DIGITAL SIGNAL PROCESSING Sampling and Reconstruction

Chapter 2:
Analog Signal Processing • 1. Introduction
Sampling and Reconstruction • 2. Overview of Analog
Reference: • 3. Sampling theorem
S J.Orfanidis, ”Introduction to Signal Processing”, Prentice –Hall , 1996,ISBN 0-13-209172-0
M. D. Lutovac, D. V. Tošić, B. L. Evans, “Filter Design for Signal Processing Using MATLAB • 4. Sampling of Sinusoids
and Mathematica”, Prentice Hall, 2001
Lectured by Prof. Dr. Thuong Le-Tien
National Distinguished Lecturer
• 5. Spectra of Sampled Signals
Tel: 08-38654184; 0903 787 989
Email: [email protected], • 6. Analog signal reconstruction
[email protected]

Dated on January 2024

1. Introduction
Three steps for digital signal processing of 2. Review of Analog signals
analog signals
 FOURIER Transform X() of x(t) is the spectrum of the
 Step 1: Digitizing of analog signals: analog signal: 
Sampling, Quantization – Analog to Digital
Conversion (ADC).

X ()  x(t )e  jt dt

(2.1)

Where  is the radian frequency (rad/s).

 Step 2: Implementing digital signal


and  = 2f (2.2)

processor for discrete samples
Definition of Laplace Transform:
 Step 3: Reconstructing the analog signal


 (2-3)
after processing – Digital to Analog X (s)   x(t ).est dt

Conversion (DAC)
3 4
 Response of a linear system
x(t) y(t)  H() is the Fourier transform of h(t)
Linear system
h(t)
input output H ( )   h( t )e  jt dt
 The system is characterized by impulse response h(t). The  The steady state response of a sinusoid:
output y(t) is obtained by the time domain convolution :
 x(t) = exp(jt) Linear system y(t) = H()exp(jt)
y(t )   h(t  t ' ) x(t ' )dt' Sinusoid in H()
Sinusoid out

 Or frequency domain:  Output is a sinusoid with frequency (),
Y ()  H (). X () amplitude equal to the signal amplitude multiplied
by MagH(), and phase shift equal to arg(H()):
where H() is the frequency response of the system.
x(t )  e jt  y(t )  H ()e jt | H () | .e jt  j arg H (  )

5 6

 The result is presented in frequency domain

X(  ) Y(  )
 Linear superposition: Signals x(t) has two frequency A1 A2

components H(  ) A 1 H(  )
A 2 H(  )
j1t j 2 t
x(t )  A1e  A2 e  

 After filtering  Spectrum of X()

X ()  2A1 (  1 )  2A2 (   2 )
y(t )  A1 H ()e j1t  A2 H ()e j2t
 Spectrum of Y()
 Note: Filtering only change the magnitudes but not Y ()  H () X ()  H ()(2A1 (  1 )  2A2 (   2 ))
the frequencies  2A1H (1 ) (  1 )  2A2 H ( 2 ) (   2 )

7 8
 The spectrum of the sampled sinusoid x(nT)
3. Concept of Sampling theorem


will be periodic replication of the original

spectral line at intervals fs=1/T
 Sampling process in Fig. 3.1. x(t) is sampled
by period T, t=nT where n=0,1,2,…
 Many high frequency components appear
in the signal spectrum
 Two questions are often provided for
1. What is the effect of sampling on the
original frequency spectrum?
2. How should one choose the sampling
interval T? Figure 3.1 Ideal Sampler

9 10

Sampling theorem
 For accurate representation of a signal x(t) by its
time samples x(nT), two conditions must be met:
1: x(t) is bandlimited
2: Sampling frequency must be chosen to be
at least twice the maximum frequency fmax,
Figure 3.2. Spectrum replication caused by sampling. fs  2fmax:
With the replicated spectrum of the sampled signal, one fs = 2fmax is the Nyquist rate.
cannot tell uniquely What the original frequency was. It
fs/2 is the Nyquist frequency or folding
could be any one of the replicated frequencies namely
f’=f+mfs. This potential confusion of the original frequency frequency
with another is known as aliasing and can be avoided if one
satisfies the condition of the sampling theorem

11 12
 Typical sampling rate for some common applications Antialiasing Prefilter
(An Approximation)  Signal must be bandlimited therefore need to pass
through a low pass filter namely prefilter before sampling
Input spectrum
Prefiltered spectrum

prefilter

f f
0 -fs/2 fs/2

Replicated
spectrum

f
-fs 0 fs

Bandlimited
x(t) signal x(nT)
Analog lowpass Sampler and
To DSP
Analog filter x(t) quantizer digital
signal signal

