Unit III

IIR Digital Filters

05-06-2021 1
Types of Filters
Based on frequency response filters can be classified as
• Low pass
• High pass
• Band pass
• Band stop/Band reject

05-06-2021 2
Types of Filters (Contd…)

05-06-2021 3
Comparison of Analog and Digital Filters
Analog Filter Digital Filter
1. Analog filter process analog inputs and 1. A digital filter process and generates
generates analog outputs digital data

2. Analog filters are constructed from 2. A Digital filter consists of elements like
active and passive electronic adder, multiplier and delay
components
3. Analog filter is described by a 3. Digital filter is described by a
differential equation difference equation
4. The frequency response of an analog 4. The frequency response can be
filter can be modified by changing the changed by changing the filter
components coefficients
05-06-2021 4
Advantages of Digital Filters
• Unlike analog filter, the digital filter performance is not influenced by
component ageing, temperature and power supply variations
• A digital filter is highly immune to noise and possesses considerable
parameter stability
• Digital filter afford a wide variety of shapes for the amplitude and
phase responses.
• There are no problems of input or output impedance matching with
digital filters
• Digital filters can be operated over a wide range of frequencies
• The coefficients of digital filter can be programmed and altered any
time to obtain the desired characteristics
• Multiple filtering is possible only in digital filters
05-06-2021 5
Disadvantage of Digital Filters
• The quantization error arises due to finite word length
in the representation of signals and parameters

05-06-2021 6
Design of Digital Filters
• Infinite Impulse Response (IIR) Filter
• Finite Impulse Response (FIR) Filter

05-06-2021 7
IIR Filters
• The impulse response h(n) for a realizable filter is
h( n)  0 for n  0
• For stability it must satisfy the condition


 h( n)  
n 0

• IIR digital filters have the transfer function of the form

M

 b z k
k

H ( z )   h( n) z  n  k 0
N
n 0
1   ak z  k
k 1

• Design of an IIR filter for the given specifications is to find the filter
coefficients ak and bk
05-06-2021 8
Analog Lowpass Filter
The most general form of analog filter transfer function is
M

N (s) i
b s i

H (s)   i 0
N
1   ai s i
D( s)
i 1

H(s) is the Laplace transform of the impulse response h(t), i.e.,


H (s)  

h(t )e  st dt

N  M must be satisfied.
For a stable analog filter, the poles of H(s) lie in the left half of the s-
plane
05-06-2021 9
Types of Analog Filters
• Butterworth
• Chebyshev

05-06-2021 10
 Comparison of Butterworth and Chebyshev Filter
Butterworth Filter Chebyshev Filter
For a particular desired specification of a digital filter the For a particular desired specification of a digital filter the
order of butterworth filter will be higher than chebyshev order of chebyshev filter will be lower as compared to
filter butterworth filter
For a particular specification butterworth filter requires For a particular specification chebyshev filter requires less
more hardware hardware
This filter has cutoff frequency not equal to passband This filter has cutoff frequency equal to passband
frequency 1
frequency Ωc = Ωp
c  p 1 A
2
p
2N 1

The transition band of butterworth filter is wider as The transition band of chebyshev filter is narrow as
compared to chebyshev filter compared to butterworth filter

All poles of filter will always lie on circle having radius r = All the poles of a filter will lie on ellipse having major axis
Ωc 'R', 'ζ', minor axis 'r'
Butterworth filter has no ripples either in passband or Chebyshev filter will have ripples either in stop band or
stopband passband
Butterworth filter does not have any type It has two type i.e Type 1 = Ripple in passband and Type 2
= Ripple in stopband
05-06-2021 11
Expression for Order of the Filter
The expression to determine the order of the analog Butterworth
lowpass filter is given by

 
0.1 p 0.5
log     10 1
N  

  10 
 s  0.1 s 0.5
log 
  
1
 p
If 
A and

p
k
s log  A 
Then N
1
log  
k
05-06-2021 12
Problems on Determination of Order
Q. Given the specifications
 p  1 dB ;  s  30 dB; p  200 rad/sec ; s  600 rad/sec
Determine the order of the filter
Sol:

 
0.1 p 0.5
  10 1 log  
 
N
 s 
  10 
0.1 s 0.5
1 log 
  
 p

05-06-2021 13
Expression of Cut off Frequency
The cutoff frequency expression for analog Butterworth lowpass filter is
given by
p s
c  
10  10 
1 1
0.1 p 0.1 s
1
2N
1 2N

