CHAPTER 7

Digital Filter Design

Wang Weilian
[email protected]
School of Information Science and Technology
Yunnan University
 Outline
• About Digital Filter Design

• Bilinear Transformation Method of IIR Filter Design

• Design of Lowpass IIR Digital Filters

• Design of Hignpass, Bandpass, and Bandstop IIR Digital Fil

ter
• FIR Filter Design Based on Windowed Fourier Series

• Computer-Aided Design of Digital Filters

• Digital Filter Design Using MATLAB

 About Digital Filter Design
• The most important step in the development of a digital filter :Determine a r
ealizable transfer function G(z)

• Digital Filter Specifications

(1) magnitude response specifications in the passband and the stopband ar

e given with some acceptable tolerances

(2) A transition band is specified between the passband and the stopband t
o permit the magnitude to drop off smoothly
About Digital Filter Design
 About Digital Filter Design
• Passband edge frequency p
• Stopband edge frequency s

• Peak ripple value of passband p

• Peak ripple value of stopband
s
• Peak passband ripple p
• Minimum stopband attenuation s

• Sample frequency FT
 About Digital Filter Design

 P  20 log10 (1   P )dB

 S  20 log10 ( s )dB
 P / 20
 P  1  10
 S / 20
 S  10
P 2FP
P  FT  FT
S 2FS
S  FT  FT
 About Digital Filter Design
• Selection of the Filter Type

(1)The objective of digital filter design is to develop a causal tr

ansfer function H(z) meeting the frequncy specifications.

(2)FIR and IIR Digital Filter

N
H ( z )   h[ n] z
n
FIR Digital Filter
n0

p0  p1 z 1  p2 z 2   p M z  M
IIR Digital Filter H ( z)  d 0  d1 z 1  d 2 z 2   d N z  N
 About Digital Filter Design

FIR IIR
Impulse Response finite infinite

System Function H(z)=P(z) H(z)=P(z)/D(z)

Structure diagram Have feedback No feedback

Phase response Exact linear phase ________________

h[n]=  h[n-N]

Zero-poles Only have zeros Both zeros and poles

 About Digital Filter Design

The order NFIR of an FIR filter is higher than the

order NIIR of an equivalent IIR filter meeting the

same magnitude specifications

The ratio NFIR/ NIIRis typically of the order of 10 or mo

re (the IIR filter usually is computationally more ef
ficient)
 About Digital Filter Design

• Basic Approaches to Digital Filter Design

Step1:convert the digital filter specifications into an

alog lowpass prototype filter specifications

Step2:determine the analog lowpass filter transfer f

unction Ha(s)

Step3:transform Ha(s) into the desired digital filter tr

ansfer function G(z)
 About Digital Filter Design

• Why analog?

(1)Analog approximation techniques are highly

advanced

(2)They usually yield closed-form solutions

(3)Extensive tables are available for analog filter

design

(4)Many applications require the digital simulation

of analog filters
 About Digital Filter Design
• How to convert an analog prototype transfer function Ha(s) in
to a digital IIR transfer function G(z)?

(1)the imaginary(j  ) axisin the s-plane be mapped onto the unit

circle of the z-plane

(2)A stable analog transfer functon be transformed into a stable

digital transfer function
 About Digital Filter Design
• Estimation of the Filter Order
IIR: The order of G(z) is determined from the transformation bei
ng used to convert Ha(s) into G(z)(The determination of Ha(s)
is refered to Eq.(5.33),(5.41),or(5.51)
FIR(lowpass digital filter):

20 log10 (  p s ) 13

N 14.6 ( s  p ) / 2

 20 log10 ( s )  0.22
For narrowband filter N ( s  p ) / 2

 20 log10 ( p )  5.94
For wideband filter N 27 ( s  p ) / 2
 Bilinear Transformation Method of IIR Filter
Design
• Bilinear transformation is more commonly used to design II
R digital filters based on the conversion of analog prototype
filters
• The Bilinear Transformation

S-plane to z-plane

G(z)= Ha(s) | s  2 ( 1 z 1 )
T 1 z 1

The transformation is a one-to-one mapping. It maps a singl

e point in the s-plane to a unique point in the z-plane
Bilinear Transformation Method of IIR Filter
Design
Bilinear Transformation Method of IIR Filter
Design
 Bilinear Transformation Method of IIR Filter
Design
• Digital filter design procedure:

Step1 the invert bilinear transformation is applied

to the digital filter specifications to arrive at the specifica

tions of the analog filter function

Step2 the bilinear transformation is employed to obtain the de

sired digital transfer function G(z) from the analog transfer f
unction Ha(s) desired to meet the analog filter specification
s
Bilinear Transformation Method of IIR Filter
Design
• When T=2(T has no effect on the G(z))

s  j 0
z 1 s
1 s
z 1 j 0
1 j 0

s   0  j 0

(1 0 )  j 0 2 (1 0 ) 2  (  0 ) 2
z (1 0 )  j 0
z  (1 0 ) 2  (  0 ) 2
If 0  0 then z <1
If  0 >0 then
z >1
Bilinear Transformation Method of IIR Filter
Design

j
• When s  j and ze

1 e  j
j  1 e j  j tan( )

