DSPchapter 3
DSPchapter 3
we have
−n
x[n] z . (3.8)
n =−
3
So, ROC is determined by Eq.(3.8). Based on (3.8), we find that ROC
Figure 3.2: The ROC as a ring in the z-plane. For specific cases, the inner boundary can extend inward to the origin, and the
ROC becomes a disc. For other cases, the outer boundary can extend outward to infinity.
if a −1z 1 , or z a .
not exists.
1
▪ For a 1 , its FT exists and is X (e j ) = − j
. Note that it is the
1 − ae
same as Eq.(3.14).
7
Ex. 3.3: Sum of Two Exponential Sequences
1 1
x[n] = ( )n u[n] + ( − )n u[n]
2 3
1 −1 n 1 −1 n 1 1 2 z ( z − 121 )
X ( z ) = ( z ) + (− z ) = 1 −1
+ 1 −1 =
n =0 2 n =0 3 1− 2 z 1+ 3 z ( z − 12 )( z + 13 )
1 −1 −1 −1 1 1 1
if z 1 and z 1 ; i.e., z and z ; ROC is z
2 3 2 3 2
8
Figure 3.5: Pole–zero plot and ROC for the individual terms and the sum of terms in Examples 3.3 and 3.4.
(a) 1/(1 − 1/2z−1), |z| > 1/2. (b) 1/(1 + 1/3z−1), |z| > 1/3. (c) 1/(1 − 1/2z−1) + 1/(1 + 1/3z−1), |z| > 1/2.
9
Ex. 3.5: Two-sided Exponential Sequence
1 1
x[n] = ( − )n u[n] − ( ) n u[−n − 1]
3 2
1 1 2 z( z − 121 ) 1 1
X ( z) = + = , if z and z
1 + 13 z −1 1 − 12 z −1 ( z + 3 )( z − 2 )
1 1
3 2
1 1
So, its ROC is the ring z .
3 2
10
Table 3.1 lists some common z-transform pairs
11
3.2 The properties of ROC
annulus, i.e., 0 rR z rL
2. The FT of x[n] converges absolutely iff the ROC of its z-transform includes
the unit circle.
3. ROC can not contain poles.
4. The ROC of the z-transform of a finite-duration sequence is the whole
z-plane except it may not contain z = 0 or . Specifically, let x(n)≠0
12
5. The ROC of a right-sided sequence is z Pi , where Pi is the largest pole.
starts at n = 0.
6. The ROC of a left-sided sequence is z Pj , where Pj is the smallest pole.
The z-transform accompanying with ROC can uniquely specify a sequence. So,
a pole-zero pattern associated with different ROC’s will specify different time
sequences.
13
Figure 3.8 Examples of four z-transforms with the same pole–zero locations, illustrating the
different possibilities for the ROC, each of which corresponds to a different sequence: (b) to a
right-sided sequence, (c) to a left-sided sequence, (d) to a two-sided sequence, and (e) to a
two-sided sequence. 14
For an LTI system, the z-transform of its impulse response h[n] is called
the system function H(z). Some properties of a system can implicitly
determine the form of its ROC, e.g.
‧ stable ROC contains the unit circle FT exists
‧ causal ROC is of the form z Pi right-sided sequence
15
Practically, we use the inspection method, partial fraction method, and
power series expansion to calculate the inverse z-transform.
1
−a n u[−n − 1] ⎯
Z
→ , z a
1 − az −1
P( z ) k
b z −k
X ( z) = = k =0
N (expressed in the polynomial form of z-1)
a
Q( z )
k z −k
k =0 16
M
Z N
bz k
M −k
= k =0
N (N poles and M zeros)
Z M ak z N − k
k =0
bo (1 − c z k
−1
)
= k =1
N
(1 − d z
ao −1
k )
k =1
Ak = (1 − dk z −1 ) X ( z) z =d k
17
Example 3.9: 2nd-oder z-transform
1 1
X ( z) = z (3.44)
(1 − z 1
4
−1
)(1 − z 1
2
−1
) 2
A1 A2 −1 2
X ( z) = + = +
(1 − 1
4
z −1
) (1 − 1
2
z −1
) (1 − 1
4
z −1
) (1 − 1
2
z −1 )
1 1
x[n] = 2( )n u[n] − ( )n u[n]
2 4
k =1 1 − d k z
−1
where
1 d s −m −1
Cm = s −m [(1 − d i w) X ( w )]
s
( s-m)!( −d i ) s −m dw w=di−1
19
Ak
Ak (d k ) u[n] if d k inside the inner boundary of ROC
n
1 − d k z −1 − Ak (d k ) u[−n − 1] if d k outside the outer boundary of ROC
n
▪
Cm
▪ −1 m
can be obtained based on the property of “Differentiation
(1 − di z )
, |z|>1
A1 A2
= B0 + +
1- 12 z −1 1-z −1
20
where B0 = 2 is obtained by the long-division,
5z −1-1 10 − 1
A1 = = = −9
1 − z −1 z=
1 1− 2
2
5z −1-1 5 −1
A2 = = =8
1 − 12 z −1 z =1
1− 2
1
−1
Since 2 ⎯⎯
Z
→ 2 [n] ,
1 Z −1 1 n 1
1 −1
⎯⎯ → ( ) u[n] |z|>1> ,
1- 2 z 2
2
1 z−1 → u[n]
⎯⎯ |z|>1
1 − z −1
1
We hence obtain x[n] = 2 [n] − 9( ) n u[n] + 8u[n]
2
21
Figure 3.11: Pole–zero plot for the z-transform in Example 3.10.
