and Systems
Syllabus:
Introduction, Continuous Time and discrete time signals, classification of signals,
simple manipulations of discrete time signals, amplitude and phase spectra,
classification of systems, analog to digital conversion of signals.
I-----------------------------------------------;-----------------------------------------------------------------------------Contents
Page No.
1.1 Introduction
1-2
1 -2
1-3
1 -4
1 -5
1-5
1-6
1 -6
1 -11
1.10
1 -28
Representation of DT Signals
1.11Basic Sequences
1-30
1.12
1-36
1.13
1-56
1.14 1.1
introduction :_______________________________________________
1.15
The world of science and engineering is filled with signals such as images from remote
space probes, voltages generated by the heart and brain and countless other applications.
1.16
1.1.1
What is DSP?
1.17
Digital signal processing is used in a wide variety of applications. It is hard to get exact
definition of DSP.
1.18
1.19
Digital: Operating by the use of discrete signals to represent data in the form of numbers.
1.20
1.21 Processing : To perform operations on data according to programmed instructions. This leads to
a simple definition of DSP.
1.22 Definition of DSP : DSP is defined as changing or analysing information which is measured as
discrete sequences of numbers.
1.23
1.2
Concept of Signal and Signal Processing :
In a communication system, the word 'signal' is very commonly used. Therefore we must know
its exact meaning.
Mathematically, signal is described as a function of one or more independent variables.
Basically it is a physical quantity. It varies with some dependent or independent variables.
So the term signal is defined as "A physical quantity which contains some information and which
is function of one or more independent variables."
When the function depends on a single variable, the signal is said to be one dimensional.
Example of one dimensional signal is speech signal whose amplitude varies with time.
1.25
Multidimensional signals:
When the function depends on two or more variables, the signal is said to be multidimensional.
The example of a multidimensional signal is an image because it is a two dimensional signal with
horizontal and vertical co-ordinates.
1.2.1 System :
1.
2.
3.
Signals and systems have several applications. Some of the important types of systems are as
follows :
Communication system.
4. Biomedical signal processing.
Control system.
5. Auditory system.
Remote sensing system.
____________________________________________________________________________________1.3
Fig. 1.3.1 shows that the basic elements of digital signal processing system
Fig. 1.3.1
1.
Input signal:
It is the signal generated from some transducer or from some communication system. It
may be biomedical signal like ECG or EEG. Generally input signal is analog in nature. It is
denoted by x(t).
2.
Anti-aliasing filter:
Anti aliasing filter is basically a low pass filter. It is used for the following purposes :
3.
As the name indicates; this block takes the samples of input signal. It keeps the voltage
level of input signal relatively constant which is the requirement of ADC.
Sometimes amplifiers are used to bring the voltage level of input signal upto the required
voltage level of ADC.
4.
As the name indicates; this block is used to convert analog signal into digital form. This is
required because digital signal processor accepts the signal which is digital in nature.
5.
It processes input signal digitally. In a simple languages processing of input signal making
modifying the signal as per requirement. For this purpose DSP processors like ADSP 2100 or
TMS 320 can be used.
1.
1.
Versatility : Digital systems can be reprogrammed for other applications (where programmable
DSP chips are used). Moreover, digital systems can be ported to different hardware.
2.
Repeatability : Digital systems can be easily duplicated. These systems do not depend upon
component tolerances and temperature.
Accuracy : To design analog system; analog components like resistors, capacitors and inductors
are used. The tolerance of these components reduce accuracy of analog system. While in case of
DSP ; much better accuracy is obtained.
4.
Remote processing : Analog signals are difficult to store because of problems like noise and
distortion. While digital signal can be easily stored on storage media like magnetic tapes, disks
etc. Thus compared to analog signals; digital signals can be easily transposed. So remote
processing of digital signal can be done easily.
5.
6.
Easy upgradations : Because of the use of software; digital signal processing systems can be
easily upgraded compared to analog system.
7.
Compatibility : In case of digital systems; generally all applications needs standard hardware.
Thus operation of dsp system is mainly dependent on software. Hence universal compatibility is
possible compared to analog systems.
8.
Cheaper : In many applications; the digital systems are comparatively cheaper than analog
systems.
9.
1.5
11.
1.
System complexity : The digital signal processing system, makes use of converters like ADC
and DAC. This increases the system complexity compared to analog systems. Similarly in many
applications; the time required for this conversion is more.
2.
Bandwidth limitation : In case of DSP system; if input signal is having wide bandwidth then it
demands for high speed ADC. This is because, to avoid aliasing effect, the sampling rate should
be atleast twice the bandwidth. Thus such signals require fast digital signal processors. But
always there is a practical limitation in the speed of processors and ADC.
3.
Power consumption : A typical digital signal processing chip contains more than 4 lakh
transistors. Thus power dissipation is more in dsp systems compared to analog systems.
4.
Cost: For small applications digital signal processing systems are expensive compared to analog
systems.
12.__________________________________________________________________________________1
.6
Table 1.6.1 shows comparison between digital and analog signal processing
14.
15.
1.7 Applications of
DSP :
16._____________________________________________________________
An incoming signal may come from a digital or analog source. If it is coming from a digital
source then it is in the right form for processing digitally.
But input signal can be analog in nature, (e.g. speech signal or video signal).
Then it has to be converted into digital form before it can be processed by a digital system. This
type of conversion is performed using analog to digital converters. (A/ D)
17.
18.
19.
20. 1.8.1
Sampling:
In order to represent the original message signal "faithfully" (without loss of information), it is
necessary to take as many samples of the original signal as possible.
Higher the number of samples, closer is the representation. The number of samples depends on the
"sampling rate" and the maximum frequency of the signal to be sampled.
Sampling theorem was introduced to the communication theory in 1949 by Shannon. Therefore this
theorem is also called as "Shannon's sampling theorem".
The statement of sampling theorem in time domain, for the bandlimited signals of finite energy is as
follows :
21. Statement:
22. (i)
If a finite energy signal x(t) contains no frequencies higher than "W" Hz (i.e. it is a bandlimited
signal) then it is completely determined by specifying its values at the instants of time which are spaced
(1/2W) seconds apart. ii)
"W" Hz then it may be completely recovered from its samples which are spaced (1/2W) seconds apart.
23.
24. 1.8.2
Quantization :
Quantization is a process of approximation or rounding off. The sampled signal is applied to the
quantizer block.
Quantizer converts the sampled signal into an approximate quantized signal which consists of only a
finite number of predecided voltage levels.
Each sampled value at the input of the quantizer is approximated or rounded off to the nearest
standard predecide voltage level.
25.
The input signal x (t) is assumed to have a peak to peak swing of VL to VH volts. This entire
voltage range has been divided into "Q" equal intervals each of size "S".
At the center of these steps, the quantization levels q0, qv ... q7 are
located.
xq (t) represents the quantized version of x (t). We obtain x q (t) at the output of the quantizer.
When x (t) is in the range A0, then corresponding to each value of x (t), the quantizer output will
be equal to "q0". Similarly for all the values of x (t) in the range A l5 the quantizer output is
constant equal to "qj". Thus in each range from A0 to A7 , the signal x (t) is rounded off to the
nearest quantization level and the quantized signal is produced.
The quantized signal x (t) is thus an approximation of x (t). The difference between them is called
as quantization error or quantization noise. This error should be as small as possible. To minimize
the quantization error we need to reduce the step size "S" by increasing the number of
quantization levels Q.
27. Why is quantization required ?
If we do not use the quantizer block, then we will have to convert each and every sampled value
into a unique digital word.
This will need a large number of bits per word (N). This will increase the bit rate and hence the
bandwidth requirement of the channel.
To avoid this, if we use a quantizer with only 256 quantization levels then all the sampled values
will be finally approximated into only 256 distinct voltage levels.
So we need only 8 bits per word to represent each quantized sampled value.
Thus the number of bits per word can be reduced. This will eventually reduce the bit rate and
bandwidth requirement.
The difference between the instantaneous values of the quantized signal and input is called as
quantization error or quantization noise.
e =
xq(t)-x(t)
...(1.8.2)
The quantization error is shown by shaded portions of the waveform in Fig. 1.8.2.
The maximum value of quantization error is S/2 where S is step size. Therefore to reduce the
quantization error we have to reduce the step size by increasing the number of quantization levels
i.e. Q.
S2
...(1.8.3)
The relation between the number of quantization levels Q and the number of bits per word (N)
in
the transmitted signal can be found as follows :
Because each quantized level is to be converted into a unique N bit digital word, assuming a
binary coded output signal.
2N
I
...(1.8.4)
/i
Q=
Thus if N = 4 i.e. 4 bits per word then the number of quantization levels will be 2 i.e. 16.
1.8.3 Encoding:
Our final aim is to convert the signal into the binary form. So after quantizing, the signal is
applied to encoder block.
Encoder assigns unique binary number to each quantization level. That means each quantization
level is converted into the binary digits.
