Lec 8 S12

Outline Introduction Relation Between LT and ZT Analyzing LTI Systems with ZT Geometric Evaluation Unilateral ZT
Signals and Systems

Lecture 8: Z Transform
Farzaneh Abdollahi
Department of Electrical Engineering
Amirkabir University of Technology
Winter 2012
Farzaneh Abdollahi Signal and Systems Lecture 8 1/29
Introduction
Relation Between LT and ZT
ROC Properties
The Inverse of ZT
ZT Properties
Analyzing LTI Systems with ZT
Geometric Evaluation
LTI Systems Description
Unilateral ZT
Z transform (ZT) is extension of DTFT
Like CTFT and DTFT, ZT and LT have similarities and dierences.
We had dened x[n] = z

n
as a basic function for DT LTI systems,s.t.
Z
n
H(z)z
n
In Fourier transform z = e
j
, in other words, |z| = 1
In Z transform z = re
j
By ZT we can analyze wider range of systems comparing to Fourier

Transform
The bilateral ZT is dened:

X(z) =
x[n]z
n
X(re
j
) =
x[n](re
j
)
n
=
{x[n]r
n
}e
j n
= F{x[n]r
n
}
Region of Convergence (ROC)
Note that: X(z) exists only for a specic region of z which is called
Region of Convergence (ROC)
ROC: is the z = re
j
by which x[n]r
n
converges:
ROC : {z = re
j
s.t.
n=
|x[n]r
n
| < }
Roc does not depend on
Roc is absolute summability condition of x[n]r

n
If r = 1, i,e, z = e
j
X(z) = F{x[n]}
ROC is shown in z-plane

Example
Consider x[n] = a
n
u[n]
X(z) =
n=
a
n
u[n]z
n
=
n=0
(az
1
)
n
If |z| > |a| X(z) is bounded
X(z) =
z
za
, ROC : |z| > |a|
Example
Consider x[n] = a
n
u[n 1]
X(z) =
n=
a
n
u[n 1]z
n
1
n=0
(a
1
z)
n
If |a
1
z| < 1 |z| < |a|, X(z) is bounded
X(z) =
z
za
, ROC : |z| < |a|
In the recent two examples two dierent signals had similar ZT but with
dierent Roc
To obtain unique x[n] both X(z) and ROC is required
If X(z) =
N(z)
D(z)
Roots of N(z) zeros of X(z); They make X(z) equal to zero.
Roots of D(z) poles of X(z); They make X(z) to be unbounded.

Relation Between L and ZT
In LT: x(t)
L
X(s) =
_
x(t)e
st
dt = L{x(t)}
Now dene t = nT:

X(s) = lim
T0
n=
x(nT)(e
sT
)
n
.T =
lim
T0
T
n=
x[n](e
sT
)
n
In ZT: x[n]
Z
X(z) =
n=
x[n]z
n
= Z{x[n]}
by taking z = e
sT
ZT is obtained from LT.
j axis in s-plane is changed to unite circle in z-plane

z-Plane s-Plane
|z| < 1 (insider the unit circle) Re{s} < 0 (LHP)
special case: z = 0 special case: Re{s} =
|z| > 1 (outsider the unit circle) Re{s} > 0 (RHP)
special case: z = special case: Re{s} =
|z| = cte (a circle) Re{s} = cte (a vertical line)
ROC Properties
ROC of X(z) is a ring in z-plane centered at origin
ROC does not contain any pole
If x[n] is of nite duration, then the ROC is the entire z-plane, except
possibly z = 0 and/or z =
X(z) =
N
2
n=N
1
x[n]z
n
If N
1
< 0x[n] has nonzero terms for n < 0, when |z| 0 negative power
of z will be unbounded
If N
2
> 0x[n] has nonzero terms for n > 0, when |z| positive power
of z will be unbounded
If N
1
0 only negative powers of z exist z = ROC
If N
2
0 only positive powers of z exist z = 0 ROC
Example:
ZT LT
[n] = 11 ROC: all z (t)1 ROC: all z
[n 1]z
1
ROC: z = 0 (t T)e
sT
ROC: Re{s} =
[n + 1]z ROC: z = (t + T)e
sT
ROC: Re{s} =
ROC Properties
If x[n] is a right-sided sequence and if the circle [z] = r

0
is in the ROC,
then all nite values of z for which |z| > r
0
will also be in the ROC.
If x[n] is a left-sided sequence and if the circle [z] = r

