1-Digital Filters (FIR)

AGC
DSP
Professor A G Constantinides
1
Digital Filter Specifications
Only the magnitude approximation problem
Four basic types of ideal filters with magnitude
responses as shown below (Piecewise flat)
AGC
DSP
2
These filters are unealisable because (one of
the following is sufficient)
their impulse responses infinitely long non-
causal
Their amplitude responses cannot be equal
to a constant over a band of frequencies
Another perspective that provides some
understanding can be obtained by looking
at the ideal amplitude squared.
AGC
DSP
3
Consider the ideal LP response squared
(same as actual LP response)

AGC
DSP
4
The realisable squared amplitude response
transfer function (and its differential) is
continuous in
Such functions
if IIR can be infinite at point but around
that point cannot be zero.
if FIR cannot be infinite anywhere.
Hence previous defferential of ideal response
is unrealisable

e
AGC
DSP
5
A realisable response would effectively
need to have an approximation of the
delta functions in the differential
This is a necessary condition
AGC
DSP
6
For example the magnitude response
of a digital lowpass filter may be given as
indicated below
AGC
DSP
7
In the passband we require
that with a deviation

In the stopband we require
that with a deviation
1 ) ( ~
e j
e G
0 ) ( ~
e j
e G
s
o
p
o
p
e e s s 0
t e e s s
s
p p
j
p
e G e e o o
e
s + s s , 1 ) ( 1
t e e o
e
s s s
s s
j
e G , ) (
AGC
DSP
8
Filter specification parameters
- passband edge frequency
- stopband edge frequency
- peak ripple value in the
passband
- peak ripple value in the
stopband
p
e
s
e
s
o
p
o
AGC
DSP
9
Practical specifications are often given
in terms of loss function (in dB)

Peak passband ripple
dB
Minimum stopband attenuation
dB
) ( log 20 ) (
10
e
e
j
e G = G
) 1 ( log 20
10 p p
o o =
) ( log 20
10 s s
o o =
AGC
DSP
10
In practice, passband edge frequency
and stopband edge frequency are
specified in Hz
For digital filter design, normalized bandedge
frequencies need to be computed from
specifications in Hz using

T F
F
F
F
p
T
p
T
p
p
t
t
e 2
2
= =
O
=
T F
F
F
F
s
T
s
T
s
s
t
t
e 2
2
= =
O
=
s
F
p
F
AGC
DSP
11
Example - Let kHz,
kHz, and kHz
Then

7 =
p
F 3 =
s
F
25 =
T
F
t
t
e 56 . 0
10 25
) 10 7 ( 2
3
3
=
=
p
t
t
e 24 . 0
10 25
) 10 3 ( 2
3
3
=
=
s
AGC
DSP
12
The transfer function H(z) meeting the
specifications must be a causal transfer
function
For IIR real digital filter the transfer function
is a real rational function of

H(z) must be stable and of lowest order N or
M for reduced computational complexity
Selection of Filter Type
1
z
N
N
M
M
z d z d z d d
z p z p z p p
z H

+ + + +
+ + + +
=
2
2
1
1 0
2
2
1
1 0
) (
AGC
DSP
13

FIR real digital filter transfer function is a
polynomial in (order N) with real
coefficients

For reduced computational complexity,
degree N of H(z) must be as small as possible
If a linear phase is desired then we must
have:
(More on this later)
=
=

N
n
n
z n h z H
0
] [ ) (
] [ ] [ n N h n h =
1
z
AGC
DSP
14
Advantages in using an FIR filter -
(1) Can be designed with exact linear phase
(2) Filter structure always stable with
quantised coefficients
Disadvantages in using an FIR filter - Order of
an FIR filter is considerably higher than that
of an equivalent IIR filter meeting the same
specifications; this leads to higher
computational complexity for FIR
AGC
DSP
15
FIR Design
FIR Digital Filter Design
Three commonly used approaches to
FIR filter design -
(1) Windowed Fourier series approach
(2) Frequency sampling approach
(3) Computer-based optimization
methods
AGC
DSP
16
Finite Impulse Response
Filters

The transfer function is given by

The length of Impulse Response is N
All poles are at .
Zeros can be placed anywhere on the z-
plane
=

=

1
0
). ( ) (
N
n
n
z n h z H
0 = z
AGC
DSP
17
FIR: Linear phase

For phase linearity the FIR transfer
function must have zeros outside
the unit circle

AGC
DSP
18
FIR: Linear phase
To develop expression for phase
response set transfer function (order n)

In factored form

Where , is
real & zeros occur in conjugates
n
n
z h z h z h h z H

+ + + + = ... ) (
2
2
1
1 0
) 1 ( ). 1 ( ) (
1
2
1
1
1
1
=
[

[
= z z K z H
i
n
i
i
n
i
| o
1 , 1 > <
i i
| o
K
AGC
DSP
19
FIR: Linear phase
Let
where

Thus
) ( ) ( ) (
2 1
z N z KN z H =
) 1 ln( ) 1 ln( ) ln( )) ( ln(
2
1
1
1
1
1
+ =
=