The bits in the binary digit are denoted by 'b'. The number of bits in the binary digit depends on
the number of levels (L). This relation is 2 > L
Thus
b > log2L
Ex. 1.8.1 :
Two signals x^t) = cos 20 nt and x2 (t) = cos 100 ret are sampled with sampling
frequency 40 Hz. Obtain the associated discrete time signals x,(n) and x2(n) and comment on the result.
Soln. :
= cos 20nt
xt(t)
...(1)
x^t)
= cos2rcF1t
...(2)
.-.
(ii)
= coslOOJtt
x2(t)
...(3)
= cos2rcF2t
x2(t)
...(4)
Compare it with,
.-.
Now discrete time signal x2(n) is obtained by putting t = T. Thus Equation (4) becomes,
x2(n) = cos 2 Tt ( 7 ) n
x2(n)
(5)1
Comment:
Given sampling frequency, fs = 40 Hz. Thus the frequency contained in signal should be less than I
fs or equal to ~x; that means < 20 Hz. But this is not the case in this example. So aliasing
takes place. Here I
both the sequences Xj(n) and x2(n) are equal; due to aliasing effect.
Ex. 1.8.2 :
xa(t)
...(1)
F1 = 25Hz,
F2=150Hz
xa(t)
...(2)
and
F3 = 50Hz.
____________________________________________________________________________________
There are various types of signals. Every signal is having its own characteristic. The
processing of signal mainly depends on the characteristics of that particular signal. So classification of
signal is necessary. Broadly the signals are classified as follows :
Mathematical expression :
+ 6) Here
x ( t ) = Asin(ci)t
A = Amplitude of signal
9 = Phase shift
Characteristics:
In this case the value of signal is specified only at specific time. So the signal represented at
"discrete interval of time" is called as discrete time signal.
The discrete time signal is generated from continuous time signal by using the sampling
operation. This process is shown in Fig. 1.9.2.
Consider a continuous analog signal as shown in Fig. 1.9.2(a). This signal is continuous in nature
from - to + .
The sampling pulses are shown in Fig. 1.9.2(b). These are train of pulses. Here the samples are
taken at Ts, 2 Ts, 3 Ts... and Ts is the sampling time.
Fig. 1.9.2(c) shows discrete time signal. Observe that this signal takes the value, only where the
sampling pulse is present. In between the two sampling pulses the signal is absent. So this is
called as discrete time signal.
In Fig. 1.9.2(a), on X-axis time (t) is plotted. On Y-axis the amplitude is plotted. So continuous
time signal is represented by x (t). Observe Fig. 1.9.2(c). On X-axis index n is plotted. Here n is
the number of corresponding sample. So discrete time signal is denoted by x ( n).
For signal in Fig. 1.9.2(a), the expression is,
x ( t ) = A cos tot and for
signal shown in Fig. 1.9.2(c), the expression is,
x (n) = A cos con
Characteristics:
Discrete time sinusoidal signals are identical when their frequencies are separated by integer
multiple of 2 n.
If the frequency of discrete time sinusoidal is a rational number, then such signal is periodic in
nature.
For the discrete time sinusoidal, the highest oscillation is obtained when angular frequency to =
n.
If the variation in the amplitude of signal is continuous then, it is called as continuous valued
signal. Such signal may be continuous or discrete in nature.
If the variation in the amplitude of signal is not continuous; but the signal has certain discrete
amplitude levels then such signal is called as discrete valued signal.
Such signal may be again continuous or discrete in nature as shown in Figs. 1.9.3(a) and 1.9.3(b).
Periodic signal:
A signal which repeats itself after a fixed time period or interval is called as periodic signal.
The
periodicity of continuous time signal can be defined mathematically as,
x(t) x ( t + T0)
...(1.9.1)
x ( n + N)
x(n) =
...(1.9.2)
Here number 'N' is the period of signal. The smallest value of N for which the condition of
periodicity exists is called as fundamental period.
(b) Discrete
time periodic
signal Fig.
1.9.4
Non-periodic signal:
called as
A signal which does not repeat itself after a fixed time period or does not repeat at all is
non-periodic or aperiodic signal. Thus mathematical expression for non-periodic signal is,
..
.
(1.9.3) ...
This is
(1.9.4)
Sometimes it is said that non-periodic signal has a period T = as shown in Fig. 1.9.4(c).
exponential signal having period, T = .
A discrete time sinusoidal signal is periodic only if its frequency(f 0) is rational. That means
frequency f0 should be in the form of ratio of two integers.
Proof:
For the discrete signal, the condition of periodicity is,
x(n +
N) = x(n)
...(1.9.5)
x(n) =
A cos(2rcf0 n + 6)
...(1.9.6)
Here A = Amplitude
and 0 = Phase shift
Now the equation of x(n + N) can be obtained by replacing 'n' by 'n + N' in Equation (1.9.6).
x(n + N) = A cos[2nf0 (n + N) + 0 ]
.-.
...(1.9.7)
According to condition of periodicity Equation (1.9.5); we can equate Equations (1.9.5) and
(1.9.7).
cos[27tf0 (n + N) + 0 ] = A cos(27tf0 n + 0)
A cos(2nf0 n + 2nf0 N + 0) = A cos(2rcf0 n + 0)
...(1.9.S
= 27tk
2rcf0N
...(1.9.9
where k is an integer
....Proved
...(1.9.10*
Here k and N both are integers. Thus discrete time (DT) signal is periodic if its frequency f 0 is
rational.
Here input sequence x(n) is expressed as summation of two discrete time sequences. We can
calculate the values of fj and f2 corresponding to Xj(n) and x2(n).
Let Xj(n) and x2(n) both be periodic discrete time signals (sequences).
f, = ^
kj
and
k2
f2 = -^
N,
The resultant signal x(n) is periodic if "j^~ is ratio of two integers. The period of x(n) will be
least
common multiple of Nj and N2.
Similarly if continuous time signals is,
We can calculate the values of T; and T2 corresponding to Xj(t) and x2(t). Then the resultant
T
Solved examples:
Ex. 1.9.1 :
Fig. P. 1.9.1
5-: n. : The sinewave shown in the Fig. P. 1.9.1 can be mathematically represented as,
= A sin 0)o t
x (t)
...(1)
Now, let us test if it satisfies the condition for periodicity i.e. if,
x(t)
= x(t + T0)
...(2)
Asin[o)0t + co0T0]
=
.-(3)
But 0)o = 27t f0 and T0 = f. Therefore Cfl0 T0 = 2n f0 x T = 2n. Substitute this in Equation (3), to
= A
...(4)
Fig. P. 1.9.2
Soln.: The exponential signal shown in Fig. P. 1.9.2 is expressed mathematically as,
= e-at
x(t)
...(1)
,T ,
-a(t + T)
x(t + T0) = e
=e
-at -oT
ButT0 = oo
= e
=0 .-.
x(t) * x(t
Ex. 1.9.3: What is the fundamental frequency of the waveform shown in Fig. P. 1.9.3, in Hz
and rad/sec ?
Soln.:
1
.-. Frequency f0 = -j- = TTZ = 5 Hz
1
...Ans.
Ex. 1.9.4 :
...Ans.
What is the fundamental frequency of the D.T. square wave shown in Fig. P. 1.9.4.
Fig. P. 1.9.4
Soln. :
22
" " N
When N = a positive integer indicating number of samples in one cycle.
...Ans.
Ex. 1.9.5 : State whether the following signals x(t) is periodic or not, giving reasons. If it is periodic,
find the corresponding period, x (t) = 2 cos 100 n t + 5 sin 501 Sofa.: The given signal is,
x(t)
...(1)
Let
x(t)
= X j W + x-jCt)
...(2)
and
= 5 sin 501
Her
...(3)
x2(t)
...(4)
...(5)
2TI 6 '
2rc~l
"
t[
T' = | = l
-<*>
co2 = . 18 7t,
.-. 2jtf2 = 18 rc
<> - -
2 7t
'
-(6)
"t2 = 9
The resultant signal x (t) is periodic if Tf is the ratio of two integers. From Equations (5)
and (6)
-eget,
li
T2
1/6_I 9 _ 9
1/9 " 6' 1 ~ 6
It is the ratio of two integers. Thus x (t) is periodic. Now the fundamental period of x (t) is
least
Given
x (t) =
3sin4t
...(7)
x(t) =
sin cot
...(8)
'
co =
.-. 2 n f =
-'-
It is not the ratio of integer values. Thus this signal is non-periodic in nature.
x(t) = 3 + t2
Given
...(9)
...(10)
>
...(11)
For any value of 'T0' Equations (9) and (11) cannot be made equal. Thus given signal is
-periodic.
