of signals

What is meant by a filter!

The DTFT is remembered again:

X (e )
jw

x[n]e

jwn

1
x[n]
2

jw
jwn
X
(
e
)
e
dw

x[n] is expressed as a summation of sinusoids with

scaled amplitude.
Using a system with a frequency selective to these
inputs, then it is possible to pass some frequencies
and attenuate the others.
Such a system is called a Filter.
Jan 6, 2016

What is meant by a filter!

The function of a filter is to remove
unwanted parts of the signal, such
as random noise, or
to extract useful parts of the signal,
such as the components lying within
a certain frequency range.
Unfiltered signal
or raw signal

Jan 6, 2016

Filtered signal

Example.1
Choose frequency response of a system such that

1
H (e )
0
jw

If

for c
for c

Almost = 0

x[n] A cos(1n) B cos(2 n) for 0 1 c 2

y[n] A H (e j ) cos 1n (1 ) B H (e j ) cos 2 n (2 )
1

y[n] A H (e ) cos 1n (1 )
j1

Which indicating the LPF effect of the LTI system

Jan 6, 2016

Adawy

Example.2
h[n]=[ a b a]

Design a HP digital filter that passes

the 0.4rad/sec, and stops the 0.1
rad/sec frequency.

H (e ) h[0] h[1]e
j

h[2]e
a be ae
j
j
e

j
j 2
j
j
a(1 e ) be 2a
e

be

j
(2a cos b)e
j

H (e j ) 2a cos b

H (e j 0.1 ) 2a cos(0.1) b 0

j 2

j 2

()= -

H (e j 0.4 ) 2a cos(0.4) b 1

Solving for the two equations gives

a= -6.76195,

y[n]=h[n]*x[n]
b=13.456335
y[n]=h[0]x[n]+h[1]x[n-1]+h[2]x[n-2]=ax[n]+bx[n-1]+ax[n-2]
Jan 6, 2016

y[n]= -6.76195 ( x[n]+x[n-2] )+13.456335 x[n-1]

If x[n]={cos(0.1n)+cos(0.4n)}u(n)
x1+x2

Transient

Output is
almost
equal to x2,
the high
frequency

Output of
the filter
x1

Jan 6, 2016

Classification of filters

Analog and digital

Classification of filters as
analog or digital
Analog filters

Digital filters

An analog filter processes

analog inputs and generates
. analog outputs

A digital filter processes

. and generates digital data

Analog filters are constructed

from passive or active
electronic components such as
resistors, capacitors and
opamps to produce the required
. filtering effect

A digital filter consists of

elements like adder,
multiplier and delay
element

An Analog filter is described by

.a differential equation

Digital filter is described by

.difference equation

Jan 6, 2016

10

Classification of filters as
analog or digital
Analog filters

Digital filters

The frequency response of an

analog filter can be modified by
.changing the components

The frequency response of

digital filter can be changed by
changing the filter coefficients

Such filter circuits are widely

used in such applications as
noise reduction, video signal
enhancement, graphic
equalizers in hi-fi systems, and
.many other areas

A digital filter uses a digital

processor to perform numerical
calculations on sampled values
. of the signal

Jan 6, 2016

The processor may be a

general-purpose computer
such as a PC, or a specialized
DSP (Digital Signal Processor)
.chip
11

Jan 6, 2016

12

Classification of filters

Based on Frequency
response

Classification of filters According to

frequency response

A analog filter is a network used to shape the frequency

spectrum of an electrical signal.
These networks are essential parts of communication and
control systems.
Filters are classified as low pass, high pass, band pass and
band reject, amplitude equalizers and delay equalizers.
H(ej)

H(ej)

LPF

HPF

H(e )

H(ej)

BPF

0
Jan 6, 2016

c1 c2

c1 c2

BSF

14

Practical analog filter

specifications

Practical Analog Filter specifications

Lowpass
filter

Highpass
filter
p

Bandpass
filter

Jan 6, 2016

Bandstop
filter
2

U
16

Practical analog Low pass filter

specifications

Jan 6, 2016

17

Practical analog low pass filter

specifications
0

Magnitude (dB)

