Z Transforms
Z Transforms
we will restrict access to material only to the DUT community: academic staff, staff
related to teaching & learning and students enrolled in courses. Students are advised
to limit their use of material provided to them only for the purpose of their own private
study, and refrain from sharing with persons not enrolled for the class. We will maximize
the full extent of compliance in accordance with Copyright law.
Z-Transforms
1 Introduction
In certain physical systems, the functions which describe the system are discrete. A Digital
System is one such example.
For discrete functions, both the domain and range consist of discrete values.
The z-transform, transforms sequences of discrete values and is the analogous discrete transform
to the Laplace Transform of piecewise continuous functions.
1. If xk = 3k , k ∈ Z, then
{xk } = {· · · , 3−3 , 3−2 , 3−1 , 30 , 31 , 32 , 33 , · · · }
= {· · · , 312 , 312 , 31 , 30 , 31 , 32 , 33 , · · · }
1 1 1
= {· · · , 27 , 9 , 3 , 1, 3, 9, 27, · · · }
and
{xk · uk } = {· · · , 0, 0, 0, 0, x0 , x1 , x2 , x3 , x4 , · · · }
{xk } = {x0 , x1 , x2 , x3 , x4 , · · · }
is the sequence consisting of the elements of the causal sequence {xk }, shifted c places to the
left , and
0 , 0 , · · · , 0 , x0 , x1 , x2 , · · · }
{xk−c } = {|{z}
x−c
k=0
consistes of the elements of the causal sequence {xk } shifted c places to the right.
Example
If xk = 2−k then
{xk } = {1 , 2−1 , 2−2 , 2−3 , 2−4 , 2−5 , · · · }
{xk+3 } = {2−3 , 2−4 , 2−5 , · · · }
{xk−3 } = {0, 0, 0, 1, 2−1 , 2−2 , 2−3 , · · · }
Using our knowledge of the sum of a geometric series, the z-transforms of most sequences can
be expressed in a closed form F (z), i.e
Z {xk } = X(z) where |z| > c or |z| < c for some constant c ∈ R
The inequality |z| > c or |z| < c for some constant c ∈ R, is referred to as the Radius of
Convergence (ROC).
∞
X xk
The radius of convergence is the region in the complex plane in which the infinite sum
k=0
zk
converges to the sum X(z).
Recall that
∞
X 1 − rk 1
rk = 1 + r + r2 + r3 + · · · = lim = = (1 − r)−1 provided |r| < 1
k→∞ 1 − r 1−r
k=0
The table on page 4, shows a list of Standard Forms of z-transforms which can be derived
from First Principles.
We will now use first principles to prove some of the above standard forms;
1. Z {δk } = 1
0 0 0
Z {δk } = 1 + + 2 + 3 + ···
z z z
=1 for all z
z
{uk } = {1, 1, 1, 1, 1 · · · } |z| > 1
z−1
cz
{c} = {c, c, c, · · · } (c is a constant) |z| > 1
z−1
z
{k} = {0, 1, 2, 3, 4, · · · } |z| > 1
(z − 1)2
z(z + 1)
{k 2 } = {0, 12 , 22 , 32 , · · · } |z| > 1
(z − 1)3
z(z 2 + 4z + 1)
{k 3 } = {0, 13 , 23 , 33 , · · · } |z| > 1
(z − 1)4
z
{ck } = {1, c, c2 , c3 , · · · } (c is a con- |z| > |c|
z−c
stant)
cz
{kck } = {0, c, 2c2 , 3c3 , · · · } (c is a con- |z| > |c|
(z − c)2
stant)
z
{e−ck } = {1, e−c , e−2c , e−3c , · · · } |z| > e−c
z − e−c
z(z − cos ω)
{cos ωk} = {1, cos ω, cos 2ω, · · · } |z| > 1
z 2 − 2z cos ω + 1
z sin ω
{sin ωk} = {0, sin ω, sin 2ω, · · · } |z| > 1
z 2 − 2z cos ω + 1
z(z − cosh ω)
{cosh ωk} = {1, cosh ω, cosh 2ω, · · · } |z| > e|ω|
z2 − 2z cosh ω + 1
z sinh ω
{sinh ωk} = {0, sinh ω, sinh 2ω, · · · } |z| > e|ω|
z 2 − 2z cosh ω + 1
z 1
= provided < 1 ⇒ |z| > 1
z−1 z
z
3. Z {k} = where |z| > 1
(z − 1)2
∞
X k 1 2 3
Z {k} = k
= 0 + + 2 + 3 + ···
k=0
z z z z
2 3 !
