Z-Transforms

and Transfer Functions

Outline
Signals and Systems
Z-Transforms
How to do Z-Transforms
How to do inverse Z-Transforms
How to infer properties of a signal from its
Z-transform

Transfer Functions
How to obtain Transfer Functions
How to infer properties of a system from its
Transfer Function

Signals
The signals we are studying in this
course Discrete Signals
A discrete signal takes value at each nonnegative time instance
18
16
14
12
10
8
6
4
2
0

10

12

Example of a System
18

18

16

16

14

14

12

12

10

10

Filter

2
0

10

2
0

12

raw readings from

a noisy temperature
sensor
- Input Signal

y(k)

u(k 1) u(k 2) u(k 3)

3

10

smooth temperature
values after filtering
- Output Signal

A (SISO) system takes an input signal, manipulates it

and gives a corresponding output signal.

12

Control System

Reference
Input

Control Controller Control

error
Input

Transduced
Output

Transducer

Target
System

Measured
Output

Common Signals

exponential

0.8

impulse

0.5

0.7

|a|<1

0.5

0.3

|a|>1

a=1.2

0.6

0.4

0
-1

(ak)

0.9

0.2

0.1
0

delayed impulse

0.5

10

0
-1

15

sin(k*pi/6)

0.5

0
-1

sine

9
-0.5

-1

10

12

14

16

18

step

0.5

cos(k*pi/6)

0.5

0
-1

cosine

9
0

-0.5

4
3

ramp

-1

10

12

14

16

18

exponentially
modulated
cosine/sine

1
0.8
0.6

u(k)=cos(k*pi/6)*0.9k

0.4
0.2
0

-0.2
-0.4
-0.6

0
-1

-0.8

-1

10

12

14

16

18

Z-Transform of a Signal
Z
u(k)
U(z)
Z-1
u(0)

u(0) z0

u(1)

+u(1) z-1

u(2)

+u(2) z-2

u(3)

+u(3) z-3

u(4)

+u(4) z-4

U(z) u(k) z k
k 0

Z-Transform Contd
Mapping from a discrete signal to a
function of z
Many Z-Transforms have this form:
n

U(z)

a z
i 0
m

Rational Function of z

b z
Helps intuitively derive the signal
properties
j0

Does it converge?
To which value does it converge?
How fast does it converges to the value?

Z Transform of Unit Impulse Signal

Z
Z-1

uimpulse(k)

Uimpulse(z)

u(0) = 1

1 z0

u(1) = 0

+0 z-1

u(2) = 0

+0 z-2

u(3) = 0

+0 z-3

u(4) = 0

+0 z-4

Uimpuls e(z) 1

0.5

0
-1

Delayed Unit Impulse Signal

Z
udelay(k)
Udelay(z)
Z-1
u(0) = 0

0 z0

u(1) = 1

+1 z-1

u(2) = 0

+0 z-2

u(3) = 0

+0 z-3

u(4) = 0

+0 z-4

Udelay (z) z

0.5

0
-1

Z-Transform of Unit Step Signal

Z
ustep(k)
Ustep(z)
Z-1
u(0) = 1

1 z0

u(1) = 1

+1 z-1

u(2) = 1

+1 z-2

u(3) = 1

+1 z-3

u(4) = 1

+1 z-4

Ust ep(z) 1 z 1 z 2 z 3 ...

0.5

0
-1

Unit Step Signal - continued

A little bit more math
(1 a)(1 a a 2 ... a n )
1a
n 1
1a

1a

1 a a 2 ... a n

n ,

assuming |a| 1,
(1 a)(1 a a 2 ... a n )
1 a a ... lim
n
1a
1 a n 1
lim
n 1 a
1

1a

U step (z) 1 z

1
...
1- z -1

Z-Transform of Exponential
Signal
Z
uexp(k)
Uexp(z)
Z-1
u(0) = 1

1 z0

u(1) = a

+a z-1

u(2) = a2

+a2 z-2

u(3) = a3

+a3 z-3

u(4) = a4

+a4 z-4

U exp (z) 1 az 1 a 2z 2 a 3z 3 ...

a=1.2

0
-1

Remember
this!

