np
Date: 2070/4/20
Digital control system: The rapid increase in the use of digital controller in the controlled
system is due to its achievement in the optimum performance. Digital control system provides
optimal performance in the form of maximum productivity, maximum profit, minimum cost
or minimum energy use etc. The application of computer control has made possible the
intelligent motion in industrial robots, the optimization of the fuel economy in automobiles
and refinement in the operations of house hold appliances and machines such as microwave
ovens, washing machine, Airconditioning. Decision making capability and flexibility in the
control programs are major advantages of digital control system.
The current trend towards rather then analog control system is mainly due do the
availability of low cost digital computers and the advantages found in working with digital
signals rather then continuous time signals.
Basic Blocks of Digital control system:
S/H

Digital
and
Control
ADC
+
Hold
System
DAC
Circuit
Actuator
Plant or
Process
Digital
Clock
Control
Transducer
Figure 1 shows the basic block diagram or principle of DCS . The controller operation is
performed or controlled by the clock. In such a DCS points of the systems pass signals of
varying amplitude either in continuous time or discrete time or in numerical code.
1. Sample and Hold ( S/H) : It is the circuit that receives an analog input signal and holds
this signals at a constant value for a specified period of time. Usually the signal is
electrical but it may be optical or mechanical.
2. ADC: ADC also called an encoder is a device that converts an analog signal into a digital
signal, usually a numerically coded signal in binary form. Such a converter is need as an
interface between an analog component and the digital component. Basically ADC
involves sampling , quantizing and encoding.
3. Digital Computer: The digital computer processes the sequences of numbers by mean of
an algorithm an produces an new sequences of numbers.
4. DAC: DAC also called an decoder is a device that converts a digital signal ( Numerically
coded data) into an analog signal. It acts as an the interfacing device between the digital
component and an analog component. The real time clock in the computer synchronizes
the events. The output of the hold circuit which is continuous time signal is fed the plant
either directly or through the actuator which controls the dynamics of the system ( i.e it
smoothens the slope of the signal)
5. Plant or process: A plant is a physical object to be controlled. The examples are a
furnace, chemical reactors and a set of machine parts functioning together to perform a
particular operations such as servo system etc.
6. Transducer/sensor: A transducer is a device that converts an input signal into an output
signal of a another form such as device that converts a temperature into a voltage output (
thermistor or thermocouple ), an optical signal into voltage ( phototransistor )
Downloaded from www.bhawesh.com.np /  1
Downloaded from www.bhawesh.com.np
Discrete times control system: Discrete time control system is control system in which one
or more variable can change only at discrete instants of time. These instants which are
denoted by KT or tk ( k =0, 1, 2, ………), specify the times at which some physical
measurements is performed. The time interval between two discrete instants is taken to be
sufficiently short that the data for the time between them can be approximated by simple
interpolation.
Date: 2065/4/21
Data Acquisition, Conversion and Distribution:
Data Acquisition system:
Downloaded from www.bhawesh.com.np /  2
Downloaded from www.bhawesh.com.np
Fig.1 shows the diagram of data acquisition system. The basic parameters are explained
below:
1. Physical variable: The input to the system is a physical variable such as position,
velocity, acceleration, temperature, pressure etc.
2. Transducer amplifier and low pass filter: The physical variables (which are generally
in nonelectrical form) is first converted into an electrical signal (a voltage or a current
signal) by a suitable transducer. Amplifier then amplifies the voltage output of the
transducer (i.e the signal have rises to the necessary level). The LPF follows the amplifier
which attenuates the high frequency signal components such as noise signals which are
random in nature. The o/p of LPF is an analog signal. The signal is then fed to an analog
multiplexer.
Downloaded from www.bhawesh.com.np /  3
Downloaded from www.bhawesh.com.np
3. Analog Multiplexer: It is a device that performs the function of time sharing and ADC
among many analog channels. It is a multiple switch (usually an electronic switch) that
switches sequentially among many analog input channels in some prescribed fashion. The
no of channels may be 4,8,16.
4. Sample and hold circuit: A simpler in a digital system converts an analog signal into an
train of amplitude modulated pulses. The hold circuit holds the value of the sampled pulse
signal over a specified period of time. It is necessary in the AD converter to produce a
number that accurately represents the i/p signal at the sampling instant.
Assignment: 1. Draw a circuit of sample and hold and explain its basic operation.
5. ADC: The output of sample and hold is then fed to the AD converter. The o/p of the
converter is the signal in digital form which is fed to the digital controller.
In this way data acquisition system is held.
Data Distribution:
Downloaded from www.bhawesh.com.np /  4
Downloaded from www.bhawesh.com.np
1. Register: The o/p of digital controller is then stored for a certain period of time in a
memory device called register.
2. Multiplexer: The demultiplexer , which is synchronized with the i/p sampling signal,
separates the composite o/p signal which is in the from of digital data from the digital
controller into the original channels. Each channel is connected to DAC to produce the
o/p analog signal for that channel.
