Scribd
Upload
2 views

unit3_6

The document discusses the sampling theorem, which states that a band-limited signal can be completely represented by its samples taken at specific intervals, allowing for perfect reconstruction if the sampling rate meets certain criteria. It details different sampling methods, including instantaneous, natural, and flat top sampling, and emphasizes the importance of the Nyquist rate for avoiding distortion in signal recovery. Additionally, it explains the role of low-pass filters in reconstructing the original signal from its samples.

Uploaded by

Ashish Goyal
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

unit3_6

The document discusses the sampling theorem, which states that a band-limited signal can be completely represented by its samples taken at specific intervals, allowing for perfect reconstruction if the sampling rate meets certain criteria. It details different sampling methods, including instantaneous, natural, and flat top sampling, and emphasizes the importance of the Nyquist rate for avoiding distortion in signal recovery. Additionally, it explains the role of low-pass filters in reconstructing the original signal from its samples.

Uploaded by

Ashish Goyal
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

SAMPLING THEOREM AND METHODS OF SAMPLING

Due to rise in digital technology, the inexpensive, portable, programmable


and easily reproducible discrete time systems are available.. Thus processing
of the discrete signals is more preferable to processing of continpus time
signals.
This requires the knowledge of the sampling theorem
The sampling theorem : Analog information is formatted using three
separate Processes : sampling , quantization and encoding .
For the purpose of sampling a number of samples of the signal should be
taken so that the original signal is represented completely in its samples
and it should be possible to reconstruct the original signal completely from
them.
Statement of the sampling Theorem : (i) A band limited signal of finite
energy having a highest frequency component fo is completely defined by
its sample values at uniform intervals < or equal to (1/2fo) apart
(ii) A band limited signal of finite energy that has no frequency components
> fm may be completely recovered from a knowledge of the samples taken
at the rate of 2fm/ second
A continous time signal may be completely represented in its samples and recovered
back if the sampling frequency fs> or equal to fm
Proof : Consider a signal x(t) that is band limited to fm . Thus

Sampling at a rate of fs Hz is achieved by multiplying x(t) by an impulse


pulse train resulting in the sampled signal g(t)

The trignometric Fourier series of the impulse train is :


The spectrum G(ω) consists of X(ω) repeating periodically with frequency ωs. For
reconstruction of X (ω) from G(ω) we see in Fig f in last slide that

As long as the condition in eq (10.6) is satisfied G (ω) will have non overlapping repetitions
of X (ω). Then x(t) can be recovered from its sampled signal (t) by passing g(t) through an
ideal LPF with BW of fm. This proves the sampling theorem
 The ideal BW of the sampled signal is infinite.But our purpose is to extract the
original X(ω) out of G(ω)
 X(ω) is centered at ω=0 and has a BW of ω m which is also the cut off
frequency of the LPF. Thus all frequencies greater than ω m will be rejected.
 If fs> 2fm the successive cycles of G(ω) do not overlap and hence X(ω) can be
recovered
 If fs=2fm though the successive cycles just touch each other. The original
spectrum can be recovered from the sampled spectrum using a LPF with cut-
off ω m . Type equation here.
 If fs<2fm th successive cycles of sampled spectrum will overlap ach other and
recovery is not possible
Hence for recovery without distortion

The minimum sampling rate is called as Nyquist rate and the corresponding frequency as
Nyquist frequency which is given as :
The Reconstruction Filter: The LPF is also called as interpolation filter. In an ideal
case the amplitude response is sharp at cut off frequency which is not possible in a
practical design.
Figure below shows the practical filter where one can observe a transition band

Signal Reconstruction : The process of signal reconstruction is also called as


interpolation.
To recover x(t) the sampled signal must be passed through a LPF with BW fs and gain Ts.
The transfer function of the interpolating filter is :
 Fig 10.6(b) shows h(t). Also h(t)
=0 at all Nyquist sampling
instants t=Ɫ (n/2fm) except at
t=0.
 The filter output is x(t). Each
sample in g(t) being an impulse
produces a sinc pulse with a
height equal to the sample
strength.
 Addition of all the sinc pulses
produced by all samples results
in x(t)
Eq 10.17 is called as interpolation formula which provides the values of x(t) as
samples as a weighted sum of sample values.
Practically x(t) is not band limited which means that fm cannot be predicted.
Hence it is not possible to select a suitable sampling frequency fa
Sampling Methods : The following are the sampling techniqus
(i) Instantaneous Sampling
(ii) (ii) Natural Sampling and (iii)
(iii) Flat top Sampling

Let us consider instantaneous sampling :


 This is also called as impulse sampling.
 This is produced by using a switch sampler. ( refer to the circuit on next slide )
 Since the switch is open only for a very short time’t’ tending to zero , the
output g(t) of the circuit will contain only instantaneous value of x(t).
 The sampling method gives a train of impulses of height equal to the
instantaneous value of the input
Natural sampling :The width of the pulses τ in the previous case approaches zerodu to which
the power content in these sampled pulses is negligible. Hnce this method is not suitable for
transmission purpose. The following assumptions are made :
(i)Consider that fs is higher than the Nyquist rate so that sampling theorm is satisfied
(ii) Sampling function is a train of pulses c(t) of width τ with frequency fs

Thus pulse train in the above equation can be written as


Since c(t) is a rectangular pulse Cn for this waveform will be expressed as
Now substituting the value of c(t) from Eq (10.32) into the equation for c(t)

X(f) is periodic in fs and fig ( next slide) shows an arbitrary spectra x(t) and also
G(f).
Flat top Sampling : This is also a practically sampling method.. The top of samples
remains constant and is equal to the instantaneous value of baseband signal x(t) at
the start of sampling.

Convolution of any function with delta function is equal to


that function
 From Fig C above it means that the pulse width in g(t) is determined by the width of
h(t) and the sampling instant is determined by the delta function. From Fig (b) we can
write

 For any function f(t)

 For flat top sampling we apply the modified form of the above Eq (10.40).
 Therefore on convolution of s(t) and h(t) we get a pulse whose duration is h(t)
only but the amplitude is defined by s(t).

You might also like