EEC 430/530

Digital Signal Processing

Unit 6
Z-Transforms:
Signals and Systems

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 116

Transfer Function Description
• In order to get to the Z-transfer function description of a DT system, we
must first look at the Z-transform of a DT signal.
• Once the Z-transform of a DT signal is clear, we can apply it to the signals
(in and out) of a DT system to produce the system’s transfer function.
Z-Transform of a Signal
• The Z-transform of a time domain DT signal 𝑥 𝑛 is defined as
$

𝑋 𝑧 = & 𝑥 𝑛 𝑧 #!
!"#$
• Why do this? Because it leads to a transfer function description of a DT
system that is a ratio of polynomials in Z – and that (as with Laplace
transfer functions for CT systems) is very useful.
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/520 - 117
Discrete-Time
Z-Transforms
Signals

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 118

Example 1: Example 3:
𝑥 𝑛 = decaying exponential 𝑥 𝑛 =𝛿 𝑛
𝛼!, 𝑛 ≥ 0 1, 𝑛 = 0
= = 𝛼!𝑢 𝑛 =
0, 𝑛 < 0 0, otherwise
𝑋 𝑧
$ $ $ $

= 9 𝛼 ! 𝑢 𝑛 𝑧 #! = 9 𝛼 ! 𝑧 #! = 9 𝛼𝑧 #& ! 𝑋 𝑧 = 9 𝛿 𝑛 𝑧 #! = 1
!"#$ !"% !"% !"#$
1 𝑧
𝑋 𝑧 = =
1 − 𝛼𝑧 #& 𝑧 − 𝛼 Example 4:
𝐴Ω! cos(𝜔% + 𝜙), 𝑛 ≥ 0
Example 2: 𝑥 𝑛 =
0, 𝑛<0
Ω cos 𝜔% − 𝜙
1, 𝑛 ≥ 0 𝐴𝑐𝑜𝑠 𝜙 𝑧 𝑧 −
𝑥 𝑛 = unit step = =𝑢 𝑛 cos 𝜙
0, 𝑛 < 0 𝑋 𝑧 =
!
𝑧 𝑧 ' − 2Ω𝑐𝑜𝑠𝜔% 𝑧 + Ω'
𝒵𝑥 𝑛 =𝒵 𝛼 𝑢 𝑛 R = poles are at: 𝑧 = Ω𝑒 ±)*! = Ω∠ ± 𝜔%
+"& 𝑧−1

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 119

• Knowing what the Z-transform of a signal is, we turn to the Z-transfer
function of a system.
• The Z-transfer function of a system is derived by applying the Z-
transform to the system’s input and output signals.

• What we’ll show is – if the system (the “?” above) is a discrete-time

linear, time-invariant system (which most of our systems will be), then
the ratio of 𝑌 𝑧 to 𝑋 𝑧 will turn out to be a ratio of polynomials in 𝑧,
and this will provide
o An easy way to think about the dynamics of systems.
o An easy way to combine systems.
o An easy way to manipulate system for design purposes.
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 120
• It will turn out that
𝑌 𝑧
𝐻 𝑧 =
𝑋 𝑧
is a ratio of polynomials in 𝑧; and 𝐻 𝑧 is part of the “language” of DSP.

• One uses 𝐻 𝑧 just like one used 𝐻 𝑠 for CT systems.

• As an example, consider a system is described by

𝑦(𝑛) = 𝑦 𝑛 − 1 + 0.5𝑦 𝑛 − 2 + 𝑥 𝑛 + 2𝑥 𝑛 − 2

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 121

• If two time functions are equal, their Z-transforms will be equal as well:
𝒵 𝑦
𝑛 = 𝒵 𝑦 𝑛 − 1 + 0.5𝑦 𝑛 − 2 + 𝑥 𝑛 + 2𝑥 𝑛 − 2
𝑌 𝑧
= 𝒵 𝑦 𝑛 − 1 + 𝒵 0.5𝑦 𝑛 − 2 + 𝒵 𝑥 𝑛 + 𝒵 2𝑥 𝑛 − 2
𝑌 = 𝑧 !"𝑌 𝑧 + 0.5𝑧 !#𝑌 𝑧 + 𝑋 𝑧 + 2𝑧 !#𝑋 𝑧
𝑧
𝑌 𝑧1 − 𝑧 !" − 0.5𝑧 !# = 𝑋 𝑧 1 + 2𝑧 !#
𝑌 𝑧 1 + 2𝑧 !# 𝑧# + 2
𝐻 𝑧 = = !" !# = #
𝑋 𝑧 1 − 𝑧 − 0.5𝑧 𝑧 − 𝑧 − 0.5
• In general, if
& (
𝑦 𝑛 = 1 𝑎$ 𝑦 𝑛 − 𝑘 + 1 𝑏$ 𝑥 𝑛 − 𝑘
$%" $%'
then
𝑌 𝑧 ∑(
$%' 𝑏$ 𝑧 !$
𝐻 𝑧 = =
𝑋 𝑧 1 − ∑&
$%" 𝑎 $ 𝑧 !$