3. DAC: At the o/p of the digital controller, the digital must be converted to an analog
n
signal by the process called D/A conversion. For the full range of digital i/p, there are 2
different analog values , including zero.
4. Hold: The sampling operation produces an amplitude modulated pulse signal. The
function of hold operation is to reconstruct the analog signal that has been transmitted as a
train of pulse samples. The purpose of hold operation is to fill the spaces between the
sampling periods and thus roughly reconstruct the original analog input signal which is
then fed to the actuator which smoothens the slope of signal.
Downloaded from www.bhawesh.com.np /  5
Downloaded from www.bhawesh.com.np
5. Plant or process: A plant is a physical object to be controlled. The examples are a
furnace, chemical reactors and a set of machine parts functioning together to perform a
particular operations such as servo system etc.
Data conversion Process:
SIGNAL SAMPLING ,QUANTIZING AND ENCODING
Signal sampling is the first step of transmission of analog signal over digital signal.
1. Sampling: Sampling is the process of conversion of continuous time analog signal into
discrete time analog signal. The discrete signal obtained after sampling is called sampled
signal.
Sampling Theorem: It states “Analog signal can be reproduce from an appropriate set of its
samples taken at some fixed intervals of time.” This theorem has made possible to transmit
only samples of analog signal by changing or encoding this samples into block of code words
suitable for digital control systems.
If
fs = sampling frequency
fx = maximum frequency component of the i/p signal, then the
distortion less recovery of the signal fs ≥ 2fx
If the signal x(t) to be sampled is band limited , then the sampled signal can be represented
as:
xs(t) = x(t) × g(t)
Where, g(t) is the sampling function (rectangular pulse train) which be represented as shown
below.
g(s)
Ts
 τ/2
τ
τ/2 Fig.1
Rectangular pulse train
Where ,
Ts = Sapling period.
τ = duration of sampling pulse= pulse
width Sampler can be implemented as:
xs(t)= x(t)×g(t)
x(t)
g(t)
Fig.2 Implementation of sampler
Proof of sampling theorem: The gate function g(t) can be expressed interms of fourier
series as
g(t) = co + ∑
∞
=
2cn cos(n
s
t)
n1
Where, co = τ/ Ts = τfs
Downloaded from www.bhawesh.com.np /  6
Downloaded from www.bhawesh.com.np
Cn = fs τ sinc[nfs τ
] = co sinc[nfsτ]
ωs =2 πfs
The signal xs(t) can be expressed
as xs = x(t)×g(t)
= x(t)×[ co + ∑ n 1 2cn cos(n s t) ]
= cox(t)+2c1x(t)cos ωnt+2c2x(t)cos2
ωst+……….+2cnx(t)cosn ωst+…………
The fourier transform of above series as
xs(f) =co(f)+2c1x(ffs )+2c2x(f2fs)+…………+2cn x(f+nfs)+…….
The above series can be graphically represented as:
∞ =
x(f)
fx
fx
Fig.3. Message spectrum
Downloaded from www.bhawesh.com.np /  7
Downloaded from www.bhawesh.com.np
It is clear from fig.4 that the spectrum of the sampled signal contains the spectrum of the
sampled signal contains the spectrum of the original message signal.
Date:2065/4/26
It is evident that for distortion less recovery of original message signal, from the spectrum of
the sampled signal, the following condition should be met.
fs ≥ fx
In this case the original message signal spectra can be recovered by passing the sampled
signal through low pass filter with bandwidth equaling to ± fx
Distortion will occur while recovering the message spectrum if. fs ≤ fx
The distortion in the above case is caused by the overlapping of side bands and message
spectra.
Downloaded from www.bhawesh.com.np /  8
Downloaded from www.bhawesh.com.np
Xs(f)
o
Xs(f)
aliasign ditortion
(fs<2fx)
o
The minimum sampling rate: fs min = 2fs is called Nyquists’s sampling rate for distortion less
recovery of one message spectrum.
The minimum interval of the sampling for a real signal is
Ts min = 1/2fx(min)
Where
fx(min) = maximum frequency in the message spectrum.
Quantizing and Quantization error:
Downloaded from www.bhawesh.com.np /  9
Downloaded from www.bhawesh.com.np
x(t)
2
1
(a)
xs(n) 1
2
3
4
5 6
1.8
0.9
7
t
1.8
1.7
0.9
(b)
1.7
0.5
q
1 2 3
4
5 6
7
4
5 6 7
n
x (n)
2
1
(c)
1 2
3
n
Fig. 1(a) continuous
 continuous
value
signal 1(b) Discrete time
 continuous value signal
1(c) Discrete time
 discrete value signal.
Quantization is the process of representing the analog sample values by a finite set of levels.
The sampling process coverts a continuous time signal to a discrete time signal with
amplitude and that can take any values from zero to maximum level and the quantization
process converts continuous amplitude samples to a finite set of discrete amplitude values.