05-06-2021 14
Steps to design an Analog Butterworth Low pass
Filter
1. From the given specifications find the order of the filter N
2. Round off it to the next higher integer
sk  e j k

3. Find the poles using the formula

k 


 2k  1 
, k  1, 2,..., N
2 2N
4. Find the transfer function H(s) for Ωc = 1 rad/sec for the
value of N
5. Calculate the value of cutoff frequency Ωc
6. Find the transfer function Ha(s) for the above value of Ωc
by substituting s s/Ωc in H(s)
05-06-2021 15
Problems on Analog Butterworth Filter Design
Q. For the given specifications design an analog Butterworth filter
0.9  H ( j)  1 for 0    0.2
H ( j)  0.2 for 0.4    
Sol:

05-06-2021 16
Problems on Analog Butterworth Filter Design

1.Order is given by log  
N  
 s 
log 
  
 p
2. Poles are given by
jk
sk  e

k 


 2k  1 
, k  1, 2,..., N
2 2N

3. Denominator of H(s) is (s-s1)(s-s2)(s-s3)(s-s4)

05-06-2021 17
Problems on Analog Butterworth Filter Design
1
H (s) 
( s  s1 )( s  s2 )( s  s3 )( s  s4 )
4. Cut-off frequency
p s
c  
10   
1 1
0.1 p
1
2N
100.1 s  1 2N

5. Replace s by s/Ωc in H(s)

05-06-2021 18
Analog Lowpass Chebyshev Filter
Type I
• These are all pole filters
• It exhibits equiripple behaviour in the passband
• It exhibits a monotonic characteristics in the stopband

Type II
• These filters contains both poles and zeros
• It exhibits a monotonic behaviour in the passband
• It exhibits equiripple behaviour in the stopband

05-06-2021 19
Type I Chebyshev Filter
 Expression for finding the order of the filter
0.1 s
1 10 1 1
cosh cosh A
10
0.1 p
1 or N
N 1  1 

1  s
 cosh  
cosh 
   k
 p
100.1 s  1  p 
where A 0.1 p
and k   
10 1  s 

05-06-2021 20
Type I Chebyshev Filter (Contd…)
 Expression for finding the poles of the filter
  1 N   1 N 
sk  a cos k  jb sin k a  p  
 2 
  1 N   1 N 
b  p  
 2 
  2k  1 
k   k  1, 2,..., N
2 2N
 
0.1 p 0.5
    1 
1 2
  10 1
05-06-2021 21
Steps to Design Analog Chebyshev Lowpass Filter
1. From the given specifications find the order of the filter N
2. Round off it to the next higher integer
3. Using the following formulas find the values of a and b, which are
the minor and major axis of the ellipse respectively
  1 N   1 N    1 N   1 N 
a  p   b  p  
 2   2 

 
0.1 p 0.5
    1 
1 2
  10 1
 p  Passband frequency
 p  Maximum allowable attenuation in the passband
For normalized Chebyshev filter  p  1 rad/sec
05-06-2021 22
Steps to Design Analog Chebyshev Lowpass
Filter (Contd…)
4. Calculate the poles of the Chebyshev filter which lie on an
ellipse by using the formula
sk  a cos k  jb sin k
  2k  1 
k   k  1, 2,..., N
2 2N
5. Find the denominator polynomial of the transfer function
using the above poles
6. The numerator of the transfer function depends on the
value of N
05-06-2021 23
Steps to Design Analog Chebyshev Lowpass
Filter (Contd…)
a) For N odd, substitute s = 0 in the denominator polynomial
and find the value, This value is equal to the numerator.
(For N odd the magnitude response H ( j) starts at 1)

b) For N even, substitute s = 0 in the denominator polynomial

and divide the result by 1   2 . This value is equal to the
numerator.