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 122

Impulse Response
• The response of a system to an impulse – its “ impulse response” –
is important because:
o Knowing the response to an impulse, one can determine the response to
any arbitrary input.
o Knowing the impulse response, one can determine the system’s transfer
function, 𝐻 𝑧 .
o A system’s impulse response is part of the “language” of DSP.
• If the input 𝑥 𝑛 is an impulse (𝑥 𝑛 = 𝛿 𝑛 ), then the output 𝑦 𝑛
is the impulse response. Because of its importance, it is given a
special symbol, ℎ 𝑛 .
• Thus, when 𝑥 𝑛 = 𝛿 𝑛 , then 𝑦 𝑛 = ℎ 𝑛 .
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 123
• Suppose 𝑥 𝑛 = 𝛿 𝑛 . Then 𝑋 𝑧 = 1.
o Now 𝑌 𝑧 = 𝐻 𝑧 𝑋 𝑧 = 𝐻 𝑧 ⋅ 1 = 𝐻 𝑧
• But, in this case, 𝑦 𝑛 is the system’s impulse response, i.e.,
𝑦 𝑛 =ℎ 𝑛 .
o Thus 𝑌 𝑧 = 𝒵 𝑦 𝑛 =𝒵 ℎ 𝑛 =𝐻 𝑧 .
• So the Z-transform of the system’s impulse response is equal to the
system’s transfer function.
• One generally doesn’t determine 𝐻 𝑧 by hitting a system with an
impulse, obtaining ℎ 𝑛 , and then Z-transforming ℎ 𝑛 to get 𝐻 𝑧 .
• Rather, one generally gets 𝐻 𝑧 by Z-transforming both sides of
the system’s difference equation, as we did two pages earlier.
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 124
Example: Suppose a FIR system is described by

𝑦 𝑛 = 2𝑥 𝑛 + 𝑥 𝑛 − 1 + 3𝑥 𝑛 − 2

Let 𝑥 𝑛 = 𝛿 𝑛 , then

𝑦 0 =2⋅1+0+3⋅0=2=ℎ 0
𝑦 1 =2⋅0+1+3⋅0=1=ℎ 1
𝑦 2 =2⋅0+0+3⋅1=3=ℎ 2
𝑦 𝑛 = 0 = ℎ 0 ,𝑛 > 2

So, 𝑦 𝑛 = ℎ 0 𝑥 𝑛 + ℎ 1 𝑥 𝑛 − 1 + ℎ 2 𝑥 𝑛 − 2
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 125
• In general, for a FIR system that has an impulse response of 𝑀 + 1
points long,
𝑦 𝑛 = ℎ 0 𝑥 𝑛 + ℎ 1 𝑥 𝑛 − 1 + ⋯ + ℎ 𝑀 𝑥(𝑛 − 𝑀)
V

or 𝑦 𝑛 = ? ℎ 𝑘 𝑥(𝑛 − 𝑘) ← Difference Equation for a FIR

STU
• The coefficients in a FIR’s difference equation are equal to its
impulse response coefficients.
• The results of the design of any FIR (a filter, say, or a controller) is
the determination of its impulse response coefficients.
• A FIR is implemented by calculating its difference equation each
time a new 𝑥 𝑛 is obtained.
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 126
 Example: Moving Average FIR

1
𝑦 𝑛 = [𝑥 𝑛 + 𝑥 𝑛 − 1 + … + 𝑥(𝑛 − 𝑀)]
𝑀+1
W
= VXW ∑V
STU 𝑥(𝑛 − 𝑘)

• The advantage of this kind of filter (also called a comb filter

because of the shape of its frequency response) is that it can be
computed fast: M adds plus one multiply.
o Note that for a general FIR, there are 𝑀 adds and 𝑀 + 1 multiplies.

• Now, let’s try to interpret the Moving Average FIR as an IIR:

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 127
 ' '
1 𝑥 𝑛 1
𝑦 𝑛 = & 𝑥(𝑛 − 𝑘) = + & 𝑥(𝑛 − 𝑘)
𝑀+1 𝑀+1 𝑀+1
%"& %"(
'
1
𝑦 𝑛−1 = & 𝑥[ 𝑛 − 1 − 𝑘]
𝑀+1
%"&
')( '
1 1 𝑥 𝑛−𝑀−1
= & 𝑥 𝑛−𝑘 = &𝑥 𝑛−𝑘 +
𝑀+1 𝑀+1 𝑀+1
%"( %"(
𝑥 𝑛 𝑥 𝑛−𝑀−1
𝑦 𝑛 = + 𝑦 𝑛−1 −
𝑀+1 𝑀+1
1
𝑦 𝑛 =𝑦 𝑛−1 + 𝑥 𝑛 −𝑥 𝑛−𝑀−1
𝑀+1
2 adds, 1 multiply!
Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 128
Example: Getting Difference Equation from Z-Transfer Function
!!
• Suppose 𝐻 𝑧 = "
!"# !"!

What is the system’s difference equation?

𝑌 𝑧 𝑧$ 𝑧$
𝐻 𝑧 = = =
𝑋(𝑧) (𝑧 − 1) 𝑧 − 1 3 1
𝑧$ − 𝑧 +
2 2 2
𝑌 𝑧 1
=
𝑋 𝑧 3 1
1 − 𝑧 "# + 𝑧 "$
2 2
3 "# 1 "$
𝑌(𝑧) 1 − 𝑧 + 𝑧 = 𝑋 𝑧 (1)
2 2
3 "# 1 "$
𝑌 𝑧 − 𝑧 𝑌 𝑧 + 𝑧 𝑌 𝑧 =𝑋 𝑧
2 2
3 1
𝑦 𝑛 − 𝑦 𝑛−1 + 𝑦 𝑛−2 =𝑥 𝑛
2 2

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 129

 End of Unit 6 – Z-Transforms: Signals and Systems

Donoghue/Hizlan/Zhang Cleveland State University EEC 430/530 - 130