0
is in the ROC,
then all nite values of z for which |z| < r
0
will also be in the ROC.
If x[n] is a two-sided sequence and if the circle [z] = r

0
is in the ROC,
then ROC is a ring containing |z| = r
0
.
If X(z) is rational
The ROC is bounded between poles or extends to innity,
no poles of X(s) are contained in ROC
If x[n] is right sided, then ROC is in the out of the outermost pole
If x[n] is causal and right sided then z = ROC
If x[n] is left sided, then ROC is in the inside of the innermost pole
If x[n] is anticausal and left sided then z = 0 ROC
If ROC includes |z| = 1 axis then x[n] has FT

The Inverse of Z Transform (ZT)
By considering r xed, inverse of ZT can be obtained from inverse of FT:
x[n]r
n
=
1
2
_
2
X(re
j
..
z
)e
j n
d
x[n] =
1
2
_
2
X(re
j
)r
n
e
(j )n
d
assuming r is xed dz = jzd
x[n] =
1
2j
_
X(z)z
n1
dz
Methods to obtain Inverse ZT:

1. If X(s) is rational , we can use expanding the rational algebraic into a
linear combination of lower order terms and then one may use
X(z) =
A
i
1a
i
z
1
x[n] = A
i
a
i
u[n] if ROC is out of pole z = a
i
X(z) =
A
i
1a
i
z
1
x[n] = A
i
a
i
u[n 1] if ROC is inside of z = a
i
Do not forget to consider ROC in obtaining inverse of ZT!
2. If X is nonrational, use Power series expansion of X(z), then apply
[n + n
0
]z
n
0
Example: X(z) = 5z
2
z + 3z
3
x[n] = 5[n + 2] [n + 1] + 3[n 3]

3. If X is rational, power series can be obtained by long division
Example: X(z) =
1
1az
1
, |z| > |a|
1
1az
1
1+az
1
+(az
1
)
2
+...
1 + az
1
az
1
az
1
+ a
2
z
2
.
.
.
x[n] = a
n
u[n]
Example: X(z) =
1
1az
1
, |z| < |a|
X(z) = a
1
z(
1
1a
1
z
)
1
1a1z
1+a1z+(a
1
z)
2
+...
1 + a1z
a1z
a1z + az
2
z
2
.
.
.
x[n] = 1a
n
u[n 1]
ZT Properties
Linearity: ax
1
[n] + bx
2
[n]aX
1
(z) + bX
2
(z)
ROC contains: R
1
R
2
If R
1
R
2
= it means that Lp does not exit
By zeros and poles cancelation ROC can be lager than R

1
R
2
Time Shifting:x[n n
0
]z
n
0
X(z) with ROC=R (maybe 0 or is
added/omited)
If n
0
> 0z
n
0
may provide poles at origin 0 ROC
If n
0
< 0z
n
0
may eliminate infrty from ROC
Time Reversal: x[n]X(

1
z
) with ROC=
1
R
Scaling in Z domain: z
n
0
x[n]X(
z
z
0
) with ROC = |z
0
|R
If X(z) has zero/poles at z = aX(

z
z
0
) has zeros/poles at z = z
0
a
Dierentiation in the z-Domain: nx[n]z

dX(z)
dz
with ROC = R
Convolution: x
1
[n] x
2
[n]X
1
(z)X
2
(z) with ROC containing R
1
R
2
ZT Properties
Conjugation: x
[n]X
(z
) with ROC = R
If x[n] is real
X(z) = X
(z
If X(z) has zeros/poles at z

0
it should have zeros/poles at z
0
as well
Initial Value Theorem: If x[n] = 0 for n < 0 then x[0] = lim

z
X(z)
lim
z
X(z) = lim
z
n=0
x[n]z
n
= x[0] + lim
z
n=1
x[n]z
n
=
x[0]
For a casual x[n], if x[0] is bounded it means # of zeros are less than or
equal to # of poles (This is true for CT as well)
Final Value Theorem: If x[n] = 0 for n < 0 and x[n] is bounded when
n then x() = lim
z1
(1 z
1
)X(z)
Analyzing LTI Systems with ZT
ZT of impulse response is H(z) which is named transfer function or

system function.
Transfer fcn can represent many properties of the system:
Casuality: h[n] = 0 for n < 0 It is right sided
ROC of a causal LTI system is out of a circle in z-plane, it includes
Note that the converse is not always correct
For a system with rational transfer fcn, causality is equivalent to ROC being
outside of the outermost pole (degree of nominator should not be greater
than degree od denominator)
Stability: h[n] should be absolute summable its FT converges
An LTI system is stable i its ROC includes unit circle
A causal system with rational H(z) is stable i all the poles of H(z) are
inside the unit circle
DC gain in DT is H(1) and in CT is H(0)