1 1 1 1
= 1+2 +3 +4 + ···
z z z z
" #
1 1 1
= where < 1 (see (*) in the note below)
z 1 2 z
1− z
z
= where |z| > 1
(z − 1)2
Note: Differentiating both sides of the identity of the sum of the infinite geometric series
yields;
d 1 d
= (1 + r + r2 + r3 + · · · ) where |r| < 1
dr 1−r dr
1
2
= 1 + 2r + 3r2 + 4r3 + · · · where |r| < 1 (∗)
(1 − r)
z
4. Z {ck } = where |z| > |c|
z−c
1 c
= c where <1
1− z
z
z
= where |z| > |c|
z−c
z
5. Z {e−ck } = , where |z| > e−c
z − e−c
∞ −ck
X e e−c e−2c e−3c
Z {e −ck
}= =1+ + 2 + 3 + ···
k=0
zk z z z
1 2 3
e−c e−c e−c
=1+ + + + ···
z z z
1 e−c
= −c where <1
1 − ez z
z
= where |z| > e−c
z − e−c
z(z − cos ω)
6. Z {cos ωk} = , where |z| > 1
z 2 − 2z cos ω + 1
∞ ∞
Re{ejωk }
X
X cos ωk
Z {cos ωk} = =
k=0
zk k=0
zk
z(z − cos ω)
= where |z| > 1
z2 − 2z cos ω + 1
Exercises
Use the definition of the z transform to find
z sin 5
1. Z {sin 5k} ans: , 1 < |z|
z2 − 2z cos 5 + 1
z
2. Z {e4k } ans: 4
, e4 < |z|
z−e
z sinh ω
1. Use the linearity property to show that; Z {sinh ωk} =
z2 − 2z cosh ω + 1
where |z| > e|ω|
eωk − e−ωk
Z {sinh ωk} = Z 1
Z {eωk } − Z {e−ωk }
= 2
Linearity property
2
z z
= 1
2
− where |z| > |eω | and |z| > |e−ω |
z−e ω z − e−ω
z sinh ω
= where |z| > e|ω|
z2 − 2z cosh ω + 1
2. Find Z {3 · 4 } + {2 · k}
k
Z {3 · 4 } + {2 · k}
k
= Z 3{4 } + 2{k}
k
= 3Z {4k } + 2Z {k}
z z
=3 +2
z−4 (z − 1)2
Exercises
ejx −e−jx ejx +e−jx
1. Use the identities j sin x = sinh jx = 2
and cos x = cosh jx = 2
, and the
linearity property to find
2. Use the definition of cosh ωk, and the linearity property to find Z {cosh ωk}
Proof:
∞
X xk+m
Z {xk+m } =
k=0
zk
Example
If {xk } = {k} = {0, 1, 2, 3, 4, · · · }, then X(z) = Z {k} = z
(z−1)2
{xk+4 } = {k + 4} = {4, 5, 6, 7, · · · }, i.e the sequence, {xk } = {k} shifted 4 places to the left
(advanced 4 places). Therefore, using the First Shift Theorem,
Z {k + 4} = Z {xk+4 } = z 4 X(z) − [z 4 x0 + z 3 x1 + z 2 x2 + z x3 ]
z
= z4 2
− [z 4 (0) + z 3 (1) + z 2 (2) + z(3)]
(z − 1)
z5
= − z 3 − 2z 2 − 3z
(z − 1)2
4z 2 − 3z
= |z| > 1
(z − 1)2
Exercises
Use the First Shift Theorem to find
1. Z {2k+3 } ans: 8z
z−2
= z −m X(z)
Example
z
1. If {xk } = {5k }, then X(z) = Z {5k } = where |z| > 5
z−5
Therefore,
Z {xk−3 } = z −3 X(z)
z
= z −3 ·
z−5
1
= where |z| > 5
z 2 (z− 5)
Z {xk−5 } = z −5 X(z)
z sinh 3
= z −5 · where |z| > e3
z2 − 2z cosh 3 + 1
sinh 3
= where |z| > e3
z 4 (z 2 − 2z cosh 3 + 1)