Amax=p

Amin=s

Passband

0
Jan 6, 2016

Transition
band
p
frequency

Stopband

18

Practical analog Low pass filter

specifications
The basic function a of LOW PASS filter is to pass LOW
frequencies with very little loss and to attenuate high
frequencies.
It is required to pass signals from DC up to pass band
edge frequency p with at most Amax(p)dB of
attenuation.
The frequencies above stop band edge frequency s
are required to have atleast A min(s)dB of attenuation.
The band of frequencies from 0 to p is called the pass
band.
The band of frequencies from s to infinity is called the
stop band.
The frequency band from p to s is referred to as
transition band.
Jan 6, 2016

19

Practical analog high pass filter specifications

The basic function a of HIGH PASS filter is to pass
HIGH frequencies with very little loss and to
attenuate low frequencies.
It is required to pass signals from pass band edge
frequency p up to infinity with at most Amax(p) dB
of attenuation.
The frequencies below stop band edge frequency s
are required to have atleast Amin(s)dB of
attenuation.
The band of frequencies from p to infinity is called
the pass band.
The band of frequencies from zero to s is called the
stop band.
The frequency band from s to p is referred to as
transition band.
Jan 6, 2016

20

Practical analog band pass filter

specifications

Magnitude (dB)

Amax=p
Amin=s

Amin=s
Passband

Stopband

Jan 6, 2016

Stopband

21

Practical analog Band pass filter specifications

The basic function a of BAND PASS filter is to pass MIDDLE
frequencies with very little loss and to attenuate low and high
frequencies.
It is required to pass signals from lower pass band edge
frequency L to upper pass band edge frequency u with at
most Amax(p) dB of attenuation.
The frequencies below lower stop band edge frequency 1 and
above upper stop band edge frequency 2 are required to
have atleast Amin(s)dB of attenuation.
The frequency band from L to U is called the pass band.
The band of frequencies from 0 to 1 and 2 to infinity are
called the stop bands.
The band of frequencies from 1 to L and U to 2 are
referred to as transition bands.

Jan 6, 2016

22

Practical analog Band stop filter specifications

The basic function a of BAND STOP filter is to attenuate
MIDDLE frequencies and to pass low and high frequencies with
very little loss .
It is required to attenuate signals from lower stop band edge
frequency 1 to upper stop band edge frequency 2 with at
least Amin(s)dB of attenuation.
The frequencies below lower pass band edge frequency L and
above upper pass band edge frequency U are required to
have at most Amax(p) dB of attenuation.
The frequency band from 1 to 2 is called the stop band.
The band of frequencies from 0 to L and U to infinity are
called the pass bands.
The band of frequencies from L to 1 and 2 to U are
referred to as transition bands.

Jan 6, 2016

23

Design of digital filters from analog filters

The most common techniques used for designing
IIR digital filters known as indirect method, involves
first designing an analog prototype filter and then
transforming the prototype to a digital filter.
For the given specifications of a digital filter, the
derivation of the digital filter transfer function
requires three steps
1. Map the desired digital filter transfer function into
equivalent analog filter.
2. Derive the analog transfer function for the analog
prototype.
3. Transform the transfer function from the analog prototype
into an equivalent digital transfer function.

Jan 6, 2016

24

Advantages of digital filters

1. Unlike analog filters, the digital filters performance
is not influenced by component aging, temperature
and power supply variations.
2. A digital filter is highly immune to noise and
posses considerable parameter stability.
3. Digital filters afford a wide variety of shapes for the
amplitude and phase responses.
4. There are no problems of input or output
impedance matching with digital filters.
5. Digital filters can be operated over a wide range of
frequencies.