05-06-2021 24
Problems on Chebyshev Filter
Q. Obtain an analog Chebyshev filter transfer function that satisfies the
constraints 1
 H ( j)  1; 02
2
H ( j)  0.1; 4
Sol:
Step 1: From the given data
1 1 1
 and  0.1
1  2
2 1  2

 p  2, s  4

05-06-2021 25
Problems on Chebyshev Filter
  1,   9.95
Order of the filter is given by

cosh 1 cosh 1  9.95 
N    2.269

1  s
 cosh  2 
1

cosh 
  
 p
Step 2: Rounding off N to next higher integer we get N=3
Step 3:     1    2.414
1 2

  1 N   1 N    2.414  13   2.414 13 

a  p   =2   =0.596
 2   2 
 
  1 N   1 N    2.414  13   2.414 13 
b  p  2    2.087
 2   2 
 
05-06-2021 26
Problems on Chebyshev Filter
Step 4: To calculate poles of chebyshev
  2k  1 
k   k  1, 2,3
2 2N
  4 2
1    
2 6 6 3
 3  
2     
2 6 2 2
 5 8 4
3    
2 6 6 3
We know sk  a cos k  jb sin k
05-06-2021 27
Problems on Chebyshev Filter
 2   2 
s1  a cos 1  jb sin 1  0.596 cos    j (2.087) sin  
 3   3 
 (0.5960(0.5)  j (2.087)(0.866)  0.298  j1.807

s2  a cos 2  jb sin 2  0.596 cos   j (2.087) sin   0.596

 4   4 
s3  a cos 3  jb sin 3  0.596 cos    j (2.087) sin  
 3   3 
 (0.5960(0.5)  j (2.087)(0.866)  0.298  j1.807

05-06-2021 28
Problems on Chebyshev Filter
Step 5: The denominator polynomial is given by
 s  0.298  j1.807  s  0.596  s  0.298  j1.807 
  s  0.596   s  0.298   1.807  
2 2

  s  0.596   s 2  0.596 s  0.089  3.265 

  s  0.596   s 2  0.596 s  3.354 
Step 6: Since N is odd, the numerator of H(s) can be obtained by
substituting s=0 in the denominator
Numerator of H(s) is
 0  0.596  0  0  3.354  2
05-06-2021 29
Problems on Chebyshev Filter
The chebyshev filter transfer function for the given specifications is
given by 2
H (s) 
 s  0.596   s 2  0.596s  3.354 

05-06-2021 30
Design of IIR Filter from Analog Filters
The conversion techniques must posses the following desirable
properties
The jΩ axis in the s-plane should map into the unit circle in
the z-plane. Thus there will be direct relationship between the
two frequency variables in the two domains

The left half plane of the s-plane should map into the inside
of the unit circle in the z-plane. Thus a stable analog filter
will be converted to a stable digital filter

05-06-2021 31
Methods of Converting Analog Filter into
Digital Filter
Approximation of Derivatives
Impulse invariant method
Bilinear transformation method
The matched z-transformation technique

05-06-2021 32
Impulse Invariant Technique
Impulse response of digital filter is the sampled version of the analog
filter
h(n)  h(nT ), n  0,1, 2,... (1)

H ( z )   h( n) z  n
n 0

 H ( z ) z esT 
 h ( n ) e  nsT
(2)
n 0
Mapping of points from the s-plane to the z-plane is given by

ze
sT
(3)
05-06-2021 33
Impulse Invariant Technique (Contd…)
Let s    j
z  re j
Therefore
j   j T
re  e  e T .e jT (4)
From (4) we get
r  e T (5)
  T (6)
The real part of analog pole determines the radius of the z-plane pole
The imaginary part of the analog pole dictates the angle of the digital
pole
05-06-2021 34
Impulse Invariant Technique (Contd…)
Case 1: σ = 0,
r e 0.T
1
Therefore impulse invariant mapping map poles from the s-plane’s jΩ
axis to the z-plane’s unit circle

05-06-2021 35
Impulse Invariant Technique (Contd…)
Case 2:   0,
0  r 1
So, the mapping is inside the unite circle

05-06-2021 36
Impulse Invariant Technique (Contd…)
Case 3:   0 ,
r 1
The poles are mapped outside the unit circle

