EEE504:

DIGITAL SIGNAL PROCESSING (DSP)

LECTURE NOTES
z-TRANSFORM

By:
Olorunniwo O.
Dept. of Electronic & Electrical Engineering
Obafemi Awolowo University
Ile-Ife

1|Page
1. Introduction : Review of Laplace Transforms
2. z-Transform
• Derivation and Interpretation;
• Inverse z-Transform;
• Properties of z-Transform;
• Frequency Response using z-Transform;

1. INTRODUCTION: Review of Laplace Transforms

Laplace transform is a useful tool in analyzing physical systems with continuous-time signals. Laplace is a well
established mathematical technique for solving differential equations. Moreover, the Laplace transform
changes a signal time domain into s-domain or s-plane. In other words, Laplace transform allows the time
domain to be complex—represented as s + jσ .
Therefore, for any signal x(t ) to be transformable,
∞
X (s ) = ∫ x(t )e dt = L{x(t )} must be satisfied.
− st

−∞
And the inverse Laplace transform is written as,
σ + j∞
1
2πj σ −∫j∞
x(t ) = X (s )e st ds .

Figure 1 illustrates the transformation of a time-domain signal into frequency and s-domain, while Table 1
shows some important Laplace transform pairs.

Figure1: Transformation of a time-domain signal into frequency and s-domain

Why is Laplace transform of importance? The introduction of Laplace transform into electronic circuit
analysis allows for easier manipulation of integro-differential equations and the resulting waveforms,
culminating into design of analog filters.

2|Page
 Table 1

2 z-Transform: Derivation and Interpretation

The z-transform is useful tool in the analysis of discrete-time signals and systems. And, just as analog
filters are designed by using the Laplace transform, recursive digital filters are developed by the z-
transform technique. The overall strategy of these two transform are similar: evaluate the impulse
response with sinusoids and exponentials to determine the poles and zeros. However, the
Laplace transforms deal with differential equations, in the s-domains and the s-plane, while the
corresponding z-transform analyzes difference equations in the z-domain and z-plane. Figure 2
illustrates the difference between both domains.

3|Page
 Figure2: Relationship between the s-plane and the z-plane

The s-plane is a rectangular coordinate system with σ along the real axis, and ω along the imaginary
axis. In comparison, the z-plane is in polar form, with r distance from the origin, and ω is the angle
measured to the positive (+) axis. Vertical lines of poles and zeros in the s-plane correspond to
circles in the z-plane.
Hence, the z-transform of a sequence x[n] is defined as
∞
X (z ) = ∑ x[n]z
n = −∞
−n
= Z {x[n]}
jω
where, z = re .
What does it imply? This implies that the z-transform is taken to be an infinite sum, or an infinite
power series—with z being a complex variable. Therefore, it is convenient to describe and interpret
z = 1 as a circle of unit radius (See Figure3).

Figure 3: The unit circle in the z-plane

For any given sequence, the set of values of z for which the z-transform converges is called the
region of convergence (ROC). Further, Table 2 refers to some common pairs in z-transform.

4|Page
 Table 2

2.1 The Transfer Function for a Discrete System

If X (s ) is the Laplace transform of the input to a continuous system and Y (s ) the Laplace transform of
the corresponding output, then the system transfer function, T (s ) is:
Y (s )
T (s ) =
X (s )
Similarly, if the z-transform of the input sequence to a discrete system is P ( z ) and the corresponding
output sequence is Q(z ) , then the transfer function, X (z ) , of the system is:
P( z )
X (z ) =
Q( z )
Where, P ( z ) and Q (z ) are polynomials in z. The values of z for which X (z ) = 0 are called the zeros of
X (z ) , and the values of z for which X (z ) is infinite are referred to as the poles of X ( z ) . Furthermore, the
poles of X (z ) for finite values of z are the roots of the denominator polynomial.
Note: The z-transform is most useful when the infinite sum can be expressed in close (rational) form.
Three examples are illustrated: (1) the right-sided sequence (2) the left-sided sequence and (3) a finite-
length sequence.