Ex. 1.9.7 :
(i)
cos (0.01 n n )
(ii) cos (3 7t n )
(iii) sin (3 n )
Soln.:
(i)
Given sequence is
= cos (0.017tn)
x(n)
...(1)
= cos con
x (n)
...(2)
co =
0.017i
But
co = 27tf
27tf = 0.0171
"
0.017t 0.01
27t - 2
f=
...(3) I
Since frequency 'f is expressed as the ratio of two integers; this sequence is periodic. Now we I
have the condition of periodicity,
samples (ii)
Given equation is
x(n)
= cos(37tn)
...(5
/.
27tf = 3 7T
f = x cycles/sample
Since 'f is ratio of two integers; the given sequence is periodic. Comparing Equations (4) and
(6) we get,
samples (iii)
Fundamental period = N = 2
sin 3 n
x (n) =
...(7)
(0
= 3 .-. 2n f =
3
"
f =
Jn
Here 2 7t is not an integer. That means ' f' cannot be expressed as the ratio of two integers. Thus
the given sequence is non-periodic.
Even signals:
An even signal is also called as symmetrical signal. A continuous time (C.T.) signal x (t) is said
to be symmetrical or even if it satisfies the following condition :
Cor*
-t)
...forC.T. signal.
Here x ( - t) indicates that the signal is present for negative time period. That means x ( -1) is
the signal which is reflected about vertical (Y) axis. So even signals are symmetric about vertical
axis or at t = 0.
signal
Here x (-1) indicates that the signal is present for negative time period. While - x (t)
indicates r i.: the amplitude of signal negative. Thus antisymmetric signal is not symmetric about vertical
axis.
Energy signal:
The total normalized energy for a "real" signal x (t) is given by,
CO
E= f
...(1.9.12)
x (t)dt
CO
But if the signal x (t) is complex then Equation (1.9.12) is modified as,
CO
E= f|
...(1.9.13)
x( t ) | dt
oo
Note:
Ex. 1.9.8 :
What is the total energy of the rectangular pulse shown in Fig. P. 1.9.8 ?
Fig. P. 1.9.9
Deterministic Signal:
Examples:
Random signal:
Example:
Multichannel signals :
As the name indicates, multichannel signals are generated by multiple sources or multiple I
sensors.
The resultant signal is the vector sum of signals from all channels.
Example:
Multidimensional signals:
A good example of multidimensional signal is the picture displayed on the TV screen. To locate a
pixel (a point) on the TV screen two co-ordinates namely X and Y are required. Similarly this point is
a function of time also. So to display a pixel, minimum three dimensions are required; namely x, y and
t. Thus this is multidimensional signal. Mathematically it can be written as, P (
(x, y, t). -"parison of Multichannel and Multidimensional
Signal:
The discrete time sequence is denoted by x (n ). Consider such a discrete time signal as
On the X-axis index 'n' is plotted. Here 'n' is corresponding number of the sample. In the
given I diagram value of n varies from - 3 to + 3. On the Y-axis, amplitude of signal is plotted. The signal
a having some amplitude at each value of n. Now the different methods used to represent the signal x (n)
are as follows :
1. Functional Representation
3. Sequence Representation
2. Tabular Representation
1.
Functional Representation : For the signal shown in Fig. 1.10.1, the functional
representation of signal is as follows :
3.
x ( n ) = {1,2,-1, 1,2,0,1}
Here all the amplitudes of signal are written sequentially starting from
the leftmost amplitude.
ARROW ALWAYS INDICATES THE AMPLITUDE OF SAMPLE AT n = 0. If arrow is no:
shown in the sequence then by default it is at first position.
e.g.:
x ( n ) = {1,2,3,4,5}
x (n ) = {1 , 2, 3,4, 5}
Number of samples contained in the given sequence is called as the length of sample,
To adjust the length of sequence we can add any number of zeros at the beginning or at the end
sequence. This is called as ZERO PADDING.
e.g.:
I f x ( n ) = {1 ,2 ,0 ,1 ,2 }
or x ( n ) = {1, 2, 0 , 1, 2, 0, 0}
t T
(i)
x(n) = {1, 2, 0 , - 1 , 1 }
(ii)
x(n) = { 0 , 0 , - 1 , 2 , 3 }
loin. : These signals are as shown in Fig. P. 1.10.1(a), (b) and (c) respectively.
____________________________________________________________________________________1
.11
In the analysis of communication systems, standard test signals play a vital role. Such
signals are used to check the performance of a system. Applying such signals at the system; the output is
checked. Now depending on the input-output characteristics of that particular system; study of different
properties of a system can be done. Some standard test signals are as follows :
Exponential signal
Sinusoidal signal
for all
Jlfort = 0
B" N
5(t) = lOfort^O
It is as shown in Fig. 1.11.1(b).
A discrete time unit step signal is denoted by u(n). Its value is unity (1) for all positive values of
n. That means its value is one for n > 0. While for other values of n; its value is zero.
flforn>0
l0forn<0
'
u(n)
, ,
J l. f o r t > 0 U(t)
10fort<0
A discrete time unit ramp signal is denoted by ur ( n ). Its value increases linearly with sample
number n. Mathematically it is defined as,
~t>\
f n for n > 0
r < n ) = lOforn<0
ort>0 r ( t ) ~
L0fort<0
Fig. 1.11.3(b).
flf
It is as shown in
\
x(n) =
an
...(1.11.1)
reJe
Here 9 denotes the phase.
cases.
Case (i): When a > 1
Let a = 3. Thus we have,
x ( n ) = a" = 3".
Graphically such signal is represented as shown in Fig. 1.11.4(a).
x(n) =
...(1.11.2)
Case(ii) :WhenO<a<l
x ( n ) = a=(Dn
a = -3 .-.
x ( n ) = (-3)n
a = -T
A continuous time exponential signals for various values of a are as shown in Fig.
1.11.5.
(a)
(b)
Fig. 1.11.5 : Continuous time exponential signals
(c)
(d)
Fig. 1.11.5 : Continuous time exponential signals
A = Amplitude
co = Angular frequency
___________________________________________________________________________________1
1.
Time advance
Folding
Folding and delay
As the name indicates, delay means signals are not appearing instantly at the receiver. Almost :
- signal used in communication provides time delay at the output. Some common examples are TV
-s. telephone signals, radar signals etc.
In case of discrete time signals, the given sequence can be delayed by few samples. We know
:-_: ;.:crete time signal is denoted by x ( n ). Suppose we want to delay this sequence by 'k'
samples. It tZ be denoted by x (n - k)
and
Here k is an integer.
E.g.: Let the given signal be,
x ( n ) = {1,2,3,4,5}
x ( n - k ) = x ( n - 2 ) = {0,0,1,2,3,4,5}
We know that, arrow always indicate the n = 0 sample. Since delayed sequence is shifted
towards Tginr so the first sample starts at n = 2. In the original sequence it is starting at n = 0. Thus to
delay the
eccr.;e.
Shift the diagram towards right by 'k' samples.
Shift the arrow towards left by 'k'.
To adjust the length of sequence, add zeros.
version x (n - k)
2.
(b) Delayed
Fig. 1.12.1
Time advance operation means the signals are present before they are generated. In a real time
operation, this is practically impossible. But if the signals are stored in the memory of computer before
starting a particular operation then it is possible to have time advanced signal.
Time advance operation is opposite to the time delay operation. Consider the same sequence x
( n ) shown in Fig. 1.12.1(a).
If we want to advance this sequence by two samples then it is denoted by x ( n + 2 ). In this case
the diagram is shifted towards left by two samples. This sequence is shown in Fig. 1.12.1(c).
Fig. 1.12.1(c): Advanced version, x ( n + k)
From Fig. 1.12.1(c), we can write advanced sequence as,
x ( n + k) = x ( n + 2 ) = {1, 2, 3, 4, 5}
3.
Folding (FD):
Folding is also called as reflection. So folding means taking the mirror image of signal. Tha
means the signal is folded about time origin n = 0. Here independent variable 'n' is replaced by - n.
Thus if x (n) represents input signal then x (- n) represents folded input signal.
Consider the same input signal x (n) as shown in Fig. 1.12.1(a)
.-. x ( n ) =
{1,2,3,4,5}
4.
This operation is a combination of folding and delaying operation. Now let us take a quick
view it the modifications used for different operations.
I A)
To obtain delayed sequence, shift the original diagram towards right by 'k'
samples. To obtain advanced sequence, shift the original diagram towards left by 'k'
samples. To obtain the folded version, take the mirror image of the diagram at n = 0.
B)
Folding !
---------------x(-n)
> x [ - ( n - k ) ] = x ( - n + k)
But stick to the basic concepts. Delay means shift the diagram towards right by k samples.
Consider the same original sequence x ( n ) = {1 , 2, 3, 4, 5} as shown in Fig. 1.12.1(a). T
sequence x (- n + 2)
Fig. 1.12.1
x ( - n + 2 ) = {5,4,3,2,1}
5.
The advanced version of original sequence x ( n ) is denoted by x ( n + k). The folded version a
version a
advance this sequence by '2' samples, then the advanced version is denoted by x ( - n - 2 ). Such i
sequence is as shown in Fig. 1.12.1(f).