05-06-2021 37
Impulse Invariant Technique (Contd…)
Let Ha(s) is the system function of an analog filter. This can be expressed
in partial fraction form as
N
ck
H a ( s)   (5)
k 1 s  pk

where {pk} = poles of the analog filter

{ck} = coefficients in the partial fraction expansion
The inverse Laplace transform of (5) is
N
ha (t )   ck e pk t t0 (6)
k 1
If we sample ha(t) periodically at t=nT, we have
N
h(n)  ha (nT )   ck e pk nT (7)
k 1
05-06-2021 38
Impulse Invariant Technique (Contd…)
We know 
H ( z )   h( n) z n

n 0
 

 
N N
  ck e  ck  e
n
pk nT n pk T 1
z z
n  0 k 1 k 1 n 0
N
ck
 pk T 1
k 1 1  e z
N
ck
Therefore if H a ( s)   then
k 1 s  pk
N
ck
H ( z)   pk T 1
(8)
k 1 1  e z
05-06-2021 39
Impulse Invariant Technique (Contd…)
For high sampling rates (for small T), the digital filter gain is high.
Therefore instead of (8) we can use
N
Tck
H ( z)   pk T 1
(9)
k 1 1  e z

05-06-2021 40
Steps to Design using Impulse Invariant Technique
1. For the given specifications, find the transfer function Ha(s) of an
analog filter
2. Select the sampling rate of the digital filter, T sec/sample
3. Express the analog filter transfer as the sum of single poles filters
N
ck
H a (s)  
k 1 s  pk

4. Compute H(z) using the formula

N
ck
H ( z)   pk T 1
k 1 1  e z
N
Tck
For high sampling rates use H ( z)   pk T 1
k 1 1  e z
05-06-2021 41
Problems on Impulse Invariant Method
Q 1. For the analog transfer function
2
H (s) 
 s  2  s  3
Determine H(z) using impulse invariance method. Assume T=1 sec.
Sol: Given 2 A B
H (s)   
 s  2  s  3 s2 s3
2 2
 H ( s )  
s2 s3
N
ck
We know if H ( s)  then H ( z)  
N
ck
a  s  pk 1  e pk T 1
z
k 1 k 1

05-06-2021 42
Problems on Impulse Invariant Method
Here c1 = 2, c2 = -2
p1 = -2, p2 = -3
H(z) can be written as
2 2
H ( z)  2T 1

1 e z 1  e 3T z 1
2 2
For T=1 sec, H ( z)  2 1

1 e z 1  e 3 z 1
2 2
 1

1  0.1353 z 1  0.0498 z 1
0.171z 1

1  0.1851z 1  0.0067 z 2
05-06-2021 43
Problems on Impulse Invariant Method (Contd…)
Q 2. For the analog transfer function
1
H (s) 
 s  1  s 2  s  1
Determine H(z) using impulse invariance method. Assume T=1 sec.
Sol: Given
1 A B C
H (s)    
 s  1  s  s  1 s  1 s  0.5  j 0.866 s  0.5  j 0.866
2

1 0.5  0.288 j 0.5  0.288 j

 H ( s )   
s  1 s  0.5  j 0.866 s  0.5  j 0.866
N
ck
H ( z)  
N
ck
H a ( s)   pk T 1
We know if k 1 s  pk then k 1 1  e z
05-06-2021 44
Problems on Impulse Invariant Method (Contd…)
Here c1 = 1, c2 = -0.5+0.288j, c3 = -0.5-0.288j
p1 = -1, p2 = -0.5-j0.866, p3 = -0.5+j0.866
H(z) can be written as
1 0.5  0.288 j 0.5  0.288 j
H ( z)  T 1
 ( 0.5 j 0.866)T 1

1 e z 1 e z 1  e( 0.5 j 0.866)T z 1
For T=1 sec,
1 0.5  0.288 j 0.5  0.288 j
H ( z)  1 1
 0.5  j 0.866 1

1 e z 1 e e z 1  e 0.5e j 0.866 z 1
1 1  0.66 z 1
 1

1  0.368 z 1  0.768 z 1  0.368 z 2

05-06-2021 45
Disadvantages of Impulse Invariant Technique
There are infinite number of s-plane poles that map to
the same location in the z-plane
The s-plane poles having imaginary parts greater than
π/T or less than - π/T cause aliasing
Impulse invariant method is suitable for the design of
lowpass and bandpass filters only.
It is not suitable for designing highpass filters

05-06-2021 46
Bilinear Transformation Technique
It is a conformal mapping
It transforms the jΩ axis into the unit circle only once
thus avoids the aliasing of frequency components
All LHP poles of ‘s’ are mapped inside the unit circle in
z-plane
All RHP poles of ‘s’ are mapped outside the unit circle

05-06-2021 47
Bilinear Transformation Technique
Let us consider an analog filter with system function
b
H ( s)  (1)
sa
Y (s) b
 