5|Page
 2.1.1 Right-Sided Sequence
Consider a signal represented as the signal x[n] = a u[n] , this is referred to as a right-sided
n

sequence, since the sequence is non-zero for n ≥ 0 . That is,

∞ ∞
X (z ) = ∑ a u[n]zn −n
(
= ∑ az −1 )−n

n = −∞ n=0

The infinite series converges to the rational form,

1 z for ∞
X (z ) = −1
= ∑ az −1 −n
<∞
1 − az z−a n=0

This is illustrated in Figure4.

Figure4: The pole-zero plot for x[n ] = a n u [n ]

2.1.2 Left-Sided Sequence

Consider the sequence x[n] = −a u[− n − 1] . The sequence is non-zero for n ≤ 1 and is referred
n

to as a left-sided sequence, such that,

∞ −1 ∞ ∞
X (z ) = ∑ − a u[− n − 1]z
n −n
= − ∑ a n z − n = − ∑ a − z n =1 − ∑ az −1
n
( )
n

n = −∞ n = −∞ n =1 n=0

The series converges to a rational function,

1 1 z for a −1 z < −1 (See Figure 5)
X (z ) = 1 − = =
1 − a −1 z 1 − az −1 z − a

Figure5: The pole zero plot for x[n] = −a n u[− n − 1]

Note: Sections 1.21 and 1.22 are two different sequences, however with similar expression for X (z ) and
pole-zero plot. The z-transform differs only in the region of convergence.

6|Page
 2.1.3 Finite-Length Sequence
A finite-length sequence is typical of,
a n 0 ≤ n ≤ N −1
x[n] =  , such that
0 otherwise
N −1 ∞
( ) ( )
1 − az −1
N
1 z N − aN
∑a z = ∑ az −1
n
X (z ) = n −n
= = .
n =0 n=0 1 − az −1 z N −1 z − a
N −1
−1 − n
Further, the finite series converges for the set of value z for which ∑ az
n =0
< ∞ . Figure 6 illustrates

the pole-zero plots for N=7, with radius a real for 0<a<1.

Figure6: The pole-zero plot for a finite sequence with

N=7 and a=0.5

3 Inverse z-Transform
The inverse z-transform of a sequence is a method that is use to invert the z-transform and recover
the sequence x(n) from X ( z ) . Three possible approaches include: (a) The partial fraction expansion,
(b) power series expansion, and (c) the inspection methods.
3.1 Partial fraction expansion
Using partial fraction methods, the rational expression X (z ) could be made into a sum of simpler
terms, each of which is tabulated. Assume that X (z ) is expressed as ratio of polynomials in z −1 , such
that,
M M

∑b k z −n
a ∏ (1 − c n z −1 )
X (z ) = nk = 0
N
= 0 n=0
N
b0
∑a
n=0
k z −n ∏ (1 − d
n=0
n z −1 )
Where cn ’s and d n ’s are nonzero poles and zeros of X (z ) . If M<N, then
N
An
X (z ) = ∑
n =1 1 − d n z −1

Evaluating for z = d n , the coefficients An is

7|Page
 An = (1 − d n z −1 )Y ( z )
z =dn

3.2 Power series expansion

∞
The z-transform of a discrete-time signal x[n] defined by ∑ x[n]z
n = −∞
−n
is a power series where the

sequence values x[n] are coefficients of z − n . Thus,

∞
X (z ) = ∑ x[n]z −n
{
= K + x[− 2]z 2 + x[− 1]z 1 + x[0] + x[1]z −1 + x[2]z − 2 + K }
n = −∞
and any particular value of the sequence could be determined by resolving the coefficient of the
appropriate power of z −1 . Moreover, this approach is useful for finite-length sequence, where X ( z )
have no rational forms other than a polynomial in z −1 .

3.3 The inspection methods

The inspection method involves the use of the invaluable z-transform Tables, recognizing with
cursory evaluations certain transform pairs (refer to Table 2).

4 Properties of z-Transform
There are a number of important and useful z-transform properties used in studying discrete-time
signals and systems. Table 3 summarizes some of the properties of z-transform. These include:
1. Linearity;
2. Time-Shifting;
3. Multiplication by an Exponential Sequence;
4. Differentiation of X ( z ) ;
5. Conjugation of a Complex Sequence;
6. Time Reversal

Table 3

4.1 Linearity
The z-transform is a linear operator. The linearity property states that:

8|Page
 z{ax1[n] + bx2 [n]} ⇔ aX1 ( z ) + bX 2 ( z ) and ROC contains Rx1 ∩ Rx 2 .
The region of convergence is at least the intersection of the respective ROCs (it could be larger).
Moreover, this property of linearity allows for expansion of X ( z ) into simpler sums, for easier
evaluation of the inverse z-transforms using partial fractions.