2

  tan( ) 
2
 Bilinear Transformation Method of IIR Filter
Design
• Design of Digital IIR Notch Filters
Example a second-order IIR notch filter
s 2   02
Analog transfer function:
H ( s)  s 2  Bs   02
(1  02 )  2 (1  02 ) z 1  (1  02 ) z 2
Applying a bilinear transformation : G( z)  (1  02  B )  2 (1  02 ) z 1  (1  02  B ) z 2
Rewrite it:
(1 )  2  (1 ) z 1  (1 ) z 2
G( z)  2[1  (1 ) z 1 z 2 ]

  11 tan( B
tan( B
w / 2)
w / 2)

  cos 0
Notch frequency 0
Notch bandwidth
Bw
 Design of Lowpass IIR Digital Filters
• Steps of designing a lowpass IIR digital filter

Step1: get the digital filter specifications(  p  s  p s )

Step2: convert to analog filter specifications with bilineat transfor

mation

Step3: design analog transfer function Ha(s)

1 z 1
Step4: transfer Ha(s) to H(z) since s ( 2
T 1 z 1
)
 Design of Lowpass IIR Digital Filters
• Example

Passband edge frequency  pis 0.25  with a passband ripple of 0.

p
5dB

Stopband edge frequency  is 0.55  with a stopband attenuation

s
s of 15dB

Then

p
 p  tan( )  tan( 0.252  )  0.4142136
2
s
 s  tan( )  tan( 0.552  )  1.1708496
2
p
1
k
 s
 10..1708496
4142135
 2.8266809
 Design of Lowpass IIR Digital Filters
• From the passband ripple of 0.5dB obtaining  2  0.1220185

From the stopband attenuation of 15dB obtaining A2  310622777

Then A 2 1
1
K1    15.841979
N log(1 / k )
log(1 / k )  2.6586997  N  3
since 2
H a ( j p )  1 (  1
2N  11 2
p / c )

2
H a ( j s )  1 (  1 ) 2N  1
A2
s / c

then we get
 c  1.419915( p )
 Design of Lowpass IIR Digital Filters
The transfer function of third-order lowpass Butterworth is

H an ( s )  1
( s 1)( s 2  s 1)

Then we can get

H a ( s )  H an ( 0.588148
s
) 0.203451
( s  0.588148)( s 2  0.588148s  0.345918)

0.0662272(1 z 1 ) 3
G ( z )  H a ( s ) |s (1 z 1 ) /(1 z 1 )  (1 0.2593284z 1 )(1 0.6762858z 1  0.3917468z 2 )
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters

• To design IIR filters there are two approches can be followed

• First approch:

Step1: prewarp the digital frequency specifications to arrive at

the specifications of an analog filter of the same type.

1 e  j
j  1 e j
 j tan( ) 
2

  tan( 2 )
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters

• Step2: convert the frequency specifications of HD(s) into that of a

prototype analog lowpass filter HLP(S)

s  F ( sˆ)
H D ( sˆ)  H LP ( s ) |s  F ( sˆ )
H LP ( s )  H D ( sˆ) |sˆ  F 1 ( S )

( s is the Laplace transform variable of the prototype analog lowp

ŝ
ass filter HLP(S) and is the Laplace transform variable of the des
ired analog filter ) H D (sˆ)
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters

• Step3: Design the analog lowpass filter HLP(S) using the meth
od described in Section 5.4

• Step4: convert the transfer function HLP(S) into HD(S) using the
inverse of the frequency transformation used in step2

• Transform the transfer function HD(S) using the bilinear transf

ormation to arrive at the desired digital IIR transfer function G
D (Z)
1
s ( 2
T
1 z
1 z 1
)
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters
• The second approach:

Step1: prewarp the digital frequency specifications to arrive at

the specifications of an analog filter of the same type.

1 e  j
j  1 e j
 j tan( 2 )

  tan( ) 
2
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters
• Step2: convert the frequency specifications of HD(s) into tha
t of a prototype analog lowpass filter HLP(S)

s  F ( sˆ)
H D ( sˆ)  H LP ( s ) |s  F ( sˆ )
H LP ( s )  H D ( sˆ) |sˆ  F 1 ( S )

( s is the Laplace transform variable of the prototype analog lo

ŝ
wpass filter HLP(S) and is the Laplace transform variable of
) ˆ)
the desired analog filter H D (s
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters
• Step3: Design the analog lowpass filter HLP(S) using the method d
escribed in Section 5.4
• Step4: convert the transfer function HLP(S) into the transfer functi
on GLP(Z) of an IIR digital filter using the bilinear transformation

1 z 1
s ( 2
T 1 z 1
)
• Step5: transform GLP(Z) into the desired digital transfer function
GD(z) using the appropriate spectral transformation discussed in
Section 7.5
Design of Highpass, Bandpass, and Bandsto
p IIR Digital Filters
• The functions we usually used in Matlab

lp2hp: transform the lowpass analog filter to

highpass analog filter

lp2bp: transform the lowpass analog filter to ban

dpass analog filter

lp2bs : transform the lowpass analog filter to ban

dstop analog filter