Geometric Evaluation of FT by Zero/Poles Plot
Consider X
1
(z) = z a
|X
1
|: length of X
1
X
1
: angel of X
1
Now consider X
2
(s) =
1
za
=
1
X
1
(s)
|X
2
| =
1
X
1
(z)|
X
2
= X
1
For higher order fcns:

X(z) = M
R
i =1
(z
i
)
P
j =1
(z
j
)
|X(z)| = |M|
R
i =1
|z
i
|
P
j =1
|z
j
|
X(z) =
M +
R
i =1
(z
i
)
R
j =1
(z
j
)
Example:H(s) =
1
1+az
1
, |z| > |a|, a real
h(t) = a
n
u[n]
H(e
j
) =
v
1
v
2
|H(e
j
)| =
1
|v
2
|
at = 0, |H(e
j
)| =
1
1a
is max
0 < < : |H(e

j
)|
at = , |H(e
j
)| =
1
1+a
H(e
j
) = v
1
v
2
= v
2
at = 0, H(e
j
) = 0
0 < < , H(e

j
)
at = , H(e
j
) = 0
First Order Systems
a in Dt rst order systems plays a similar role of time constant in CT
|a|
|H(e
j
)| at = 0
Impulse response decays more rapidly
Step response settles more quickly
In case of having multiple poles, the poles closer to the origin, decay more
rapidly in the impulse response
Second Order Systems
consider a second order system with poles at z

1
= re
j
, z
2
= re
j
H(z) =
z
2
(zz
1
)(zz
2
)
=
1
12rcosz
1
+r
2
z
2
, 0 < r < 1, 0 < <
h[n] =
r
n
sin(n+1)
sin
u[n]
H = 2v
1
v
2
Starting from = 0 to = :
At rst v
2
is decreasing
v
2
is min at = max |H|
Then v
2
is increasing
If r is closer to unit, then |H| is greater at poles and change of H is

sharper
If r is closer to the origin, impulse repose decays more rapidly and step
response settles more quickly.
Bode Plot of H(j )
LTI Systems Description

N
k=0
a
k
y[n k] =
M
k=0
b
k
x[n k]

N
k=0
a
k
s
k
Y(z) =
M
k=0
b
k
s
k
X(z)
H(z) =
Y(z)
X(z)
=
M
k=0
b
k
z
k
N
k=0
a
k
z
k
ROC depends on
placement of poles
boundary conditions (right sided, left sided, two sided,...)

Feedback Interconnection of two LTI systems:
Y(z) = Y
1
(z) = X
2
(z)
X
1
(z) = X(z) + Y
2
(z) = X(z) + H
2
(z)Y(z)
Y(z) = H
1
(z)X
1
(z) = H
1
(z)[X(z) + H
2
(z)Y(z)]
Y(z)
X(z)
= H(z) =
H
1
(z)
1H
2
(z)H
1
(z)
ROC: is determined based on roots of 1 H

2
(z)H
1
(z)
Block Diagram Representation for Casual LTI Systems
We can represent a transfer fcn by dierent methods:
Example: H(z) =
12z
1
1
1
4
z
1
= (
1
1
1
4
z
1
)(1 2z
1
)
Unilateral ZT
It is used to describe casual systems with nonzero initial conditions:

X(z) =
x[n]z
n
= UZ{x[n]}
If x[n] = 0 for n < 0 then X(z) = X(z)
Unilateral ZT of x[n] = Bilateral LT of x[n]u[n]
If h[n] is impulse response of a casual LTI system then H(z) = H(z)
ROC is not necessary to be recognized for unilateral ZT since it is always

outside of a circle
For rational X(z), ROC is outside of the outmost pole

Similar Properties of Unilateral and Bilateral ZT
Convolution: Note that for unilateral ZT, If both x

1
[n] and x
2
[n] are zero
for t < 0, then X(z) = X
1
(z)X
2
(z)
Time Scaling
Time Expansion
Initial and Finite Theorems: they are indeed dened for causal signals
Dierentiating in z domain:
The main dierence between UL and ZT is in time dierentiation:
UZ{x[n 1]} =
0
x[n 1]z
n
= x[1] +
n=1
x[n 1]z
n
=
x[1] +
m=0
x[m]z
m1
UZ{x[n 1]} = x[1] + z