Remember the basic rule. Advancing the sequence means shifting the diagram towards left by 'k'
dimples. From Fig. 1.12.1(f), we can write the sequence x (- n - 2) as,
x ( - n - 2 ) = {5,4,3,2,1,0,0}
As the name indicates, time scaling operations are related to the change in time scale. There are '': n,pes of
time scaling operations. Down scaling (Compression) Up scaling (Expansion)
Consider the same sequence x ( n ) = {1 , 2, 3, 4, 5}. It is shown in Fig. 1.12.1(a). Let this
T
n i be input to some device which produces output y (n ). Also say,
y(n)
= x ( 2 n ) Now
x(l) =2x (2 )
= 3
x( 3) = 4
x(4) = 5
x(5) = 0...
y(n) = x ( 2 n )
.-. y(0) = x ( 0 ) = l
y(l) = x ( 2 ) = 3
y(2) = x ( 4 ) = 5
y(3) = x ( 6 ) = 0
.-. y(n) = x ( 2 n ) = {l,3,5,0,..........................}
2.
Up Scaling or Expansion :
y(0)
= x() =x(0)=l
y(l)
= x ( x J > No sample
y(2)
= x(f) =x(l)=2
y(3)
y(4)
= x(|) =x(2) = 3
y(5)
y(6)
= x() =x(3) = 4
y(7)
= x ( x J = x ( 3.5 ) No sample
y(8)
= x() = x ( 4 ) = 5
Thus
y(n)
= xf|j = {1 , 0, 2, 0, 3, 0,4, 0, 5}
upscaling operation.
As the name indicates, in case of amplitude scaling operations, amplitude of signal is char,;:
Different amplitude scaling operations are as follows :
(Attenuation)
Addition Multiplication
1.
Here
Now let,
x(0) = 1,
x(l) = 2,
x(3) = 4,
x(4) = 5.
x (2) = 3,
y(n) = 2 x ( n )
L e t y ( n ) = ^f
2
t
3.
Addition:
t
and x2(n) = {2,2, 0,2,2}
t
Let y (n ) = Xj (n) + x2 (n )
Multiplication:
Consider
X j ( n ) = {1,1,0,1,1}
Let y(n) =
x 1 ( n ) x x 2 ( n ) .-. y(n) =
{2,2,0,2,2}
Ex. 1.12.1 :
A
di
s
cr
et
e
ti
m
e
si
g
n
al
is
gi
v
e
n
b
y,
x(
n)
=
{1
,1
,
1,
1,
2
}
(a) x ( n - 2 )
(b) x ( n + 1 )
(c) x ( 3 - n )
(d) x ( n ) u
(n-1)
(e) x ( n - 1 )
5(n-1)
(f) Even
samples of x ( n )
(g) Odd
samples of x ( n )
(a)
x(n-2):
x(3-n):
:x(n)u(n-1):
u (n) means unit step and u (n - 1 )=> unit step delayed by 1 sample.
x(n-1)8(n-1):
Even samples of x ( n ):
I means we have to find out x (2n).
Given sequence is, x ( n ) = { 1, 1 , 1, 1, 2 }
ex(-l)=l, x ( 0 ) = l , x ( l ) = l , x ( 2 ) = l , x(3) = 2
r
At
want
x
n
2n
=
).
-1
Putting
different
x(2n)
At n = 0,
x (2n)
= x (0) = 1
At n = 1, x ( 2 x l ) = x ( 2 )
At n = 2, x ( 2 x 2 ) = x ( 4 )
=1
=0
values
of
=
we
x(-2)=
get,
0
2n) = { 0 , 1 , 1 , 0 , 0 . . . . }
T
This sequence is shown in Fig. P. 1.12.1(g).
Fig. P. 1.12.1(g): Even samples of x (n )
(g)
at
n = -1,
x(2n+l)=x(-
2+l)=x(-l)=l
at n=0,
x(2n+l)=x(0+l)=x(l)=l
at
n k 1, x ( 2 n + l ) = x ( 2 + l )
at
n = 2, x ( 2 n + l ) = x ( 4 + l ) = x ( 5 ) = 0
=x(3)
=2
.-. x ( 2 n + l ) = {1, 1 , 2 , 0 . . . }
y,(n)
= 2 x( n) + 5(n) (ii)
y2 (n)
= x (n) u (2 - n)
x(n) = 2""
for-2<n< 2
That means the range of 'n' is from -2 to + 2. Putting these values of 'n' in the
-2
"4
at
n=-2,
x(n) = 2
at
n = -1,
x(n) = 2~" = 2+
=2
at
n = 0,
x(n) = 2~" = 2 = 1
-n
-1_ 1
=2 _ ^
at
n = 1,
x (n) = 2
at
n = 2,
=|
Fig. P. 1.12.2(a):
Discrete time sigj
x(n) = 2_n
lil
yi
5(n) = {1}
T
(')
that means u ( - n + 2). Now u (- n) indicates folded unit step as shown in Fig. P. 1.12.2(d). While u (- n
+ 2). Indicates that we have to delay u (- n) by two samples. That means shift u (- n) towards right by
two samples. This is shown in Fig. P. 1.12.2(e). x (n> u ( - n + 2) indicates multiplication of x (n) and u
(- n + 2). This operation is shown in Fig. P. 1.12.2(f).
Soln.:
(i)
5(n) = 1
for n= 0
= 0 otherwise
Let the signal to be generated be x(n). The duration of this signal is from n = 0 to n = N and
the amplitude of every sample is 1. That means the required sequence is,
x(n) = {1,1,1,1,.... 1}
n=0
T
n=N-l
(ii) Unit step starts from n = 0 and every sample is having magnitude equals to ltill n = . But
w want the sequence having samples of magnitude 1 till n = N. This is obtained as shown i Fig.
P. 1.12.3.
x(n-3)
(ii)
x(3-n)
x(n)u(3-n)
Fig. P. 1.12.4(a)
Here x (n - 3) indicates delay operation. Delay means shifting the diagram towards right.
Thus I (n - 3) is obtained by shifting x(n) towards right by '3' positions. It is shown in Fig. P.
1.12.4(b).
It can be also written as. x (- n + 3). Brace x (- n) indicates folding operation. Thus x (- n +
3) indicates delay of folded signal, sy 3 positions. This operation is shown in Fig. P. 1.12.4(c).
(ii)
x(2n)
follows :
For
n = -2
=>x(-4) = 0
For n = -1
=> x(-2)= j
For
For
For
n = 0 => x(0) = 1
n = 1
=> x(2)
n = 2
=> x (4)
= 1
= 0
(iv)
Fig. P. 1.12.4(d)
x (n) x(3 - n)
as shown in Fig. P.
______________________________________________________________________________1.
Unit step u (t) and unit impulse 8 (t) of CT can be related as follows :
Mathematically
Fig. 1.13.1
We know that uA (t) is continuous function and hence it is differentiable. Derivative of uA (t) as I
shown in Fig. 1.13.1(c).
As A - 0 Fig. 1.13.1(c) becomes unit impulse.
->0~dT"
"
5t
Or
Importance of impulse:
To convert CT-signal into DT-signal samples of CT signals are required to take at regula interval of
time.
For sampling of CT signal a train of unit impulses is required.
Let x (t) and y (t) be two continuous time signals with y (t) derived from x (t) through i
combination of time shifting and time scaling operations.
Let the relation between x (t) and y (t) be given mathematically as follows :
x(at-b)
_
y(t) =
...(1
Let the relation between x (t) and y (t) satisfy the following two
conditions :
Condition -1
y (0) = x (- b)
and
Condition-2 y (b/a) = x
(0)
In order to obtain y (t) correctly from x (t), it is necessary to perform the time shifting and
time scaling operations in the correct order.
The proper order is decided on the basis of the fact that in the scaling operation "t" is
replaced by "at" and in the time shifting operation "t" is replaced by (t - b). So the time shifting
operation is carried out on x (t) to get an intermediate signal y' (t) as,
y'(t) =
x(t-b)
...(1.13.2)
Then the time scaling operation is performed on y' (t) by replacing t by "at". The result is,
I
desired signal y (t).
y(t)
= y'(t)|t=at
.-.
y(t) = x(at-b)
...(1.13.3)
If the rule is not followed, then we do not obtain the desired signal y (t). This is
demonstrated as
First perform the time scaling operation on x (t) to get y' (t)
y'(0 = x(at) Tben perform the time
shifting operation on y' (t) by replacing t by (t - b).
.-. y(t) = x [ a ( t - b ) ] = x [ a t a b ] "* 'e have not obtained the desired signal y (t) = x (at - ab). T:
understand the precedence rule, solve the following example :
:"
If x (t) = rect(t/3) then obtain y (t) = x (2t - 3) first by following the precedence rule
. .
J.