X (s) s  a
 sY ( s )  aY ( s )  bX ( s ) (2)
The system can be characterized by the differential equation
dy (t )
 ay (t )  bx(t ) (3)
dt
y(t) can be approximated by the trapezoidal formula
t
y (t )   y ' ( )d  y (t0 ) (4)
05-06-2021 t0 48
Bilinear Transformation Technique
where y’(t) denotes derivative of y(t)
The approximation of integral in (4) by the trapezoidal formula
at t=nT and t0=nT-T yields
T '
y (nT )   y (nT )  y ' (nT  T )  +y (nT  T ) (5)
2
Differential equation in (3) evaluated at t=nT yields
y (nT )  ay (nT )  bx(nT )
'
(6)
Substituting (6) in (5) we get,
T
y (nT )   ay (nT )  bx(nT )  ay (nT  T )  bx(nT  T )   y (nT  T ) (7)
2
05-06-2021 49
Bilinear Transformation Technique
aT aT bT bT
y (nT )  y (nT )  y (nT  T )  y (nT  T )  x(nT )  x(nT  T )
2 2 2 2
aT aT bT
(1  ) y (nT )  (1  ) y (nT  T )   x(nT  x(nT  T )  (8)
2 2 2
with y (n)  y (nT ) and x(n)  x(nT ) we get
aT aT bT
(1  ) y (n)  (1  ) y ( n  1)   x(n)  x(n  1)  (9)
2 2 2
z-transform of (9)
aT aT 1 bT
(1  )Y ( z )  (1  ) z Y ( z)   X ( z )  z 1 X ( z ) 
2 2 2
 aT  1  aT   bT
=>Y(z) 1    z  1   = 1  z 1
 X(z)
05-06-2021  2   2  2 50
Bilinear Transformation Technique
Y ( z)
bT
2
1  z 1


X ( z) aT  aT  1
1  1  z
2  2 
bT
2
1  z 1
 bT
2
 1  z 1 
H ( z)  
aT  aT  1 aT 1 aT 1
1  1  z 1   z  z
2  2  2 2
bT
2
1  z 1 
b
  (10)
1  z   2 1  z   1   a  1
1 aT 1 2 1  z
T  1 z 
05-06-2021 51
Bilinear Transformation Technique
Comparing (1) and (10) , the mapping from s-plane to the z-plane is
2  1 z  1
s  1 
(11)
T  1 z 

The relationship between s and z in (11) is known as

bilinear transformation

05-06-2021 52
Characteristics of Bilinear Transformation
Technique
Let z  re j (12)
s    j (13)
Eq. (11) can be expressed as
2  z  1 2  re  1 2  r cos   jr sin   1 
j

s   
T  z  1 T  re  1 T  r cos   jr sin   1 
j

2  r cos   1  jr sin   2  r cos   1  jr sin    r cos   1  jr sin  

   

T  r cos   1  jr sin   T  r cos   1  jr sin    r cos   1  jr sin  
2  r 2 cos 2   r 2 sin 2   1  j 2r sin  
  2
T  r cos 2   2r cos   1  r 2 sin 2  
2  r 2 (cos 2   sin 2  )  1  j 2r sin   2  r 2  1  j 2r sin  
  2  
2 2 
T  r (cos   sin  )  2r cos   1  T  1  r 2  2r cos  
05-06-2021 53
Characteristics of Bilinear Transformation
Technique
2 r 2 1 2r sin  
s  j  (14)
T 1  r  2r cos 
2
1  r  2r cos  
2

Comparing (13) and (14) we obtain

2 r 2 1
 (15)
T 1  r 2  2r cos 
2 2r sin 
 (16)
T 1  r 2  2r cos 

05-06-2021 54
Characteristics of Bilinear Transformation
Technique
Case 1: when r<1,
Then   0 ,
LHP of s maps inside the unit circle.