4.2 Time Shifting

The shifting (delaying or advancing) of a sequence is achieved by the convolution of the z-transform
with the power of z. The time-shifting property states:

z{x[n − n0 ]} ⇔ z − n0 X ( z ) n0 is an integer, and ROC is Rx

If n0 is taken to be positive, the original sequence x[n] is shifted right, otherwise it is shifted left.
Further, shifting a sequence does not affect its summation, and therefore does not change the ROC.
For example, the z-transform of x[n] and x[n − n0 ] have the same region of convergence.

4.3 Multiplication by an Exponential Sequence

The exponential multiplication property is expressed as,
{ }
z z 0n x[n ] ⇔ X ( z z 0 ) , ROC = z0 Rx

The ROC = z0 Rx implies that the ROC ( Rx ) is scaled by z0 . This corresponds to a scaling of the
z-plane.

4.4 Differentiation of X ( z )
If X ( z ) is the z-transform of x[n] , the z-transform of nx[n] is stipulated as,
dX ( z )
z{x[n ]} ⇔ − z
dz
4.5 Time Reversal
If x[n] has a z-transform X ( z ) with a ROC, such that α < z < β , then the z-transform of the
time-reversed sequence x[− n] is
( )
z{x[n]} = X z −1
1
Further, the ROC is the set of values of z for which 1 β < <1 α
Rx
4.6 Conjugation of a Complex Sequence
If X ( z ) is the z-transform of x[n] , the z-transform of the complex conjugate of x[n] is given by
{ }
z x ∗ [n ] = X ∗ z ∗( ) , and ROC= R x
Moreover, if x [n] = x[n] , then X ( z ) = X
∗ ∗
(z )
∗

4.7 Convolution of Sequences

The convolution theorem states: The convolution in the time domain is mapped into multiplication
in the frequency domain. Hence, given the dicrete-time signals x1 [n] and x 2 [n] , the convolution
theory states that:

9|Page
 z{y[n] = x1 [n] ∗ x 2 [n]} ⇔ Y ( z ) = X 1 ( z )X 2 ( z ) and ROC contains R x1 ∩ R x2
Note: The region of convergence of Y ( z ) may be larger, if there exist pole-zero cancellation in the
product X 1 ( z ) X 2 ( z )

5 Frequency Response using z-Transform

5.1 Frequency Response of Continuous System
The frequency response of a system is the way in which the gain of the system, and the phase difference
between the I/O signals, depend on the signal frequency.

Review:

Fig.7: Typical phasor diagram

Let a sinusoid signal input be represented as,
z = A cos (ω t + φ ) = A ∠ φ
Where,
A=amplitude;
φ =phase angle; and
ω =angular frequency.

Thus, if an input signal changes both in the amplitude and phase , the steady-state output is simply
the product of the input and the transfer function in complex representation, such that:
Ao (ω )
A(ω ) = = The magnitude frequency response; and
Ai (ω )
φ (ω ) = ∠φo (ω ) − ∠φi (ω ) =the phase frequency response.
The block-diagram representation is aptly illustrated in Figure 8.

Fig.8: Phasor block-diagram representation

The frequency response of the system is the combination of the magnitude and phase response
represented as, A(ω )∠ φ (ω ) .

5.2 Frequency Response of Discrete Systems

10 | P a g e
 As with continuous systems, the frequency response of discrete system could be determined using
the p-z diagram or from the transfer function. In the z-plane, the discrete signal moves as a
‘frequency point’ around the unit circle in the anticlockwise direction, as the signal frequency
increases (See Figure 9).

Figure9: Frequency response using p-z diagram

The point starts z=+1 (d.c.) and ends at z=-1 (the Nyquist frequency). Moreover, the gain is given
by,
∏ zero dis tan ces
Gain = k
∏ pole dis tan ces
and the phase difference is,
Phase angle = ∑ zero angles − ∑ pole angles

Consider the p-z plot of the discrete time system with the transfer function T ( z ) = (z − 0 .1) ( z − 0 .3 )
and is used with sampling frequency of 1 kHz. Figure 10 illustrate the pole-zero diagram of the filter.

