Digital Signal Processing Reference
In-Depth Information
þ
c
;
/
¼
X
2
þ
p
2
where c = p is to compensate for the change of sign in sin
ð
2
X
Þ:
The magnitude and phase responses are plotted in Fig.
2.28
. It is seen that the
digital differentiator is a high-pass filter.
Frequency Domain Approach
From Fourier transform Tables one can find the transfer function of the continu-
ous-time differentiator as:
H
a
ð
x
Þ¼
jx
ð
2
:
25
Þ
which can be transformed into the digital domain as follows:
H
ð
e
jX
Þ¼
jX
=
T
s
;
p
X\p
:
ð
2
:
26
Þ
Figure (
2.29
) shows the magnitude and phase responses of the above transfer
functions (compare with Fig. (
2.28
)). Note that Fig. (
2.29
a) represents a theoretical
magnitude response, as practical differentiators are band-limited within a cutoff
frequency x
c
.
The impulse response of the above filter can be found to be:
Z
p
h
ð
n
Þ¼
1
2p
jX
T
s
e
jnX
dX
p
ð
2
:
27
Þ
cos
ð
np
Þ
n
sin
ð
np
Þ
pn
2
¼
1
T
s
;
1
\n\
1
For a practical design, the above impulse response should be shifted by 2M ? 1
samples and then truncated. The shifting operation is described by:
h
ð
n
Þ¼
1
T
s
cos
½ð
n
M
Þ
p
ð
n
M
=
2
Þ
sin
½ð
n
M
Þ
p
p
ð
n
M
Þ
2
;
1
\n\
1;
ð
2
:
28
Þ
while the truncation window w(n) is specified by:
h
w
ð
n
Þ¼
w
ð
n
Þ
h
ð
n
Þ
;
0\n\2M
ð
2
:
29
Þ
If a linear phase is required, then the truncation window should be symmetric so
that one can get h(n) =-h(2M - n). The filter in (
2.29
) can be implemented
using the approach in
Sect. 2.6.4.6
.
Search WWH ::
Custom Search