EE 102 spring 2001-2002 Handout #23

Lecture 11
The Fourier transform
denition
examples
the Fourier transform of a unit step
the Fourier transform of a periodic signal
properties
the inverse Fourier transform
111
The Fourier transform
well be interested in signals dened for all t
the Fourier transform of a signal f is the function
F() =
_

f(t)e
jt
dt
F is a function of a real variable ; the function value F() is (in
general) a complex number
F() =
_

f(t) cos t dt j
_

f(t) sint dt
|F()| is called the amplitude spectrum of f;

F() is the phase
spectrum of f
notation: F = F(f) means F is the Fourier transform of f; as for
Laplace transforms we usually use uppercase letters for the transforms
(e.g., x(t) and X(), h(t) and H(), etc.)
The Fourier transform 112
Fourier transform and Laplace transform
Laplace transform of f
F(s) =
_

0
f(t)e
st
dt
Fourier transform of f
G() =
_

f(t)e
jt
dt
very similar denitions, with two dierences:
Laplace transform integral is over 0 t < ; Fourier transform integral
is over < t <
Laplace transform: s can be any complex number in the region of
convergence (ROC); Fourier transform: j lies on the imaginary axis
The Fourier transform 113
therefore,
if f(t) = 0 for t < 0,
if the imaginary axis lies in the ROC of L(f), then
G() = F(j),
i.e., the Fourier transform is the Laplace transform evaluated on the
imaginary axis
if the imaginary axis is not in the ROC of L(f), then the Fourier
transform doesnt exist, but the Laplace transform does (at least, for
all s in the ROC)
if f(t) = 0 for t < 0, then the Fourier and Laplace transforms can be
very dierent
The Fourier transform 114
examples
one-sided decaying exponential
f(t) =
_
0 t < 0
e
t
t 0
Laplace transform: F(s) = 1/(s + 1) with ROC {s | s > 1}
Fourier transform is
_

f(t)e
jt
dt =
1
j + 1
= F(j)
one-sided growing exponential
f(t) =
_
0 t < 0
e
t
t 0
Laplace transform: F(s) = 1/(s 1) with ROC {s | s > 1}
f doesnt have a Fourier transform
The Fourier transform 115
Examples
double-sided exponential: f(t) = e
a|t|
(with a > 0)
F() =
_

e
a|t|
e
jt
dt =
_
0

e
at
e
jt
dt +
_

0
e
at
e
jt
dt
=
1
a j
+
1
a + j
=
2a
a
2
+
2
1/a
1/a
0
1
t
f
(
t
)
a a
0
2/a

F
(

)
The Fourier transform 116
rectangular pulse: f(t) =
_
1 T t T
0 |t| > T
F() =
_
T
T
e
jt
dt =
1
j
_
e
jT
e
jT
_
=
2 sin T

T T
0
1
t
f
(
t
)
/T /T
0
2T

F
(

)
unit impulse: f(t) = (t)
F() =
_

(t)e
jt
dt = 1
The Fourier transform 117
shifted rectangular pulse: f(t) =
_
1 1 T t 1 + T
0 t < 1 T or t > 1 + T
F() =
_
1+T
1T
e
jt
dt =
1
j
_
e
j(1+T)
e
j(1T)
_
=
e
j
j
_
e
jT
e
jT
_
=
2 sin T

e
j
/T /T
0
2T

|
F
(

)
|
/T /T

0
(phase assuming T = 1)

F
(

)
The Fourier transform 118
Step functions and constant signals
by allowing impulses in F(f) we can dene the Fourier transform of a step
function or a constant signal
unit step
what is the Fourier transform of
f(t) =
_
0 t < 0
1 t 0
?
the Laplace transform is 1/s, but the imaginary axis is not in the ROC,
and therefore the Fourier transform is not 1/j
in fact, the integral
_

f(t)e
jt
dt =
_

0
e
jt
dt =
_

0
cos t dt j
_

0
sin t dt
is not dened
The Fourier transform 119
however, we can interpret f as the limit for 0 of a one-sided decaying
exponential
g