1- az -1

LTI Systems
Linear, Time Invariant (LTI) System
Many systems we analyze or design are or
can be approximated by LTI systems
We have a well-established theory for LTI
system analysis and design

Example - A simple moving average

y(k)=[u(k-1)+u(k-2)+u(k-3)]/3
u(k)

3-MA

y(k)

Control System

Reference
Input

Control Controller Control

error
Input

Transduced
Output

Transducer

Target
System

Measured
Output

What does Linear mean exactly?

Scaling

u(k)

3-MA

y(k)

u(k)

3-MA

y(k)

Superposition

u1(k)

3-MA

y1(k)

u2(k)

3-MA

y2(k)

u1(k)+u2(k)

3-MA

y1(k)+y2(k)

Time Invariance
u(k)

3-MA

y(k)

u(k)=u(k-n)

3-MA

y(k)=y(k-n)

Idiom:
u(k-n) is u(k)
delayed by n
time units!

y(k+1)=[u(k)+u(k-1)+u(k-2)]/3

y(k+1-n)=[u(k-n)+u(k-1-n)+u(k-2-n)]/3
y(k+1)=[u(k)+u(k-1)+u(k-2)]/3

Reality Check
Typically speaking, are computing
systems linear? Why?
Consider saturation

Typically speaking, are computing

systems time-invariant? Why?

Unit Impulse Response

3-MA

uimpulse(k)

yimpulse(k)

0.5

0.5

0
-1

10

y(k)

0
-1

u(k 1) u(k 2) u(k 3)

3

Claim:
If we know yimpulse(k), we can obtain y(k)
corresponing to ANY input u(k)!
yimpulse(k) contains ALL information about
the input-output relationship of an LTI system.

An Example: 3-MA
3MA

uimpulse(k)

yimpulse(k)

0.5

0.5

0
-1

0
-1

3MA

u (k)

6x

9
8

y (k) ?

u(k) =

5
4
3

0
-1

uimpulse(k-1)

0.5
0
-1

uimpulse(k-2)

1
0.5
0
-1

0.5

9x

3x

uimpulse(k)

0
-1

An Example: 3-MA
3MA

uimpulse(k)

yimpulse(k)

0.5

0.5

0
-1

0
-1

3MA

u (k)

6x

9
8

y (k) ?

y(k) =

5
4
3

0
-1

yimpulse(k-1)

0.5
0
-1

yimpulse(k-2)

1
0.5
0
-1

0.5

9x

3x

yimpulse(k)

0
-1

Convolution
y(5)= u(0) yimpulse(k)
+ u(1) yimpulse(k-1)
+ u(2) yimpulse(k-2)
+ u(3) yimpulse(k-3)
+ u(4) yimpulse(k-4)
u(0) x

9
8

y(k) [u(i) y impuls e(k i)]

i 0

u(k)* y impuls e(k)

yimpulse(k)

1
0.5

0
-1

y(k) =

5
4
3

u(1) x

yimpulse(k-2)

1
0.5
0
-1

0.5
0
-1

u(2) x

yimpulse(k-1)

0
-1

k 1

Important Theorem
Time Domain

u(k)

v(k)

y(k)

(convolution)

Z Z-1
U(z)

Z Z-1

(multiplication)

Z Domain

V(z)

Z Z-1
=

Y(z)

Z-Transform/Inverse Z-Transform
LTI: yimpuse(k)=0.3k-1

u (k)=0.7k
1

0.9

0.9

0.8
0.7
0.6
0.5

0.3

0.6
0.5

0.5

0.2
0.1
0

Z
1
1 0.7z1

0.6

0.3

15

0.7

0.4

0.1

10

0.8

0.3

0.1
5

0.9

0.4

0.2

0.7

0.2

0.8

(convolution)

0.4

y (k)?

