ELC 4351: Digital Signal Processing

Liang Dong

Electrical and Computer Engineering

Baylor University
liang [email protected]

April 13, 2017

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 1 / 20
Frequency-domain Analysis of LTI Systems

1 Linear Time-Invariant Systems as Frequency-Selective Filters

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 2 / 20
Linear Time-Invariant Systems as Frequency-Selective
Filters

A LTI system performs a type of

discrimination or filtering among the
various frequency components at its
input.

The nature of this filtering action is

determined by the frequency response
characteristics H(ω).

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 3 / 20
Linear Time-Invariant Systems as Frequency-Selective
Filters

By proper selection of the coefficients

ak ’s and bk ’s, we can design
frequency-selective filters.

These filters pass signals with frequency

components in some bands while they
attenuate signals containing frequency
components in other frequency bands.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 4 / 20
Ideal Filter Characteristics
A filter with frequency response

Ce −jωn0 , ω1 < ω < ω2


H(ω) =
0, otherwise

where C and n0 are constants.

Y (ω) = X (ω)H(ω) = CX (ω)e −jωn0

y (n) = Cx(n − n0 )

The filter output is simply a delayed and amplitude-scaled version of

the input signal.
A pure delay is usually tolerable and is not considered a distortion of
the signal. Neither is amplitude scaling.
Therefore, ideal filters have a linear phase characteristic within their
passband, that is,
Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 5 / 20
Ideal Filter Characteristics
A filter with frequency response

Ce −jωn0 , ω1 < ω < ω2


H(ω) =
0, otherwise

where C and n0 are constants.

Ideal filters have a linear phase characteristic within their passband, that is,

Θ(ω) = −ωn0

Group delay of the filter

dΘ(ω)
τg (ω) = −
dω
Linear phase = group delay is constant. In this case, all frequency
components of the input signal undergo the same time delay.
Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 6 / 20
Ideal Filter Characteristics

“Ideal” filter:
Impulse response is a sinc function.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 7 / 20
Ideal Filter Characteristics

“Ideal” filter:
Impulse response is a sinc function.
This filter is not causal and it is not absolutely summable and
therefore unstable.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 7 / 20
Ideal Filter Characteristics

“Ideal” filter:
Impulse response is a sinc function.
This filter is not causal and it is not absolutely summable and
therefore unstable.

Design some simple digital filters by the placement of poles and zeros
in the z -plane.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 7 / 20
Ideal Filter Characteristics

“Ideal” filter:
Impulse response is a sinc function.
This filter is not causal and it is not absolutely summable and
therefore unstable.

Design some simple digital filters by the placement of poles and zeros
in the z -plane.
The location of poles and zeros affects the frequency response
characteristics of the system.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 7 / 20
The Pole-Zero Placement Method

The basic principle underlying the pole-zero placement method:

Locate poles near points of the unit circle corresponding to
frequencies to be emphasized, and

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 8 / 20
The Pole-Zero Placement Method

The basic principle underlying the pole-zero placement method:

Locate poles near points of the unit circle corresponding to
frequencies to be emphasized, and
Locate zeros near the frequencies to be deemphasized.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 8 / 20
The Pole-Zero Placement Method

The basic principle underlying the pole-zero placement method:

Locate poles near points of the unit circle corresponding to
frequencies to be emphasized, and
Locate zeros near the frequencies to be deemphasized.
All poles should be placed inside the unit circle in order for the filter
to be stable.
However, zeros can be placed anywhere in the z -plane.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 8 / 20
The Pole-Zero Placement Method

The basic principle underlying the pole-zero placement method:

Locate poles near points of the unit circle corresponding to
frequencies to be emphasized, and
Locate zeros near the frequencies to be deemphasized.
All poles should be placed inside the unit circle in order for the filter
to be stable.
However, zeros can be placed anywhere in the z -plane.
All complex zeros and poles must occur in complex-conjugate pairs in
order for the filter coefficients to be real.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 8 / 20
The Pole-Zero Placement Method

The system function:

QM
k=1 (1 − zk z −1 )
H(z) = b0 QN
k=1 (1 − pk z −1 )

Usually, b0 is selected such that |H(ω0 )| = 1. ω0 in the passband of the

filter.
N ≥ M.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 9 / 20
Lowpass, Highpass, and Bandpass Filters
Design of lowpass digital filters: the poles should be placed near the
unit circle at points corresponding to low frequencies (near ω = 0)
and zeros should be placed near or on the unit circle at points
corresponding to high frequencies (near ω = π).
Design of highpass digital filters: The opposite.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 10 / 20
A Simple Lowpass-to-Highpass Filter Transformation
Frequency translation of π rad:

Hhp (ω) = Hlp (ω − π)

Therefore,
hhp (n) = e jπn hlp (n) = (−1)n hlp (n)

e.g., Lowpass filter by difference eqn.