Draw x
(t): is
given by,
Step 5 : Time-scaling :
Refer Fig. P. 1.13.1(e). The time scaled signal is y' (t) = x (2t).
Step 6 : Time-shifting :
Refer Fig. P. 1.13.1(f). The intermediate signal y'(t) has been delayed by 3 time units. I obtain y
(t).
But y (t) = 2y' (t) = x [ 2 (t - 3) ] = x [ 2t - 6 ] which is not the desired signal.
Questions
Define the term signal and classify it. With the help of block
Continuous
With the help of neat diagram, explain the quantization process. Explain
the different methods of representing D.T. signals. Z. ' I
different standard signals used in DSP.
o_____
\T^/
Fourier Analysis of
^*^^H|
Periodic and Aperiodic Continuous 1
Time Signalsand Systems
|
syllabus:
Introduction, trigonometric Fourier series, complex or exponential form of
Fourier series, Parsevals identify for Fourier series, Power spectrum of a periodic
function. Fourier transform and its properties, Fourier transforms of some important
signals, Fourier transforms of power and energy signals.
Contents
Page No.
2.1
Amplitude
2-2
2.2
CT Fourier
Series
2-5
2,3
Transform
CT. Fourier
2-27
____________________________________________________________________________________2
.1
All the signals till now were drawn with respect to time. That means time "t" was considered as a
variable. The representation of signal with respect to time is called as its time domain
representation.
The time domain representation of the signal is not sufficient for its analysis. Hence we have to
use the frequency domain representation of the signal for the sake of analysis.
In the frequency domain representation, the variable plotted on the X-axis is frequency "f' rathe:
than "t".
The signal represented in the frequency domain is called as the line spectrum. The line spectrum
consists of two graphs namely,
The signal x(t) and its line spectrum are shown in Fig. 2.1.1.
The graph of instantaneous signal voltage versus time is called as the time domain representation]
It is as shown in Fig. 2.1.1(a). The time domain representation gives us the follow* information:
Shape of the signal
Its frequency
Type of the signal (periodic or nonperiodic).
One cycle period.
But we can not know anything about what frequency components are present and in proportion
they have been mixed in order to obtain the particular shape of the signal.
All this information can be obtained from the line spectrum of a signal.
Line spectrum [Fig. 2.1.1(b)] is the representation of the same signal x(t), now in the frequea
domain.
It can be obtained by using either Fourier series or Fourier transform. It consists of the amplai and
phase spectrums of the signal.
The line spectrum indicates the amplitude and phase of various frequency components present in
the given signal. The line spectrum enables us to analyze and synthesize a signal.
2.1.1
The line spectrum is useful in understanding the existence and amplitudes/phases of various
frequency components present in a waveform.
The important conventions about the line spectra are as follows :
1.
2.
In all the spectral drawings, the independent variable plotted on the x-axis is frequency f
in Hz and not co.
Phase angle is always measured with respect to the cosine waves. That means it is
measured with respect to the positive real axis of the phasor diagram. Hence it is
necessary to convert sinewaves to cosines using the following standard identity :
sin cot =
...(2.1.1)
-A cos tot
= A cos (tot 180)
...(2.1.2)
The additional phase change of 180 converts the negative amplitude "- A" to positive
amplitude "+ A". We can choose either + 180 or - 180, as the effect is going to be the same.
2.1.1:
-.
In the given signal, the first term represents a dc term which has a zero frequency. The other two
ens can be written as follows :
First term :
3 = 3 cos 2m 01
as f = 0
Second term : - 5 cos (40 rc t - 30 ) = 5 cos ( 2 n 201 - 30 + 180) = 5 cos ( 2 n 201 + 150 )
Thus the negative amplitude has been made positive by adding a phase angle of 180.
Third term : 4 sin 120 n t = 4 sin 2 n 601 = 4 cos ( 2 n 601 - 90 ).
Thus the sine term has been converted to the cosine term by adding a phase shift of - 90.
The amplitudes, frequencies and phase angles of the three terms are listed in the Table P. 2.1.1.
Table P. 2.1.1
spectrum
Fig. P. 2.1.1
Conclusions:
1.Looking at Fig. 2.1.2(a) i.e. the amplitude spectrum, we conclude that in the single a
spectrum there is only one frequency component present at f = f0 with an amplitude A.
Whereas in double sided line spectra, two frequency components f0 and - f0 are present i
amplitude (A/2) but no change in polarity.
2.The single sided phase spectrum contains only one component at f0 with phase <|>. But tkJ
sided spectrum contains two components at f0 and - f0 with phases equal to <J> anij
respectively. Thus phase shift remains unchanged but they have opposite signs.
3.
From Fig. 2.1.2(a) it is clear that the double sided amplitude spectrum has an even
symmeo^
Fig. 2.1.2(b) shows that the double sided phase spectrum has an odd symmetry.
The double sided line spectrum representation is very useful in mathematical anal negative
frequency components present in the double sided spectrums are not practically present.
The polar fourier series is derived from the trigonometric fourier series by combining the sua
and cosine terms of same frequency. The polar fourier series representation of x( t) is as follows :
E spectrum :
The line spectrum of x( t) can be plotted using Equation (2.2.8). A line spectrum of x( t) with
arbitrary values of amplitudes and phases is shown in Fig. 2.2.1.
Line spectrum of x( t)
-.5 seen from Fig. 2.2.1(b) the frequency spectrum of a continuous signal is discrete in nature. The
frequency components f0, 2 f0, 3 f0.... etc. are called as the "spectral components".
The adjacent spectral components are spaced by "f0" from each other. As the spectrum is nsisting
of vertical lines, (Cv C2, ....) this spectrum is called as the line spectrum.
(b)
Substituting the sine and cosine functions in terms of exponential function in the expression for
the quadrature fourier series, Equation (2.2.1), we can obtain another type of fourier series called
the exponential fourier series.
A periodic signal x (t) is expressed in the exponential fourier series form as follows
...
(2.2.12)
As seen from the Equation (2.2.11), this fourier series consists of exponential terms. Therefore it I
This theorem relates the average power P of a periodic signal to its "fourier series" coefficients.
The Parseval's theorem states that the total average power of a periodic signal x (t) is equal to the I
sum of the average powers of the individual fourier coefficients i.e. Cn.
..
...(2.2.13)1
...(2.2.14
Proof:
The total average normalized power of signal x (t) is given by the following equatio
The term inside the square bracket in Equation (2.2.18) is nothing but "Cn ".
The interpretation of the Parseval s theorem is that the total average power of the signal x (t)
can found by squaring and adding the heights | C | of the amplitude lines in the spectrum of the
periodic signal x (t).
Thus the Parseval's theorem implies "superposition" of the average powers.
The fourier series stated in the Equations (2.2.20) and (2.2.21) will exist if and only if the
I periodic signal x( t) satisfies the following conditions. These are known as the "Dirichlet" conditions. I
They are as follows :
1. The periodic signal x(t) and its integrals are finite and single valued in the interval I (t to t + T0),
i.e. over a period of one cycle T0.
2. x( t) must have only finite number of discontinuities in the given interval of time.
3. x( t) should have only finite number of maxima and minima in the given interval of time.
4. The function x( t) is absolutely integrable, that is
Ex. 2.2.1 :
Obtain the quadrature fourier series for the rectangular pulse
train shown r I Fig. P. 2.2.1.
Fig. P. 2.2.1
Soln.: Let us obtain the quadrature fourier series for the given rectangular pulse. The
quadrature fo series is given by the following expression :
To find the fourier coefficients, we must consider one complete cycle of x (t) for
integration. I
T0
T0
Here we will consider one cycle from t = - ~z~ to t = ~x~. Let us obtain the fourier coefficients now.
Ex. 2.2.2:
Obtain the exponential fourier series for the rectangular pulse train
shown in Fig. P. 2.2.2(a) and sketch the spectrum.
(ii)
ni
Spectrum of the signal x (t): From the value of Cn in Equation (4), it is clear that Cn does not
have any imaginary part, "berefore the amplitude spectrum of x (t) is given as,
Thus the amplitude spectrum of a rectangular pulse of duration X is a sine function. The spectrum
. -own in Fig. P. 2.2.2(b).
The imaginary part of Cn is zero therefore the phase spectrum is zero for all the
values of f. The _>e spectrum is shown in Fig. P. 2.2.2(c).
Conclusions:
Important points related to the amplitude and phase spectrums of Fig. P. 2.2.2(c) are :
3. Zero crossings occur in the envelope of the amplitude spectrum at frequencies of f = 1/T. ~ 1
4.
The phase spectrum takes on values of 0 corresponding to the positive values of | Cn |. Hon it
takes on values of 180 corresponding to the negative values of | Cn | e.g. bet* f = 1/T,
2/x. The negative values in the amplitude spectrum are made positive by assigl phase shifts of
180.
5.