The output sates of each quantized sample is then described by a numerical code. The
process of representing a sampled value by a numerical code is called encoding . Thus
encoding is a process of assigning a digital word to each discrete states.
In uniform quantization it is assumed that the range of input sample is –xmax to xmax and
n
the number of quantization level, known as Qlevel, N=2 . Where,
n = is the number of bits per source sample , then the step size on length of the Q
level is assumed to be,
n
n1
= 2xmax / N = 2xmax/2 = 2xmax/ 2
n1
= xmax/2
The step size ‘
is also called ‘quantum’.
Downloaded from www.bhawesh.com.np /  10
Downloaded from www.bhawesh.com.np
Quantization error:
Since the number of bits in the digital word is finite , ie the digital output can assume only a
finite number of levels and therefore analog number must be round off to the nearest digital
level. Hence any ADC involves quantization error. It is evident that the maximum Q error
could be only /2. In uniform quantization the steps size is constant for the entire dynamic
range of the input discrete signal level. Q error depends on the fineness of the quantization
level and can be made as small as desire by making the quantization level smaller or by
increasing the number of bits ‘n’. In practice there is a maximum for n and so there is always
some error due to quantization.
The uncertainty present in the quantization process results quantization noise.
Signal to quantization noise ratio (SQNR): It is evident that the Qerror ( i.e qe) lies
between –
by,
/2 to
/2 in random manner , the average power of Qnoise is therefore given
∆
2
q 2 dq
e
∆ ∫∆ e
−
= Pq
=
2
∆2
…………………..(i)
P =1
12 q
It is seen form equation (1) that Qnoise is dependent on stepsize ‘ ’ only. Reducing the
stepsize or increasing the no of representation level, we can reduce pq and hence Qerror.
From equ (1)
Pq =
2
/12 But,
= xmax /2
n1
n1
2
n
Pq = (xmax/2 )/12 = xmax /3×4 ……….(ii)
2
Assume that the average signal power is x , SQNR for uniform quantization will be,
SQNR = average power of signal/ average power of noise.
2
n
QSNR = xˆ × 3 × 4 ……………………(iii)
Where,
x2
xˆ 2 = x2 = Normalized signal power.
Again , equation (iii) can be reproduce in dB, as
3
n
(QSNR)dB = Px (dB) +10 log10 +10 log10 (4)
= Px (dB) + 4.8 +10n
log10 Where,
2
Px (dB) =10(log10 (xˆ )dB
Thus,
(QSNR)dB = Px+4.8+6n………………….(iv)
Also,
N
SQNR(dB) = Px +4.8+20 log10 …………….(v ) [since, N=
n
2 ] And for N>> 1
Downloaded from www.bhawesh.com.np /  11
Downloaded from www.bhawesh.com.np
SQNR = 20 log10N …………….(vi)
Which is approximated SQNR for uniform quantization interms of level of quantization “N”.
Assignmet:1
1. Simple and hold ckt.
Analog multiplexer and Demultiplexer.
Date:2065/5/1
Review of Z transform: mathematical tool commonly used for the analysis and synthesis of
discrete time control system is the ztransform. In considering the ztransform of a time
function x(t) , we consider only the sampled values of x(t) , i.e x(0), x(T) , x(2T),………,
where ‘T’ is sampling period.
The ztransform of a time function x(t) , where ‘t’ is non –negative or of a sequence of
values x(kT) , where k = 0,1,2,3 …
∞
X(z) = z[x(t)] = ∑x(kT )z
−k
…………….(i)
k =0
Or , X(z) = z[x(t)] =
∑
∞
x ( k ) z −1
…………….(ii), for T=1
k=0
The ztransform defined by equation (i) and (ii) is onesided ztransform .
The both sided ( or double sided) ztransform is defined by:
∞
X(z) = z[x(t)] = z[x(k)] =
∑x(kT )z −1
k =−∞
ztransform of Elementary function:
(i)
Step unit function: The unit step function defined by
1
for t ≥ 0
x(t) =
otherwise
0
∞
Thus , X(z) = z[x(t)] = z[1(t)]= ∑1.z
k =0
1
−k
2
3
= 1 + z +z +z +……….
=
1
1 + z −1
1
Therefore, z[1(t)] = 1 + z −1
Or, z[1(t)] =
z
, z >1
z −1
2. Unit Ramp Function: The unit ramp function is defined by
Downloaded from www.bhawesh.com.np /  12
Downloaded from www.bhawesh.com.np
for t ≥ 0
t
x(t) =
0
otherwise
Or, x(KT) = KT for k = 0,1,2,3………….
Thus,
∞
X(z) = z[x(t)] = z[t] = ∑t.z
−k
k =0
∞
Or, X(z) = ∑(KT )z
k =0
−k
1
2
3
= k(z +2z +3z ……..)