1
X(z)
UZ{x[n 2]} = x[2] + z

1
x[1] + z
2
X(z)
UZ{x[n + 1]} =
n=0
x[n + 1]z
n
=
m=1
x[m]z
m+1
x[0]z
UZ{x[n + 1]} = zX(z) zx[0]
UZ{x[n + 2]} = z
2
X(z) z
2
x[0] zx[1]
Follow the same rule for higher orders

Example
Consider y[n] + 2y[n 1] = x[n], where y[1] = , x[n] = u[n]
Take UZ:
Y(z) + 2[ + z
1
]Y(z) = X(s)
Y(z) =
2
1 + 2z
1
. .
ZIR
+
X(z)
1 + 2z
1
. .
ZSR
Zero State Response (ZSR): is a response in absence of initial values
H(z) =
Y(z)
X(z)
Transfer fcn is ZSR
ZSR: Y
1
(z) =
2
1+2z
1
y
1
[n] = 2(2)
n
u[n]
Zero Input Response (ZIR): is a response in absence of input (x[n] = 0)
ZIR: Y
2
(z) =

(1+2z
1
)(1z
1
)
y[n] = y
1
[n] + y
2
[n]

Lec 8 S12

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lec 8 S12

Uploaded by

Copyright:

Available Formats

Outline Introduction Relation Between LT and ZT Analyzing LTI Systems with ZT Geometric Evaluation Unilateral ZT

Signals and Systems

Z transform (ZT) is extension of DTFT

Like CTFT and DTFT, ZT and LT have similarities and dierences.

We had dened x[n] = z

By ZT we can analyze wider range of systems comparing to Fourier

The bilateral ZT is dened:

Roc does not depend on

Roc is absolute summability condition of x[n]r

ROC is shown in z-plane

If |z| > |a| X(z) is bounded

To obtain unique x[n] both X(z) and ROC is required

Roots of N(z) zeros of X(z); They make X(z) equal to zero.

Roots of D(z) poles of X(z); They make X(z) to be unbounded.

Now dene t = nT:

j axis in s-plane is changed to unite circle in z-plane

ROC of X(z) is a ring in z-plane centered at origin

ROC does not contain any pole

If x[n] is a right-sided sequence and if the circle [z] = r

If x[n] is a left-sided sequence and if the circle [z] = r

If x[n] is a two-sided sequence and if the circle [z] = r

The ROC is bounded between poles or extends to innity,

no poles of X(s) are contained in ROC

If x[n] is causal and right sided then z = ROC

If x[n] is anticausal and left sided then z = 0 ROC

If ROC includes |z| = 1 axis then x[n] has FT

By considering r xed, inverse of ZT can be obtained from inverse of FT:

assuming r is xed dz = jzd

Methods to obtain Inverse ZT:

x[n] = 5[n + 2] [n + 1] + 3[n 3]

By zeros and poles cancelation ROC can be lager than R

Time Reversal: x[n]X(

If X(z) has zero/poles at z = aX(

Dierentiation in the z-Domain: nx[n]z

If X(z) has zeros/poles at z

Initial Value Theorem: If x[n] = 0 for n < 0 then x[0] = lim

ZT of impulse response is H(z) which is named transfer function or

Transfer fcn can represent many properties of the system:

Casuality: h[n] = 0 for n < 0 It is right sided

ROC of a causal LTI system is out of a circle in z-plane, it includes

Note that the converse is not always correct

Stability: h[n] should be absolute summable its FT converges

An LTI system is stable i its ROC includes unit circle

DC gain in DT is H(1) and in CT is H(0)

For higher order fcns:

0 < < : |H(e

0 < < , H(e

a in Dt rst order systems plays a similar role of time constant in CT

Impulse response decays more rapidly

Step response settles more quickly

consider a second order system with poles at z

If r is closer to unit, then |H| is greater at poles and change of H is

boundary conditions (right sided, left sided, two sided,...)

Feedback Interconnection of two LTI systems:

ROC: is determined based on roots of 1 H

We can represent a transfer fcn by dierent methods:

It is used to describe casual systems with nonzero initial conditions:

If x[n] = 0 for n < 0 then X(z) = X(z)

Unilateral ZT of x[n] = Bilateral LT of x[n]u[n]

If h[n] is impulse response of a casual LTI system then H(z) = H(z)

ROC is not necessary to be recognized for unilateral ZT since it is always

For rational X(z), ROC is outside of the outmost pole

Convolution: Note that for unilateral ZT, If both x