N
X M
X
y (n) = − ak y (n − k) + bk x(n − k)
k=1 k=0

A highpass filter can be derived: (How?)

N
X M
X
y (n) = − (−1)k ak y (n − k) + (−1)k bk x(n − k)
k=1 k=0

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 11 / 20
Digital Resonator

A digital resonator is a special two-pole

bandpass filter with the pair of
complex- conjugate poles located near
the unit circle.
The filter has a large magnitude
response (i.e., it resonates) in the
vicinity of the pole location.
The angular position of the pole
determines the resonant frequency of
the filter.
Digital resonators are useful in many
applications, including bandpass
filtering and speech generation.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 12 / 20
Digital Resonator

A resonant peak at or near ω = ω0 ,

p1,2 = re ±jω0 , 0 < r < 1

We can select up to two zeros –

One choice is to locate the zeros at the origin.

The other choice is to locate a zero at z = 1 and a zero at z = −1. This

choice completely eliminates the response of the filter at frequencies ω = 0
and ω = π.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 13 / 20
Digital Resonator

Digital resonator with zeros at the origin:

b0
H(ω) = −jω
(1 − jω
re e )(1
0 − re −jω0 e −jω )

We select b0 so that |H(ω0 )| = 1.

b0
H(ω0 ) =
(1 − re jω0 e −jω0 )(1 − re −jω0 e −jω0 )
b0
=
(1 − r )(1 − re −j2ω0 )
b0
|H(ω0 )| = p =1
(1 − r ) 1 + r 2 − 2r cos 2ω0
p
b0 = (1 − r ) 1 + r 2 − 2r cos 2ω0

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 14 / 20
Digital Resonator

Digital resonator with zeros at the origin:

b0
H(ω) = −jω
(1 − jω
re e )(1
0 − re −jω0 e −jω )

b0
|H(ω0 )| =
U1 (ω)U2 (ω)
∠H(ω) = 2ω − Φ1 (ω) − Φ2 (ω)
q
U1 (ω) = 1 + r 2 − 2r cos(ω0 − ω)
q
U2 (ω) = 1 + r 2 − 2r cos(ω0 + ω)

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 15 / 20
Digital Resonator

q
U1 (ω) = 1 + r 2 − 2r cos(ω0 − ω)
q
U2 (ω) = 1 + r 2 − 2r cos(ω0 + ω)

1 + r2
 
−1
min U1 (ω)U2 (ω) =⇒ ωr = cos cos ω0
ω 2r

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 16 / 20
Digital Resonator

Digital resonator with zeros z = 1 and z = −1:

(1 − e −jω )(1 + e −jω )

H(ω) = b0
(1 − re jω0 e −jω )(1 − re −jω0 e −jω )

p
2(1 − cos 2ω)
|H(ω)| = b0
U1 (ω)U2 (ω)

The actual resonant frequency is altered.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 17 / 20
All-Pass Filters

|H(ω)| = 1, 0 ≤ ω ≤ π
e.g.,
1 a pure delay system H(z) = z −k .
2

PN
k=0 ak z −N+k
H(z) = N
, a0 = 1
−k
P
k=0 ak z
A(z −1 )
= z −N
A(z)
PN −k .
where A(z) = k=0 ak z

|H(ω)|2 = H(z)H(1/z)|z=e jω = 1

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 18 / 20
All-Pass Filters

If z0 is a pole of H(z), then 1/z0 is a zero of H(z).

The poles and zeros are reciprocals of one another.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 19 / 20
All-Pass Filters

All-pass filter with real coefficients:

NR NC
Y z −1 − αk Y (z −1 − βk )(z −1 − βk∗ )
Hap (z) =
1 − αk z −1 (1 − βk z −1 )(1 − βk∗ z −1 )
k=1 k=1

where there are NR real poles and zeros and NC complex-conjugate pairs
of poles and zeros.

For causal and stable systems, 1 < αk < 1 and |β| < 1.

Q: What is all-pass filter for?

A: All-pass filters find application as phase equalizers. When placed in
cascade with a system that has an undesired phase response, a phase
equalizer is designed to compensate for the poor phase characteristics of
the system and therefore to produce an overall linear-phase response.

Liang Dong (Baylor University) Frequency-domain Analysis of LTI Systems April 13, 2017 20 / 20