Jan 6, 2016

25

Advantages of digital filters

6. The coefficients of digital filter can be programmed
and altered any time to obtain the desired
characteristics.
7. Multiple filtering is possible only in digital filters.

Disadvantage of digital filters

1. The quantization error arises due to finite word
length in the representation of signals and
parameters.

Jan 6, 2016

26

Analog filter
approximations

Analog filter approximations

The rational function low pass approximations which we describe
in this have the general form.

Magnitude function

V
H ( S ) H ( j) OUT
VIN
2

1
1 K ( j )

1
1

N ( j ) 2
D ( j )

(1)

Where H(S) is the desired magnitude function and K(S) is the

rational function in S.
The function K(S) is chosen such that
its magnitude is small in pass band to make the magnitude
of H(S) close to UNITY.
Its magnitude is large in the stop band to make the
magnitude of H(S) close to ZERO.
In particular K(S) may be chosen to be a polynomial of the
form
K ( S ) PN ( S ) a0 a1S a2 S 2 ... a N S N (2)
Where the coefficients of the nth order polynomial Pn(S) are
chosen so that the corresponding loss function satisfies the
given filter requirements.
Jan 6, 2016

28

Low pass filter approximation

In particular K(S) may be chosen to be a polynomial of the form

In the pass band

i.e as 0, K ( j) 0

VIN 2
VOUT

1 K ( j) 2 1 VOUT VIN

As expected in the pass band of Low pass filter (near to DC) no

loss of signal the signal. But practically there will be some loss.
In the stop band

i.e as , K ( j)

VIN 2
VOUT

1 K ( j) 2 VOUT 0

As expected in the stop band of Low pass filter (high frequencies)

no pass of signal the signal. But practically there will be some
Pass of signal

Jan 6, 2016

29

Low pass filter approximation

There are four types of polynomials which satisfy these conditions.

They are
1. Butterworth Polynomial
(Maximally flat approximation)
1. Chebyshev polynomial
2. Elliptic polynomials
3. Bessel Polynomials
We are going to study

Butterworth filter approximation and

Chebyshev filter approximation

Jan 6, 2016

30

IIR Filter Types

Butterworth

Chebyshev Type I

Elliptic
Jan 6, 2016

Chebyshev Type II

Bessel

31

Butterworth filter Approximation

(Maximally flat approximation)

K ( S ) PN ( S )

S N
SP

N
P

or

N
C

(4)

The corresponding magnitude function is

H ( j )

Where

N
C

Jan 6, 2016

N
P

VOUT
1

VIN
1 C

2N

for N 1,2,3,... (5)

N is the order of the filter

is a constant
is the operating frequency and
C is the cutoff frequency
P is the pass band edge frequency

1/ N
P

or

P 1 / N C ( 6)
32

Butterworth filter Approximation

(Maximally flat approximation)

At DC means near =0
So

1
2

2N
P

1
2

1 12 2

2N
P

2N
P

18 4

4N
P

161 6

6N
P

...

This expression shows that the first 2N-1 derivatives are zero at
=0.
Since K(S) was chosen to be an nth order polynomial this
is the maximum number of derivatives that can be made zero.
Thus the slope is as flat as possible at DC.
For this reason the butter worth approximation is also known as
the Maximally flat Approximation.
Jan 6, 2016

33

Butterworth filter Approximation

(Maximally flat approximation)

The loss in dB is given from equ i.e

as

A() 10 log10 1

V
H ( j) OUT
VIN
2

dB (8)

At pass band edge frequency

1
1

2N
P

i.e at P

2N
P

(7)

the loss is Amax ( P )

10
1 (9)
P A( P ) 10 log10 1 2
It is the parameter related to pass band
At high frequencies the loss asymptotically approaches 20 log10 N
0.1 P

because

10 log10 1

10 log
2N
P

10

2
N
P

20 log10

N
P

These loss is seen to increase with the order N. At high frequencies

the slope is 6N dB/Octave. Therefore the stop band loss increases
with the order N.
34
Jan 6, 2016