10

12

14

16

(multiplication)

18

Transfer
Function
-1

z
1 0.3z 1

10

15

Z-1

z 1
(1 0.3z 1)(1 0.7z 1)

Delay the Unit Step Signal

y(k)=u(k-1)
LTI: yimpuse(k)
=udelayed(k)

u (k)

y (k)

0.5

0.5

0.5

0
-1

ustep (k)

(multiplication)

udelayed(k)

(convolution)
Transfer
Function

Z
1
1 z 1

0
-1

0
-1

udstep(k)

Z
z-1

z -1
1 z 1

Delayed Unit Step Signal Contd

Z
Z-1

udstep(k)
u(0) = 0

0 z0

u(1) = 1

+1 z-1

u(2) = 1

+1 z-2

u(3) = 1

+1 z-3

u(4) = 1

+1 z-4

Remember
this!

Uds t ep(z) z 1 z 2 z 3 ...

0.5

0
-1

Udstep(z)

z -1
1

1- z -1
z 1

Transfer Function
Transfer function provides a much more
intuitive way to understand the input-output
relationship, or system characteristics of an
LTI system

Stability
Accuracy
Settling time
Overshoot

Signals and Systems in Computer

Systems
Spike, one-time fluctuation in input/output,
or disturbance

0.5

0
-1

Change of reference value

0.5

0
-1

12

10

y(k)=u(k-n)

10

Multiple changes of reference value

-Sum of delayed step signals
ustep(k)+8ustep(k-3)-4ustep(k-6)
Input got delayed for n time units

n-Delay
y(k)=u(k-n)
k n
1,
yimpuls e(k) undelay (k)
0, otherwise

Yimpulse(z) z n

Y(z) z nU(z)
U(z) u(0) u(1)z 1 u(2)z 2 u(3)z 3 ...

Y(z) y(0) ... y(n 1)z n1 y(n)z n y(n 1)z n1 y(n 2)z n2 y(n 3)z n3 ...
0 ... 0z n1 u(0)z n u(1)z n1 u(2)z n2 u(3)z n3 ...
z -nU(z)

Transfer function: z-n

Unit Shift and n-Shift

y(k)=u(k+1)
1

0.5

0.5

0
-1

0
-1

Caveat: u(0) disappears

U(z) u(0) u(1)z 1 u(2)z 2 u(3)z 3 ...
Y(z) y(0) y(1)z1 y(2)z2 y(3)z3 ...

Y(z) zU(z) zu(0)

u(1) u(2)z1 u(3)z2 u(4)z3 ...

y(k)=u(k+n)

Y(z) z nU(z) z nu(0) ... zu(n - 1)

Other properties of Z-Transform

Linearity
Time Domain

Z-Transform

Scaling

y(k)=au(k)

Y(z)=aU(z)

Superposition

y(k)=u(k)+v(k)

Y(z)=U(z)+V(z)

sin? cos?
1

sin(k*pi/6)

0.5

-0.5

-1

10

12

14

16

18

10

12

14

16

18

cos(k*pi/6)

0.5

-0.5

-1

From Exponential to Trigonometric

Uexp(z) 1 az1 a2 z 2 a3 z 3 ...

1
1- az-1

Euler Formula:

Z[cos(k)]?
Z[sin(k)]?

ei cos isin
e cos( ) isin( ) cos isin

e i e i
cos
2

e i e i
sin
2i

Z-Transform of sin/cos
Z-Transform

Time Domain
u(k) e ik

U(z)

u(k) e -ik

u(k) cos(k )

U(z)

ik

e
2

ik

1- e z -1

1
1- e-i z -1

1
1
U(z) (

)/2
i
-1
i
1- e z 1 e z 1
1
1
(

)/2
1
1
1
1 cos z isin z
1 cos z isin z 1
1 cos z 1

(1 cos z 1)2 (sin z 1)2

eik e ik
u(k) sin(k )
2i

1
i

1 cos z 1
1 2cos z 1 z 2

U(z)

sin z -1
1- 2cos z -1 z 2

Exponentially Modulated sin/cos

1
0.8
0.6

u(k)=cos(k*pi/6)*0.9k

0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1

10

(aei )k (ae i )k
uexpcos (k) a cos(k )
2
k

a sin z -1
U(z)
1- 2a cos z -1 a2 z 2

12

14

16

18

(aei )k (ae i )k
uexps in(k) a sin(k )
2i
k

a sin z -1
U(z)
1- 2a cos z -1 a2 z 2

A damped oscillating signal a typical output of a second order system

An LTI System Discrete Integrator

y(k)=y(k-1)+u(k-1)
Y(k)=u(k-1)+u(k-2)++u(1)+u(0)

