You are on page 1of 10

EECE 301

Signals & Systems


Prof. Mark Fowler
Note Set #18
D-T Signals: Frequency-Domain Analysis

1/10

Fourier Analysis of D-T Signals


We now develop Fourier ideas for D-T signals like we did for C-T signals:
Define a D-T FT (DTFT) for D-T signals and see that it works pretty much like the
FT for C-T signals (CTFT)

But we also do something we cant do for CTFT-based ideas:


Develop a computer-processing version of the DTFT called the Discrete Fourier
Transform (DFT) that will allow you to use the computer to numerically compute a
view of the DTFT
But to make this DFT useful well need to understand the relationships between the
DFT, the DTFT, and the CTFT!

2/10

Recall: Sampling Analysis


ADC

x(t)

DAC

Hold

x[n] = x(nT)

Sample at
t = nT
A

B
A/T

2Fs

Fs

Impulse
Gen

CT LPF
~
x (t )

x (t )

X( f )
B

~
X( f )

Fs

2Fs

As long as Fs 2B then we can clearly see


~
a view of X( f ) in X ( f )
But we did this using a FT of a signal inside the DAC
Is there some other way to do this by using the samples?
3/10

Motivation for D-T Fourier Transform (DTFT)


x(t)

ADC

DAC
x[n]

Hold
Sample at
t = nT

Impulse
Gen

CT LPF
~
x (t )

x (t )

~
X ( f ) ?????

We know that if sampling has been done perfectly that:

X ( f ) shows the original signals FT X( f ) with no aliasing


But that is only something that helps us conceptually but not
really numerically

So that raises this question:


Since x (t ) is completely determined by x[n] can we use those
samples to actually compute X ( f ) ????

4/10

Recall Fourier Transform of ~


x (t )
~
x (t ) x (t )

(t nT )

x (t ) T (t )
FS of T(t)

1
~
x (t )
T

jk 2Fs t
x
(
t
)
e

FT & Mod. Prop

1
~
X ( f ) X ( f kFs )
T k
~
Tells what X ( ) looks like!
5/10

Take An Alternate Path to the DTFT!

~
x (t ) x (t )

(t nT ) x(t ) T (t )

~
x (t )

FS of T(t)

1
~
x (t )
T

x[n] (t nT )

x(t )e jk 2F t
s

FT & Mod. Prop

~
X ( ) F x[n ] (t nT )
n

x[n]F (t nT )

1
~
X ( ) X ( k 2Fs )
T k

~
X ( )

x[n]e jnT

Uses
Samples!!

~
Tells what X ( ) looks like!

~
Tells how to compute X ( ) !

6/10

Re-Define to Get The DTFT!


~
X ( )

x[n]e jnT

Let = T where T = 1/Fs

DTFT:

X ()

x[n]e

jn

is called D-T Frequency


= T: (rad/sec) (sec/sample) = rad/sample

~
X ( ) and X () are really the same thing...
just " plotted" w.r.t. a different unit

: rad/sec
:

rad/sample
7/10

DTFT X()

shows

X ( f )

If sampling was
done right!!!

which shows
CTFT X( f )

A/T

2Fs

Fs F /2
s
A/T

X( f )

CTFT of x(t)

B
~
X( f )

~(t )
CTFT of x

Fs/2 Fs
X ()

2Fs

DTFT of x[n]
2

Only Need to Look Here!!!

8/10

Physical Relationship of DTFT


ADC
DAC
x[n] = x(nT) Impulse

x(t)

Hold

Gen

Sample at
t = nT
A

A/T

A/T

2Fs

Fs

Fs/2

X( f ) CTFT

~
X( f )

Fs/2

~
x (t )

x (t )

of x(t)

X ()

CT LPF

DTFT of x[n]
2

~(t )
CTFT of x
Fs

2Fs

f
9/10

Motivating D-T System Analysis using DTFT


x(t)

x[n]

ADC

D-T
System

CTFT of x(t)

y[n]

DAC

CTFT of y(t)

X( f )
B

X( f )
B

DTFT of x[n]

DTFT of y[n]

X ()

y(t)

X ()

10/10