Antialiasing prefilter
14

What happens if we do not sample in 4. Sampling of sinusoid: x(t) = cos(2ft)

accordance with the sampling theorem? The number of samples per is given by the quantity fs/f:

 Missing important time variations between sampling instants f s samples / sec samples
 
 May arrive at the erroneous conclusion that the samples f cycles / sec cycle
represent a signal which is smoother than it actually is
 Be confusing the true frequency content of the signal with a
lower frequency content. Such confusion of signals is called
aliasing

Aliasing in
Special case with multiple frequency components in the x(t)
The time domain

16
 Analog reconstruction and aliasing
LPF as an ideal
reconstructor

Define also the following family of sinusoids, for m in integer

And its sampled version

Using the property fsT=1 and the trigonometric identity

x m (nT )  e 2j ( f  mfs )Tn  e 2jfTn e 2jmfsTn  e 2jfTn  x(nT )

f , f  f s , f  2 f s ,..., f  mf s ,...
Note that xm(t) are different from each other
but they have same samples:
17 18

 Example: 5 signals are sampled by the rate 4Hz:

 sin(14t ), sin(6t ), sin(2t), sin(10t), sin(18t) (t second)
Let prove they are aliased each other due to their same
samples.
 Example  Sol: The frequencies of the signals: -7, -3, 1, 5, 9 Hz. They
 As sinusoid f=10 Hz, sampled by fs=12Hz. The sampled have the same periodic replication in multiples of fs=4Hz.
signal consists of periodic frequencies 10+m.12Hz, m = 0, Writing the five frequencies compactly:
1, 2,… or: …, -26, -14, -2, 10, 22, 34, 46, … but only fa fm=1+4m, m=-2, -1, 0, 1, 2.
= 10 mod(12) = 10 – 12 = -2 Hz in the range of Nyquist xm (t )  sin(2f mt )  sin( 2 (1  4m)t ), m  -2,-1,0,1,2
interval [-6,6] Hz. So the reconstructed signal with –2 Hz x m ( nT )  sin(2 (1  4m )nT )  sin(2 (1  4m )n / 4)
is not as the original one with 10 Hz.
 sin(2n / 4  2mn)  sin(2n / 4)

19 20
Example: x(t)=4+3cos(t)+2cos(2t)+cos(3t) t: in ms
Determine the min sampling rate without any aliasing effects
Supposed the signal sampled at half its Nyquist rate.
Determine xa(t) that would be aliased with x(t).
Sol:
Freq. of 4 terms: f1=0, f2=0.5kHz, f3=1kHz,f4=1.5kHz

Example: The square wave sampled at rate fs; t in seconds For fs =4kHz, the aliased signal will be

Determine the xa(t) that will appear at the output of the

reconstructor for 2 cases fs=4Hz and 8Hz.
Sol: For fs =8kHz, the aliased signal will be
Fourier series of square wave contains odd harmonics at freq.
•The first case: Sketch for xa(t)
Condition xa(t)=x(nT) evalued at n=1 implies A=1 Example: A given x(t), t in ms and a block of DSP

•The second case: xa(t)=Bsin(n/4)+Csin(3n/4)

Condition xa(t)=x(nT) at n=1,2 give two equations Determine the y(t) and ya(t) in the following cases:
a. When there is no prefilter, that is, H(f)=1 for all f
b. When H(f) is the ideal filter with cutoff fs/2=20kHz
c. When H(f) is a practical prefilter as follows,

Sol: Six terms of freq. in x(t) Case a.

Case b.
 Case c.
5. Spectra of sampled signals

 Sampled signal: xˆ ( t )   x(nT ) (t  nT )
n  

 In practical sampling, the sampled signal:


x flat ( t )   x(nT ) p(t  nT )
n  

 where, p(t) is flat-top pulse of duration  second.

Ideal sampling with  toward 0.
x ( nT ) ( t  nT )
xˆ ( t ) xflat (t) x ( nT ) p( t  nT )

0 T 2T …. nT t
0 T 2T …. nT t
30

Discrete Time Fourier Transform DTFT Spectrum Replication

or

This approximation become exact if

Practical approximation
32
Aliasing caused by overlapping spectral replicas Practical antialiasing prefilter

Ideal antialiasing prefilter

Attenuation in dB

6. Analog signal reconstruction Reconstructed analog signal


yˆ (t )   y(nT ) (t  nT )
n  


y a (t )   y(nT )h(t  nT )
n  


Staircase reconstructor y a (t )   y(nT )h(t  nT )
n  

Y a ( f )  H ( f )Yˆ ( f )

Replicated spectrum
1 
Y ( f )   Y ( f  mfs )
ˆ
Analog reconstructor as a low pass filter T m
35
36
Ideal reconstructor Staircase reconstructor

Anti-image postfilter Digital equalization filter for D/A conversion