Figure10: frequency response of T ( z ) = ( z − 0 .1) ( z − 0 .3 ) at f=0Hz

11 | P a g e
 At f=0 Hz the magnitude is given by,

∏ zero dis tan ces 0 .9

Magnitude = k = ≈ 1 .29 (2 .18 dB ) , and the phase is calculated to be
∏ pole dis tan ces 0 .7
Phase angle = ∑ zero angles − ∑ pole angles = ∠ 0 o − ∠ 0 o = 0 o

At f=¼f=250Hz, the magnitude is given by,

∏ zero dis tan ces 1 2 + 0 .1 2 1 .005

Magnitude = k = = ≈ 0 . 96 (− 0 .33 dB )
∏ pole dis tan ces 2
1 + 0 .3 2 1 . 044

Phase angle = ∑θ z − ∑ θ p = ∠ 95 .71 o − ∠106 .70 o ≈ 11 o (See Figures 11 and 12)

Figure10: frequency response of T ( z ) = ( z − 0 .1) ( z − 0 .3 ) at f=250Hz

Notice that the frequency response is only displayed up to the Nyquist frequency on both of the freqz
and the bode plot. The response does not stop at this frequency point, but continues from fN to fs, , as
the frequency point proceeds around the unit circle. Furthermore, due to the symmetry of the p-z plot
about the real abscissa, the portion of the magnitude response is taken to be a mirror image up to fN..
For this reason there is little interest in the frequency response above the Nyquist frequency, as the
responses repeats itself as the frequency point makes revolutions of the unit circle.

5.3 Frequency Response of Discrete Time Systems using Transfer Functions

It is possible to calculate the frequency response of a discrete system directly from its transfer function.
1. It is required to convert from the z-domain to the s-domain by substituting z = e sT where
s = σ + jω
Where, z = e (σ + jω )T = cos ωT + j sin ω T

12 | P a g e
 Figure11: The freqz plot for T ( z ) = ( z − 0 .1) (z − 0 .3 )

Figure12: The bode plot for T ( z ) = ( z − 0 .1) ( z − 0 .3 )

The σ part of the s produces the decaying response. Therefore assuming σ=0,
( jω )T
z=e = cos ω T + j sin ω T .

2. Substitute the z-value into the transfer function, X ( jω ) allows the gain and phase to be obtained
from ω = 0 to π T (Nyquist angular frequency).

Consider the T (z ) = (z − 0 .1) ( z − 0 .3 ) .

13 | P a g e
 (a) At f=0Hz, z = e ( jω )T = cos ω T + j sin ω T = 1 (See Figure 10)
Substituting z = 1 + j 0 , T ( z ) = (1 − 0 .1) (1 − 0 .3 ) = 0 .9 = 1 .29 ( 2 . 18 dB )
0 .7
Also, the phase angle is ∠ θ = 0
(b) At f=250Hz, z = e ( jω )T = cos 90 o + j sin 90 o = j (See Figure 10)

Substituting z = j , T ( z ) = ( j − 0 .1) = (0 .1)2 + 12 1 .005

= = 0 . 96 ( − 0 .33 dB )
( j − 0 .3 ) (0 .3 )2 + 12 1 .044

phase angle = ∠95 .71 − ∠106 .70 ≈ 11o

(c) Further, At f=375Hz, z = e ( jω )T = cos 67 .5o + j sin 67 .5o = 0.3827 + j 0.9239

Substituting z = 1 + j 0 , T ( z ) = ([0 .3827 − 0 .1] + j 0 . 9239 )

=
0 .93333
= 1 .04161 ( 0 .35 dB )
([0 .3827 − 0 .3 ] + j 0 . 9239 ) 0 .86025
phase angle = ∠ 72 .98 − ∠84 .88 ≈ 12 o (See Figure 11)

5.4 Stability and the z-plane

Figure 13 illustrate the boundary stability in the s-plane. Moreover, the boundary between the areas of
stability and instability is defined by the boundary s = jω . However, if this region is mapped onto the z-
plane with z = e ( jω )T , it defines a circle centered on the origin, and with a radius of 1 with the regions of
instability as shown.

