Chapter 5

THE APPLICATION OF THE Z TRANSFORM

5.1 Introduction
5.2 The Discrete-Time Transfer Function

Copyright c 2005 Andreas Antoniou

Victoria, BC, Canada
Email: [email protected]

July 14, 2018

Frame # 1 Slide # 1 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Introduction

I Through the use of the z transform, a discrete-time system

can be characterized by a discrete-time transfer function.

Frame # 2 Slide # 2 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Introduction

I Through the use of the z transform, a discrete-time system

can be characterized by a discrete-time transfer function.
I The discrete-time transfer function plays the same key role as
the continuous-time transfer function in an analog system.

Frame # 2 Slide # 3 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Introduction

I Through the use of the z transform, a discrete-time system

can be characterized by a discrete-time transfer function.
I The discrete-time transfer function plays the same key role as
the continuous-time transfer function in an analog system.
I It can be used to obtain the time-domain response of a
system to any excitation or its frequency-domain response.

Frame # 2 Slide # 4 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Introduction

I Through the use of the z transform, a discrete-time system

can be characterized by a discrete-time transfer function.
I The discrete-time transfer function plays the same key role as
the continuous-time transfer function in an analog system.
I It can be used to obtain the time-domain response of a
system to any excitation or its frequency-domain response.
I In this presentation, the definition, derivation, and properties
of the discrete-time transfer function are examined.

Frame # 2 Slide # 5 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Discrete-Time Transfer Function

I The transfer function of a discrete-time system is the ratio of

the z transforms of the response and the excitation.

Frame # 3 Slide # 6 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Discrete-Time Transfer Function

I The transfer function of a discrete-time system is the ratio of

the z transforms of the response and the excitation.
I Consider a linear time-invariant discrete-time system and let
– x(nT ) be the excitation (or input)
– y (nT ) be the response (or output)
– h(nT ) be the impulse response

Frame # 3 Slide # 7 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Discrete-Time Transfer Function Cont’d

I The convolution summation gives

∞
X
y (nT ) = x(kT )h(nT − kT )
k=−∞

Frame # 4 Slide # 8 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Discrete-Time Transfer Function Cont’d

I The convolution summation gives

∞
X
y (nT ) = x(kT )h(nT − kT )
k=−∞

I From the real-convolution theorem (see Chap. 3), we have

Zy (nT ) = Zx(nT )Zh(nT )

Frame # 4 Slide # 9 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Discrete-Time Transfer Function Cont’d

I The convolution summation gives

∞
X
y (nT ) = x(kT )h(nT − kT )
k=−∞

I From the real-convolution theorem (see Chap. 3), we have

Zy (nT ) = Zx(nT )Zh(nT )

I Therefore,
Y (z)
= H(z)
X (z)
In effect, the transfer function is also the z transform of the
impulse response of the system.

Frame # 4 Slide # 10 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation of Transfer Function from Difference Eqn.

I A noncausal, linear, time-invariant, recursive discrete-time

system can be represented by the difference equation
N
X N
X
y (nT ) = ai x(nT − iT ) − bi y (nT − iT )
i=−M i=1

where M and N are positive integers.

Frame # 5 Slide # 11 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation of Transfer Function from Difference Eqn.

I A noncausal, linear, time-invariant, recursive discrete-time

system can be represented by the difference equation
N
X N
X
y (nT ) = ai x(nT − iT ) − bi y (nT − iT )
i=−M i=1

where M and N are positive integers.

I The z transform gives
N
X N
X
Y (z) = Zy (nT ) = Z ai x(nT −iT )−Z bi y (nT −iT )
i=−M i=1

Frame # 5 Slide # 12 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

···
N
X N
X
Y (z) = Z ai x(nT − iT ) − Z bi y (nT − iT )
i=−M i=1

Using the linearity and time-shifting theorems of the z transform,

we get
N
X N
X
−i
Y (z) = ai z Zx(nT ) − bi z −i Zy (nT )
i=−M i=1
N
X N
X
= ai z −i X (z) − bi z −i Y (z)
i=−M i=1

Frame # 6 Slide # 13 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

···
N
X N
X
Y (z) = ai z −i X (z) − bi z −i Y (z)
i=−M i=1

Solving for Y (z)/X (z) and then multiplying the numerator and
denominator polynomials by z N , we get
PN −i
PN N−i
Y (z) i=−M ai z i=−M ai z
H(z) = = =
X (z) 1+ N −i zN + N N−i
P P
i=1 bi z i=1 bi z
a(−M) z M+N + a(−M+1) z M+N−1 + · · · + aN
=
z N + b1 z N−1 + · · · + bN

