CHAPTER 3

SAMPLING PROCESS AND THEOREMS

1
IMPULSE SAMPLING
■ The impulse sampler is a fictitious sampler whose output is
considered as a ‘train of impulses’.
■ The impulse have the magnitude of 𝑓(𝑡) at the
corresponding instant of time 𝑓(𝑡) was sampled.

If 𝑓(𝑡) is the sampled signal (at T sampling period), then the

f(t) f*(t)
T

2
If we define a train of unit impulses as 𝛿𝑇 𝑡 , where:-
∞

𝛿𝑇 = ෍ 𝛿(𝑡 − 𝑘𝑇)
𝑘=0
then the sampler output is equal to the product of the continuous-time input; 𝑓(𝑡), and the
train of unit impulses, 𝛿𝑇 (𝑡).
∞

𝑓 ∗ 𝑡 = ෍ 𝑓(𝑘𝑇)𝛿(𝑡 − 𝑘𝑇)
𝑘=0

𝑓 ∗ 𝑡 = 𝑓 0 𝛿 𝑡 + 𝑓 𝑇 𝛿 𝑡 − 𝑇 + ⋯ + 𝑓 𝑘𝑇 𝛿 𝑡 − 𝑘𝑇 + ⋯

Consequently, the sampler may be considered a modulator with the input 𝑓(𝑡) as the
modulating signal and the train of unit impulse 𝛿𝑇 as the carrier.

f(t) f*(t) Figure 3.1

3
The Laplace Transform of 𝑓 ∗ 𝑡 is given by:-

𝐹 ∗ 𝑠 = 𝐿 𝑓 ∗ 𝑡 = 𝑓 0 𝐿 𝛿(𝑡) + 𝑓 𝑇 𝐿 𝛿 𝑡 − 𝑇 + 𝑓 2𝑇 𝐿 𝛿 𝑡 − 2𝑇 +⋯
𝐹 ∗ 𝑠 = 𝑓 0 + 𝑓 𝑇 𝑒 −𝑇𝑠 + 𝑓 2𝑇 𝑒 −2𝑇𝑠 + ⋯
∞

𝐹 ∗ 𝑠 = ෍ 𝑓(𝑘𝑇)𝑒 −𝑘𝑇𝑠
𝑘=0
1
If we define 𝑒 𝑇𝑠 = 𝑧 or s = ln 𝑧; then
𝑇
∞

𝐹 ∗ (𝑠)| 1 = ෍ 𝑓(𝑘𝑇)𝑧 −𝑘
𝑠= ln 𝑧
𝑇
𝑘=0

Then,
1
𝐹 ∗ (𝑠)| 1 = 𝐹∗ ln 𝑧 = 𝐹 𝑧
𝑠= ln 𝑧
𝑇 𝑇

∴ 𝐹 ∗ (𝑠) = 𝐹(𝑧)

The Laplace Transform of the impulse-sampled signal 𝑓 ∗ (𝑡) is the same as the z transform of
signal 𝑓 𝑡 if 𝑒 𝑇𝑠 is defined as z, where T is the sampling period.

4
DATA HOLD CIRCUITS
▪ Data hold is a process of generating a continuous-time signal ℎ 𝑡 from a discrete-
time sequence 𝑓 𝑘𝑇 .
▪ A data hold circuit converts the sampled signal into a continuous-time signal, which
approximately reproduces the signal applied to the sampler.
The signal ℎ 𝑡 during the time interval 𝑘𝑇 ≤ 𝑡 < 𝑘 + 1 𝑇 may be approximated by a
polynomial in 𝜏 as:-
ℎ 𝑘𝑇 + 𝜏 = 𝑎𝑛 𝜏 𝑛 + 𝑎𝑛−1 𝜏 𝑛−1 + ⋯ + 𝑎1 𝜏 + 𝑎0
where 0 ≤ 𝜏 < 𝑇.

Signal ℎ(𝑘𝑇) must equal 𝑓(𝑘𝑇), thus;

5
ZERO ORDER HOLD (ZOH)
The simplest data hold is obtained when 𝑛 = 0, that is when ℎ 𝑘𝑇 + 𝜏 = 𝑓(𝑘𝑇), where
0 ≤ 𝜏 < 𝑇 and 𝑘 = 0,1,2, ⋯.

The circuit holds the amplitude of the sample from one sampling instant to the next.

The input signal 𝑓(𝑡) is sampled at discrete instants and sampled signal is passed
through the ZOH. The ZOH circuit smooths the sampled signal to produce the signal
ℎ 𝑡 , which is constant from the last sampled value until the next sample is available.

ℎ 𝑘𝑇 + 𝜏 , for 0 ≤ 𝑡 < 𝑇

Figure 3.2

6
The fact that the integral of an impulse function is constant, we may assume that the ZOH is
an integrator and the input to the ZOH is a train of impulse. A mathematical model for the
real sampler and ZOH may be constructed as shown.

a)

b)

Figure 3.3

7
Consider the sampler and ZOH in Figure 3.3(a)
(a)

(b)

Figure 3.3

Assuming the signal 𝑓 𝑡 = 0 for 𝑡 < 0, the output ℎ1 (𝑡) is related to 𝑓(𝑡) as follows:-

ℎ1 𝑡 = 𝑓 0 1 𝑡 − 1(𝑡 − 𝑇) + 𝑓 𝑇 1 𝑡 − 𝑇 − 1 𝑡 − 2𝑇
+𝑓 2𝑇 1 𝑡 − 2𝑇 − 1 𝑡 − 3𝑇 +⋯
∞