Case 2: when r>1,

Then   0 ,
RHP of s maps outside the unit circle

Case 3: when r=1,

Then   0
05-06-2021 55
Characteristics of Bilinear Transformation
Technique
and
2 2sin 

T 2  2 cos 
 
2sin cos
2 sin  2 2 2  2 tan 
    (17)
T 1  cos  T 2  T 2
2 cos
2
 T
tan 
2 2
1 T
  2 tan (18)
2
05-06-2021 56
The Warping Effect
We have
2 
  tan
T 2
For small value of 
2 
 .  For small  , tan = 
T 2
   T

05-06-2021 57
The Warping Effect (Contd…)

05-06-2021 58
The Warping Effect (Contd…)
For low frequencies the relationship between w and Ω is
linear
As a result both analog and digital filter has same frequency
response
For higher frequencies relationship between w and Ω is non-
linear
It introduces distortion in frequency scale of digital filter to
that of the analog filter.
This is known as warping effect.
05-06-2021 59
Steps to Design Digital filter using Bilinear
Transformation
1. From the given specifications, find the prewarping analog
frequencies using formula
2 
 tan
T 2
2. Using the analog frequencies find H(s) of the analog filter
3. Select the sampling rate of the digital filter, call it T seconds per
sample
2  1  z 1 
4. Substitute s    into the transfer function found in step
1
2 T  1 z 
05-06-2021 60
Problem on Bilinear Transformation
Q 1. Apply bilinear tansformation to
2
H (s) 
( s  1)( s  2)
With T=1 sec and find H(z)
Sol: Given
2
H (s) 
( s  1)( s  2)
Substitute 2  1  z 1  in H(s) to get H(z)
s  
T  1  z 1  2
H ( z) 
 2  1  z 1    2  1  z 1  
  1 
 1   1 
 2
T  1  z   T  1  z  
05-06-2021 61
Problem on Bilinear Transformation (Contd…)
For T=1, 2
H ( z) 
  1  z 1     1  z 1  
2  1 
 1  2  1 
 2
  1 z     1 z  
2 1  z 
1 2
1 z  1 2

 
3  z   4 6  2z 
1 1

0.166 1  z  1 2


1  0.33z  1

05-06-2021 62
Problem on Bilinear Transformation (Contd…)
Q2. Using the bilinear transformation, design a highpass filter,
monotonic in passband with cutoff frequency of 1000 Hz and down
10dB at 350 Hz. The sampling frequency is 5000 Hz.
Sol: Given
 p  3 dB;  p  c  2   1000  2000 rad/sec
 s  10 dB; s  2    350  700 rad/sec
1 1
T   2  104 sec
f 5000

The characteristics are monotonic in both passband and stopband.

Therefore, the filter is Butterworth filter.
05-06-2021 63
Problem on Bilinear Transformation (Contd…)
Prewarping the digital frequencies we have
2
 p  tan
 pT

2 
tan
2000  2  10 4

4
T 2 2 10 2
 104 tan  0.2   7265 rad/sec

2
 s  tan
 sT

2 
tan
700  2  10 4

4
T 2 2 10 2
 104 tan  0.07   2235 rad/sec
First design a lowpass filter and then use suitable transformation to
obtain transfer function of highpass filter.
05-06-2021 64
Problem on Bilinear Transformation (Contd…)
100.1 s  1 100.1(10)  1
log 0.1 p log
N 10  1  10 0.1(3)
 1  log 3  0.4771
 s   7265  log 3.25 0.5118
log  log  
    2235 
 p
 N  0.932
Therefor, N=1
Pole
s1  e j  1
1
Transfer function of first order analog Butterworth filter is H ( s) 
s 1

05-06-2021 65
Problem on Bilinear Transformation (Contd…)
The highpass filter for c   p  7265 rad/sec can be obtained by using
the transformation
c
s
s
7265
i.e., s 
s
The transfer function of highpass filter
 1  s
H (s)    
 s  1  s  7265 s  7265
s

05-06-2021 66
Problem on Bilinear Transformation (Contd…)
Using bilinear transformation
 s 
H ( z )   H ( s ) s  2  1 z 1  

T  1 z 1   s  7265  s  2  1 z 1 
 
2104  1 z 1 

 1  z 1 


10000 
 1 z 1 

0.5792 1  z 1

1
 1 z 
1
1  0.1584 z
10000  1 
 7265
 1 z 

05-06-2021 67
Frequency Transformation in Analog Domain
1. Lowpass to Lowpass:
p
s s
 '
p

 p  passband frequency of lowpass filter

  passband frequency of new lowpass filter
'
p

05-06-2021 68
Frequency Transformation in Analog Domain
2. Lowpass to highpass:

 p '

s
p

s
 p  passband frequency of lowpass filter
  passband frequency of new highpass filter
'
p