Butterworth filter Approximation

(Maximally flat approximation)

At stop band edge frequency

S A( S ) 10 log10
10

0.1 S

2 S 2 N
P

S 2 N
P

i.e at S

2 S 2 N
P

10 0.1 S 1
0.1 P

10
1

the loss is Amin ( S )

S N
P

10 0.1 S 1

100.1 P 1

100.1 1
log
log10

10
0.1
1

10

1
log10
log10
S

Where

log

S
10 P

N log

10

100.1
100.1

100.1 1 (10)
S

is a Parameter related to stop band

Since this expression normally does not result in an integer value

we therefore round off N to the next higher integer to satisfy the
minimum required specifications.
N
Jan 6, 2016

log10

10
P

log

(11)
35

Stable normalized Butterworth

filter design
The magnitude function of the Butterworth low pass filter is given by
H ( j )

2N

1
2

(11)

The magnitude squared function of a normalized Butterworth low pass

filter with c=1 rad/sec is given by
1
2
H ( j )
(12)
2N
1
The function is monotonically decreasing. The maximum response is
zero at =0. The response approaches ideal characteristics as the
order N increases.
Jan 6, 2016

36

Stable normalized Butterworth

filter design
At =c the curve
passes through 0.707
which corresponds to
3dB point.

Now let us derive the transfer function of a stable filter.

For this purpose substitute =S/j in equ.12 then we have
1

H ( j ) H ( j ) H ( j )

Jan 6, 2016

H ( S ) H ( S )

S 2N
j

(13)

1
(14)
1 (1) N S 2 N
37

Stable normalized Butterworth

filter design
1.2

1
N=100
N=2

0.8
N=1

N=200

0.6

0.4

0.2

Jan 6, 2016

10

20

30

40

50

60

70

80

90

100

38

Stable normalized Butterworth

filter design
The above relation tells us that this function has poles in the LHP as well as in
the RHP because of the presence of two factors H(S) and H(-S).
If H(S) has poles in the LHP then H(-S) has the corresponding poles in the
RHP.
These roots we can get by equating the denominator to zero i.e

1 ( S 2 ) N 0 (15)
For N odd it reduces to

S 2 N 1 e j 2k S k e

For N even it reduces to S

2N

1 e

j ( 2 k 1)

j Nk

(16)

Sk e

j ( 2 kN1)

(17)

The solution of the above equation is

Sk e
Jan 6, 2016

j 2

2 k NN 1

for

k 1,2,...,2 N (18)
39

Stable normalized Butterworth

filter design
These 2N roots are located on the unit circle and are equally spaced at
/N radians intervals
The S domain magnitude function is therefore given by
H (S )

1
(19)
(
S

S
)

Im

Z-Plane

Where Sj are the left half plane poles.

Poles of Butterworth filter are located

on the circle in the S-plane and are
equally spaced at /N radians intervals

Jan 6, 2016

r
Re

40

Stable normalized Butterworth filter design

Example.3

Find the Butterworth approximation

function for the 3rd order Normalized low Pass Filter

Solution: N=3 for third order and k=1 to 2N = 1 to 6.

j 2 2 k NN 1
j 2 2 k3 2
j k31
Sk e
e
e
S1 e
S2 e

S3 e

23

33

cos 23 j sin 23 0.5 j 0.866

e j cos j sin 1
j 43
4
4

cos

j sin 3 0.5 j 0.866

Im

Z-Plane

The S domain magnitude function is therefore given by

H (S )

1
1

( S S1 )( S S 2 )( S S3 ) S (0.5 j 0.866) S (1) S (0.5 j 0.866)

H (S )

1
1
2
2
( S 0.5) ( j 0.866) S 1 ( S S 0.25 0.75)( S 1)

