Chapter 14

Finite Impulse Response (FIR) Filters

Learning Objectives

Introduction to the theory behind FIR

filters:

Chapter 14, Slide 2

Properties (including aliasing).

Coefficient calculation.
Structure selection.

Implementation in Matlab, C, assembly

and linear assembly.

Dr. Naim

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.

Chapter 14, Slide 3

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

x[n]
bk
y[n]
N

Chapter 14, Slide 4

represents the filter input,

represents the filter coefficients,
represents the filter output,
is the number of filter coefficients
(order of the filter).

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

x(n)

-1

-1

-1

b2

FIR equation
z

-1

N-1

y(n)

Filter structure
Chapter 14, Slide 5

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

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

[n] then:
y n

N 1

b n k
k

k 0

y 0 b0 0 b1 1 bk N

Chapter 14, Slide 6

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

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

[n] then:
y n

N 1

b n k
k

k 0

y n b0 n b1 n 1 bk n N

Chapter 14, Slide 7

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

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

[n] then:
y n

N 1

b n k
k

k 0

1 for n k
n k
0 for n k
Chapter 14, Slide 8

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

Finally:
b0 h 0
b1 h1

bk h k

Chapter 14, Slide 9

Dr. Naim

Properties of an FIR Filter

Filter coefficients:
N 1

y n bk x n k
k 0

With:

Chapter 14, Slide 10

bk h k

The coefficients of a filter are the same as

the impulse response samples of the filter.

Dr. Naim

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

Chapter 14, Slide 11

n 0

Dr. Naim

Frequency Response of an FIR Filter

Since e-j2k = 1 then:

N 1

N 1

n 0

n 0

H z z e 2 h n e jn 2 h n e jn

Therefore:

H e j 2 k H e j

FIR filters have a periodic frequency

response and the period is 2 .

Chapter 14, Slide 12

Dr. Naim

Frequency Response of an FIR Filter

Frequency
response:

H e j 2 k H e j

FIR

y[n]

x[n]

y[n]

x[n]

Freq

Fs/2

Chapter 14, Slide 13

Freq

Fs/2

Dr. Naim

Frequency Response of an FIR Filter

Solution: Use an anti-aliasing filter.

x(t)

ADC

x[n]

FIR

y[n]

y[n]

x(t)

Analogue
Anti-Aliasing

Fs/2

Chapter 14, Slide 14

Freq

Fs/2

Freq

Dr. Naim

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
0

n+1

2n

2n+1

Even symmetry
Chapter 14, Slide 15

n
0

n-1

n+1

2n-1

2n

Odd symmetry
Dr. Naim

Phase Linearity of an FIR Filter

A causal FIR filter whose impulse response is symmetrical (ie h[n] = h[N-1n] for n = 0, 1, , N-1) is guaranteed to have a linear phase response.

Condition
h n h N n 1

Positive Symmetry

Chapter 14, Slide 16

Phase k

N 1

Phase Property

Filter Type
Odd Symmetry Type 1

Linear phase

Even Symmetry Type 2

Dr. Naim

Phase Linearity of an FIR Filter

Application of 90 linear phase shift:

I

Signal separation

Q
I A cos f t B sin r t
Q A sin f t B cos r t

Chapter 14, Slide 17

90

IH

delay

delay

90o

QH

IH A cos f t B sin r t
2
2

A sin f t B cos r t

Reverse

Forward

IH Q 2 B cos r t

QH I 2 B sin f t

Dr. Naim

Design Procedure

To fully design and implement a filter five steps are required:

(1)
(2)
(3)
(4)
(5)

Filter specification.
Coefficient calculation.
Structure selection.
Simulation (optional).
Implementation.

Chapter 14, Slide 18

Dr. Naim

Filter Specification - Step 1

|H(f)|

pass-band

stop-band

f c : cut-off frequency

f(norm)

f s /2
(a)

|H(f)|
(dB)

pass-band

transition band

|H(f)|
(linear)

stop-band

1 p

1
1 p

pass-band
ripple

-3

stop-band
ripple

s
f sb : stop-band frequency
f c : cut-off frequency
f pb : pass-band frequency

f s /2

f(norm)

(b)

Chapter 14, Slide 19

Dr. Naim

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.

Chapter 14, Slide 20

Dr. Naim

Window Method

First stage of this method is to calculate the coefficients of the ideal filter.
This is calculated as follows:

1
hd n
2

1
2

H e jn d

1 e jn d

2 f c sin n c

n c

2 fc
Chapter 14, Slide 21

for n 0
for n 0
Dr. Naim

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
Width (f(Hz))

Passband Ripple(dB)

Stopband Attenuation
(dB)

Rectangular

0.9
N

0.7416

21

Hanning

3.1
N

0.0546

44

Hamming

3.3
N

0.0194

53

Blackman

5.5
N

0.0017

74

0.0274

50

0.000275

90

Window Type

Kaiser

2.93
4.54
N
5.71
8.96
N

Using the Hamming

Window:
Chapter 14, Slide 22

3.3
3.3

8kHz 132
f 1.2 1.4 kHz
Dr. Naim

Window Method

The third stage is to calculate the set of truncated or windowed impulse

response coefficients, h[n]:

h n hd n W n

Where:

Chapter 14, Slide 23

for

N 1
N 1
n
for N odd
2
2
N
N
for N even
n
2
2

W n 0.54 0.46 cos

0.54 0.46 cos

2n

N
2n

133

for

66 n 66

Dr. Naim

Window Method

Matlab code for calculating coefficients:

close all;
clear all;
fc = 8000/44100;
N = 133;
n = -((N-1)/2):((N-1)/2);
n = n+(n==0)*eps;