05-06-2021 69
Frequency Transformation in Analog Domain
3. Lowpass to bandpass:

s  l u
2
s  p
s  u  l 
u  upper cutoff frequency
l  lower cutoff frequency

05-06-2021 70
Frequency Transformation in Analog Domain
4. Lowpass to bandstop:

s  u  l 
s  p 2
s  l u
u  upper cutoff frequency
l  lower cutoff frequency

05-06-2021 71
Frequency Transformation in Digital Domain
1. Lowpass to Lowpass:
1 z 1
z 
1   z 1
sin  p   p'  2 
where  
sin  p   p  2 
 '

 p  passband frequency of lowpass filter

 'p  passband frequency of new lowpass filter

05-06-2021 72
Frequency Transformation in Digital Domain
2. Lowpass to highpass:
1 z 1
z 
1   z 1
cos  p   'p  2 
where  
cos  p   p  2 
 '

 p  passband frequency of lowpass filter

 'p  passband frequency of highpass filter

05-06-2021 73
Frequency Transformation in Digital Domain
3. Lowpass to bandpass:
 2 2 k 1 k  1 
 z  z  
 
z 1   
1 k k 1
k  1 2 2 k 1
z  z 1
k 1 1 k
cos u  l  2   u  l  p
where   and k  cot   tan
cos u  l  2   2  2
u  upper cutoff frequency
l  lower cutoff frequency
05-06-2021 74
Frequency Transformation in Digital Domain
4. Lowpass to bandstop:
 2 2 1 1  k 
z  z  
 
z 1   
1 k 1 k
1  k 2 2 1
z  z 1
1 k 1 k
cos u  l  2   u  l  p
where   and k  tan   tan
cos u  l  2   2  2
u  upper cutoff frequency
l  lower cutoff frequency
05-06-2021 75
Problems on Frequency Transformation
0.5 1  z 1 
Q. Convert the single pole lowpass filter with system function H ( z )  1  0.302 z 1

into bandpass filter with upper and lower cutoff frequencies u and l
respectively. The lowpass filter has 3 dB bandwidth  p   and u  3 ,
 6 4
l 
4

Sol: The digital-to-digital transformation from filter to a bandpass filter is

05-06-2021 76
 Problems on Frequency Transformation (Contd…)
 2 k 1 k  1 
  z 2  z  
 
z 1   
1 k k 1
k  1 2 2 k 1
z  z 1  3  
k 1 1 k  4  4 
 3   cos  
 u  l  p  4 4   2   cos 
k  cot   cos u  l  2   
where  tan cot   tan    2 0
 2  2  2  12 cos u  l  2   3    cos 
   4  4  4
cos   
  
 cot   tan  0.268  2 
4 12  

05-06-2021 77
Problems on Frequency Transformation (Contd…)
Substituting the values of α and k in the transformation
 2 0.268  1 
 z  
1  0.268  1 
z 
0.268  1 2
z 1
0.268  1
1
  z  0.577 
2

 z 
0.577 z 2  1
2
1  z  0.577
 z 
1  0.577 z 2
05-06-2021 78
Problems on Frequency Transformation (Contd…)
Now the transfer function of bandpass filter can be obtained by
substituting 1  z 2  0.577 in H(z)   z 2
 0.577 
z  0.5 1 
1  0.577 z 2 2 
 1  0.577 z 
H ( z) 
  z 2  0.577 
1  0.302  2 
 1  0.577 z 
 1.577 1  z 2  
 0.5  2

 0.82575  0.275 z 
0.955 1  z 2 

1  0.333 z 2
05-06-2021 79
Applying the frequency Transformation
For impulse invariance
i. Convert the analog lowpass filter into digital lowpass filter
ii. Perform the frequency transformation in the digital domain

For bilinear transformation

First method:
i. Perform frequency transformation in analog domain
ii. Convert the analog filter into digital filter
Second method:
i. Convert the analog lowpass filter into digital lowpass filter
ii. Perform the frequency transformation in the digital domain
05-06-2021 80
Additional Problems on IIR Filter Design
Q 1. Design a digital Butterworth filter satisfying the constraints

0.707 | H  e j
 | 1 for 0   
2
3
| H  e j  | 0.2 for  
4
with T=1 sec using (a) the bilinear transformation (b) Impulse
invariance.
Sol:
(a) Bilinear transformation
1 1  3
 0.707;  0.2;  p  ; s 
1  2 1 2 2 4