Frame # 7 Slide # 14 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

···
a(−M) z M+N + a(−M+1) z M+N−1 + · · · + aN
H(z) =
z N + b1 z N−1 + · · · + bN
If M = N = 2, we have

N(z) a(−2) z 4 + a(−1) z 3 + a0 z 2 + a1 z + a2

H(z) = =
D(z) z 2 + b1 z + b2

Frame # 8 Slide # 15 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

···
a(−M) z M+N + a(−M+1) z M+N−1 + · · · + aN
H(z) =
z N + b1 z N−1 + · · · + bN
If M = N = 2, we have

N(z) a(−2) z 4 + a(−1) z 3 + a0 z 2 + a1 z + a2

H(z) = =
D(z) z 2 + b1 z + b2

Note: In noncausal systems, the degree of the numerator

polynomial exceeds the degree of the denominator polynomial.

Frame # 8 Slide # 16 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

I For a causal system, M = 0 and hence

a(−M) z M+N + a(−M+1) z M+N−1 + · · · + aN

H(z) =
z N + b1 z N−1 + · · · + bN
a0 z N + a1 z N−1 + · · · + aN
= N
z + b1 z N−1 + · · · + bN

Frame # 9 Slide # 17 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function from Difference Equation Cont’d

I For a causal system, M = 0 and hence

a(−M) z M+N + a(−M+1) z M+N−1 + · · · + aN

H(z) =
z N + b1 z N−1 + · · · + bN
a0 z N + a1 z N−1 + · · · + aN
= N
z + b1 z N−1 + · · · + bN
I Since some of the numerator coefficients can be zero, we
conclude that in causal recursive systems the numerator
degree is equal to or less than the denominator degree.

Frame # 9 Slide # 18 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Representation by Zero-Pole Plots

I By factorizing the numerator and denominator polynomials,

the transfer function of a noncausal system can be expressed
as
H0 Zi=1 (z − zi )mi
Q
N(z)
H(z) = = QP
D(z) i=1 (z − pi )
ni

where
– z1 , z2 , . . . , zZ are the zeros of H(z)
– p1 , p2 , . . . , pP are the poles of H(z)
– mi is the order of zero zi
– ni is the order PZ of pole pi
– M + N = i=1 mi is the order of N(z)
PP
– N = i=1 ni is the order of D(z)
– H0 is a multiplier constant

Frame # 10 Slide # 19 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Representation by Zero-Pole Plots Cont’d

···
H0 Zi=1 (z − zi )mi
Q
N(z)
H(z) = = QP
D(z) i=1 (z − pi )
ni

I The order of a discrete-time transfer function is the order of

N(z) or D(z), whichever is larger, i.e., M + N if M > 0 or N
if M = 0.

Frame # 11 Slide # 20 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Representation by Zero-Pole Plots Cont’d

···
H0 Zi=1 (z − zi )mi
Q
N(z)
H(z) = = QP
D(z) i=1 (z − pi )
ni

I The order of a discrete-time transfer function is the order of

N(z) or D(z), whichever is larger, i.e., M + N if M > 0 or N
if M = 0.

I Discrete-time systems can be represented by zero-pole plots.

Frame # 11 Slide # 21 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Representation by Zero-Pole Plots Cont’d

jIm z z plane

Re z

Frame # 12 Slide # 22 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function in Nonrecursive Systems

I A nonrecursive system can be represented by the difference equation

N
X
y (nT ) = ai x(nT − iT )
i=0

Frame # 13 Slide # 23 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function in Nonrecursive Systems

I A nonrecursive system can be represented by the difference equation

N
X
y (nT ) = ai x(nT − iT )
i=0

I Hence the transfer function assumes the form

N
Y (z) X
= H(z) = ai z −i
X (z)
i=0
PN N−i
i=0 ai z
=
zN

Frame # 13 Slide # 24 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Transfer Function in Nonrecursive Systems

I A nonrecursive system can be represented by the difference equation

N
X
y (nT ) = ai x(nT − iT )
i=0

I Hence the transfer function assumes the form

N
Y (z) X
= H(z) = ai z −i
X (z)
i=0
PN N−i
i=0 ai z
=
zN

I Evidently, the poles of nonrecursive systems are all located at the

origin of the z plane.