LTI: yimpuse(k)
=udstep(k)

u (k)

y (k)

3
2
1

1
1

0.5
0.5

0
-1

ustep(k)

*
Transfer

Z
1
1 z 1

0
-1

udstep(k)

(convolution)

Function

(multiplication)

0
-1

Z
z -1
1 z 1

uramp(k)

Z-1

z -1
(1 z 1)2

Inverse Z-Transform
Z
u(k)
Z-1?

U(z)

Table Lookup if the Z-Transform looks

familiar, look it up in the Z-Transform table!
u(k) 3us t ep(k) 2uramp (k)

Z-1?

Long Division
Partial Fraction Expansion

3
2z 1
U(z)

1 z 1 (1 z 1)2
3Us t ep(z) 2Uramp(z)

Long Division
Sort both nominator and denominator with
descending order of z first

3 z 1
U(z)
1 2z 1 z 2

u(0)=3, u(1)=5, u(2)=7, u(3)=9, , guess:

u(k)=3ustep(k)+2uramp(k)

Partial Fraction Expansion

Many Z-transforms of interest can be
expressed as division of polynomials of z
n

U(z)

i
a
z
i
i 0
m

j
b
z
j
j0

cj

j 1

z pj

U(z) c0
May be trickier:
complex root
duplicate root

z 1

1
1 p j z 1

b0 b1z b2 z 2 ... bmz m

bm(z p1)(z p2 )...(z pm )

u(k) c0uimpuls e(k) up j dexp (k)

j 1

where

k 1

upj dexp(k) p j ,

k>0

An Example
3z 2 14z 14
U(z)
z 2 6z 8

U(z) c0

c1
c
2
z 2 z 4

(z-2)(z-4)
U1(z)=c0
c1
z 2
c
U3(z) 2
z 4
U2(z)

Z-1
Z-1
Z-1

k0
c ,
u(k) 0 k 1
k 1
c

4
, k0
2
1

u1(k)=c0*uimpulse(k)
u2(k)=c1*2k-1, k>0
u2(k)=c2*4k-1, k>0
c0? c1? c2?

Get The Constants!

3z 2 14z 14
U(z)
z 2 6z 8

U(z) c0

c1
c
2
z 2 z 4

(z-2)(z-4)
U(z) c0

c1
c
2 , z , U(z) c ,
0
z 2 z 4

K(z) (z - 4)U(z) (z 4)c0

c1(z 4)
c2 ,
z 2

3z 2 14z 14
K(4) c2
|z 4 3
z 2

3z 2 14z 14
c0 lim 2
3
z
z 6z 8

Partial Fraction Expansion contd

n

U(z)

a z

b z

i 0
m

j0

c0 limU(z)
z

cj Upj (pj )

cj

j 1

z pj

U(z) c0

How to get c0 and cjs ?

define

Upj (z) (z - pj )U(z)

An Example Complete Solution

3z 2 14z 14
U(z)
z 2 6z 8

U(z) c0

c1
c
2
z 2 z 4

3z 2 14z 14
c0 lim U(z) lim 2
3
z
z
z 6z 8
3z 2 14z 14
z 2 6z 8
3z 2 14z 14

z-4

U2(z) (z 2)

3z 14z 14
z 2 6z 8
3z 2 14z 14

z-2

U4(z) (z 4)

U(z) 3

1
3

z 2 z 4

3 2 2 14 2 14
c1 U2(2)
1
2-4
c2 U4(4)

3 4 2 14 4 14
3
4-2

k0
3,
u(k) k 1
k 1
2 3 4 , k 0

Solving Difference Equations

y(k) a1y(k 1) ... any(k n) b1u(k 1) ... bmu(k m)

Z
Y(z) a1z 1Y(z) ... an z nY(z) b1z 1U(z) ... bm z mU(z)

b1z 1 ... bm z m
Y(z)
U(z)
1
n
1 a1z ... an z
Z-1

y(k) ...

