5.
4 SAMPLING CONTINUOUS-TIME SIGNAL
The sampling of continuous-time signals is an important topic. The analysis and design of
digital communication systems, digital controllers, etc., requires an understanding of the
sampling process and its results.
In sampling, we convert continuous–time analog signals (signals that are defined at all instants
of time and have amplitudes that may take any real value) to discrete–time analog signals
(signals that are defined at specific instants of time but still have amplitudes that may take any
real value).
The sampling theorem states that a band-limited signal x(t) with a bandwidth B can be
reconstructed from its sample values x(nTs) if the sampling frequency fs =1/Ts is greater than
twice the bandwidth B of x(t). Otherwise, aliasing would result in the reconstructed x(t).
The minimum sampling rate fs = 2B for an analog band-limited signal is called the Nyquist rate.
Sampling
If we want to sample x(t) periodically every Ts then we multiply the signal x(t) by a train of delta
functions that occur every Ts seconds. A train of delta function Ts(t) that occur every Ts is given by
( )= ( − )
Therefore, the sampled signal ( ) is given by
( )= ( ) ( − )
= ( ) ( − )
= ( ) ( − )
Hence, the sampled signal is a sum of delta functions that have magnitudes equal to the value of x(t)
at the time instants that the delta functions occur. Figure 1 shows a sampled signal and its Fourier
spectra.
1
Figure 1: Sampled signal and its Fourier spectra
We can get the spectrum of ( ) by finding the spectrum of the train of delta functions and
convolving it with ( ),
From chapter 4,
1
( )= …+ + + + + +⋯
1
=
Its Fourier transform is
2
ℱ ( ) = ( − )
Now using Frequency-convolution property: ( ) ( )↔ ( )∗ ( )
We obtain,
2
1 2
( )= ( )∗ ( − ) = ( − )
2
Where =
This indicates that the spectrum of sampled signal is simply scaled copies of the spectrum of the
original continuous–time signal at multiples of the sampling frequency as shown in Figure 1(e).
The original signal can be extracted from the sampled signal by using a LPF with bandwidth equal to
the bandwidth of the original signal x(t), which is 2B rad/s in this case. However, this is true only if
the signal was sampled at a sampling rate that is greater than twice the bandwidth of the signal.
Aliasing Phenomenon
If the signal is sampled at a sampling rate lower than 2 times the bandwidth of the signal, reconstructing
the original signal will be impossible and the reconstructed signal will suffer from aliasing as illustrated
in Figure 2. In aliasing, a high-frequency signal will look like a low-frequency signal after being
sampled at a low sampling rate.
Figure 2: Aliasing
Q1 Specify the Nyquist rate and the Nyquist interval for each of the following signals:
(a) ( ) = sin(200 )
(b) ( ) = sin (200 )
(c) ( ) = sin(200 ) + sin (200 )
(a) Nyquist rate fs = 200 Hz, Nyquist interval Ts = 5 ms.
(b) Nyquist rate fs = 400 Hz, Nyquist interval Ts = 2.5 ms.
(c) Nyquist rate fs = 400 Hz, Nyquist interval Ts = 2.5 ms.
3
Q2 The signal ( ) = 1.5 + 4 cos 10 + 2 cos 20
is sampled at a rate of 30 samples/sec. Plot the magnitude spectra of the sampled signal.
Solution
The highest frequency component in this signal is B = 10 Hz. The minimum sampling rate is 20 Hz.
With fs = 30 Hz there is no aliasing in the sampled signal.
( )=3 ( ) + 4 [ ( − 10 ) + ( + 10 )] + 2 [ ( − 20 ) + ( + 20 )]
X( )
-20 -10 10 20
The sampled signal is
( )= ( − )