ℎ1 𝑡 = ෍ 𝑓(𝑘𝑇) 1 𝑡 − 𝑘𝑇 − 1(𝑡 − 𝑘 + 1 𝑇)
𝑘=0

𝑒 −𝑘𝑇𝑠
Since 𝐿 1(𝑡 − 𝑘𝑇) = , then;
𝑠
∞
𝑒 −𝑘𝑇𝑠 𝑒 −(𝑘+1)𝑇𝑠
𝐿 ℎ1 (𝑡) = 𝐻1 𝑠 = ෍ 𝑓 𝑘𝑇 −
𝑠 𝑠
𝑘=0

∞
1 − 𝑒 −𝑇𝑠
𝐻1 (𝑠) = ෍ 𝑓(𝑘𝑇)𝑒 −𝑘𝑇𝑠
𝑠
𝑘=0

8
Consider the mathematical model in Figure 3.3(b)
(a)

(b)

Figure 3.3
The output of this model must be the same as that of the real ZOH.

𝐿 ℎ2 (𝑡) = 𝐻2 𝑠 = 𝐻1 (𝑠)

Thus;
∞
1 − 𝑒 −𝑇𝑠
𝐻2 𝑠 = ෍ 𝑓(𝑘𝑇)𝑒 −𝑘𝑇𝑠
𝑠
𝑘=0

We know that;
∞

𝐹 ∗ 𝑠 = ෍ 𝑓(𝑘𝑇)𝑒 −𝑘𝑇𝑠
𝑘=0
1 − 𝑒 −𝑇𝑠 ∗
𝐻2 𝑠 = 𝐹 𝑠
𝑠

From Figure 3.3(b), we know that 𝐻2 𝑠 = 𝐺ℎ0 (𝑠)𝐹 ∗ 𝑠

Then, the transfer function of the ZOH may be given by;

1 − 𝑒 −𝑇𝑠
𝐺ℎ0 𝑠 =
𝑠

9
First Order Hold (FOH)

10
RECONSTRUCTING ORIGINAL
SIGNALS FROM SAMPLED SIGNALS
Sampling Theorem
To reconstruct the original signal from a sampled signal, there is a certain
minimum frequency that the sampling operation must satisfy. Such a
minimum frequency is specified in the sampling theorem.

The sampling theorem states that:-

11
Let’s say we have the sampled signal 𝑓 ∗ (𝑡). The Laplace Transform of 𝑓 ∗ (𝑡) is given by:-

∞
1
𝐹 ∗ 𝑠 = ෍ 𝐹(𝑠 + 𝑗𝜔𝑠 𝑘)
𝑇
𝑘=−∞

To obtain the frequency spectrum, substitute 𝑠 with 𝑗𝜔;

∞
1
𝐹 ∗ 𝑗𝜔 = ⋯ + ෍ 𝐹(𝑗𝜔 + 𝑗𝜔𝑠 𝑘)
𝑇
𝑘=−∞

1 1 1
𝐹 ∗ 𝑗𝜔 = ⋯ + 𝐹 𝑗 𝜔 − 𝜔𝑠 + 𝐹 𝑗𝜔 + 𝐹 𝑗 𝜔 + 𝜔𝑠 +⋯
𝑇 𝑇 𝑇

The frequency spectrum of the impulse sampled signal is reproduced at infinite number of
1
times and is attenuated by the factor of . Thus the process of impulse modulation of the
𝑇
continuous time signal produces a series of sidebands.

12
 Figure 3.3 : Plots of frequency spectra 𝐹 ∗ (𝑗𝜔) vs .

Figure (b); 𝜔𝑠 > 2𝜔1

No two components of 𝐹 ∗ (𝑗𝜔) will overlap, and the sampled frequency spectrum will be
repeated every s rad/sec.

Figure (c); 𝜔𝑠 < 2𝜔1

The original shape no longer appears in the plot because of the superposition of the spectra.

The continuous time signal 𝑓(𝑡) can be reconstructed from the impulse-sampled signal 𝑓 ∗ (𝑡) by
filtering if and only if 𝜔𝑠 > 2𝜔1 .

13
FOLDING
This is the phenomenon where part of the frequency spectrum of the
continuous time signal folds over into the original spectrum.
𝜔𝑠
The frequency is called the folding frequency or Nyquist frequency 𝜔𝑁 .
2
1 𝜋
𝜔𝑁 = 𝜔𝑠 =
2 𝑇
In practice, signals in control systems have high frequency components and
some folding effect will almost always exist.

14
ALIASING
In the frequency spectra of an impulse-sampled signal 𝑓 ∗ (𝑡), where 𝜔𝑠 < 2𝜔1 ; aliasing is a
phenomenon where the frequency component 𝑛𝜔𝑠 ± 𝜔2 , where 𝑛 is an integer, shows up at
frequency 𝜔2 when the signal 𝑓(𝑡) is sampled.
This frequency, 𝑛𝜔𝑠 ± 𝜔2 is called an alias of 𝜔𝑠 .
The sampled signals are the same if the two frequencies differ by an integral multiple of the
sampling frequency, 𝜔𝑠 .
If a signal is sampled at a low frequency such that the sampling theorem is not satisfied, then
high frequencies are folded in and appear as low frequencies.
To avoid aliasing:-
1. Choose sampling frequency that is high enough ( 𝜔𝑠 > 2𝜔1 , where 𝜔1 is the highest
frequency component present in the signal).
2. Use a pre-filter ahead of the sampler to reshape the frequency spectrum of the signal (so that
𝜔
the frequency spectrum for 𝜔 > 2𝑠 is negligible) before the signal is sampled.

15