05-06-2021 81
Additional Problems on IIR Filter Design(Contd…)
The analog frequency ratio is
2  3
tan s tan
s T 2  8  2.414

p 2 p 
tan tan
T 2 4
The order of the filter

log
N 
s
log
p
  4.898  1
log 4.898
N  1.803
log 2.414
05-06-2021 82
Additional Problems on IIR Filter Design
Rounding N to nearest higher value we get N=2
p
c   p
 
1
N

2 p 
 tan  2 tan  2 rad/sec
T 2 4
The transfer function of second order normalized Butterworth filter is
1
H (s)  2
s  2s  1
Ha(s) for c  2 rad/sec can be obtained by substituting s  s 2 in H(s)

05-06-2021 83
AdditionalProblems
Additional Problemsonon
IIRIIR Filter
Filter Design
Design(Contd…)
1 4
H a (s)  
 s 2  2 1 s 2  2.828s  4
2
 2. s
By using bilinear transformation H(z) can be obtained as
4 [since T=1]
H ( z )  H ( s ) | 2  1 z 1   2 |  1 z 1 
s 
T

1 
s  2.828s  4 s  2 1 
 1 z   1 z 

4
 H ( z )  2
 1 z  1
 1  z 1 
4 1 
 2.828  2  1 
4
 1 z   1 z 
4 1  z 
1 2
0.2929 1  z 
1 2

 
4 1  z 
1 2
 5.656 1  z 2
  4 1  z  1 2 1  0.1716 z 2
05-06-2021 84
AdditionalProblems
Additional Problemsonon
IIRIIR Filter
Filter Design
Design(Contd…)
(b) Impulse invariant method
The relationship between analog and digital frequencies in impulse
invariant method is   T
Given T=1 sec
Therefore      p   p and  s  s

log
The order of the filter N 
s
log
p
  4.898  1
log 4.898
N  3.924
3
log 4

05-06-2021 2 85
AdditionalProblems
Additional Problemsonon
IIRIIR Filter
Filter Design
Design(Contd…)
Therefore N=4
The transfer function of a 4th order normalized Butterworth filter is
1
H (s)  2
 s  0.76537 s  1  1.8477 s  1
s 2

As   1;  p   c  0.5  1.57
H a (s)  H (s) | s
s
1.57

1.57 
4


 s 2
 1.202 s  2.465   2.902s  2.465
s 2

05-06-2021 86
AdditionalProblems
Additional Problemsonon
IIRIIR Filter
Filter Design
Design(Contd…)
1.57 
4

H a (s) 
 s 2
 1.202 s  2.465   2.902s  2.465
s 2

1.57 
4


 s  1.45  j 0.6  s  1.45  j 0.6  s  0.6  j1.45  s  0.6  j1.45 
1.57 
4

Let
 s  1.45  j 0.6  s  1.45  j 0.6  s  0.6  j1.45  s  0.6  j1.45 
A A* B B*
   
 s  1.45  j 0.6   s  1.45  j 0.6   s  0.6  j1.45  s  0.6  j1.45 
A  0.7253  j1.754; B  0.7253  j 0.3
05-06-2021 87
AdditionalProblems
Additional Problemsonon
IIRIIR Filter
Filter Design
Design(Contd…)
0.7253  j1.754 0.7253  j1.754 0.7253  j 0.3 0.7253  j 0.3
H a (s)    
 s  1.45  j 0.6   s  1.45  j 0.6   s  0.6  j1.45   s  0.6  j1.45 
We know for T=1 sec
N
ck
H ( z)   pk 1
k 1 1  e z

Therefore
0.7253  j1.754 0.7253  j1.754 0.7253  j 0.3 0.7253  j 0.3
H ( z)  1.45  j 0.6 1
 1.45 j 0.6 1
 0.6 j1.45 1

1 e e z 1 e e z 1 e e z 1  e 0.6e j1.45 z 1
1.454  0.1839 z 1 1.454  0.2307 z 1
 1 2

1  0.387 z  0.055 z 1  0.1322 z 1  0.301z 2
05-06-2021 88
Additional Problems on IIR Filter Design (Contd…)
Q 2. Design a digital second order lowpass Butterworth filter with cutoff
frequency 2.2 kHz using Bilinear transformation. Sampling rate 8 kHz

05-06-2021 89