=

n
i
i
n
i
i
z z K z H | o
) 1 ( ) (
1
1
1
1
=
[
= z z N
i
n
i
o ) 1 ( ) (
1
2
1
2
=
[
= z z N
i
n
i
|
AGC
DSP
20
FIR: Linear phase
Expand in a Laurent Series
convergent within the unit circle
To do so modify the second sum as

)
1
1 ln( ) ln( ) 1 ln(
2
1
1
2
1
1
2
1
z z z
i
n
i
i
n
i
i
n
i
|
| |

+
=
AGC
DSP
21
FIR: Linear phase
So that

Thus

where
)
1
1 ln( ) 1 ln( ) ln( ) ln( )) ( ln(
2
1
1
1
1
2

+
+ =
= =

n
i
i
n
i
i
z z z n K z H
|
o
m
N
m
m
m
N
m
z
m
s
z
m
s
z n K z H
2
1
1
2
) ln( ) ln( )) ( ln(

+ + =
=
=
1
1
1
n
i
m
i
N
m
s o

=
=

1
1
2
n
i
m
i
N
m
s |
AGC
DSP
22
FIR: Linear phase
are the root moments of the
minimum phase component
are the inverse root moments of
the maximum phase component
Now on the unit circle we have
and

u j
e z =
) (
) ( ) (
u | u
u
j j
e A e H =
1
N
m
s
2
N
m
s
AGC
DSP
23
Fundamental Relationships

hence (note Fourier form)

u u u
u
jm
N
m
m
jm
N
m
j
e
m
s
e
m
s
jn K e H
2
1
1
2
) ln( )) ( ln(

+ + =
) ( )) ( ln( ) ) ( ln( )) ( ln(
) (
u | u u
u | u
j A e A e H
j j
+ = =
u u m
m
s
m
s
K A
N
m
m
N
m
cos ) ( ) ln( )) ( ln(
2
1
1
+ + =
u u u | m
m
s
m
s
n
N
m
m
N
m
sin ) ( ) (
2
1
1
2
=
AGC
DSP
24
FIR: Linear phase

Thus for linear phase the second term in the
fundamental phase relationship must be
identically zero for all index values.
Hence
1) the maximum phase factor has zeros
which are the inverses of the those of the
minimum phase factor
2) the phase response is linear with group
delay (normalised) equal to the number of
zeros outside the unit circle
AGC
DSP
25
FIR: Linear phase

It follows that zeros of linear phase FIR
trasfer functions not on the
circumference of the unit circle occur in
the form

| |
1
i
j
i
e
u
AGC
DSP
26
FIR: Linear phase

For Linear Phase t.f. (order N-1)

so that for N even:

) 1 ( ) ( n N h n h =
=

=
=

1
2
1
2
0
). ( ). ( ) (
N
N
n
n
N
n
n
z n h z n h z H
=

=

=

1
2
0
) 1 (
1
2
0
). 1 ( ). (
N
n
n N
N
n
n
z n N h z n h
| |
=

=

1
2
0
) (
N
n
m n
z z n h n N m = 1
AGC
DSP
27
FIR: Linear phase

for N odd:

I) On we have for N even,
and +ve sign
| |
|
.
|
\
|

+ =

=
|
.
|
\
|

1
2
1
0
2
1
2
1
). ( ) (
N
n
N
m n
z
N
h z z n h z H
1 : = z C
|
.
|
\
|
|
.
|
\
|

=

=
|
.
|
\
|

1
2
0
2
1
2
1
cos ). ( 2 . ) (
N
n
N
T j
T j
N
n T n h e e H e
e
e
AGC
DSP
28
FIR: Linear phase
II) While for ve sign

[Note: antisymmetric case adds rads
to phase, with discontinuity at ]
III) For N odd with +ve sign

|
.
|
\
|
|
.
|
\
|

=

=
|
.
|
\
|

1
2
0
2
1
2
1
sin ). ( 2 . ) (
N
n
N
T j
T j
N
n T n h j e e H e
e
e
2 / t
0 = e
|
.
|
\
|

=
(
2
1
) (
2
1
N
h e e H
N
T j
T j
e
e
|
.
|
\
|

+

=
2
3
0
2
1
cos ). ( 2
N
n
N
n T n h e
AGC
DSP
29
FIR: Linear phase
IV) While with a ve sign

[Notice that for the antisymmetric case to
have linear phase we require

The phase discontinuity is as for N even]
|
.
|
\
|

=

=
(
2
3
0
2
1
2
1
sin ). ( . 2 ) (
N
n
N
T j
T j
N
n T n h j e e H e
e
e
. 0
2
1
=
|
.
|
\
|
N
h
AGC
DSP
30
FIR: Linear phase

The cases most commonly used in filter
design are (I) and (III), for which the
amplitude characteristic can be written
as a polynomial in
2
cos
T e
AGC
DSP
31
Design of FIR filters: Windows