Note that the choice of phase shift + 180 or - 180 is arbitrary. However we have used : :<:
them to preserve the "antisymmetry" of the phase spectrum.
Ex. 2.2.3 :
durand 50 msec and are separated by intervals of 500 msec. Assuming that the centre
ol is located at t = 0, obtain the fourier series of the above signal and sketch the specrjJ
Soln.:
It is given that:
50 msec.
T0 = 500 msec.
A = 10 volts, ;jr=0.1
x=
(i)
The other values of I Cn I are found using Equation (1), as shown in the Table P. 2.2.3.
Using the Equation (5), we can calculate the amplitude response as shown in Table P.
2.2.3. Table P. 2.2.3 : Amplitude and phase spectrums for positive value of "n"
To obtain the values of | Cn | for negative values of "n", substitute negative "n" in
Equation (5).
Thus the values of amplitude spectrum for various negative values of "n" will be same as that
fori the corresponding positive values. The amplitude and phase spectrums are as shown in Fig. P.
2.2.3(b).
Ex. 2.2.4 :
Obtain the fourier series of the unit impulse train shown in Fig. P. 2.2.4(a).
Also plot thai amplitude and phase spectrums for the same.
In the Ex. 2.2.2, we have already obtained the exponential fourier series for a train of rectangulJ
pulses of duration x and period T0. Its value is,
.-.
...(2)
As
...(3)
Henc
e a rectangular pulse will be equivalent to a unit impulse as its area i.e. "AT" approaches
or." and as its width x approaches zero.
Ax-*landx->0
Ji
...(4)
To obtain Cn for the unit impulse train : We can obtain the value of "Cn" for the unit impulse
This is the value of Cn for the unit impulse train. To obtain the exponential fourier series :
Substitute the value of "Cn" from Equation (5) into the standard expression of exponential
fourier
This is the required fourier series for the unit impulse train
Amplitude spectrum :
Amplitude spectrum:
This means that for every value of "n" the value of Cn is going to be the same, equal to ( IT The
amplitude spectrum also is train of impulses each having amplitude of ( 1/T0 ), as showx 1 Fig. P.
2.2.4(c).
Phase spectrum :
The phase spectrum <|)n = arg ( Cn ) = 0 as Cn is constant. The phase spectrum is as shown in
Fig. P. 2.2.4(c).
Ex. 2.2.5 :
Obtain the fourier series of the sawtooth waveform shown in Fig. P. 2.2.5(a)
and plot si spectrum.
Sain.:
The sawtooth signal x (t) shown in Fig. P. 2.2.5(a) can be represented over one cycle as,
.
To find Cn:
In order to represent the signal in the form of an exponential fourier series, let us find "Cn" using IK
following equation :
.
r " ""
Ex. 2.2.6 :
Fig. P. 2.2.6(a)
(i)
Describe analytically that the above pulse is a rectangular pulse using follow
equation.
(ii)
Determine the fourier series expansion for the signal x( t). (iii)
:r :
7i f - J represents a rectangular pulse centered about t = 0. The width of this rectangular pulse
The given waveform of Fig. P. 2.2.6(a) can be obtained by adding infinite number of
shifted rectangular pulses which are of width x and centered around t = T0, 2 T0 .... as shown
in Fig. P. 2.2.6(c). The mathematical expressions for such shifted rectangular pulse are written in
Fig. P. 2.2.6(c) itself.
Fig. P. 2.2.6
Addition of the waveforms in Fig. P. 2.2.6(b) and (c) will give us the waveform in Fig. P. 2.2.6(a).
I
(ii)
For the fourier series and fourier coefficients refer Ex. 2.2.1.
Ex. 2.2.7:
Find the quadrature fourier series for the full wave rectified sine wave shown
in I Fig. P. 2.2.7.
Soln.:
So, we will have to find the values of the fourier coefficients aQ, an
and bn.
Phase spectrum:
Phase spectrum
12.3
C.T.
Fourier
Transform
:
_____________________________________________________________
Till now we have seen how to represent the periodic signals extended over the interval (- , ),
using the fourier series. Non-periodic time limited signal can also be represented by the fourier
series.
.................................................lines at f0 2f0,
Fig. 2.3.1
However the non-periodic signals which extend from - to can be represented more
conveniently using the "Fourier Transform" in the frequency domain.
It is possible to find the fourier transform of periodic signal as well. For the periodic signals
T0 > . Hence the frequency f0 = 7f> 0. Therefore the difference between the spectral
components which is f0 (as seen in the line spectrum) becomes extremely small and they come
very close to each other. Due to this the frequency spectrum appears to be continuous as shown
in Fig. 2.3.1(a) and (b).
Thus fourier transform can be used for the "analysis" of a signal. It is used for transform a 3t I
from the time domain to frequency domain.
The F.T. can also be used for analysis of LTI systems.
The signal x (t) can be obtained back from fourier transform X (f) by using the inverse few
transform. The inverse fourier transform (IFT) is defined as follows :
Representation :
The signal x(t) and its fourier transform X(f) form a fourier transform pair which car
represented as,
- X(f)
x(t)
:.
= F[x(t)]
X(t)
J
= F_1|X(f)|
X(t)
U
X(f)|-e
X(f)
j e ( f)
In this expression:
| X (f) | =
9 (f)
The amplitude spectrum is a graph of amplitude versus frequency. Whereas the phase
spectrum is 4)h of phase angle versus frequency.
These conditions should be satisfied by a signal x (t), then only it is possible to obtain
the fourier
transform of x (t).
For the periodic signals the integration is obtained over one period however for the
periodic
The signal x(t) will have to satisfy the following conditions so that it's fourier transform
can be
obtained:
1. The function x (t) should be single valued in any finite time interval T.
3.The function x (t) should have a finite number of maxima and minima in any finite
interval of time T.
oo
means J |x(t)|dt<
That
...(2.3.7)
oo
The conditions stated above are sufficient conditions, but they are not the necessary
conditions.
The amplitude and phase spectrums are continuous rather than being discrete in nature.
Out of them, the amplitude spectrum of a real valued function x (t) exhibits an even symmetry.
X(f) = X(-f)
.-.
...(2.3.8)
6(f) =
-0(-f)
...(2.3.9)
h. 2.3.1 :
Find the fourier transform of the decaying exponential pulse shown in Fig. P. 2.3.1.
Soln.:
The exponential pulse shown in the Fig. P. 2.3.1 can be represented mathematically as
follows:
x(t) = e~at
= 0 fort<0
fort>0
x(t) = e~atu(t)
The meaning of both the Equations (1) and (2) is the same. This is because u (t) = 1 for t > 1
multiplying by u (t) does not affect the original function.
Ex. 2.3.2 :
Find the fourier transform of the exponential pulse shown in Fig. P.
2.3.2(a). Asel amplitude and phase spectrums for the same.
Fig. P. 2.3.2(a)
Soln.:
= 0
for t > 0
0(f) = 0.
The amplitude and phase spectrums are plotted as shown in Fig. P. 2.3.3(b).
------------------------------------------------------------------------------------------------------------Ex 2.3.4 :
Obtain the fourier transform of the delta function shown in Fig. P.
2.3.4(a).
X(f) =
...(1)
J x(t)e-j2,rftdt= J 8(t)e*"*<k
OO
OO
We cannot substitute the value of 8 (t) directly in the Equation (1) because it is infinitely large at =
Therefore let us use the sifting property of the delta function.
Sifting property of delta function :
The sifting property states that
oo
j f (t) 8 (t
- td) dt = f(td)
...(2)
OO
(iii)
.-.
X(f) =
oo
J e"j2rft-5(t-0) .-.
OO
X(f) = e"j2nft<i,but td = 0
.-.
X(f) = e'i2m'=l
...(3)
Thus (Sit) - I
__________________i
I
The amplitude spectrum of the delta function is a
shown in the Fig. P. 2.3.4(b) This shows that the delt
function contains all the frequencies from - to wit]
equal amplitudes. The fourier transform of a delt
function is a dc signal.
Ex. 2.3.5 :
Obtain the FT. of the antisymmetrical pulse shown in the Fig. P. 2.3.5.
Fig. P. 2.3.5
Soln.:
t>0
= |1|
t=0
= -eat
t <0
Ex. 2.3.7 :
Soln.:
=u
eisewnei
Ex. 2.3.8:
Obtain the fourier transform of a rectangular pulse of duration T and
amplitude A as shown in Fig. P. 2.3.8(a).
Soln.:
The rectangular pulse shown in Fig. P. 2.5.8(a) can be expressed mathematically as,
= 0
elsewhere
...by definition of FT
je -jo
eJ -e J
sine =
we get,
...(2)3
t
sinc(fT) =0
.\ AT sine (0)
= AT
The sine function will have zero value for the following
values of "fT" :
forfT = 1,2,3,........................
o_____
p-^
Application of Laplace
^^^| Transform to System Analysis
I
abus:
Introduction, definition, region of convergence (ROC) LT of some important
functions, Initial and final value theorems, convolution integral, Table of Laplace
transforms, partial fraction expansions, network transfer function. S-plane Poles
and zeros. LT of periodic functions. Application of LT in analysing networks.