X(z) =
z −1
−1 2
= T. (1 − z )
−1
T.z
−1
(1 − z
2
)
3. Polynomial Function: The polynomial function is defined by
k
k = 0,1,2,3.......
x(t) =
a
K<0
0
Where a is constant.
∞
Then, X(z) = z[x(k)] = z[aK] = ∑x(k)z
−k
k =0
∞
X(k) = ∑a k z
k =0
−k
1
2 2
3 3
= 1 + az +a z + a z +……………
k
1
Z[a ] =
−1
1 − az
1
k
Or, z[a ] = 1 − az
−1
4. Exponential function:
at
for t ≥ 0
x(t) =
e
otherwise
akT
Where, x(kT) = e
, K = 0,1,2………..
We have,
0
∞
X(z) = z[e ] = ∑e
at
−akT
z
k =0
aT 1
Or, X(z) = 1 + e z +e
1
−k
2aT 2
z +e
3aT 3
z ………
z
−aT
X(z) = 1 − e−aT z −1 = z − e
5. Sinusoidal Function:
Downloaded from www.bhawesh.com.np /  13
Downloaded from www.bhawesh.com.np
sin
for
t
t≥0
x(t) =
0
Noting , e
jwt
e
t<0
= coswt +jsinwt
jwt
= coswtjsinwt
we have ,
Sinwt = 1/2j[e
& coswt = 1/2[e
jwt jwt
e
jwt
+e
jwt
]
]
We know that,
at
Z[e ] =
1
1 − e−at z −1
1
Z[e ] =
1 − e jwT z −1
jwt
1
Z[e ] =
jwt
1 − e− jwT z −1
jwt jwt
Therefore, z[sinwt] = z[ 1/2j(e e )]
jwt
jwt
= 1/2j[z(e )z(e )]
1
1
= 1/2j
jwt −1 −
− jwt −1
1−e z
1− e z
−1
z sin wt
Z[sinwt] = 1 − 2z
−1
cos wt + z
−2
Similarly,
z
Z[coswt] = 1 − 2z
−1
−1
cos wt
cos wt + z
−2
Important properties of ztransform:
1. Multiplication by a constant:
If x(t)
Z
X(z), then,
Z[ax(t)] = az[x(t)] = ax(z)
Where ‘a’ is a constant.
2. Linearity of Ztransform:
If z[f(t)] = z[f(kT)] = z[f(k)] = F(z)
Z[g(t)] = z[g(kT)] = z[g(k)] = G(z)
And a and b be scalers then then x(k) formed by liner combination,
x(k) = a f(k) + b g(k)
Has the ztransfom
X(z) = a F(z) +G(z)
Downloaded from www.bhawesh.com.np /  14
Downloaded from www.bhawesh.com.np
k
3. Multiplication by a :
If z[x(k)] = X(z) , then
k
1
Z[a x(k)] = X(a z)
Proof:
k
Z[ a x(k)] =
∞
∞
∑a
k
x(k)z
−k
= ∑x(k)(a
k =0
−1
z)
−k
k =0
1
= X(a z)
4. Shifting theorem:
If x(t) = 0 for t< 0 and
x(t)
Z
X(z)
n
Then, z[x(tnT)] = z X(z) ……………(i)
And , z[x(t+nT)] = z
n
n−1
X (z)
− ∑x(KT )z
−k
…………..(ii)
k =0
Where, n is zero or +ve integer.
Proof:
For equation (i)
We know that , z[x(tnT)] = z[x(KTnT)] =
z[x(kn)T]
∞
Or, z[x(tnT)] = ∑x(k − n).Tz
−k
k =0
∞
∑x(k − n).T.z −(k −n) .z −n
=
k =0
∞
= z
−n
.∑x(k − n).T.z
−(k −n)
k =0
Let us define kn = m , therefore,
∞
Z[x(mT)] = z
−n
. ∑x(mT ).z
−m
m=−n
Since ‘m’ must be zero and nonnegative integer
n
Therefore, z[x(mT)] = z X(z)
For equation (ii)
Z[x(t+nT) = z[ x(KT+nT)]
= z[x(K+n)T]
∞
Or, z[x(t+nT)] = ∑x(k + n).Tz
−k
k =0
∞
= ∑x(k + n).T.z
−(k +n)
.z
n
k =0
Downloaded from www.bhawesh.com.np /  15
Downloaded from www.bhawesh.com.np
∞
= z n .∑x(k + n).T.z
−(k +n)
k =0
Let us define, K+n = m
∞
Z[x(mT)] = z
−n
.∑x(mT ).z
−m
m=n
n−1
= z
∑x(mT ).z
n
− ∑x(mT ).z
m=0
−m +
m=0
∑∞ x(mT ).z −m
= zn
∞
n−1
−m
−
m=0
∑x(mT ).z −m
.
m =n
n−1
∑x(mT ).z −m
.