H (S )
Jan 6, 2016

Re

1
( S 1)( S 2 S 1)
41

Stable normalized Butterworth filter design

Example.4

Find the Butterworth approximation function for the

4th order Normalized low Pass Filter

Solution: N=4 for fourth order and k=1 to 2N = 1 to 8.

j 2 2 k NN 1
j 2 2 k43
Sk e
e
S1 e

78

S2 e

S3 e

S4 e

58

98

cos 58 j sin 58 0.3827 j 0.9239

cos 78 j sin 78 0.9239 j 0.3827

Z-Plane

cos 98 j sin 98 0.9239 j 0.3827

j 118
11
11

cos

j sin 8 0.3827 j 0.9239

The S domain magnitude function is therefore given by

1
H (S )
( S S1 )( S S 2 )( S S3 )( S S 4 )
H (S )

S (0.3827 j 0.9239) S (0.9239 j 0.3827) S (0.9239 j 0.3827) S (0.3827 j 0.9239)

H (S )

Im

1
( S 0.3827) ( j 0.9239) ( S 0.9239) 2 ( j 0.3827) 2

H (S )

r
Re

1
( S 2 0.76536 S 1)( S 2 1.84776S 1)

Jan 6, 2016

42

List of Butterworth polynomials

N

Denominator of H(S)

(S+1)

(S2+2S+1)

(S+1) (S2+S+1)

(S2+0.76537S+1) (S2+1.84776S+1)

(S+1) (S2+0.61803S+1) (S2+1.61803S+1)

(S2+1.931855S+1)(S2+2S+1) (S2+0.51764S+1)

(S+1) (S2+1.80194S+1)(S2+1.247S+1) (S2+0.445S+1)

Jan 6, 2016

43

Chebyshev filter Approximation

The main feature of the Butterworth approximation is that the loss
is maximally flat at the origin.
Thus the approximation to a flat pass band is very good at the
origin but it gets progressively poorer as frequency approaches
pass band edge.
Moreover the attenuation provided in the stop band is less than
that attainable using some other polynomial types, such as
Chebyshev polynomial.
There are two types of Chebyshev filters .i.e. Type-I and Type-II
Type-I are all-pole filters that exhibits equiripple behavior in the
Pass band and a monotonic characteristics in the stop band.
Type-II contains both poles and zeros and exhibits a monotonic
behavior in the pass band and an equiripple behavior in the
stop band.
Jan 6, 2016

44

Chebyshev filter Approximation

TYPE-I

Jan 6, 2016

45

Chebyshev filter Approximation

The magnitude square response of Nth order Type-I filter can be
expressed as
2

VOUT
1
H ( j )

VIN
1 2C N2
2

for N 1,2,3,... (1)

Where
and CN

is a filter parameter related to the ripple in the pass band.

is the Nth order Chebyshev polynomial defined as
cos( N cos 1 )

C N ()

1
cosh(
N
cosh
)

passband

stopband

(2)

It can be expressed by recursive formula from

C N 1 () CN 1 () cos ( N 1) cos 1 cos ( N 1) cos 1

2 cos( N cos 1 ) cos(cos 1 )
2 cos( N cos 1 ) 2C N ()

as
Jan 6, 2016

C N 1 () 2C N () CN 1 () (3)

46

Chebyshev filter Approximation

We know that

C0 ( ) 1

and

C1 ()

Then from

C N 1 () 2C N () C N 1 ()
C2 () 2C1 () C0 () 2 2 1
C3 () 2C2 () C1 () 2 2 2 1 43 3
C4 () 2C3 () C2 () 2 43 3 2 2 1 8 4 8 2 1

C5 () 2C4 () C3 () 2 8 4 8 2 1 43 3 165 20 4 5

Jan 6, 2016

47

Chebyshev filter Approximation

Jan 6, 2016

48

Chebyshev filter Approximation

Chebyshev Polynomial has the following properties
1.

C N () for N odd
C N ( )
C N () for N even

2.

0 for N odd
C N ( 0)
(1) for N even

3.