Frame # 13 Slide # 25 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation of Transfer Function from a Network

I The unit delay, adder, and multiplier are characterized by the

equations
K
X
y (nT ) = x(nT − T ), y (nT ) = xi (nT ), y (nT ) = mx(nT )
i=1

Frame # 14 Slide # 26 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation of Transfer Function from a Network

I The unit delay, adder, and multiplier are characterized by the

equations
K
X
y (nT ) = x(nT − T ), y (nT ) = xi (nT ), y (nT ) = mx(nT )
i=1

I Hence if we apply the z transform, we get

K
X
Y (z) = z −1 X (z), Y (z) = Xi (z), Y (z) = mX (z)
i=1

Frame # 14 Slide # 27 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation of Transfer Function from a Network

I The unit delay, adder, and multiplier are characterized by the

equations
K
X
y (nT ) = x(nT − T ), y (nT ) = xi (nT ), y (nT ) = mx(nT )
i=1

I Hence if we apply the z transform, we get

K
X
Y (z) = z −1 X (z), Y (z) = Xi (z), Y (z) = mX (z)
i=1

I By using these relations, H(z) can be obtained directly from a

network representation.

Frame # 14 Slide # 28 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Example

Find the transfer function of the system:

W(z)
X(z) Y(z)

1
2

−1
4

Frame # 15 Slide # 29 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Example Cont’d

W(z)
X(z) Y(z)

1
2

−1
4

Solution By inspection

W (z) = X (z) + 12 z −1 W (z) − 41 z −2 W (z)

and
Y (z) = W (z) + z −1 W (z)

Frame # 16 Slide # 30 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Example Cont’d

···
W (z) = X (z) + 12 z −1 W (z) − 41 z −2 W (z)
and
Y (z) = W (z) + z −1 W (z)
Hence
X (z)
W (z) = 1 −1
, Y (z) = (1 + z −1 )W (z)
1− 2z + 14 z −2

If we eliminate W (z) in the right-hand equation, we obtain

z(z + 1)
H(z) =
z2 − 12 z + 41

Frame # 17 Slide # 31 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation from a State-Space Representation

I A discrete-time system can be represented by the state-space

representation

q(nT + T ) = Aq(nT ) + bx(nT ) (A)

y (nT ) = cT q(nT ) + dx(nT ) (B)

Frame # 18 Slide # 32 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation from a State-Space Representation

I A discrete-time system can be represented by the state-space

representation

q(nT + T ) = Aq(nT ) + bx(nT ) (A)

y (nT ) = cT q(nT ) + dx(nT ) (B)

I Applying the z transform to Eq. (A), we get

Zq(nT + T ) = AZq(nT ) + bZx(nT )

or zQ(z) = AQ(z) + bX (z)

Frame # 18 Slide # 33 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation from a State-Space Representation

I A discrete-time system can be represented by the state-space

representation

q(nT + T ) = Aq(nT ) + bx(nT ) (A)

y (nT ) = cT q(nT ) + dx(nT ) (B)

I Applying the z transform to Eq. (A), we get

Zq(nT + T ) = AZq(nT ) + bZx(nT )

or zQ(z) = AQ(z) + bX (z)

I

Hence zIQ(z) = AQ(z) + bX (z)

or Q(z) = (zI − A)−1 bX (z) (C)

where I is the identity matrix.

Frame # 18 Slide # 34 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation from a State-Space Representation Cont’d

··· q(nT + T ) = Aq(nT ) + bx(nT ) (A)

T
y (nT ) = c q(nT ) + dx(nT ) (B)

or Q(z) = (zI − A)−1 bX (z) (C)

I Now from Eq. (B)

Y (z) = cT Q(z) + dX (z) (D)

Frame # 19 Slide # 35 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

Derivation from a State-Space Representation Cont’d

··· q(nT + T ) = Aq(nT ) + bx(nT ) (A)

T
y (nT ) = c q(nT ) + dx(nT ) (B)

or Q(z) = (zI − A)−1 bX (z) (C)

I Now from Eq. (B)

Y (z) = cT Q(z) + dX (z) (D)

I If we now eliminate Q(z) using Eq. (D), we have

Y (z) N(z)
H(z) = = = cT (zI − A)−1 b + d
X (z) D(z)

where N(z) and D(z) are polynomials in z.

Frame # 19 Slide # 36 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2

 This slide concludes the presentation.
Thank you for your attention.

Frame # 20 Slide # 37 A. Antoniou Digital Signal Processing – Secs. 5.1, 5.2