m=0
4. Complex translation Theorem:
If x(t)
Z
X(z), then,
at
aT
e x(t)
Z
X(ze )
Proof:
∞
∑e−akT x(kT ).z −k
at
Z[e x(t)] =
k =0
∞
=
∑x(kT ).(eaT z)−k
k =0
aT
= X(e z)
aT
= X (ze )
5. Initial value Theorem:
lim
If x(t) has the ztransform X(z) and if z →∞ X (z) exist then, the initial value x(0)
of x(t) is given by ,
lim
x(0) = z →∞ X (z)
6. Final Value Theorem:
lim
k →∞
−1
lim
[
x(k) = z →1 (1
]
− z ) X (z)
Assignment # 02:
1. Prove all the important properties of ztransform.
1
Example: 01: Obtain the ztransform of x(s) = s(s +1)
Solution:
1
Given, x(s) = s(s +1)
= A+ B
s
s +1
Downloaded from www.bhawesh.com.np /  16
Downloaded from www.bhawesh.com.np
1
.
Where, A = s(s +1) s s = 0
A=1
1
And, B = s(s +1) .(s +1) s = −1
B = 1
1
1
Therefore, x(s) = −
s s +1
t
Taking inverse Laplace Transform, we have, x(t) = 1 – e
Again taking ztransform on both sides,
t
Z[x(t)] = z [ 1 – e ]
t
= z[1] – z[e ]
1
1
=
−1 −
1−z
1 − e−T z −1
1/− e
=
−T
(1 − z
−1
=
z
−1
−1/+ z
−1
−1
)(1 − e−T z
) (1 − e−T )z −1
−1
−T
−1
(1 − z )(1 − e z )
Example: 02 Obtain the ztransform of
k −1
k = 0,1,2,3.......
a
f (a) =
K≤0
0
∞
Z[f(a)] =
∑ f (a).z −k
k =0
1
−1 = X (k)
1 − az
k1
1
Then, z[x(k1)] = z[a ] = z .X(z)
1
Here, z[x(k)] = z[ak] =
k1
Therefore, z[a
k1
Z[a
1
−1
] = z 1 − az
z −1
−1
] = 1 − az
Example:03 obtained the ztrasform
[i] e
at
sinωt
at
[ii] e
cosωt
Date:2065/5/3
at
[i] e sinωt
We know that
Z
−1
−1
sin T
X (z)
−1
cos wT + z =
Z[sinwt] = 1 − 2z
We know that,
at
akT
Z[e x(kT)] = X[ze ]
Downloaded from www.bhawesh.com.np /  17
Downloaded from www.bhawesh.com.np
e
at
−aT
Therefore , z[e sinwT] = 1 − 2e
−aT
z
−1
z
−1
sin wt
cos wt + e
−2aT
z
−2
Example 04: Consider the function y(k) , which is sum of functions x(h) where, h = 0, 1, 2
…….k , Such that
k
y(k) = ∑x(h), k = 0,1,2.....
h=0
Where y(k) = 0 for k < 0 . Obtained the ztransform of y(k).
Solution: Given,
k
y(k) = ∑x(h)
h=0
or, y(k) = x(0) + x(1) + x(2) + x(3)+………….+ x(k)(i)
similarly,
y(k1) = x(0)+ x(2)+ x(3)+ ……….+x(k1) (ii)
Now from (i) and (ii)
gives. y(k) – y(k1) = x(k)
Now taking ztransform on both sides,
Z[y(k) – y(k1)] = z[x(k)]
1
Y(z) – z Y(z) = X(z)
1
Y(z) [1z ] = X(z)
1
Therefore, Y(z) = X(z)/ ( 1z ) Ans.
at
Example 05: Obtained the ztransform of t.e .
Given,
at
x(t) = t e
at
x(kT) = kT e
1
1 2
We know z –transform of t is given by, z[t] = Tz /(1z )
Using complex translational theorem,
at
at 1
at 1 2
Z[e .t] = Te .z /(1e z )
Example 06: Determine the initial value x(0) if,
T
1
T 1
X(z) = (1e )/(1z )(1e z )
Solution:
T
1
T 1
Given, X(z) = (1e )/(1z )(1e z )
We know initial value theorem is given by ,
lim
→∞X (z)
X(0) = z
−T −1
(1 − e )z
lim
= z→∞
= 0 Ans.
(1 − z
−1
−T −1
)(1 − e z
)
1
Example 07: Determine the final value x(∞) of X(z) =
1
−
1
− z −1 1 − e−aT z −1
Downloaded from www.bhawesh.com.np /  18
a>0
Downloaded from www.bhawesh.com.np
Solution:
From final value theorem,
lim
x(∞) = z →1[(1
lim
=
−1
− z )X (z)]
1
−1
z →1 (1 − z
)
−
−aT
1−e
aT
1− e z
−1
−1
z
z −1
lim
= z →1 1
−
1−z
−1
1−z
lim
= z →1 1
1
−1
−
z−e
−aT
= 10
= 1 Ans.
