EE3054

Signals and Systems

Frequency Response of
Continuous Time LTI
Systems
Yao Wang
Polytechnic University
Most of the slides included are extracted from lecture presentations prepared by
McClellan and Schafer
3/28/2008 2003, JH McClellan & RW Schafer 2
License Info for SPFirst Slides
This work released under a Creative Commons License
with the following terms:
Attribution
The licensor permits others to copy, distribute, display, and perform
the work. In return, licensees must give the original authors credit.
Non-Commercial
The licensor permits others to copy, distribute, display, and perform
the work. In return, licensees may not use the work for commercial
purposesunless they get the licensor's permission.
Share Alike
The licensor permits others to distribute derivative works only under
a license identical to the one that governs the licensor's work.
Full Text of the License
This (hidden) page should be kept with the presentation
3/28/2008 2003, JH McClellan & RW Schafer 3
LECTURE OBJECTIVES
Review of convolution

THE
THE operation for
LTI
LTI Systems
Complex exponential input signals
Frequency Response
Cosine signals
Real part of complex exponential
Fourier Series thru H(j)
These are Analog Filters
3/28/2008 2003, JH McClellan & RW Schafer 4
LTI Systems
Convolution defines an LTI system
Response to a complex exponential gives
frequency response H(j)
y(t) = h(t) x(t) = h( )

x(t )d
3/28/2008 2003, JH McClellan & RW Schafer 5
Thought Process #1
SUPERPOSITION (Linearity)
Make x(t) a weighted sum of signals
Then y(t) is also a sumsame weights
But DIFFERENT OUTPUT SIGNALS usually
Use SINUSOIDS
SINUSOID IN GIVES SINUSOID OUT
Make x(t) a weighted sum of sinusoids
Then y(t) is also a sum of sinusoids
Different Magnitudes and Phase
LTI SYSTEMS: Sinusoidal Response
3/28/2008 2003, JH McClellan & RW Schafer 6
Thought Process #2
SUPERPOSITION (Linearity)
Make x(t) a weighted sum of signals

Use
Use
SINUSOIDS
SINUSOIDS
Any Any x(t) = weighted sum of sinusoids x(t) = weighted sum of sinusoids
HOW? HOW? Use FOURIER ANALYSIS INTEGRAL Use FOURIER ANALYSIS INTEGRAL
To find the weights from x(t) To find the weights from x(t)
LTI SYSTEMS:
Frequency Response changes each sinusoidal
component
3/28/2008 2003, JH McClellan & RW Schafer 7
Complex Exponential Input
t j j t j j
e Ae j H t y e Ae t x

) ( ) ( ) ( = = a

= =

d e Ae h t h t x t y
t j j ) (
) ( ) ( * ) ( ) (
Frequency
Response

=

d e h j H
j
) ( ) (
t j j j
e Ae d e h t y


|
|

\
|
=

) ( ) (
3/28/2008 2003, JH McClellan & RW Schafer 8
When does H(j) Exist?
When is ?
Thus the frequency response exists if the LTI
system is a stable system.
H( j) = h( )

e
j
d h( )

e
j
d
H( j) h()

d <
H( j) <
3/28/2008 2003, JH McClellan & RW Schafer 9
Suppose that h(t) is:
h(t) = e
t
u(t)
H( j) = e
a
u(

)e
j
d = e
(a+ j)
d
0

H( j) =
e
(a+ j)
(a + j)
0

=
e
a
e
j
(a + j)
0

=
1
a + j
h(t) = e
at
u(t) H( j) =
1
a + j
a = 1
a > 0
3/28/2008 2003, JH McClellan & RW Schafer 10
Magnitude and Phase Plots
1
1+ j
=
1
1+
2
H( j) = atan()
H( j) =
1
1+ j
H( j) = H

( j)
3/28/2008 2003, JH McClellan & RW Schafer 11
Freq Response of
Integrator?
Impulse Response
h(t) = u(t)
NOT a Stable System
Frequency response H(j) does NOT exist
h(t) = e
at
u(t) H( j) =
1
a + j

1
j
?
Need another term
a 0
Leaky Integrator (a is small)
Cannot build a perfect Integral
) r! (integrato ) ( ) ( * ) ( ) (

= =
t
d x t h t x t y
3/28/2008 2003, JH McClellan & RW Schafer 12
Example: Rectangular pulse
h(t)=u(t)-u(t-10)
Show H(jw) is a sinc function
) average! or past in the interval short a over r (integrato
) ( ) ( * ) ( ) (
10

= =
t
t
d x t h t x t y
3/28/2008 2003, JH McClellan & RW Schafer 13
( )
t j t j t t j
t j
e e e t y
e t x
d d