CN (1) 1 for

all N

1 for N odd
4. CN (1)
1 for N even
5. C N () oscillates with equal ripple between 1 for 1.
6. For all N 0 C N () 1 for 0 1 and C N () 1 for 1
7. CN () is monotonically increasing for 1 for all N .
8. Every coefficient is an integer and the one associated with N
is 2 N 1.
Jan 6, 2016

49

Chebyshev filter Approximation

Chebyshev Type-I

Jan 6, 2016

50

Chebyshev filter Approximation

The loss in dB is given from equ.1 i.e
as

VOUT
1
H ( j )

VIN
1 2CN2
2

A() 10 log10 1 2CN2 dB (4)

At pass band edge frequency i.e at P the loss is Amax ( P )

2
2
2

A
(

10
log
1

C
1

10
log
1

P
P
10
N
10

At stop band edge frequency i.e at S

2
2

A
(

10
log
1

S
S
10
N

S
P

100.1 1 (5)
P

the loss is Amin ( S )

10 log 1
10

cosh 2 N cosh 1 ( )
S

100.1 1
100.1 1
2
1
1
1
cosh N cosh ( ) N cosh ( ) cosh
2

100.1 1
S

100.1 S 1
1
cosh
cosh

0.1 P
10
1
(6)
N
cosh 1 ( PS )
cosh 1 PS

Jan 6, 2016

Where 100.1 1 (7)

S

51

Chebyshev Approximation provides 6(N-1)dB more

attenuation than Butterworth for the same order
The attenuation in STOP BAND of Butterworth filter for
in dB is given by

A() 10 log10 1

2 S 2 N
P

10 log

2 S 2 N
P

10

20 log I

But for

S
P

A() 20 log10

S N
P

S
P

10 log C 20 log C II
2

10

N 1 S N
P

S
N P

So above equation-II reduces to

A() 20 log10 C N

S N
P

10

The attenuation in STOP BAND of Chebyshev filter for

in dB is given by

A() 10 log10 1 2C N2

20 log

10

2
N

S
P

10

S
P

III

N 1 S N
P

20 log

20( N 1) log10 2 20 log10

10

S N
P

S
P

20 log10 2 N 1

6( N 1) IV

Comparing above equ.I and IV it is seen that the Chebyshev

approximation provides 20 log10 2 N 1 6( N 1)dB more attenuation
than a Butterworth of the same order.
Jan 6, 2016

52

Stable normalized Chebyshev filter

design
To find the poles of the Chebyshev approximation transfer function
Take the denominator of the equ(1) substitute P 1 to get
normalized function and equate it to zero. .i.e.
2

H ( j ) H ( j ) H ( j )

1
1 2C N2 ()

j
2 2
1 C N () 0 C N ()

It can be proved that the roots of above equation are

S k k j k for k 1,2,...,2 N

Where

k sin 2 2 kN1 sinh N1 sinh 1 ( 1 )

k cos 2 2 kN1 cosh N1 sinh 1 ( 1 )

further
Jan 6, 2016

k
k

1 1
1 1
1
1

sinh
sinh
(
)
cosh
sinh
( )
N

N

1
53

Stable normalized Chebyshev filter

design
These 2N roots are located on the ellipse in the s-plane spaced at
/N radians intervals
The S domain magnitude function is therefore given by
H0
H (S )
(S S j )

cosh N1 sinh 1 ( 1 )

Im
Z-Plane

Where Sj are the left half plane poles and

H0 is the order dependent constant

R
e

It can be found from

H ( S ) S 0

Jan 6, 2016

1
1
1 2

for N odd
for N even

sinh N1 sinh 1 ( 1 )
54

Stable normalized Chebyshev filter design

Example.5

Find the Chebyshev approximation function order

for the filter requirements

p=200 rad/s, s=600 rad/s, p=0.5dB, s=20dB.

Solution: Given p=200 rad/s, s=600 rad/s, p=0.5dB, s=20dB.