The inverse ztransform:
(1) Direct division method:
Example 01: Find x(k) for k = 0, 1,3,4. When X(z) is given by ,
X(z) = (10z+5)/(z1)(z0.2)
Solution:
X(z) = (10z+5)/(z1)(z0.2)
2
= (10z+5)/(z  1.2z+0.2)
1
2
1
2
= (10z + 5z )/(11.2z +0.2z )
10z
17z
17z
−1
+ 5z
−1
−12z
−2
− 2z
−2
− 20.4z
−3
+ 3.4z
−4
+18.4z
10z
−2
−3
+ 3.4z
−4
−2
+ 2z
−3
−3
18.4z
−3
− 22.08z
18.68z
18.68z
18.736z
−5
−4
+ 3.68z
−5
−4
− 3.68z
−5
−4
− 22.416z
− 3.7362z
−5
+ 3.7362z
−6
−6
Quotient of the x(z) gives the inverse ztransform is given by
X(0) = 0
X(1) = 10
X(2) = 17
X(3) = 18. 4
X(4) = 18.68
Example 02: Find x(k) when X(z) is given by X(z) =
1
1
1/(z+1) Example 03: Obtained x(k) for X(z) = z / (1 az )
Downloaded from www.bhawesh.com.np /  19
Downloaded from www.bhawesh.com.np
1
1
X(z) = z /(1az )
1
1
=z
−1
1 − az
1
= z [X(z)]
1
Where, Y(z) = 1/(1az )
1
The inverse ztransform of Y(z) is z [Y(z)] =
k
a So, the inverse ztransform of (z) = z
1
1
k1
[Y(z)] z [X(z)] = a = y(k1)
k1
Therefore, x(k) = y(k1) = a
2. partial fraction method: We use the following formula for the function written in the
form
X (z) = a1 = a2 + ............ an
z−p
z−p
z−p
Z
1
ai = (zpi)
(i)
2
n
X (z)
Z z = pi
Again if,
C1
C2
X (z)
Then,
=
+
Z
(z − p ) (z − p )
2
1
2
C1 = (zp1) X (z)
Z z = p1
And , C
d
2 X (z)
(z
−
z Z =P
dz
In general , if there are n multiplication roots then,
1 d n−1
X (z)
n
Cn =
n−1 (z − p1 )
(n −1)! dz
z Z =P
2=
p1 )
1
n
−aT
1 − e .z
Example 01: Given X(z)
−aT
(z −1)(z − e ) Where ‘a’ is a constant and T is the sampling
=
period. Determine the inverse ztransform x(KT) by the use of partial fraction expansion
method.
Solution:
−aT
(1 − e )Z
−aT
X(z) = (z −1)(z − e )
X (z)
1 − e−aT
=
−aT
(z −1)(z − e )
Z
X (z)
A
B
=
+
Z
z −1 z − e−aT
−aT
1−e
X (z)
A = (z1)
= (z −1)
−aT
Z z =1
(z −1)(z − e ) z =1
=
1−e
−aT
=1
1 − e−aT
Downloaded from www.bhawesh.com.np /  20
Downloaded from www.bhawesh.com.np
Threfore, A = 1.
−aT
X (z)
Similarly B =
=
−aT
−aT
z=e
Z
1−e
= (z − e
1−e
−aT
−aT
) (z −1)(z − e ) z = e
−aT
−1)
(e−aT
B = 1
X (z) = 1 − 1
Z
z −1
z − e−aT
z
z
X (z) =
−
z −1
z − e−aT
1
1
−1
X (z) = 1 − z −
−aT −1
1 1 − e z
Therefore, z [X(z)] = 1eaT= 1keakT
k
Therefore, x(t) = 1 – e
akT
ans.
2
z +z+2
2
X (z) = (z −1)(z − z +1) by partial fractional
Example 02: Obtained the ztransform of
expression method.
Solution:
z2 + z + 2
= A
+ Bz + c
2
(z −1)(z − z +1) (z −1) z 2 − z +1
2
2
z +z+2
A(z − z +1) + (Bz + C)(Z −1)
=
2
(z −1)(z 2 − z +1)
(z −1)(z − z +1)
2
2
Z + z +2 = (A+B)z  (A+BC)z +(AC)
2
0
Comparing the coefficient of z , z and z we get,
B+A = 1
(A+BC) = 1
AC = 2
Solving the above equation we get,
A = 4 , B = 3, C = 2
A
Bz + c
Thus , X (z) =
+ 2
z − z +1
(z −1)
4
− 3z +1
X (z) =
+ 2
(z − 1) z − z +1
−1
−2
4z −1 + − 3z + 2z
X (z) =
−1
(1 − z ) 1 − z −1 + z −2
−1
1
1 − 0.5z
−1
−1
X (z) = 4z
−1 − 3z
−1
−2
− z1 − z
+z
1
−1
−1
1
1 − 0.5z
1/ 6.z
−1
−1
−1
X (z) =
X (z) = 4z
−1
1 −z
− 3z
−1
−2
+ 3z
−1
−2
1 −z +z
1 −z +z
Downloaded from www.bhawesh.com.np /  21
Downloaded from www.bhawesh.com.np
−1
X (z) = 4z
1
1
− z −1
−1
−1
− 3z
1
1 − 0.5z
− z −1 + z −2
3 −1
1 −1
2z
+
z
3 1 − z −1 + z −2
We know that
z
1
1
−1
=1
k
1 −z
−1
1 − 0.5z
1
= cos k
−2
+z
3
3
−1
z .
