Digital Signal Processing Reference
In-Depth Information
Frequency Domain
X
ð
f
Þ¼
X
ð
f
Þ
P
ð
f
Þ
¼
X
ð
f
Þ
f
s
X
1
d
ð
f
kf
s
Þ
kn
¼1
¼
f
s
X
1
X
ð
f
kf
s
Þ
d
ð
t
nT
s
Þ
n
¼1
where we have been used Tables to find that:
X
d
ð
t
nT
s
Þ
F
f
s
X
1
1
d
ð
f
kf
s
Þ;
ð
2
:
1
Þ
n
¼1
k
¼1
X
ð
f
Þ
d
ð
f
kf
s
Þ¼
X
ð
f
kf
s
Þ;
ð
2
:
2
Þ
x
ð
t
Þ
d
ð
t
nT
s
Þ¼
x
ð
nT
s
Þ
d
ð
t
nT
s
Þ:
ð
2
:
3
Þ
The sampling process in the frequency domain is illustrated in Fig. (
2.3
). From
the above equations and Fig. (
2.3
) one can deduce that X
s
(f) is essentially a col-
lection of repeated versions of X(f) scaled by f
s
. These repeated versions are often
referred to as ''images''. Hence, when a signal x(t) is sampled, its spectrum X(f) is:
1. scaled by f
s
and
2. repeated every f
s
.
Cintinuous−time Spectrum
X
(
f
)
1
f
, Hz
−B
0
B
Spectrum of Sampled Signal
X
s
(
f
)
f
s
X(f+1.f
s
)
f
s
X
(
f
)
f
s
X(f−1.f
s
)
f
s
f
, Hz
−B
0
B
f
s
−B
f
s
+B
−f
s
/2
f
s
/2
2f
s
f
s
Normalised−frequency Spectrum
of Sampled Signal
X
s
(v)
f
s
X(v + 1)
f
s
X( v )
f
s
X(v − 1)
f
s
X(v − 2)
f
s
v
−b
0
b
1−
b
1+
b
Note:
b = B / f
s
−0.5
0.5
1
2
Fig. 2.3
The discrete-time Fourier transform (DTFT) of a sampled signal
Search WWH ::
Custom Search