0.1
2
10

1
10
1
1
cosh 1
cosh
0.1
0.05
cosh 1 (28.484)
10

1
10

1
N

2.293
1
1 600
1
cosh ( )
cosh ( 200 )
cosh (3)
S

So the required order is N=3, for third order k=1 to 2N = 1 to 6.

S k k j k for k 1,2,...,2 N

Where

k sin

k cos 2 2 kN1

Jan 6, 2016

sinh
cosh

2 k 1
N

1
N

1
N

sinh ( 1 )
1

sinh 1 ( 1 )

55

Stable normalized Chebyshev filter design

Example.6

obtain an analog Chebyshev filter transfer function

that satisfies the constraints
1
2

H ( j) 0.1 for 4

H ( j) 1 for 0 2

Solution: In general specifications are given as

1
1 2

H ( j) 1 for 0 p H ( j)
P 2
1
1 2
1
1 2

S 4

1
2

1
1 2

for s

1 2 1
2

0.1 1 2 100 99 9.95

0.1
10
1
cosh 1
1
1 9.95
0.1

1 2.269
cosh
cosh
10
1

cosh 1 ( )
cosh 1 ( ) cosh 1 ( 42 )
S

Order to be selected is N=3 and

Jan 6, 2016

k=1 to 2N = 1 to 6.
56

Stable normalized Chebyshev filter design

Example.6

obtain an analog Chebyshev filter transfer function

that satisfies the constraints

H ( j) 1 for 0 2

1
2

Solution cont.d:
Where

1
N

H ( j) 0.1 for 4

S k k j k for k 1,2,...,2 N

k sin 2 2 kN1 sinh N1 sinh 1 ( 1 )

k cos 2 2 kN1 cosh N1 sinh 1 ( 1 )

sinh 1 ( 1 ) 13 sinh 1 ( 11 ) 0.29379

A sinh N1 sinh 1 ( 1 ) sinh 0.29379 0.298

B cosh N1 sinh 1 ( 1 ) cosh 0.29379 1.043

Jan 6, 2016

57

Stable normalized Chebyshev filter design

Example.6 cont.d

k sin 2 2 k31 A

k cos 2 2 k31 B

1 sin 2 13 (0.298) 0.149

1 cos 2 13 (1.043) 0.903

2 sin 2 33 (0.298) 0.298

2 cos 2 33 (1.043) 0

3 sin 2 53 (0.298) 0.149

3 cos 2 53 (1.043) 0.903

Left half plane Poles are given by

S1 1 j1 0.149 j 0.903

S 2 2 j 2 0.298
S3 3 j3 0.149 j 0.903
Jan 6, 2016

58

Stable normalized Chebyshev filter design

Example.6 cont.d
Normalized Transfer function is
H (S )

H (S )
Using

H0
H0

( S S1 )( S S 2 )( S S3 ) S (0.149 j 0.903) S (0.298) S (0.149 j 0.903)

H0
H0

(S 0.149)2 ( j 0.903)2 S 0.298 (S 0.298)(S 2 0.298S 0.8388)

H ( S ) S 0 1 for N Odd

H 0 (0.298)(0.8388) 0.25 H ( S )

0.25
( S 0.298)( S 2 0.298S 0.8388)

Denormalized Transfer function with C P / 1/ N 2 / 1 2 is

H ( s) H ( S ) S

s
C

0.25
0.25

( S 0.298)( S 2 0.298S 0.8388) S s ( 2s 0.298)[( 2s ) 2 0.298( 2s ) 0.8388]

2

2
H (s)
( s 0.596)( s 2 0.596 s 3.354)
Jan 6, 2016

59

Design of Analog Butterworth LOW PASS

filter
1. From the given specifications find the order of
the filter N.
2. Round off Order N to the next higher integer.
3. Find the Normalized Transfer function H(S).
4. Calculate the value of cut off frequency c.
5. Find the De-normalized transfer function H(s) by
replacing S with s/c.