22
Ex. 3.11: Finite-Length Sequence
1
X ( z ) = z 2 1 − z −1 (1 + z −1 )(1 − z −1 ) (3.52)
2
1 1
=z2- z-1+ z-1
2 2
1 n = −2
1
− 2 n = −1
x[n ]= −1 n=0
1 n =1
2
0 otherwise
1 1
Equivalently, x[n] = [n + 2] − [n + 1] − [n] + [n − 1]
2 2
Sometimes, we can use mathematical formula to find the power series expansion
of X(z).
23
Ex. 3.12: Inverse Transform by Power Series Expansion
X(z)=log (1+az-1) |z|>|a|
n +1
( −1) x n
From log (1+x)= for |x|<1, we have
n =1 n
n +1
( −1) a n z − n
X ( z) =
n =1 n
n +1 a
n
( −1) n 1
x[n ]= n
0 n0
25
3.4.1 Linearity
ax1[n]+bx2[n] ⎯→
Z
aX1[z] +bX2[z] ROC: Rx Rx
1 2
If some poles of x1(z) or x2(z) are cancelled out in the combination, the
resulting ROC may be enlarged. (because ROC is bounded by poles)
26
3.4.3 Multiplication by an exponential sequence
z0n x[n] ⎯
Z
→X( z ) ROC=|z0| Rx (i.e., Rx is scaled by |z0|)
z0
z=z1 z0.
left-rotated by 0 .
▪ If FT exists, then we have e j nx[n] ⎯
0 F
→ X (e j (− )) .
0
27
x[n] ⎯→ X ( z ) =
Z
n =−
x[n]z −n
= Z{x[n]}
n =−
n
z x[n]z
0
−n
= x[n]( zz
n =−
−1 − n
0 ) = X ( zz ) −1
0
z0n x[n] ⎯
Z
→X( z )
z0
If X ( z ) has a pole ( zero ) at z = z1 ,
then X ( z ) has apole ( zero ) at z = z1 z0 .
z0
X ( z1 ) = 0
z1 z0
X( z ) z = z1 z0 = X( ) = X ( z1 ) = 0
z0 z0
z = z1 z0 is a zero of X ( z )
z0 28
3.4.4 Differentiation of X(z)
dX ( z )
nx[n] ⎯
Z
→− z ROC=Rx
dz
nx[n]=a(–a)n-1u[n –1] 1
a n u[n ] ⎯
Z
→ , z a
n 1 − az −1
n-1 a
∴ x[n]=(–1) u[n –1]
n
29
Ex. 3.17: 2nd-Order Pole
x[n] = n an u[n] = n (an u[n])
−1
d 1 az
X ( z) = − z ( )= , |z|>|a|
dz 1-az −1 (1-az −1)2
If some poles are canceled out in the product, then ROC may be enlarged.
x1[n] x2 [n] ⎯
Z
→ X1 ( z) X 2 ( z)
−n
−n
1 2
n =− k =−
x [ k ] x [ n − k ]
z = 1 n
k =−
x [ k ]
=−
x 2 [ n − k ] z
− ( n '+ k )
−k
−n'
= x1[k ] x2 [n ']z 1
= x [ k ] z 2 x [ n '] z
k =− n '=− k =− n '=−
= X1 ( z) X 2 ( z)
response is h[n ] .
Example 3.20: Convolution of Infinite-Length Sequences
Let h[n] = anu[n] and x[n] = Au[n] . Find y[n] .
1
H ( z) = a n z −n = z a
n =0 1 − az −1
A
X ( z ) = Az − n = , z 1
n =0 1 − z −1
A Az 2
Y ( z) = = z 1
(1 − az −1
)(1 − z −1
) ( z − a)( z − 1)
33
A 1 a
Y ( z) = − z 1
1 − a (1 − z ) (1 − az )
−1 −1
A
y n = (1 − a n +1 )u n
1− a
34
When an LTI system can be expressed by a difference equation of the form
N
ak M
bk
y n = − y n − k + x n − k (3.66)
k =1 a0 k = 0 a0
with the input being zero prior to n=0 and initial rest condition being
imposed prior to the time when the input becomes nonzero, we have
N
ak − k M
bk − k
Y ( z ) = − z Y ( z ) + z X ( z ) (3.67)
k =1 a0 k =1 a0
M −k
bk z
Y ( z ) = kN=0 X ( z)
Solving for Y(z) to obtain a z −k (3.68)
k
k =0
35
M
b z k
−k
And we have H ( z ) =
k =0
N (3.69).
a z
k =0
k
−k
N M
36
The time-shift property is different in the unilateral case because the lower
limit in the unilateral transform definition is fixed at zero.
▪ To illustrate the difference, we let y[n] = x[n −1] . Then
Y ( z ) = x n − 1 z − n .
n =0
−1
So, Y ( z ) = x[−1] + z X ( z ) . In general, if y[n] = x[n − k ], where k 0,
then
Y ( z ) = x[−k ] + x[−k + 1]z −1 + + x[−1]z − k +1 + z − k X ( z )
k
= x m − k − 1 z − m+1 + z − k X ( z )
m =1
37