Digital Signal Processing Reference
In-Depth Information
Tutorial 53
Q: Use MATLAB to design a fourth -order elliptic lowpass filter with the
following specifications:
Pass-band peak-to-peak ripple, R
p
= 0.6 dB
Minimum stop-band attenuation R
s
= 18 dB
The normalized pass-band-edge frequency w
p
= 0.63 Then, implement this filter
by changing the filter properties to fixed-point representation.
Solution: We may use the
dfilt
object for this purpose.
The first step is to design the filter using the Matlab default double-precision
format:
M
¼
4; Rp
¼ :
6; Rs
¼
18; wp
¼ :
63;
½
Den
;
Num
¼
ellip
ð
M
;
Rp
;
Rs
;
wp
Þ
;
Digital filter implementation using direct form—I:
H
¼
dfilt
:
df1
ð
Den
;
Num
Þ
Second step is to convert the filter coefficients to fixed point format:
H.Arithmetic
=
'fixed'
get(H)
H.FilterInternals
=
'SpecifyPrecision'
Note: ''SpecifyPrecision'' enables you to set your own wordlength and fraction
length for the output and accumulator-related properties.
Now set the hardware specifications:
set(H,'InputWordlength', 16,...
'InputFracLength', 13,...
'ProductWordLength', 24,...
'AccumulatorWordLength', 28,...
'OutputWordLength', 16);
H
Verify that the object
dfilt
has made an appropriate auto-scaling (Hint: check the
range of the filter coefficients).
Tutorial 54
Q: Verify the equivalence of the identities shown in the figure below. What is the
potential of these identities?
Search WWH ::
Custom Search