Transfer
Function

A Difference Equation Example

Exponentially Weighted Moving Average

y(k)=cy(k-1)+(1-c)u(k-1)

Solve it!
u (k)=0.8k

y (k)?

LTI: y(k)=0.4y(k-1)+0.6u(k-1)

Y(z) 0.4z 1Y(z) 0.6z 1U(z)

U(z)

1
1 0.8z1

0.6z1
0.6z-1
U(z)

1 0.4z1
(1- 0.4z-1)(1- 0.8z-1)
0.6z
0.6
1.2

(z 0.4)(z 0.8) z 0.4 z 0.8

Y(z)

1
0.8

0.6
0.4
0.2
0
-1

0.6z1
1.2z1

1 0.4z1 1 0.8z1

Z-1
y(k) -0.6 0.4k 1 1.2 0.8k -1

1
0.8
0.6
0.4
0.2
0
-1

Signal Characteristics from Z-Transform

If U(z) is a rational function, and
y(k) a1y(k 1) ... any(k n) b1u(k 1) ... bmu(k m)

Then Y(z) is a rational function, too

zeros

N(z)
Y(z)

D(z)

(z z )
i

i 1
m

(z p )
j

j 1

poles

Poles are more important determine key

characteristics of y(k)

Why are poles important?

Z domain
N(z)
Y(z)

D(z)

(z z )
i

i 1
m

(z p j )

cj

j 1

z pj

c0

j 1

Time domain

poles

Z-1
m

Y(k) c0 uimpuls e(k) c j pkj -1

j 1

components

Various pole values (1)

2.5

2.5
2

1.5

1.5

0.5

-0.5

p=1.1

0.5

p=-1.1

-1
-1.5
-2

0
-1

-2.5
-1

1
0.8

0.6

0.8

0.4
0.2

p=1

0.6
0.4

p=-1

0
-0.2
-0.4
-0.6

0.2

-0.8

0
-1

-1
-1

0.8
0.6

0.8

0.4
0.2

0.6

p=0.9

0.4

p=-0.9

0
-0.2
-0.4

0.2
-0.6

0
-1

-0.8

-1
-1

Various pole values (2)

1

0.9

0.8

0.8

0.6

0.7

0.4

0.6

0.2

0.5

0
0.4

p=0.9

0.3
0.2
0.1
0
-1

-0.2

p=-0.9

-0.4
-0.6
-0.8

-1
-1

0.9

0.8

0.8

0.6

0.7

0.4

0.6

0.2

p=0.6

0.5
0.4
0.3

p=-0.6

0
-0.2
-0.4

0.2

-0.6

0.1
0
-1

-0.8
0

-1
-1

0.9

0.8

0.8

0.6

0.7

0.4

0.6

p=0.3

0.5
0.4

p=-0.3

0.2
0
-0.2

0.3

-0.4
0.2

-0.6
0.1

-0.8
0
-1

-1
-1

Conclusion for Real Poles

If and only if all poles absolute values
are smaller than 1, y(k) converges to 0
The smaller the poles are, the faster
the corresponding component in y(k)
converges
A negative poles corresponding
component is oscillating, while a
positive poles corresponding
component is monotonous

How fast does it converge?

U(k)=ak, consider u(k)0 when the
absolute value of u(k) is smaller than
or equal to 2% of u(0)s absolute value
|a|k 0.02
kln|a| ln0.02 3.912
4
k
ln|a|

Remember
This!
a 0.7
k

4
4

11
ln|0.7| 0.36

1
0.9
0.8

y(k)=0.7

0.7
0.6
0.5
0.4
0.3

y(11)=0.0198

0.2
0.1
0

10

12

Example
u (k)=0.8k

y (k)?

LTI: y(k)=0.4y(k-1)+0.6u(k-1)

Z
Y(z) 0.4z 1Y(z) 0.6z 1U(z)

U(z)

1
1 0.8z1

0.6z1
0.6z-1
Y(z)
U(z)
1 0.4z1
(1- 0.4z-1)(1- 0.8z-1)
Y(z)
0.6
G(z)

-1
U(z) z 0.4

y(k) a 0.4k 1 b 0.8k 1

When There Are Complex Poles

b1z 1 ... bmz m
Y(z)
U(z)
1
n
1 a1z ... an z
(az 2 bz c)...