Jan 6, 2016

60

Design of Analog Chebyshev LOWPASS filter

1. From the given specifications find the order of
the filter N.
2. Round off Order N to the next higher integer.
3. Find the denominator of the Normalized Transfer
function H(S).
4. Calculate the value of cut off frequency c and
find numerator constant H0 depending on the value
1
for N odd

of N.

1
H ( S ) S 0
for N even
1 2
5. Find the De-normalized transfer function H(s) by
replacing S with s/c.
Jan 6, 2016

61

Frequency transformations of analog filters

The approximations described so far were directly applicable to lowpass filters.
These approximations can be adapted to high pass,
symmetrical band pass and symmetrical band reject filters from a
normalized low pass filter(c=1 rad/sec)

HP, BP or BR
requirements

Equivalent
LP
Requirements

TLP (S)

T HP (s)
TBP(s)
TBR(s)

Take the given filter requirements

Translate the given requirements to EQUIVALENT low pass
requirements.
Approximate the resulting low pass requirement using the
specified approximation method
Finally translate the low pass approximation function to the
desired HP, BP or BR approximation function
Jan 6, 2016

62

Frequency transformations of analog filters

Practical Analog Filter specifications

Low pass
filter

High pass
filter

Band pass
filter

Jan 6, 2016

Band stop
filter

U
63

Design of HP/BP/BR filters of Butterworth /

Chebyshev type Analog filter
1. Find the equivalent low pass requirements.
p

p 1 and
p
s
for
HPF

.i.e.
r
s
for BPF r u2 1l

for

BRF

u l
2 1

2. Find the normalized Low pass filter order and

Transfer function for the given approximation Type.
3. Find the required De-normalized Transfer function
by replacing S in H(S) with below transformations

H ( s) H (S ) S

for

HPF

for

BPF

H ( s) H (S ) S

for

BRF or BRF

H ( s) H (S ) S

Jan 6, 2016

C
s

s 2 u l
s ( u l )
s ( u l )
s 2 u l

64

Stable Analog filter design- Example.7

For the given specifications p=3dB, s=15dB, p=1000 rad/s
and s=500 rad/s design a butter worth approximated High pass
filter.

Solution:

p 3dB P 1000 rad / s

s 15dB S 500 rad / s

100.1 1 101.5 1 5.533

100.1 1 100.3 1 1

/ 5.533
Equivalent low pass requirements are

p 3dB s 15dB p 1 rad / s r 1000

500 2 rad / s
P
S

N
Jan 6, 2016

log10

log10

r
p

log10 5.533

2.468
log10 2
65

Stable analog filter design- Example.7: Cont.d

Solution: select N=3 and k=1 to 2N = 1 to 6.

j 2 2 k NN 1
j 2 2 k3 2
j k31
Sk e
e
e
S1 e

23

S2 e
S3 e

cos 23 j sin 23 0.5 j 0.866

33

43

e j cos j sin 1
cos 43 j sin 43 0.5 j 0.866

Normalized Transfer function of equivalent Low-pass filter is

1
1

( S S1 )( S S 2 )( S S3 ) S (0.5 j 0.866) S (1) S (0.5 j 0.866)

1
1
H (S )

( S 0.5) 2 ( j 0.866) 2 S 1 ( S 2 S 0.25 0.75)( S 1)

H (S )

H (S )
Jan 6, 2016

1
( S 1)( S 2 S 1)

66

Stable analog filter design- Example.7: Cont.d

Solution:

P 1/ N C C P / 1/ N 1000 / 1 1000

De-normalized Transfer function of required High-pass filter is

Given by

H (s) H ( S ) S

( S 1)( S 2 S 1) S 1000
s

H (s)

1
2

1000
1000 1000
1
1

s
s
s

De-normalized Transfer function of required High-pass filter is

s3
H (s)
s 103 s 2 103 s 106
Jan 6, 2016

67