(t) =
_
e
t
t 0
0 t < 0,
( > 0), which has Fourier transform
G

() =
1
a + j
=
a j
a
2
+
2
=
a
a
2
+
2

j
a
2
+
2
as 0,
a
a
2
+
2
(),
j
a
2
+
2

1
j
lets therefore dene the Fourier transform of the unit step as
F() =
_

0
e
jt
dt = () +
1
j
The Fourier transform 1110
negative time unit step
f(t) =
_
1 t 0
0 t > 0
F() =
_
0

e
jt
dt =
_

0
e
jt
dt = ()
1
j
constant signals: f(t) = 1
f is the sum of a unit step and a negative time unit step:
F() =
_

e
jt
dt =
_
0

e
jt
dt +
_

0
e
jt
dt = 2()
The Fourier transform 1111
Fourier transform of periodic signals
similarly, by allowing impulses in F(f), we can dene the Fourier transform
of a periodic signal
sinusoidal signals: Fourier transform of f(t) = cos
0
t
F() =
1
2
_

_
e
j
0
t
+ e
j
0
t
_
e
jt
dt
=
1
2
_

e
j(
0
)t
dt +
1
2
_

e
j(+
0
)t
dt
= (
0
) + ( +
0
)

F()

0

The Fourier transform 1112
Fourier transform of f(t) = sin
0
t
F() =
1
2j
_

_
e
j
0
t
e
j
0
t
_
e
jt
dt
=
1
2j
_

e
j(
0
)t
dt +
1
2j
_

e
j(
0
+)t
dt
= j(
0
) + j( +
0
)

F()

0
j
j
The Fourier transform 1113
periodic signal f(t) with fundamental frequency
0
express f as Fourier series
f(t) =

k=
a
k
e
jk
0
t
F() =

k=
a
k
_

e
j(k
0
)t
dt = 2

k=
a
k
( k
0
)

F()

0
2
0
3
0

0
2
0
3
0
2a
0
2a
1
2a
2
2a
3
2a
1
2a
2
2a
3
The Fourier transform 1114
Properties of the Fourier transform
linearity af(t) + bg(t) aF() + bG()
time scaling f(at)
1
|a|
F(

a
)
time shift f(t T) e
jT
F()
dierentiation
df(t)
dt
jF()
d
k
f(t)
dt
k
(j)
k
F()
integration
_
t

f()d
F()
j
+ F(0)()
multiplication with t t
k
f(t) j
k
d
k
F()
d
k
convolution
_

f()g(t ) d F()G()
multiplication f(t)g(t)
1
2
_

F( )G( ) d
The Fourier transform 1115
Examples
sign function: f(t) =
_
1 t 0
1 t < 0
write f as f(t) = 1 + 2g(t), where g is a unit step at t = 0, and apply
linearity
F() = 2() + 2() +
2
j
=
2
j
sinusoidal signal: f(t) = cos(
0
t + )
write f as
f(t) = cos(
0
(t + /
0
))
and apply time shift property:
F() = e
j/
0
((
0
) + ( +
0
))
The Fourier transform 1116
pulsed cosine: f(t) =
_
0 |t| > 10
cos t 10 t 10
!10 0 10
!1
!0.5
0
0.5
1
t
f
(
t
)
write f as a product f(t) = g(t) cos t where g is a rectangular pulse of
width 20 (see page 12-7)
F(cos t) = ( 1) + ( + 1), F(g(t)) =
2 sin 10

The Fourier transform 1117

now apply multiplication property
F(j) =
_

sin 10

(( 1) + ( + 1)) d
=
sin(10( 1))
1
+
sin(10( + 1))
+ 1
!1 0 1
!4
!2
0
2
4
6
8
10
12

F
(

)
The Fourier transform 1118
The inverse Fourier transform
if F() is the Fourier transform of f(t), i.e.,
F() =
_

f(t)e
jt
dt
then
f(t) =
1
2
_

F()e
jt
d
lets check
1
2
_

=
F()e
jt
d =
1
2
_

=
_
_

=
f()e
j
_
e
jt
d
=
1
2
_

=
f()
__

=
e
j(t)
d
_
d
=
_

f()( t)d
= f(t)
The Fourier transform 1119