Contents
Page No.
3.1 Introduction
3-2
3-2
3-16
3-26
3-50
3-51
3.7
3.1
Introduction :___________________________________________ I
3.8
We have studied the fourier analysis of continuous time signals and systems. We know
that I fourier transform exists if the signals have finite energy. But for the signals such as ramp, rising
exponents etc.; this condition of finite energy is not satisfied. Thus fourier transform does not exist for I
such signals.
3.9
By the use of laplace transformation; this limitation can be avoided. We know that in
fourier transform the variable 's' = jco. But in laplace transformation variable 's' can be expressed as,
3.10
Here a is real part which represents the attenuation factor and jco is imaginary part, in
which co is angular frequency.
3.11
Laplace transform exists for almost all signals of practical interest. Some of the
advantages of I Laplace transform are as follows :
1. Solution of integrodifferential equations of continuous time systems can be easily obtained.
2. Initial conditions are automatically incorporated.
3. Both complementary and particular solution can be obtained in one operation. Thus it gnJ
complete solution.
3.2
Definition of Laplace Transform :____________________________I
3.12
3.13
3.14
3.15
oo
X(jco) =
...(3.2-1
3.18 oo
3.19
We know that fourier transform exists only if x (t) has a finite energy. If the signal x (t) is
hz infinite energy then we will introduce a convergence factor, e~ ; to convert x (t) into finite enesJ
signal. Then we will denote the resulting transformation by X (a + jco). Thus Equation (3.2.1) becomoJ
3.20
Notation :
3.21
3.25
or L{x(t)} = X(s)
Here bidirectional arrow indicates that we can obtain original signal x (t) from its
oo
X(s) =
...(3.2.4)
3.31
Jx(t)e"stdt
3.32
3.33
oo
J|
<
...(3.2.5)
Now we have,
3.36
3.37
-jrot
3.38
= e
s = a+jto
-st
/0 ~ ,x
-at
.-. e
...(3.2.6)
3.39 c have e~JC0t = cos cot j sin cot. Thus the value of e~JC0t is always in the range + 1 to - 1. So
we ami -. c:fy the condition of existence of laplace transform as,
3.40 oo
3.41
x(t)e~ot|dt
3.42
3.43
3.44
< oo
I|
...(3.2.7)
oo
Equation (3.2.7) gives sufficient condition for the existence of laplace transform.
rta;
3.45 . -e range of values of a for which Equation (3.2.7) attains some finite value is called as
region ::;ence(ROC).
3.48
3.50
We can factorize the numerator and denominator. Thus after obtaining roots of
numerator and I denominator; Equation (3.2.8) can be written as,
3.51
-(s-z 1 )(s-z 2 )....(s-z n )
3.52
3.53
a0
1
Zeros:
3.54
3.55
If 's' takes values zx, z2 .... zn then according to Equation (3.2.9); laplace transform X i
vanishes. Thus such complex frequencies are called as zeros of laplace transform. Zeros are denoted i
the mark '0' in the pole zero diagram.
Poles:
3.56
3.57
When V takes the values pj, p2 .... pm then according to Equation (3.2.9); laplace
transform Xm becomes infinity. Such complex frequencies are called as the poles of laplace transform.
Poles m denoted by the mark 'X' in the pole-zero diagram.
Example:
3.58
3.59
3.61
X(s)
(s + 2)(s-3)
Zeros:
3.63
3.64
s+l
= 0
=>
s-2 = 0
=>
3.70I
function :
Fig. 3.2.2
It is defined as,
8(t) =
3.72
fort = 0
= 0
otherwise
...
(3.2.10)
3.73
X(s) =
|x(t)-e~stdt
3.74
oo
...(3.2.11)
IX(s) =
l-e=l
L.T.
.-. o(t) <-----------------------------------------------*1
Since 's- term is absent in Equation (3.2.12);
ROC is entire s-plane.
Unit step:
3.75
3.76
v________________________1
-------
3.77 ROC:
3.78
The laplace transform of unit step is - and it is for the range Re {s} > 0, that
for a > 0. Thus ROC is o>0. It has a pole at s = 0 that means at origin. The sketch of ROC is start
3.79
Fig. 3.2.4.
3.80
3.81
fort>0 = 0
otherwise
3.82
3.83
3.84
3.86
3.85
3.87
F"
3.88
/x(t)e~stdt
We have, X(s) =
...(3.2.19)
3.89
oo
3.90
Here
x(t) = e a t - u (t ) ...
(3.2.20)
3.91 v'..:tiplication by unit step u (t) indicates that the exponential pulse ea' is present in the range
3.92 "": -5 limits of integration in Equation (3.2.19) will be from t = 0 to t = .
ROC:
3.93
3.94
The laplace transform is _ . It has
pole as + 'a'.
3.95
Thus ROC is Re {s - a} > 0, that means Re {s} >
a. But Re {s} means a.
3.96
3.97
3.98 5.
3.99
This is also called as right handed
decaying exponential signal. It is given by x (t) = e~ a u
(t). Here 'a' is some positive arbitrary constant. Such a
function is shown in Fig. 3.2.9.
3.100
According to definition of laplace
transform,
3.101
We have x (t) = e at u (t). Multiplication
by unit step u (t) indicates that the signal is onl; I range t
= 0 to t = . Thus Equation (3.2.22) becomes,
3.102
3.103
3.104
3.105
3.106
3.107
.-. ROCisa>-a
3.108
3.109
3.110
3.111
3.112 In Equation (3.2.24) if the power of exponent of second term is negative then we will get
tha I means this term becomes zero. Thus we can write the laplace transform.
3.113
3.114
The pole is at s = - a
ROC:
3.115
To obtain this laplace transform, the condition
is Re {(s + a)} < 0. That means Re {s} < - a.
3.116
3.117
3.118
3.2.4
Convolution Integral:
Statement:
3.119
If x(t) i--------->X(s),
LT
ROC:^
3.120
and h (t) <-------> H (s)
3.121 LT
then x; (t) * h (t) <-------> X (s) H (s)
3.122
3.2.5
LT
ROC : R2
ROC : Intersection of R[ and R-
3.124
Satement:
3.126
3.127
If x (t) is the periodic function with fundamental period T0' that means x (t) = x (t
+ T0) then
3.128
3.129
3.130
3.131
3.133
3.134
3.135
3.136
3.137
3.138
Solved Problems:
3.139
iii
in I
3.140
3.141
3.142
that
One of the important properties of Laplace transform is Time cos property. It states
3.143
3.144
EL. 3.2.2 :
3.145
3.146
3.147
ROC of X (s) is the combination of two ROCs. Thus combined ROC is a > 0. It is
.3
3.149
Any time domain function can be expressed in terms of singular functions. There are
additions a subtractions of a singular function to an existing function. There are three possibilities.
(1) Step Step
(2) Step Ramp
(3) Ramp Ramp Case 1: Step
Step :
3.150
The addition or subtraction of a step to a step results in a step function. The
magnitude of n resultant is the algebraic addition or subtraction respectively of the two steps. The
change in ll magnitude occurs at the instant of addition.
3.151
DEMO
3.152
Consider x (t) = u
(t) + 3u (t - 2) = Xj
(t) + x2 (t) say
3.153
3.156
3.157
3.158 DEMO
3.159
Consider x (t)
3.160
= u(t) + r(t-2)
For t
< 0
x(t)
For 0
< t < 2
x(t)
=0
3.161
=l
3.162
3.163
Fig. 3.3.2
3.164
At t = 2, case 2 addition occurs. The resultant is a ramp. It is shifted by the magnitude
of >JM i.e. + 1. This is done at t = 2. This gives the wave as shown in Fig. 3.3.2(c).
3.165
3.166
The addition of a ramp to a ramp gives a ramp. The slope of this ramp is the algebraic
addition of two slopes. The change in slope occurs at the instant of addition.
3.168
A special case is an addition of two equal and opposite slopes. This results in a ramp
r (t) is shown in Fig. 3.3.3(b). At t = 3, there is a case 3 addition. The new ramp has
3.171
3.172
r^nitude [1 + (- 2) = - 1]. So the resultant ramp has a slope of- 1 units. This change in slope
occurs at x(t) = r ( t ) - 2 r ( t - 3 ) If x (t) = r (t) - r (t - 3), then the resultant would be a ramp of
slope zero [1 + (- 1) = 0] and in r: - case the line would be horizontal as shown in Fig. 3.3.3(c).
3.173
In waveforms synthesis, we have to usually decode a waveform into component
parts. The ir - :~s will illustrate this.
3.174
fc^ped Problems:
3.175
ML 13.1 :
3.176
3.177
x(t) = u(t) + u ( t - l ) - 2 u ( t - 2 )
3.178
Step 2 : We have,
3.179 L x (t)
3.180
3.181
.-.
