Digital Signal Processing Reference
In-Depth Information
It is common when specifying digital filters to give the filter-transfer function
or the difference equation. For example, the block-diagram specification for
the -filter of the last example can be any of the four forms illustrated in Figure 11.5.
The representation of the time delay by the transfer function
H(z)
a
z
-1
is based on the
property
z
[y[n - 1]] = z
-1
z
[y[n]] = z
-1
Y(z).
Three procedures for finding the inverse
z
-transform will now be discussed. The
first procedure involves the use of the complex inversion integral, from (11.2):
1
2p j
C
≠
F(z)z
n- 1
dz.
f[n] =
This integration is in the complex plane and is usually too complicated to be of prac-
tical value; hence, we will not use this approach. (See Ref. 3 for applications.)
The second method for finding the inverse
z
-transform is by partial-fraction
expansion, in which a function that does not appear in the
z
-transform tables is ex-
pressed as a sum of functions that do appear in the tables. Partial-fraction expan-
sions are presented in Appendix F; those readers unfamiliar with this topic should
read this appendix. We now illustrate the use of partial fractions to find inverse
z
-transforms.
Inverse
z
-transform by partial-fraction expansion
EXAMPLE 11.9
We solve for the time response of the
a
-filter of Example 11.8, for the unit step input. From
Table 11.2,
X(z) = z/(z - 1).
From (11.45), the transformed output is given by
0.1z
z - 0.9
z
z - 1
.
Y(z) = H(z)X(z) =
We expand
Y(z)
/
z
in partial fractions:
Y(z)
z
k
1
z - 0.9
+
k
2
z - 1
.
0.1z
(z - 0.9)(z - 1)
=
=
Then (see Appendix F)
0.1z
(z - 0.9)(z - 1)
0.1z
z - 1
B
R
k
1
= (z - 0.9)
z= 0.9
=
z= 0.9
=-0.9
and
0.1z
(z - 0.9)(z - 1)
0.1z
z - 0.9
B
R
k
2
= (z - 1)
z= 1
=
z= 1
= 1.