Exercise
Proof
∞
X ck x k
Z {c xk } =
k
k=0
zk
c0 x0 cx1 c2 x2 c3 x3
= + + 2 + 3 + ···
z0 z z z
x0 x1 x2 x3
= 0 + 1 + 2 + 3 + · · ·
z z z z
c c c c
z
=X c
Example
z
If Z {xk } = Z {k} = , then
(z − 1)2
Z {2k k} = X z
2
z
2
= 2
z
2
−1
2z z
= where > 1 ⇒ |z| > 2
(z − 2)2 2
Exercise
3z(z + 3)
1. Find Z {3k k 2 } ans: , |z| > 3
(z − 3)3
z 2 − 4z cos 2
2. Find Z {4k cos 2k} ans: , |z| > 4
z 2 − 8z cos 2 + 16
= Z {k ck }
Exercises
e2 z
1. Find Z {k e }2k
ans:
(z − e2 )2
z
2. Find Z {k cos(π k)} ans: −
(z + 1)2
Proof
lim X(z) = lim Z {xk }
z→∞ z→∞
∞
X xk
= lim
z→∞
k=0
zk
x x1 x2 x3
0
= lim 0
+ + 2 + 3 + ···
z→∞ z z z z
= x0
=1
Exercise
3z 2z 2
Find x0 if Z = + 2 ans: x0 = 5
z−4 z +1
z−1 z z−1 2z
lim · 1 = lim ·
z→1 z z− 2
z→1 z 2z − 1
2(z − 1)
= lim
z→1 2z − 1
0
= =0
1
Exercise
3z 2z 2
Find x0 if Z {xk } = + 2 ans: x0 = 5
z−4 z +1
( ∞
)
X
{xk } ∗ {yk } = xn yk−n
n=−∞
∞
! ∞
! ∞
!
X X X
= xn y0−n , xn y1−n , xn y2−n , · · · · · ·
n=−∞ n=−∞ n=−∞
| {z } | {z } | {z }
k=0 k=1 k=2
since for causal sequences, xk = 0 for k < 0 and xk−n = 0 for n > k
Therefore, the convolution product can be re-stated as;
( k )
X
{xk } ∗ {yk } = xn yk−n
n=0
( 0
! 1
! 2
! k
!)
X X X X
= xn y0−n , xn y1−n , xn y2−n , ··· , xn yk−n
n=0 n=0 n=0 n=0
Example
If {vk } = {k} ∗ {2k }, then
(∞ ) ( k )
X X
{vk } = {k} ∗ {2k } = n 2k−n = n 2k−n
n=0 n=0
· 2}0 , 0| · 21 {z
= {0| {z + 1 · 2}0 , 0| · 22 + 1 {z
· 21 + 2 · 2}0 , · · · , 0| · 2k + 1 · 2k−1 0
{z + · · · + k · 2}}
u0 u1 u2 vk
= {0, 1, 4, 11, · · · , vk , 0, 0, 0, · · · }
Convolution Theorem
If {xk } and {yk } are sequences, then
∞ ∞
!
X X
= xn yk−n z −k
k=0 n=0
∞
X ∞
X
= xn yk−n z −k
n=0 k=0
∞ ∞
!
X X
= xn z −n yk z −k (second shift theorem)
n=0 k=0
∞
! ∞
!
X X
= xn z −n yk z −k
n=0 k=0
= X(z) · Y (z)
Example
3 Inverse z Transform
If Z {xk } = X(x), then Z −1 (X(z)) = {xk }
We will consider three techniques for finding inverse z transforms; [Long Division
Partial Fractions and Convolution,].
−12 + z − 96
72
z2
0 2 12 56 240 − z + 96
56
=Z −1
− − 2 − 3 − 4 − ··· z2
z0 z z z z − 56 + 336
− 448
z z2 z3
240 448
= {0, −2, −12, −56, −240, · · · } − z2
+ z3
The long division method produces numerical values of the sequence. However, it is not always
possible to establish the closed form of the sequence. In fact the sequence found above is the
sequence {2k − 4k } = {2k } − {4k }, but this is not obvious from the sequence.