= =
=
) (
) (
) ( a
Ideal Delay:
y(t) = x(t t
d
)
H( j) = e
jt
d
H( j) = ( t
d
)

e
j
d = e
jt
d
H( j)
3/28/2008 2003, JH McClellan & RW Schafer 14
Delay system is All-pass with linear phase
3/28/2008 2003, JH McClellan & RW Schafer 15
Ideal Lowpass Filter w/ Delay
H
LP
( j) =
e
jt
d
<
co
0 >
co



f
co
"cutoff freq."
Magnitude
Linear Phase

3/28/2008 2003, JH McClellan & RW Schafer 16

Example: Ideal Low Pass
H
LP
( j) =
e
j 3
< 2
0 > 2



= = ) ( 10 ) (
5 . 1 3 /
t y e e t x
t j j
a

t j j
e e j H
5 . 1 3 /
10 ) 5 . 1 (

( )
) 3 ( 5 . 1 3 / 5 . 1 3 / 5 . 4
10 10 ) (

= =
t j j t j j j
e e e e e t y

3/28/2008 2003, JH McClellan & RW Schafer 17
Cosine Input
x(t) = Acos(
0
t +) =
A
2
e
j
e
j
0
t
+
A
2
e
j
e
j
0
t
y(t) = H( j
0
)
A
2
e
j
e
j
0
t
+ H( j
0
)
A
2
e
j
e
j
0
t
Since H( j
0
) = H

(j
0
)
y(t) = A H( j
0
) cos(
0
t + + H( j
0
))
3/28/2008 2003, JH McClellan & RW Schafer 18
Sinusoid in Gives Sinusoid out
3/28/2008 2003, JH McClellan & RW Schafer 19
Review Fourier Series
ANALYSIS
Get representation from the signal
Works for PERIODIC Signals
Fourier Series
INTEGRAL over one period
a
k
=
1
T
0
x(t)e
j
0
kt
dt
0
T
0

3/28/2008 2003, JH McClellan & RW Schafer 20

General Periodic Signals
x(t) = x(t + T
0
)
T
0
2T
0
T
0
2T
0
0
t
x(t) = a
k
e
j
0
k t
k =

a
k
=
1
T
0
x(t)e
j
0
kt
dt
0
T
0

Fundamental Freq.

0
= 2 / T
0
= 2f
0
Fourier Synthesis
Fourier Analysis
3/28/2008 2003, JH McClellan & RW Schafer 21
Square Wave Signal
x(t) = x(t + T
0
)
T
0
2T
0
T
0
2T
0
0
t
a
k
=
e
j
0
kt
j
0
kT
0
0
T
0
/ 2

e
j
0
kt
j
0
kT
0
T
0
/2
T
0
=
1 e
jk
jk
a
k
=
1
T
0
(1)e
j
0
kt
dt +
1
T
0
(1)e
j
0
kt
dt
T
0
/ 2
T
0

0
T
0
/ 2

3/28/2008 2003, JH McClellan & RW Schafer 22

Spectrum from Fourier Series

a
k
=
1 e
jk
jk
=
2
jk
k = 1,3,K
0 k = 0, 2, 4,K

0
= 2(25)
3/28/2008 2003, JH McClellan & RW Schafer 23
LTI Systems with Periodic
Inputs
By superposition,
a
k
e
j
0
kt
H( j
0
k )a
k
e
j
0
kt
y(t) = a
k
H( j
0
k )e
j
0
k t
= b
k
e
j
0
k t
k =

k=

b
k
= a
k
H( j
0
k)
Output has same frequencies
3/28/2008 2003, JH McClellan & RW Schafer 24
Ideal Lowpass Filter (100 Hz)
y(t) =
4

sin 50t ( ) +
4
3
sin 150t ( )
H( j) =
1 <
co
0 >
co



f
co
"cutoff freq."
3/28/2008 2003, JH McClellan & RW Schafer 25
Ideal Lowpass Filter (200 Hz)
H( j) =
1 <
co
0 >
co



f
co
"cutoff"
y(t) =
4

sin 50t ( ) +
4
3
sin 150t ( ) +
4
5
sin 250t ( ) +
4
7
sin 350t ( )
3/28/2008 2003, JH McClellan & RW Schafer 26
Ideal Bandpass Filter
y(t) =
2
j 7
e
j 2(175)t

2
j 7
e
j 2 (175)t
=
4
7
cos 2(175)t
1
2
( )
H( j) =
1
c
<
1
2

B
0 elsewhere



What is the ouput signal ?
Passband Passband
3/28/2008 2003, JH McClellan & RW Schafer 27
What will be the output if the
filter in high-pass?
3/28/2008 2003, JH McClellan & RW Schafer 28
Ideal Delay
y(t) = a
k
e
j
0
k t
d
e
j
0
k t
= a
k
e
j
0
k ( t t
d
)
k =

k=

b
k
= a
k
H( j
0
k) = a
k
e
j
0
kt
d
H( j) = e
j t
d

x(t) = a
k
e
j
0
k t
k =

a y(t) = b
k
e
j
0
k t
k =

y(t) = x(t t
d
)
3/28/2008 2003, JH McClellan & RW Schafer 29
Convolution GUI: Sinusoid
3/28/2008 2003, JH McClellan & RW Schafer 30
Transient and Steady State
Response
Similar to discrete case
READING ASSIGNMENTS
This Lecture:
Chapter 10, all