WINSEM2019-20 EEE2005 ETH VL2019205003575 Reference Material I 05-Feb-2020 Design of FIR Filters

Chapter 14
Finite Impulse Response (FIR) Filters

Learning Objectives
 Introduction to the theory behind FIR
filters:
 Properties (including aliasing).
 Coefficient calculation.
 Structure selection.
 Implementation in Matlab, C, assembly
and linear assembly.
Chapter 14, Slide 2 Dr. Naim Dahnoun, Bristol University, (c) Texas Instruments 2004
Introduction
 Amongst all the obvious advantages that
digital filters offer, the FIR filter can
guarantee linear phase characteristics.
 Neither analogue or IIR filters can achieve
this.
 There are many commercially available
software packages for filter design.
However, without basic theoretical
knowledge of the FIR filter, it will be
difficult to use them.
Properties of an FIR Filter
 Filter coefficients:
N 1
yn   bk  xn  k 
k 0
x[n] represents the filter input,

bk represents the filter coefficients,
y[n] represents the filter output,
N is the number of filter coefficients
(order of the filter).
N 1
yn   bk  xn  k  FIR equation
k 0
-1 -1 -1 -1
x(n) z z z z
b0 b1 b2 bN-1
x x x x
+ + + y(n)
Filter structure
N 1
yn   bk  xn  k 
k 0
 If the signal x[n] is replaced by an impulse

[n] then:
N 1
yn   bk  n  k 
k 0
y0  b0 0  b1  1    bk   N 
N 1
yn   bk  xn  k 
k 0

[n] then:
N 1
yn   bk  n  k 
k 0
yn  b0 n  b1 n  1    bk  n  N 
N 1
yn   bk  xn  k 
k 0

[n] then:
N 1
yn   bk  n  k 
k 0
1 for n  k
 n  k   
0 for n  k
N 1
yn   bk  xn  k 
k 0
 Finally:
b0  h0
b1  h1

bk  hk 
N 1
yn   bk  xn  k 
k 0
With: bk  hk 
 The coefficients of a filter are the same as

the impulse response samples of the filter.
Frequency Response of an FIR Filter
 By taking the z-transform of h[n], H(z):
N 1
H  z    hnz  n
n 0
 Replacing z by ej in order to find the

frequency response leads to:
 
N 1
H  z  z e j  H e j   hne  jn
n 0
 Since e-j2k = 1 then:
N 1 N 1
H z  z e 2   hne  jn  2   hne  jn
n 0 n 0
 Therefore:
   
H e j 2 k  H e j
 FIR filters have a periodic frequency

response and the period is 2.
 Frequency
response:
   
H e j 2k  H e j
x[n] FIR y[n]

x[n]
y[n]
Freq
Freq
Fs/2 Fs/2
 Solution: Use an anti-aliasing filter.
x[n]
x(t) ADC FIR y[n]
Analogue
Anti-Aliasing
y[n]
x(t)
Freq Freq
Fs/2 Fs/2
Phase Linearity of an FIR Filter
 A causal FIR filter whose impulse
response is symmetrical is guaranteed to
have a linear phase response.
h(n) h(n)
N = 2n + 2 N = 2n + 1
n n
0 1 n n+1 2n 2n+1 0 1 n-1 n n+1 2n-1 2n
Even symmetry Odd symmetry
 A causal FIR filter whose impulse
response is symmetrical (ie h[n] = h[N-1-n]
for n = 0, 1, …, N-1) is guaranteed to have
a linear phase response.
 N 1 
Condition Phase  k    Phase Property Filter Type
 2 
hn   hN  n  1 Odd Symmetry – Type 1

k Linear phase
Positive Symmetry Even Symmetry – Type 2
 Application of 90° linear phase shift:
o IH
I 90 +
Reverse
delay +
Signal
separation delay -
Forward
Q 90o +
QH
I  A cos f t  B sin r t     IH  Q  2 B cosr t

IH  A cos  f t    B sin r t  
 2  2
Q  Asin  f t  B cosr t   A sin  f t  B cosr t QH  I  2B sin  f t
Design Procedure
 To fully design and implement a filter five
steps are required:
(1) Filter specification.
(2) Coefficient calculation.
(3) Structure selection.
(4) Simulation (optional).
(5) Implementation.
Filter Specification - Step 1
|H(f)| pass-band stop-band
fc : cut-off frequency fs/2 f(norm)