Example 1 + z2 + z42 + 8
z3
+ 16
z4
+ ···
z z−2 z
Find Z −1 ;
z−2 z−2
2
z
Z −1 2 − z4
z−2 4
z
1 2 4 8 16 4
− 8
=Z −1
+ + + + + ··· z z2
z0 z z2 z3 z4 8
z2
8 16
= {20 , 21 , 22 , 24 , 25 , · · · } z2
− z3
16
z3
= {2k }
Exercises
Use long division to find the first 5 terms of the following inverse transforms
3
z − 2z 2 − 3
1. Z −1
ans: {1, −2, −1, 1, −3 · · · }
z3 + z − 2
z2
2. Z −1
ans: {0, 1, 3, 9, 23, · · · }
z 3 − 3z 2 + 4
In this example, the method of partial fractions was applied to X(z), and then each fraction
had to be multiplied by zz in order to get the fractions in standard form. However, a simpler
way of dealing with Z −1 (X(z)) is as follows;
−2z
Let X(z) =
z 2 − 6z + 8
X(z) −2
Then, = (divide through by z)
z (z − 2)(z − 4)
1 1
= − (by partial fractions)
z−2 z−4
z z
∴ X(z) = − , (multiply both sides by z), and then
z−2 z−4
−2z
Z (X(Z)) = Z
−1 −1
z 2 − 6z + 8
z z
=Z −1
−Z −1
z−2 z−4
z2
Let X(z) = 2
z +4
X(z) z
=
z (z − j2)(z + j2)
1 1
X(z) 2
= + 2 (partial fractions)
z z − j2 z + j2
1 z z
X(z) = +
2 z − j2 z + j2
z2
1 z z
∴ Z −1
=Z −1
+
z2 + 4 2 z − j2 z + j2
[(j2)k + (−j2)k ]
1
= 2
π π π π
= { 12 [(2ej 2 )k + (2e−j 2 )k ]} (ja = aej 2 , and − ja = ae−j 2 )
Example
z
Find Z −1
2
z + 6z + 6
z
Let X(z) = Z −1
2
z + 6z + 6
z 2 + 6z + 6
X(z) 1
= 2 =(z + 3)2 − √3 (diff. of squares)
√
z z + 6z + 6
=((z + 3) + 3)((z + 3) − 3)
X(z) 1
= √ √
z (z + (3 + 3))(z + (3 − 3))
X(z) − 2√1 3 1
√
2 3
= √ + √ (partial fractions)
z z + (3 + 3) z + (3 − 3)
Example
z
Find Z −1
√
z 2 − 3z + 1
z
Let X(z) = √
z2 − 3z + 1
√
z 2 − 3z
√ +1
X(z) 1 =(z − √23 )2 + 14
= √ √
=(z − 2√3 )2 − − 14 (diff.
z (z − ( 23 + j 21 ))(z − ( 23 − j 12 )) √
of squares)
3 3
=((z − 2 ) + j 2 )((z − 2 ) − j 12 )
1
!
X(z) 1 1 1
= √ − √ (partial fractions)
z j z− ( 23 + j 21 ) z−( 3
− j 12 )
2
!
1 z z
X(z) = √ − √
j z−( 3
+ j 21 ) z−( 3
− j 12 )
2 2
" #!