1
2
z 1 − z −1 + z −2 = sin k 3
z
−1
1 −z
1
Therefore, z = [X(z)] = x(k)
k1
= 4.1 – 3 cos(k1) π/3 + 1/√3 . sin(k1) π/3
Date: 2065/5/10
3. Inverse Integral method:
1
The inversion integral for the ztransform X(z) is given by z [x(z)] = x[kT] =
1 ∫x(z)z k −1dz ………(i)
2 j
1
Where ‘c’ is a circle with its centre at the origin of the zplane such that all poles of x(z)z
are inside it.
The equation for giving the inverse ztransform in terms of residues can
be derived by using the theory of complex variables. It can be obtained as follows.
k1
X(kT) = x1 + x2 + ……….+xn denotes the residues of x(z)z at poles z1, z2 ……..zm
respectively.
k1
In evaluating residues, if the denominator of x(z)z contains a simple
pole at z = zi , then the corresponding residue x is given by,
k −1
lim
…………….(ii)
ki = z0 → z1 [( z − zi )X (z)z
k1
If X(z) z contains a multiple pole zi or order q, then the residue k is given by ,
−1
1
dq
lim
q
k −1
x(k) =
]
]
−1
ki = (q −1)! .z →2 j
dzq [( z − zi ) X (z)z
………..(iii)
01 Obtained x(kT) by using the
inversion integral method when,
Example:
−aT
z(1 − e )
X (z) =
−aT
)
(z −1)(z − e
Solution:
−aT
X (z) =
z(1 − e
)
(z −1)(z − e
−aT
)
Downloaded from www.bhawesh.com.np /  22
Downloaded from www.bhawesh.com.np
X (z)z
k −1
z
=
kT
−aT
z(1−e
)
−aT
(z −1)(z −e )
k
−aT
z (1 − e )
X (z)z k −1 =
−aT
(z −1)(z − e )
k1
aT
For k = 0,1,2……….x(z)z has simple poles at z1 = 1and z2= e
2
Hence, x(kT) = x(k) = ∑residue of x(z)z
k1
at pole z = zi
i=1
= k1 + k2
lim
[( z − zi )X (z)z
Where k1 = z → z 1
= z →lim z1
k −1
k
( z −1)
z (1 − e
]
−aT
)
−aT
−aT
k
=
1 (1 − e )
−aT
(z − e )
(z −1)(z − e
)
K1 = 1
For K2 ,
K2 =
lim
[( z − z2 )X (z)z
z → z2
e
−akT
=
k
] −aT
)z (1 − e
)
−aT
(1 − e
e
(z − e
−aT
lim
z →e
=
−aT
k −1
−aT
(z −1)(z − e
−aT
)
)
−1
akT
Therefore , k2 =  e
X(kT) = x(k) = x1 + x2 = 1e
akT
which is required inverse ztrasnform.
Example 02: Obtained the inverse ztransform of
z2
−aT
2
x(z) = (z −1) (z − e ) by using the inverse integral method.
Solution:
z2
−aT
2
X(z) = (z −1) (z − e
)
z k +1
k1
−aT
2
X(z)z = (z −1) (z − e )
k1
aT
X(z) z has simple pole at z = z1 = e and double pole at z = z2 =1 ( i.e q =
2) The inverse ztransform is therefore given by, x(kT) = k1 +k2
Where, k1 = z →lim e
−aT
[(z − e−aT )X (z)z k −1 ]
−aT
lim
= z →e
(z − e−aT )z k +1
2
(z −1) (z − e
−aT
)
−a(k +1)T
=
e
−aT
(e
2
−1)
Downloaded from www.bhawesh.com.np /  23
Downloaded from www.bhawesh.com.np
1
And for K2
d
lim
=
, k2
(2 −1)!
d
z→1
=
.
)
−aT
dz (z − e
)
−aT
k
k +1
(z − e )(k +1)z − z
lim
(z − e
k
−aT
k +1
(k +1)k (1 − e ) −1
=
−aT
(z −1) (z − e
z k +1
z→1
=
z k +1
2
dz
lim
=
2
. (z −1)
.z→1
−aT
(z − e )2
(k −
−aT
ke
− e−aT
−aT
=
2
)
−aT
(k +1)(1 − e ) −1
(1 − e
−aT 2
)
(1 − e−aT )2
−a(k +1)T
Therefore, x(kT) k1 + k2 = e−aT
+
(e
−1)2
−aT
e−a(k +1)T + k − ke
− e−aT
X(kT) =
−aT 2
(1 − e )
k − ke−aT − e−aT
−aT
(1 − e )2
z −2
−1 3
Example 03: Obtain the inverse ztransform of x(z) = (1 − z )
k −1
1
lim
3 z.z
d2
Where k1 =
2!