(a)
|H(f)| pass-band transition band stop-band |H(f)|

(dB) (linear)
p 1   p
0 1
1   p
pass-band
-3
ripple
stop-band
ripple
s  s
fs/2 f(norm)
fsb : stop-band frequency
fc : cut-off frequency
fpb : pass-band frequency
(b)
Coefficient Calculation - Step 2
 There are several different methods
available, the most popular are:
 Window method.
 Frequency sampling.
 Parks-McClellan.
 We will just consider the window method.
Window Method
 First stage of this method is to calculate
the coefficients of the ideal filter.
 This is calculated as follows:

hd n   H  e jn d
1
2 

c
1

2  1  e jn d
 c
 2 f c sin n c 
 for n  0
 n c
 2 fc for n  0
Window Method
 Second stage of this method is to select a window
function based on the passband or attenuation
specifications, then determine the filter length based on
the required width of the transition band.
Normalised Transition Stopband Attenuation

Window Type Passband Ripple(dB)
Width (f(Hz)) (dB)
0.9
Rectangular 0.7416 21
N
3.1
Hanning N
0.0546 44
3.3
Hamming N 0.0194 53
5.5
Blackman N 0.0017 74
2.93
   4.54
N 0.0274 50
Kaiser 5.71
   8.96
N 0.000275 90
Using the Hamming N

3.3

3.3
 8kHz  132
Window: f 1.2  1.4kHz
Window Method
 The third stage is to calculate the set of
truncated or windowed impulse response
coefficients, h[n]:
N 1 N 1
 n for N  odd
hn  hd nW n for 2 2
N
 n
N for N  even
2 2
 2n 
Where: W n   0.54  0.46 cos 
 N 
for 66  n  66
 2n 
 0.54  0.46 cos 
 133 
Window Method
 Matlab code for calculating coefficients:
close all;
clear all;
fc = 8000/44100; % cut-off frequency

N = 133; % number of taps
n = -((N-1)/2):((N-1)/2);
n = n+(n==0)*eps; % avoiding division by zero
[h] = sin(n*2*pi*fc)./(n*pi); % generate sequence of ideal coefficients

[w] = 0.54 + 0.46*cos(2*pi*n/N); % generate window function
d = h.*w; % window the ideal coefficients
[g,f] = freqz(d,1,512,44100); % transform into frequency domain for plotting
figure(1)
plot(f,20*log10(abs(g))); % plot transfer function
axis([0 2*10^4 -70 10]);
figure(2);
stem(d); % plot coefficient values
xlabel('Coefficient number');
ylabel ('Value');
title('Truncated Impulse Response');
figure(3)
freqz(d,1,512,44100); % use freqz to plot magnitude and phase response
axis([0 2*10^4 -70 10]);
Window Method
Truncated Impulse Response
0.4
0.3
0.2
h(n)
0.1
-0.1
0 20 40 60 80 100 120 140
Coefficient number, n
0
Magnitude (dB)
-20
-40
-60
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Frequency (Hz) 4
x 10
0
Phase (degrees)
-2000
-4000
-6000
0 0.5 1 1.5 2
Frequency (Hz) 4
x 10
Realisation Structure Selection - Step 3
 Direct form structure for an FIR filter:
N 1
H z    bk z  k
k 0
Y z   H z   X z  yn  b0 xn  b1 xn  1  ....  bN 1 xn  N  1
-1 -1 -1
x(n) z z z
b b b b
0 1 2 N-1
+ + + y(n)
N 1
H z    bk z  k
k 0
 Linear phase structures:

N
1
 
2
 N even: H z    bk z  k  z N  k 1
k 0
N 1
N 1
 
2 
 N Odd: H z    bk z  k  z N  k 1  b N 1 z 2
k 0 2
-1 -1 -1 -1 -1
(a) N even. z z z z z
+
b0
b1
(b) N odd. +
+
b2
+
b N/2-1
+ +
y(n)
(a)
x(n) -1 -1 -1 -1 -1 -1
z z z z z z
b0
+
b1
+ +
b2
+ +
b(N-1)/2
b(N-3)/2
+ +
y(n) +
(b)
N 1
H z    bk z  k
k 0
 Cascade structures:
N 1
H z    bk z  k  b0  b1 z 1  b2 z  2  ...  bN 1 z   N 1
k 0
 b b b 
 b0 1  1 z 1  2 z  2  ...  N 1 z   N 1 
 b0 b0 b0 
 