z 1 z z
Z −1 √ = Z −1 √ − √
z2 − 3z + 1 j z−( 3
+ j 21 ) z−( 3
− j 12 )
2 2
√ √
!k !k
1 3 1 3 1
= +j − −j
j 2 2 2 2
1 jπ k π
= (e 6 ) − (e−j 6 )k
j
1
= · {2 sinh(j π6 k)}
j
= {2 sin( π6 k)}
X(z) z
=
z (z + 1)(z − 2)2
− 19 1
9
2
3
= + + (partial fractions)
z + 1 z − 2 (z − 2)2
1 z 1 z 2 z
∴ X(z) = − · + · + ·
9 z + 1 9 z − 2 3 (z − 2)2
z2
1 z 1 z 2 z
Z −1
=Z −1
− · + · + ·
(z + 1)(z − 2)2 9 z + 1 9 z − 2 3 (z − 2)2
= − 19 (−1)k + 91 (2)k + 13 (k · 2k )
Exercises
Use partial fractions to find the following inverse z transforms
7z n k o
1. Z −1
ans: 1 k
3
[2 − − 13 ]
9z 2 − 15z − 6
6z 2 − 15z
2. Z −1
1
ans: 2
[4k − uk + 3k]
2(z − 4)(z − 1)2
2z 2 − z
3. Z −1 π
ans: 2 cos 3
k
z2 − z + 1
3.3 Convolution
z2
Use the convolution theorem to find Z −1
z2 + z − 6
z2 z2
Z −1
=Z −1
z2 + z − 6 (z + 3)(z − 2)
z z
=Z −1
·
z+3 z−2
= {(−3)k } ∗ {(2)k }
−3 k+1
( " #)
−1
= 2k 2
−3
(sum of a finite geometric series (k + 1 terms))
2
−1
( " " k+1 ##)
2 −3
= 2k − −1
5 2
( " " k+1 ##)
1 k+1 −3
= 2 1−
5 2
1 k+1
− (−3)k+1
= 5
2
z2
z 1
Z −1
=Z −1
·
(z − 4)(z − 5) z (z − 4)(z − 5)
1 z z
=Z −1
· ·
z z−4 z−5
40 41 4k−1
k 0
= 5 + + + ··· + k (see note below)
50 51 52 5
4 k
( " #)
−1
= 5k−1 5
4
(k terms in the series)
5
−1
= {5k − 4k }
Note: By the second shift theorem, {xk−1 } = {4k−1 }. Since the sequence is causal, when
k = 0, x0−1 = x−1 = 0. Thus, {4k−1 } = {0, 40 , 41 , 42 , 43 , · · · } and {5k } = {50 , 51 , 52 , 53 , · · · }
Exercises
Use the convolution theorem to find;
2z 2
1. Z −1
2
2
ans: 3
[(5)k+1 − (2)k+1 ]
z − 7z + 10
z2
2. Z −1
1 k 2
2
ans: 3
(2) + 3 (−4)k
z + 2z − 8
1) Take z transforms of both sides, making use of the second shift theorem
X(z) z−1
=
z (z − 2)2 (z + 5)
6 1 6
X(z) 49 7 49
= + − (partial fractions)
z z − 2 (z − 2)2 z + 5
6 z 1 z 6 z
∴ X(z) = · + · 2
− ·
49 z − 2 7 (z − 2) 49 z + 5
6 z 1 1 2z 6 z
X(z) = · + · − ·
49 z − 2 7 2 (z − 2)2 49 z + 5
{xk } = Z −1 (X(z)) = { 49
6
· 2k + 1
14
· k 2k − 6
49
· (−5)k }
Y (z) 2z − 1
= =
z (z − 2)(z + 2)(z − 1)
3
Y (z) −5 − 13
= 4 + 12 +
z z−2 z+2 z−1
3 z 5 z 1 z
Y (z) = · − · − ·
4 z − 2 12 z + 2 3 z − 1
∴ {yk } = 34 (2)k − 12
5
(−2)k − 31 uk
Exercises
1. Verify the answers in the above two examples by generating the sequences for each problem
using
(a) the given recurrence relation, and
(b) the closed form of the sequence.
− 29 (−2)k + 29 uk + 31 k
2. Solve xk+2 + xk+1 − 2xk = uk , where x0 = 0 and x1 = 1 ans:
3. Solve qk+1 + 2qk = 3 cos(π k), where q0 = 1 ans: 3(−1)k − 2(−2)k
This suggests that if we let z = esT , there is an obvious link between the Laplace Transform
and the z Transform of a sampled function as follows;
∞
X
L {fk (t)} = f (kT )e−ksT
k=0
∞
X
= f (kT )z −k
k=0
∞
X f (kT )
=
k=0
zk
= Z {f (kT )}
This shows that the Laplace Transform of a continuous time sampled function fk (t), is equal
to the z transform of the sequence of the sampled discrete function values.
Example
A continuous function f (t) = t2 is sampled every 3 seconds. Then, T = 3 and the sampled
function is;
∞
X
L {fk (t)} = fk (3k) e−3ks
k=0
∞
X
= f (3k) z −k ( where z = e−3s )
k=0
∞
X (3k)2
=
k=0
zk
∞
X k2
=9
k=0
zk
9z(z + 1)
=
(z − 1)3
= Z {9k 2 }