If

b2 4ac 0,

b b 2 4ac
z
2a
b b 2 4ac
b b 2 4ac
az bz c a(z
)(z
)
2a
2a
2

If b2 4ac 0,

b i 4ac b 2
b i 4ac b 2
az bz c a(z
)(z
)
2a
2a
2

Or in polar coordinates,
az 2 bz c a(z r cos ir sin )(z r cos ir sin )

What If Poles Are Complex

If Y(z)=N(z)/D(z), and coefficients of both D(z)
and N(z) are all real numbers, if p is a pole,
then ps complex conjugate must also be a
pole
Complex poles appear in pairs
l

cj

j 1

z pj

cj

j 1

z pj

Y(z) c0
c0

Time domain

c
c'

z r cos ir sin z r cos ir sin

bzr sin dz(z r cos )

z 2 (2r cos)z r 2

Z-1
m

y(k) c0 uimpuls e(k) c j pkj -1 br k sink dr k cosk

j 1

An Example
Z-Domain: Complex Poles
2

Time-Domain:
Exponentially Modulated Sin/Cos

1.5

z2 z
Y(z) 2
z 0.8z 0.64
k
k
y(k) 2 0.8 k sin( ) 0.8 k cos( )
3
3

0.5

-0.5

-1

10

12

14

16

18

20

Poles Everywhere

Observations
Using poles to characterize a signal
The smaller is |r|, the faster converges the signal
|r| < 1, converge
|r| > 1, does not converge, unbounded
|r|=1?

When the angle increase from 0 to pi, the

frequency of oscillation increases
Extremes 0, does not oscillate, pi, oscillate at the
maximum frequency

Change Angles
1

1
0.8

0.8

0.6

0.8

0.4

0.6

0.2

0.4
0.2

0.6
-0.2

-0.4

-0.2

0.2

-0.6

-0.4

-0.8

-0.2

-1

0.4

-0.6
-0.8
0

10

15
-1

-0.4

10

15
1

-0.6

0.8

-0.8
0.6

-1

10

0.4

15

0.2

0.8
0

0.6

-0.2

0.4

-0.4

0.2

Im

0
-0.2
-0.4

-0.6
-0.8
-1

10

15

-0.6
-0.8
1

-1

10

0.8

15

0.6

0.8

0.4

0.6

0.2

0.4

0.2

-0.2

0
-0.4

-0.2
-0.6

-0.4
-0.8

-0.6
-1

-0.8
-1

10

10

15

15

-0.9

0.8
0.6
0.4
0.2
0

0.9

Re

1
0.8
0.6
0.4
0.2

-0.2

-0.4

-0.2

-0.6

-0.4

-0.8
-1

-0.6

10

15

-0.8
-1

10

15

Changing Absolute Value

1

0.8

0.8

0.6

0.6

0.4

0.8

0.2

0.6

0.4

-0.2

0.2

-0.4

-0.6

-0.2

-0.8

-0.8

-0.4

-1

-1

-0.6

0.4
0.2
0

-0.2
-0.4

10

-0.6

3
0

10

15

15

-0.8
-1

10

15

12
0

-1

10

1
0.8

Im

0.6
0.4
0.2
0

-2

-0.2

-3

10

12

-0.4
-0.6
-0.8
-1

10

15

Re
1

-2

-4

-6

10

12

14

14

Conclusion for Complex Poles

A complex pole appears in pair with
its complex conjugate
The Z-1-transform generates a
combination of exponentially
modulated sin and cos terms
The exponential base is the absolute
value of the complex pole
The frequency of the sinusoid is the
angle of the complex pole (divided
by 2)

Steady-State Analysis
If a signal finally converges, what value does
it converge to?
When it does not converge
Any |pj| is greater than 1
Any |r| is greater than or equal to 1

When it does converge

If all |pj|s and |r|s are smaller than 1, it
converges to 0
If only one pj is 1, then the signal converges to cj

