EECE 301 Signals & Systems Prof.
Mark Fowler
Note Set #35
D-T Systems: Z-Transform Stability of Systems, Frequency Response Reading Assignment: Section 7.5 of Kamen and Heck
1/14
Course Flow Diagram
The arrows here show conceptual flow between ideas. Note the parallel structure between the pink blocks (C-T Freq. Analysis) and the blue blocks (D-T Freq. Analysis). New Signal Models
Ch. 1 Intro
C-T Signal Model Functions on Real Line System Properties LTI Causal Etc D-T Signal Model Functions on Integers
Ch. 3: CT Fourier Signal Models
Fourier Series Periodic Signals Fourier Transform (CTFT) Non-Periodic Signals
Ch. 5: CT Fourier System Models
Frequency Response Based on Fourier Transform
Ch. 6 & 8: Laplace Models for CT Signals & Systems
Transfer Function
New System Model
New System Model
Ch. 2 Diff Eqs
C-T System Model Differential Equations D-T Signal Model Difference Equations Zero-State Response Zero-Input Response Characteristic Eq.
Ch. 2 Convolution
C-T System Model Convolution Integral D-T System Model Convolution Sum
New System Model
New Signal Model Powerful Analysis Tool
Ch. 4: DT Fourier Signal Models
DTFT (for Hand Analysis) DFT & FFT (for Computer Analysis)
Ch. 5: DT Fourier System Models
Freq. Response for DT Based on DTFT
Ch. 7: Z Trans. Models for DT Signals & Systems
Transfer Function
New System Model
New System Model 2/14
Stability of DT Systems
For systems with rational H(z):
It is " BIBO" stable if
h[n] <
n =0
Recall : H ( z ) =
B( z ) A( z )
Where A( z ) = z N + a1 z N 1 + ... + a N 1 z + a N
Any common roots in B(z) and A(z) are assumed to have been cancelled.
Let A( z ) have roots of p1 , p2 ,..., p N poles of H(z)
Then H ( z ) = B( z ) ( z p1 )( z p2 )...( z p N )
and
h[ n] = h1[n] + h2 [n] + ... + hN [n] Note : each h1[n ] will have ( pi ) n u[n ]
decays if pi < 1
3/14
Result:
h[n] <
n =0
is equivalent to
all pi < 1
System is stable
i.e., poles are inside the unit circle
System is stable
Im{z}
unit circle
Re{z}
1
For a Stable System
Poles must be inside unit circle Zeros can be anywhere
Aside: Complex poles and complex zeros must occur as conjugate pairs
4/14
Frequency Response
All the same as for the CT case! (e.g. how sinusoids go through, how general signals go through)
H () = h[n ]e jn = H ( z ) z =e j
n =0
Using Matlab to Compute Frequency Response:
b0 + b1 z 1 + b2 z 2 + ... + bN z N H ( z) = a0 + a1 z 1 + a2 z 2 + ... + a N z N
>> num = [b0 b1 ... bN ]
Some bi may be 0 Some ai may be 0
must put any zero bi into the vector must put any zero ai into the vector Pick appropriate spacing
>> den = [a0 a1 ... a N ]
>> omega = -pi : ? : pi
>> H = freqz(num, denom, omega) >> plot(omega/pi, abs(H)) >> plot(omega/pi, angle(H))
5/14
Relationship between the ZT and the DTFT
Recall: H () = H ( z ) j z =e Consider the Z-Transform given by:
Lets explore this idea with some pictures for an explicit case
This causes H(z) = 0 for z = 0
H ( z) =
(1 0.8e
j 0.3
z 1 )(1 0.8e j 0.3 z 1 )
(z 0.8e
j 0.3
)(z 0.8e
j 0.3
These cause H(z) = for z = 0.8ej0.3
Pole-Zero Plot For This H(z)
Im{z} 0.3 0.8 Re{z}
6/14
So from this pole-zero plot we can then imagine that the plot of the |H(z)| might look something like this:
Pole-Zero Plot For This H(z)
Im{z} 0.3 0.8 Re{z}
|H(z)|
And we know that the Frequency Response is just the Transfer Function evaluated on the Unit Circle.
H () = H ( z ) z =e j
7/14
Now plot just those values on the unit circle:
NowCut here and unwrap
This shows the Frequency Response H() where is the angle around the unit circle this explains why H() is a periodic function of
8/14
This shows the previous plot cut and unwrapped and plotted on the axis:
Normalized for convenience
9/14
Effect of Poles & Zeros on Frequency Response of DT filters
Im{z} {z} Im{z} {z}
Note: Including a pole or zero at the origin
doesnt change the magnitude but does change the phase
Placing a zero at
Im{z} {z} Im{z}
makes |H()| = 0
Placing more zeros/poles
{z} Im{z} {z}
gives sharper transitions.
Figure from B.P. Lathi, Signal Processing and Linear Systems
10/14
So from these plots and ideas we can see that we could design simple DT filters by deciding where to put poles and zeros.
This is not a very good design approach but this insight is crucial to understanding transfer functions.
The following charts in this set of notes shows a filter designed not by placing poles and zeros but rather by using one really good computer-based design method for designing DT filters.
11/14
A practical DT filter (Designed using MATLABs remez)
(See Digital Signal Processing course to learn the design process) Here is the impulse response h[n] it is assumed to be zero where not shown Note that it has only finite many non-zero samples
Called a Finiteimpulse Response (FIR) filter
12/14
All these zeros, right on the unit circle, pull the frequency response down to create the stop band
13/14
The effect of the zeros on the unit circle
Note, this filter has linear phase in the passband this is the ideal phase response (as we saw back in Ch. 5 for CT filters) FIR DT filters are well-suited to getting linear phase and are therefore very widely used.
14/14