M
 b0  1  bk ,1 z 1  bk , 2 z  2
k 1
N 1
H z    bk z  k
k 0
 Cascade structures:
x(n) b0 y(n)
+ + +
z -1 z -1 z -1
b 1,1 b 2,1 b M,1
+ + +
z -1 z -1 z -1
b 1,2 b 2,2 b M,2
Implementation - Step 5
 Implementation procedure in ‘C’ with
fixed-point:
 Set up the codec (\Links\CodecSetup.pdf).
 Transform: N 1
yn   bk  xn  k  to ‘C’ code.
k 0
(\Links\FIRFixed.pdf)
 Configure timer 1 to generate an interrupt at
8000Hz (\Links\TimerSetup.pdf).
 Set the interrupt generator to generate an
interrupt to invoke the Interrupt Service
Routine (ISR) (\Links\InterruptSetup.pdf).
 Implementation procedure in ‘C’ with
floating-point:
Same set up as fixed-point plus:
 Convert the input signal to floating-point
format.
 Convert the coefficients to floating-point
format.
 With floating-point multiplications there is
no need for the shift required when using
Q15 format.
 See \Links\FIRFloat.pdf
 Implementation procedure in assembly:
Same set up as fixed-point, however:
 N 1
yn   bk  xn  k 
is written in assembly.
k 0
(\Links\FIRFixedAsm.pdf)
 The ISR is now declared as external.
 Implementation procedure in assembly:
The filter implementation in assembly is
now using circular addressing and
therefore:
 The circular pointers and block size register
are selected and initialised by setting the
appropriate values of the AMR bit fields.
 The data is now aligned using:
#pragma DATA_ALIGN (symbol, constant (bytes))
 Set the initial value of the circular pointers,

see \Links\FIRFixedAsm.pdf.
b0 x0
b1 x1
b2 x2
b3 x3
y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

y[n]
time
0 1 2
Circular addressing link slide.
b0 x4
b1 x1
b2 x2
b3 x3
y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

y1 = b0*x4 + b1*x1 + b2*x2 + b3*x3
y[n]
time
0 1 2
b0 x4
b1 x5
b2 x2
b3 x3
y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

y1 = b0*x4 + b1*x1 + b2*x2 + b3*x3
y2 = b0*x4 + b1*x5 + b2*x2 + b3*x3
y[n]
time
0 1 2
FIR Code
 Code location:
 Code\Chapter 14 - Finite Impulse Response Filters
 Projects:
 Fixed Point in C: \FIR_C_Fixed\
 Floating Point in C: \FIR_C_Float\
 Fixed Point in Assembly: \FIR_Asm_Fixed\
 Floating Point in Assembly: \FIR_Asm_Float\
Chapter 14
Finite Impulse Response (FIR) Filters
- End -

WINSEM2019-20 EEE2005 ETH VL2019205003575 Reference Material I 05-Feb-2020 Design of FIR Filters

Uploaded by

Document Information

Original Title

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

WINSEM2019-20 EEE2005 ETH VL2019205003575 Reference Material I 05-Feb-2020 Design of FIR Filters

Uploaded by

Copyright:

Available Formats

Chapter 14

Finite Impulse Response (FIR) Filters

x[n] represents the filter input,

 If the signal x[n] is replaced by an impulse

y0  b0 0  b1  1    bk   N 

 If the signal x[n] is replaced by an impulse

yn  b0 n  b1 n  1    bk  n  N 

 If the signal x[n] is replaced by an impulse

 The coefficients of a filter are the same as

 Replacing z by ej in order to find the

 FIR filters have a periodic frequency

x[n] FIR y[n]

Even symmetry Odd symmetry

hn   hN  n  1 Odd Symmetry – Type 1

I  A cos f t  B sin r t     IH  Q  2 B cosr t

fc : cut-off frequency fs/2 f(norm)

|H(f)| pass-band transition band stop-band |H(f)|

Normalised Transition Stopband Attenuation

Using the Hamming N

fc = 8000/44100; % cut-off frequency

[h] = sin(n*2*pi*fc)./(n*pi); % generate sequence of ideal coefficients

[g,f] = freqz(d,1,512,44100); % transform into frequency domain for plotting

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Y z   H z   X z  yn  b0 xn  b1 xn  1  ....  bN 1 xn  N  1

 Linear phase structures:

 The ISR is now declared as external.

 Set the initial value of the circular pointers,

y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

y0 = b0*x0 + b1*x1 + b2*x2 + b3*x3

You might also like

[h] = sin(n2pifc)./(npi); % generate sequence of ideal coefficients

y0 = b0x0 + b1x1 + b2x2 + b3x3

y0 = b0x0 + b1x1 + b2x2 + b3x3

y0 = b0x0 + b1x1 + b2x2 + b3x3