If more than one real pole is 1, the signal does not

converge (e.g. the ramp signal)

-1
z
m
y(k) c0 uimpuls e(k) c j pkj -1 br k sin k dr k cos k (1 z 1)2
j 1

An Example
2z
z
3z

z 1 z 0.5 z 0.9
u(k) 2 0.5k 3 (0.9)k

U(z)

converge to 2

-1

10

20

30

40

50

60

Final Value Theorem

Enable us to decide whether a system
has a steady state error (yss-rss)

Final Value Theorem

Theorem: If all of the poles of (1 z )Y ( z ) lie within the unit circle, then
k lim y ( k ) z lim1 ( z 1)Y ( z )
0
-0.05
-0.1
-0.15
y(k)

0.11z
0.11z
Y ( z) 2

z 1.6 z 0.6 ( z 1)( z 0.6)

0.11z
( z 1)Y ( z ) |z 1
|z 1 0.275
z 0.6

-0.2
-0.25
-0.3
-0.35
0

10
k

If any pole of (1-z)Y(z) lies out of or ON the

unit circle, y(k) does not converge!

15

What Can We Infer from TF?

Almost everything we want to know
Stability
Steady-State
Transients
Settling time
Overshoot

Bounded Signals
5

5
a=0.4

5
a=0.9

a=1.2
0.5

-5

-5

-5

-0.5

5
a=-0.4

5
a=-0.9

a=-1.2

0.8

-1
0
1

0.5

0.6

-0.5
0.4

-5
0

-5
10 0

0.2

0
0

10

-5
10 0

5
15

20

10

-1
0

10

20

30

40

50

60

70

BIBO Stability
Bounded Input Bounded Output
Stability
If the Input is bounded, we want the
Output is bounded, too
If the Input is unbounded, its okay for the
Output to be unbounded

For some computing systems, the

output is intrinsically bounded
(constrained), but limit cycle may
happen

Limit Cycle
Output constrained,
But oscillating
Bad!
Imagine CPU utilization
Constantly switching from
1 to 0, 0 to 1,

Solution: make sure the system works in a linearized operating region

Are these BIBO?

Unity
P Controller
Integrator

y(k+1) = 1
y(k+1) = KP u(k)
y(k+1) = y(k) + u(k)

I Controller
M/M/1/K
Mystery

y(k+1) = y(k) + KI u(k)

y(k+1) = 0.49y(k) + 0.033u(k)
y(k+1) = -1.3y(k) + 2.3u(k)

Better Way to Decide BIBO or NOT

Theorem:
A system G(z) is BIBO stable iff all the poles of G(z) are inside the
unit circle.
System

Time domain Eq

Transfer Function

Poles

Unity

y(k+1) = 1

G(z) = 1

N/A

P
Controller

y(k+1) = KP u(k)

G(z) = KP

N/A

Integrator

y(k+1) = y(k) + u(k)

G(z) = 1/(z-1)

z=1

G(z) = KI/(z-1)

z=1

I Controller y(k+1) = y(k) + KI u(k)

M/M/1/K

y(k+1) = 0.49y(k) + 0.033u(k) G(z) = 0.033/(z-0.49) z = 0.49

Mystery

y(k+1) = -1.3y(k) + 2.3u(k)

G(z) = 2.3/(z+1.3)

z = -1.3

Example
u (k)=0.8k

LTI: y(k)=0.4y(k-1)+0.6u(k-1)

y (k)?

Z
Y(z) 0.4z 1Y(z) 0.6z 1U(z)

U(z)

1
1 0.8z1

0.6z1
0.6z-1
Y(z)
U(z)
1
1 0.4z
(1- 0.4z-1)(1- 0.8z-1)
Y(z)
0.6
G(z)

U(z) z 0.4

BIBO? only one pole at 0.4, so BIBO!