(i) Start with ideal infinite duration
(ii) Truncate to finite length. (This
produces unwanted ripples increasing in
height near discontinuity.)
(iii) Modify to
Weight w(n) is the window

{ } ) (n h
) ( ). ( ) (
~
n w n h n h =
AGC
DSP
32
Windows
Commonly used windows
Rectangular
Bartlett
Hann
Hamming

Blackman

Kaiser

2
1
<
N
n
N
n 2
1
|
.
|
\
|
+
N
n t 2
cos 1
|
.
|
\
|
+
N
n t 2
cos 46 . 0 54 . 0
|
.
|
\
|
+
|
.
|
\
|
+
N
n
N
n t t 4
cos 08 . 0
2
cos 5 . 0 42 . 0
) (
1
2
1
0
2
0
| | J
N
n
J
(
(
|
.
|
\
|
AGC
DSP
33
Kaiser window

Kaiser window
Transition
width (Hz)
Min. stop
attn dB
2.12 1.5/N 30
4.54 2.9/N 50
6.76 4.3/N 70
8.96 5.7/N 90
AGC
DSP
34
Example

Lowpass filter of length 51 and
2 / t e =
c
0 0.2 0.4 0.6 0.8 1
-100
-50
0
e/t
G
a
i
n
,

d
B
Lowpass Filter Designed Using Hann window
0 0.2 0.4 0.6 0.8 1
-100
-50
0
e/t
G
a
i
n
,

d
B
Lowpass Filter Designed Using Hamming window
0 0.2 0.4 0.6 0.8 1
-100
-50
0
e/t
G
a
i
n
,

d
B
Lowpass Filter Designed Using Blackman window
AGC
DSP
35
Frequency Sampling Method

In this approach we are given and
need to find
This is an interpolation problem and the
solution is given in the DFT part of the
course

It has similar problems to the windowing
approach
=

=

1
0
1
2
. 1
1
). (
1
) (
N
k
k
N
j
N
z e
z
k H
N
z H
t
) (k H
) (z H
AGC
DSP
36
Linear-Phase FIR Filter
Design by Optimisation
Amplitude response for all 4 types of
linear-phase FIR filters can be
expressed as

where
) ( ) ( ) ( e e e A Q H =
=
4 Type for ), 2 / sin(
3 Type for ), sin(
2 Type for /2), cos(
1 Type for , 1
) (
e
e
e
e Q
AGC
DSP
37
Modified form of weighted error function

where
)] ( ) ( ) ( )[ ( ) ( e e e e e D A Q W = E
] ) ( )[ ( ) (
) (
) (
e
e
e e e
Q
D
A Q W =
)] (
~
) ( )[ (
~
e e e D A W =
) ( ) ( ) (
~
e e e Q W W =
) ( / ) ( ) (
~
e e e Q D D =
AGC
DSP
38
Optimisation Problem - Determine
which minimise the peak absolute value
of

over the specified frequency bands
After has been determined,
construct the original and
hence h[n]
)] (
~
) cos( ] [
~
)[ (
~
) (
0
e e e e D k k a W
L
k
=
=
E
] [
~
k a
R e e
) (
e j
e A
] [
~
k a
AGC
DSP
39
Solution is obtained via the Alternation
Theorem
The optimal solution has equiripple
behaviour consistent with the total
number of available parameters.
Parks and McClellan used the Remez
algorithm to develop a procedure for
designing linear FIR digital filters.

AGC
DSP
40
FIR Digital Filter Order
Estimation
Kaisers Formula:

ie N is inversely proportional to
transition band width and not on
transition band location
t e e
o o
2 / ) ( 6 . 14
) ( log 20
10
p s
s p
N
~
AGC
DSP
41
Estimation
Hermann-Rabiner-Chans Formula:

where

with
t e e
t e e o o o o
2 / ) (
] 2 / ) )[( , ( ) , (
2
p s
p s s p s p
F D
N

~

s p p s p
a a a D o o o o o
10 3 10 2
2
10 1
log ] ) (log ) (log [ ) , ( + + =
] ) (log ) (log [
6 10 5
2
10 4
a a a
p p
+ + + o o
] log [log ) , (
10 10 2 1 s p s p
b b F o o o o + =
4761 . 0 , 07114 . 0 , 005309 . 0
3 2 1
= = = a a a
4278 . 0 , 5941 . 0 , 00266 . 0
6 5 4
= = = a a a
51244 . 0 , 01217 . 11
2 1
= = b b
AGC
DSP
42
Estimation
Formula valid for
For , formula to be used is
obtained by interchanging and
Both formulae provide only an estimate
of the required filter order N
If specifications are not met, increase
filter order until they are met
s p
o o >
s p
o o <
p
o
s
o
AGC
DSP
43
Estimation
Fred Harris guide:

where A is the attenuation in dB
Then add about 10% to it
t e e 2 / ) ( 20
p s
A
N

1-Digital Filters (FIR)

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1-Digital Filters (FIR)

Uploaded by

Copyright:

Available Formats

AGC

You might also like