X (s) then
Lx(t-a)
For x(t)=
e_asX(s)
u(t) + u ( t - l ) - 2 u ( t - 2 )
3.182
3.183
3.184
Ex. 3.3.2 :
Soln.:
3.185
.-.
x(t) = u(t) + u ( t - l ) - u ( t - 2 ) - u ( t -
3 ) Step 1:
3.186
3.187
3.188
e
Laplace of u (t - 1) = ~~~
-s
3.189
3.190
e
Laplace of - u (t - 2) = - -
-2s
3.191
-3s
3.192
e
Laplace of-u(t-3) = - -"~
3.193 X (s)vy =\ I ~~+
, I -s eI -2s
i -3s
3.194
3.195 v '
s -~e
s
s-~e
s
3.196
Ex. 3.3.3 :
3.197
3.198
fltopl:
u(t-3)
3.199
3.200
uz 2 :
x(t) = r ( t ) - 2 r ( t - l ) + r(t-2) + u ( t - 2 ) -
3.201
Soln.:
3.204
3.205
is.
Ex. 3.3.6 :
Find LT of
periodic wave of Fig. P. 3.3.6.
3.207
3.208
t lg.
F.
3.3.
6
3.209
Soln.:
3.210
Step 1: Xj (t) for one period
of given waveform is as shown in Fig. P.
3.3.6(a).
3.211
Aj \i) u yij z, u vi
i) -r u ^i t.)
3.212
3.213
3.214
^^3.7 :
3.215
3.217
^M1:
x (t) is given as,
3.218
x(t) = u(t) + u ( t - l ) - 2 u ( t - 2 )
3.219 for one wave only. Call it xt (t)
3.220 The period, T
- 3 She ". :
Laplace of x: (t)
is,
3.223
3.224 Fig. P. 3.3.8
3.225 tea.
3.226
x (t) = A sin t
3.227
= 0 for 7t<t<2rc
3.228
3.229
3.230
3.231
This method is suitable for the laplace transforms which are rational in nature. That
X(s) "
Factorize the denominator and obtain the roots. Then the denominator will be in the form.
3.235
(s-P,)(s-
Aj
A2 A^
S - P[
S - P 2 S - PN
= (s - PK) x (s)
3.242
3.243
s = pK
After calculating Aj, A2.... AN use the standard laplace transform pairs.
3.244 LT
1
3.245-------------------(i)
eatu(t)<> ROC:rj>a
3.246
s a
3.247
LT
1
3.248-----------------------(ii) -eatu(-t)< >
ROC:a<a
3.249 s a
3.250 The nature of time domain signal; that means whether it is right handed (causal) or lefthanded
3.251 (anticausal), is decided using following rules.
3.252(i)
When the poles lie to the L.H.S. of ROC then corresponding signal is right handed
3.253 (causal). That means it is multiplied by u (t). (ii) When the poles lie to the
R.H.S. of ROC then corresponding signal is left handed
3.254 (anticausal). That means it is multiplied by u (- n).
(iii) For bilateral ILT, the poles must lie on both sides of ROC
3.255
Soived Problems:
3.256While solving the numericals we will consider different cases depending on the nature of
poles.
3.257
Case t: V
3.258
3.260
3.261
3.262
3.263
Sfiep 3 : The pole-zero plot and given ROC is shown in Fig. P. 3.4.1. Here both the poles are
to the left side of ROC. Hence both terms corresponds to right handed i rial (multiplied by u
(t)).
3.264
3.265
3.266
3.267
3.268
3.270
3.269
3.271
3.272
3.273
3.274
and
position
of
poles
are
in -z P. 3.4.3(b). Both the poles lie at the R.H.S. of given ROC. nee the corresponding
time domain term are left tanoed. That means multiplied by u (-1). We have standard laplace
transform pair.
3.275
3.276
Now we will consider all possible ROC conditions. The poles are at - 1
ROC : a < - 2
Step 3 : We will calculate inverse laplace for each ROC condition
3.278
3.279
: ~ we
have,
(t)
3.281
3.282
3.283 t
the corresponding time domain sequence is left handed, that means multiplied by u (-1). We have
standard laplace transform pair,
3.284
S'ow consider the second term. It is L \s + 9| ^ere ^e P^e *s at ^2 = ~~ 2. This pole lies at
of given ROC. So the corresponding time domain sequence is right handed; that means
________J
:edbyu(t).
3.285
3.286
Condition (iii):
3.287
3.288
3.289
In this case both the poles lie at R.H.S. of given ROC. Hence corresponding time domain
terms are left handed that means multiplied by u (- t). We have standard laplace transform pair,
3.290Ex. 3.4.5 :
each cam
3.291
Specify all possible ROCS for the function x(s) given below. Also find x(t) in
3.292
3.293
3.295
Pole-zero plot:
3.294
Ail possible conditions of ROC : The poles are at - 2 and - 4. Thus
Here both poles lie at the L.H.S. of given ROC. So the corresponding time
3.296
3.297
3.298
The poles and ROC are
shown in Fig. P. 3.4.5(b)
3.299 if-41
For the first term L i~Tf Here
3.300 U + 2J
3.301 the pole is at P, = - 2. This pole
lies at R.H.S. of given ROC. So the
corresponding time domain sequence is
left handed, that means multiplied by u(t).
3.302
3.303
Here the pole is at P2 = - 4. This pole lies at L.H.S. of given ROC. So corresponding
3.306
In this case both the poles lie at R.H.S.
of given ROC. Hence corresponding time domain terms
are left handed that means multiplied by u (-1).
3.307
3.309
3.310
(a) s > 3
(b) s < -1
3.311
3.312
3.313
3.314
3.315
3.316 4
4
3.317 Step iii : ROC is not mentioned in the numerical. Here the poles are at -r and + T . We will
consider
3.318
3.319
3.320
3.321
: :ion (i)ROCa>g :
The poles and ROC are shown in Fig. P. 3.4.8
3.322
Here both poles lie at the L.H.S. of given ROC. Hence corresponding time domain
signals are
3.323
' 25 1
4
1 36
- 25 -31
u
L
=36"e
s+x
-i (51361
and L I
J
3l /x
^1 =35-6
u(t)
l -3j
25 ~3l
+
36TT
5 e4
3.324
(t)
|_36'e
4
4
Condition (ii)
ROC -3 < o < + :
3.325
3.326
The ROC and pole
plot is shown in Fig. P. 3.4.8(a).
3.327
4
Here the pole at -T is to the
left of
3.328 ROC. Hence
corresponding time domain term
is right handed.
3.329
4
3.330
r-i
f25 / 361
25
~3l ^ ' L \
u(t) S
4~l = 36e
+
3j
3.331
4 And
the pole at + T is to the
right side
3.332 of ROC. Hence
corresponding time domain term
is left handed.
3.333
f5/361
3.334
u( l)
~
_i
_5_ 3l
= "36e
T
' L i_4
3.335
4
4
3.336
25 "3'
5 3l
3.337
x(t) =
^e u(t) + 3ge u(-t)
3.338
3:
3.339
The ROC and plot of
poles is shown in Fig. P.
3.4.8(b).
3.340
Here both poles lie
towards R.H.S. of given ROC.
Hence corresponding time
domain tern
3.341
3.342
3.343
Here the poles are at P, = a + j|3 and P2 = a - jp\ That means the poles are complex conjugate of
h other. But the procedure of obtaining the time domain signal is same as that of case-I.
3.344
?! 3.4.Q ;
3.345
3.347
3.349
3.348
Here the poles are complex
conjugate of each other. tea 2 : In the P.F.E. form
the equation of X (s) can be written a
3.350
3.351
3.352
3.354
3.353
3.355
3.356_________________________________i___________,
________i
3.357
3.358
When there are repeated poles then we have to add some extra terms in the equation of
X (s). For such extra terms the coefficients AK are calculated by taking derivative of X (s). Let as solve
some numericals related to repeated poles.
3.359
Ex. 3.4.10 :
3.360
3.361
3.362
3.363
There is a single pole at + 1; while there are repeated poles at + ^. Step 2 : For the
repeated poles; in partial fraction expansion form equation of X (s) can be written as.
3.364
3.365
Note that the second term in the equation of X (s) is an added term. The coefficients
A, aaz are calculated using earlier method. But extra coefficient A2 is calculated by taking the derivalh
fnllows
3.366
3.367
3.368
3.369___________________________________________________________________________3.
We know that LTI system can be completely characterised by its impulse response h
3.371
3.372
3.373
3.374
3.375
3.376
Q. 1
transform.
3.377
Q. 2
transform.
3.378
Q. 3
3.379
(i)
8(t)
(ii) u(t)
(iii)r(t) Q. 4
3.380 (s
Obtain inverse laplace transform of x (s) = (s + 2) (s +
3.381
3.382 Q. 8
,2.