Figure 13: Stability in the z-plane

Obviously, for the z-plane: The region of instability is the area outside the unit circle, while the area
enclosed by the circle is the region of stability. The unit circle itself corresponds to marginal stability,
which is represented in s-plane by the imaginary axis.

5.5 Application of z-transform

A digital computer could be implemented as a compensator or controller in a feedback control system.
Computers receive sampled-data, and thus, utilize z-transform techniques for describing and analyzing the
performance of sampled-data systems. The transfer function of the time-series of sampled data could be

14 | P a g e
analyzed for stability and transient responses. When a process is controlled using digital signals and systems, it
is referred to as digital control.
Since 1998, there are approximately 100 million control systems implemented using mini- and micro-
computers. In addition to this, it is estimated that the sales of mini- and micro-computers accounted for over
$50 billion in the U.S in 1994 alone.
Digital control systems are used in many applications. These include: machine tools, metal working processes,
chemical processes, aircraft control, and vehicle traffic controls. Moreover, the advantages of using digital
control include: improved measurement sensitivity; the use of digitally coded signals digital sensors and
transducers, and microprocessors; reduced sensitivity to signal noise; and the capability to easily reconfigure
the control algorithm in the software. Examples of such digital systems are radar tracking and satellite
communication system.

Review Questions:
1. Find the z-transform of the expressions and analyze for region of convergence (ROC)
u[n]( 12 )
n
a.
− u[− n − 1]( 12 )
n
b.
u[n]( 12 ) + u[n]( 13 )
n n
c.
d. δ [n]
e. δ [n − 1]
f. δ [n + 1]
g.(13 )n (u[n] − u[n − 5])
2. Determine the sequence x[n] with z-transform X ( z ) = (1 + 2 z )(1 + 3z −1 )(1 − z −1 )
3. Obtain the inverse of T ( z ) defined by
11z 2 − 15 z + 6
a. T (z ) =
(z − 2)(z − 1)2
z2 + z
b. Y (z ) =
(z + 1)3
z 3 + z 2 + 3z − 1
c. Y (z ) =
(
(z − 1) z 2 − z + 1 )
4. Define the initial and final value theorem. Find the initial and final value of the causal sequence with
the z-transform,
0.792 z 2
Y (z ) =
(
(z − 1) z 2 − 0.416 z + 0.208 )
5. Find the z-transform u[n] = 2δ [n] − 3δ [n − 2] + 4δ [n − 5]
6. Sketch the pole-zero plot for each of the following z-transforms and shade the region of
convergence:
1 − 12 z −1
i. X 1 (z ) =
1 + 2 z −1
1 − 13 z −1
ii. X 2 (z ) =
( )(
1 + 12 z −1 1 − 23 z −1 )
15 | P a g e
 1 + z −1 − 2 z −2
iii. X 3 (z ) =
1 − 136 z −1 + z −2
7. Determine the inverse z-transform of each of` following. Use the specified method.
1 − 13 z −1
a. Long division; X 1 ( z ) =
1 + 13 z −1
3
b. Partial fractions; X 1 ( z ) =
z − 4 − 18 z −1
1

c. Power series; X 1 ( z ) = ln(1 − 4 z )

8. Consider the p-z plots illustrated in Figure P1 and P2 which represent the output of digital filters.

a. Determine the transfer functions

b. Evaluate the region of convergence for each X 1 ( z )
c. Assume the filters are with the sampling frequency 120Hz. Use the p-z map to estimate the
gain and phase angle at 0Hz and 60Hz.
d. Estimate the gain and phase angle at 100Hz, directly from the transfer function.
s+2
9. A continuous filter has a transfer function, T (s ) = .
(s + 1)(s + 3)
i. Assuming the sampling frequency is 10Hz. Use the p-z diagram to find the gain of
the system at 0Hz
ii. Determine the gain and the phase angle at f s 4
10. Consider Figure P3. Let x[n] be the sequence with the
p-z diagram. Sketch the pole-zero plots for:
y[n] = ( 12 ) x[n]
n
1.
 πn 
2. w[n] = cos  x[n]
 2

(10 marks)

Figure P3

16 | P a g e