Steady State Gain

yss

Steady-State Gain Contd

Which value does the output
converges to when the input is an unit
step signal?
First of all, it has to converge

Final Value
Theorem

y s s lim y(k) lim(z 1)Y(z)

k

z 1

z
lim(z 1)G(z)
z 1
z 1
lim zG(z)
z 1

G(1)

Unit Step
Input

More General Cases

y(k) a1y(k 1) ... any(k n) b1u(k 1) ... bmu(k m)

Z
b1z 1 ... bm z m
Y(z)
U(z)
1
n
1 a1z ... an z

z=1
b1 ... bm
yss
1 a1 ... an

Transfer
Function

Example
u (k)=1

LTI: y(k)=0.4y(k-1)+0.6u(k-1)

y (k)?

Z
Y(z) 0.4z 1Y(z) 0.6z 1U(z)

U(z)

1
1 z 1

0.6z1
0.6z-1
Y(z)
U(z)
1
1 0.4z
(1- 0.4z-1)(1- z -1)
Y(z)
0.6
G(z)

U(z) z 0.4
Yss? G(1)=1, so yss=1

System Orders
System Order = Number of Poles
The higher the system order is, the
more complex the system behavior is
Some poles are more important than
others
Why?
If |pi|<|pj|,|pi/pj|k-1 approaches 0 when
k is large (pik-1 converges faster than pjk-1)
m

y(k) c0 uimpuls e(k) c j pkj -1

j 1

Overshoot and Setting Time

If not all poles are positive real
numbers, overshoot may happen
Easy to figure out when the system is first
order
For higher order systems, approximation to
first order systems works under certain
conditions

Setting time
First order system
Higher order systems

ks

4
ln|p|

How fast does it converge?

U(k)=ak, consider u(k)0 when the
absolute value of u(k) is smaller than
or equal to 2% of u(0)s absolute value
|p|k 0.02
kln|p| ln0.02 3.912
4
k
ln|p|

Remember
This!
p 0.7
k

4
4

11
ln|0.7| 0.36

1
0.9
0.8

y(k)=0.7

0.7
0.6
0.5
0.4
0.3

y(11)=0.0198

0.2
0.1
0

10

12

Examples: Positive Pole

1

0.1
z 0.9

0.8
0.6
0.4
0.2
0

10

15

20

25

30

0.09
(z 0.9)(z - 0.1)

0.8
0.6
0.4
0.2
0

10

15

20

25

30

1
0.8

0.07
(z 0.9)(z - 0.3)

0.6
0.4
0.2
0

10

15

20

25

30

1
0.8

0.063
(z 0.9)(z - 0.3)(z - 0.1)

0.6
0.4
0.2
0

10

15

20

25

30

Dominant
Pole: 0.9

Examples: Negative Pole

2
1.5

1.9
z 0.9

1
0.5
0

10

15

20

25

30

2
1.5

1.71
(z 0.9)(z - 0.1)

1
0.5
0

10

15

20

25

30

2
1.5
1
0.5
0

10

15

20

25

30

1.33
(z 0.9)(z - 0.3)

2
1.5

1.197
(z 0.9)(z - 0.3)(z - 0.1)

1
0.5
0

10

15

20

25

30

Dominant
Pole: -0.9

Dominant Pole
We can approximate a high-order
system with a first-order system with
the dominant pole of the high-order
system
IF the dominant pole DOES exist
Can give a pretty good estimation of
settling time
Can give a reasonable estimate of the
maximum overshoot

Some high-order systems do not have

dominant pole! for example

No Dominant Pole
Step Response
5
pole=-0.9
pole=-0.7
poles=-0.9, -0.7

4
3

Amplitude

1
0
-1
-2

-3

10

15

20

25
Time (sec)

30

35

40

45

50

Dominant Pole Contd

If there is a dominant pole, it must be
the pole with the maximum
magnitude
The largest pole should have at least
twice the magnitude of the other poles!

If the dominant pole is real (p), the

high-order system can be
approximated by a first-order system
G' (z)

G(1)(1 p' )
z p'

4
ks
ln|p'|

Summary
Signals/Systems

An LTI system can be specified by

Difference equation
Unit impulse response
Transfer function

If one is known, how to get the other two?

Characterize a signal with Z-transform

Z-domain (poles) -> Time domain (convergence,

etc.)

Characterize a system with Transfer function

BIBO stability
Steady-State Gain
Transients: overshoot, settling time
If there exists a dominant pole