% cut-off frequency
% number of taps

[h] = sin(n*2*pi*fc)./(n*pi);
[w] = 0.54 + 0.46*cos(2*pi*n/N);
d = h.*w;

% generate sequence of ideal coefficients

% generate window function
% 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)));
axis([0 2*10^4 -70 10]);

% plot transfer function

figure(2);
stem(d);
xlabel('Coefficient number');
ylabel ('Value');
title('Truncated Impulse Response');
figure(3)
freqz(d,1,512,44100);
axis([0 2*10^4 -70 10]);
Chapter 14, Slide 24

% avoiding division by zero

% plot coefficient values

% use freqz to plot magnitude and phase response

Dr. Naim

Window Method
Truncated Impulse Response

0.4
0.3

h(n)

0.2
0.1
0

Magnitude (dB)

-0.1

20

0.6

0.8
1
1.2
Frequency (Hz)

100

120

140

-20
-40
-60
0.2

0.4

0
Phase (degrees)

60
80
Coefficient number, n

1.4

1.6

1.8

2
4

x 10

-2000
-4000
-6000

Chapter 14, Slide 25

40

0.5

1
Frequency (Hz)

1.5

2
4

x 10

Dr. Naim

Realisation Structure Selection - Step 3

Direct form structure for an FIR filter:

N 1

H z bk z k
k 0

y n b0 x n b1 x n 1 .... bN 1 x n N 1

Y z H z X z

x(n)

-1

Chapter 14, Slide 26

-1

-1

N-1

y(n)

Dr. Naim

Realisation Structure Selection - Step 3

Direct form structure for an FIR filter:

N 1

H z bk z k
k 0

Linear phase structures:

N even:

N Odd:

N
1
2

H z bk z k z N k 1
k 0

N 1
2

H z bk z k z N k 1 b N 1 z
k 0

Chapter 14, Slide 27

N 1
2

Dr. Naim

Realisation Structure Selection - Step 3

(a) N even.
(b) N odd.

-1

-1

-1

-1

-1
b0

b1

b2

N/2-1

+
+

y(n)
(a)

x(n)

-1

-1

-1

-1

-1

-1
b0

b1

b2

+
+

b(N-1)/2

+
y(n)

(N-3)/2

+
(b)

Chapter 14, Slide 28

Dr. Naim

Realisation Structure Selection - Step 3

Direct form structure for an FIR filter:

N 1

H z bk z k
k 0

Cascade structures:
H z

N 1

k 0

bk z k b0 b1 z 1 b2 z 2 ... bN 1 z N 1

bN 1 N 1
b1 1 b2 2
b0 1 z z ...
z

b
b
b
0
0
0

b0

1 b
M

k 1

Chapter 14, Slide 29

1
2
z

b
z
k ,1
k ,2

Dr. Naim

Realisation Structure Selection - Step 3

Direct form structure for an FIR filter:

N 1

H z bk z k
k 0

Cascade structures:
x(n)

+
z -1

z -1
1,1

z -1

1,2

+
z -1

2,1

z -1
b

Chapter 14, Slide 30

M,1

y(n)

z -1
b

2,2

M,2

Dr. Naim

Implementation - Step 5

Implementation procedure in C with fixed-point:

Set up the codec (\Links\

[Link]).
).
(\Links\[Link]

Transform:

(\Links\[Link]
\Links\[Link]))
Configure timer 1 to generate an interrupt at 8000Hz (\Links\
[Link]).
).
(\Links\[Link]
Set the interrupt generator to generate an interrupt to invoke the Interrupt Service Routine (ISR) (\Links\
[Link]).
).
(\Links\[Link]

to C code.

N 1

y n bk x n k
k 0

Chapter 14, Slide 31

Dr. Naim

Implementation - Step 5

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\[Link]
\Links\[Link]

Chapter 14, Slide 32

Dr. Naim

Implementation - Step 5

Implementation procedure in assembly:

Same set up as fixed-point, however:

is written in assembly.
(\Links\[Link]
\Links\[Link]))

The ISR is now declared as external.

N 1

y n bk x n k
k 0

Chapter 14, Slide 33

Dr. Naim

Implementation - Step 5

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:

Set the initial value of the circular pointers, see \Links\[Link]

\Links\[Link]..

#pragma DATA_ALIGN (symbol, constant (bytes))

Chapter 14, Slide 34

Dr. Naim

Implementation - Step 5
b0
b1
b2
b3

+ b1*x1 + b2*x2 + b3*x3

y[n]

y0 = b0*x0

x0
x1
x2
x3

time

Circular addressing link slide.

Chapter 14, Slide 35

Dr. Naim

Implementation - Step 5
b0
b1
b2
b3
y0 = b0*x0

+ b1*x1 + b2*x2 + b3*x3

+ b1*x1 + b2*x2 + b3*x3

y[n]

y1 = b0*x4

x4
x1
x2
x3

time

Circular addressing link slide.

Chapter 14, Slide 36

Dr. Naim

Implementation - Step 5
b0
b1
b2
b3
y0 = b0*x0

+ b1*x1 + b2*x2 + b3*x3

+ b1*x1 + b2*x2 + b3*x3
+ b1*x5 + b2*x2 + b3*x3

y[n]

y1 = b0*x4
y2 = b0*x4

x4
x5
x2
x3

time

Circular addressing link slide.

Chapter 14, Slide 37

Dr. Naim

FIR Code

Code location:

Projects:

Fixed Point in C:
Floating Point in C:
Fixed Point in Assembly:

Floating Point in Assembly: \FIR_Asm_Float\

Chapter 14, Slide 38

Code\Chapter 14 - Finite Impulse Response Filters

\FIR_C_Fixed\
\FIR_C_Float\
\FIR_Asm_Fixed\

Dr. Naim

Chapter 14
Finite Impulse Response (FIR) Filters
- End -