.z→1.
2
( z −1)
dz
(z − 1)
2
1 d
k
= 2! . 2 [z ]
dz
z =1
3
1 d k −1
] z =1
= 2 . dz [kz
1
k −2
= 2 . k ( k −1)z
[
=
] z =1
k(k −1)
2
1
Therefore, z (x(z)) = x(kT) = k(k1)/2
Solution of difference equation by ztransform method:
Example 01: Solve the following difference equation by use of ztransform method of
x(K+2)+ 3x(k+1)+2x(k) = 0
Taking ztransform on both the sides , we have,
Z[x(k+2)]+3z[x(k+1)]+3z[x(k+1)] +2z[x(k)] = 0
2
2
Z x(z)z x(0)zx(1) + 3[zx(z)zx(0)] +2x(z) = 0
2
2
X(z) [z +3z+2] – (z +2)x(0) – zx(1) – zx(1) = 0
But, x(0) = 0 and x(1) = 1
2
Therefore, x(z) [z +3z+2] – z = 0
z
z
A
B
X(z) = 2
=
=
+
(z + 3z + 2) (z +1)(z + 2) z + 1 z + 2
Downloaded from www.bhawesh.com.np /  24
Downloaded from www.bhawesh.com.np
x(z)
1
A
B
=
=
+
z
(z +1)(z + 2) z +1 z + 2
A = (z +1)
x(z)
z z = −1
A = 1.
Similarly, B = 1,
x( z)
z
1
+1 − z + 2 x(z)
z
z
=z+ 1−z+ 2
=z
1
x(z) =
1 − 1
1 + z −1 1 + 2z −1
−
x(z) =
1
1
1 − (− 1)z −1 1 − (− 2)z −1
The inverse ztransform of x(z) is
k
k
X(z) = (1) – (2) , k = 0,1,2………..
Example 02: Obtained the solution of the following difference equation in terms of x(0) and
x(1).
X(K+2) + (a+b)x(k+1) + ab x(k) = 0
2
2
Z x(z) – z x(0) – zx(1) + (a+b) [z x(z) – z x(0)] + ab x(z) =
2
2
0 X(z) [z + (a+b)z+ab] = [z + (a+b)z] x(0) + z x(1)
[z 2 + (a + b)z]x(0) + zx(1)
X(z) =
z 2 + (a + b)z + ab
B
X(z) = [z + a + b]x(0) + x(1) = A +
z+a z+b
(z + a)(z + b)
ax(0) + x(1)
bx(0) + x(1)
Where, A =
and
B=
(a − b)
(b − a)
bx(0) + x(1)
x(z) =
z
x(z) =
ax(0) + x(1)
(b − a) +
(a − b)
(z + a)
(z + b)
bx(0) + x(1) ax(0) + x(1)
(b − a)
+
−1
1 − (−a)z
bx(0) + x(1)
X (z) =
(b − a)
(a − b)
−1
1 − (−b)z
ax(0) + x(1) k
k
(−b)
(−a) +
(a − b)
[ for the case a ≠
b]
Now for a = b,
2
(z + 2az)x(0) + zx(1)
X(z) =
2
2
z + 2az + a
x(z) =(z + 2a)x(0) + x(1)
z
2
(z + a)
Downloaded from www.bhawesh.com.np /  25
Downloaded from www.bhawesh.com.np
x(z) = A + B
2
z
(z + a) (z + a)
zx(0) + ax(0) + x(1)
−1
−1 2
(1 + az ) (1 + az )
We get , x(z) = x(z) =
k
Therefore, x(k) = x(0)(a) + [ ax(0)+x(1)][(a)
Assignment:
2
1. Obtained the ztransform of (a) k
(b)
k
1
z ∑x(h) =
−1 x(z)
2. Show that
1 −z
n=0
k −1
−1
z
z ∑x(h) =
−1 x(z)
1 −z
n=0
∞
] for case a = b
k ak1
lim
∑x(k) = z
And ,
k1
→1X (z)
k =0
Also show that,
k
∑
x(h)
1
=
−h
i−1
−1
X (z) −∑x(h)z
(1− z
)
Where 1 ≤ i ≤ (k1)
h=i
h=0
3.Obtained the ztransform of
Figure:
4. Obtained the ztransform of
3
2z + 2
(a) x(z) =
(b)
x(z) =
2
(z + 2)
(z − 2).z
2
(z − 2) (z −1)
10
z(z + 2)
(c ) x(z) =
(d) x(z) =
2
(z −1)
(z − 2)(z −1)
−2
(e) x(z) = 1 + 6z + z −3
−1
−1
(1 − z )(1 − 0.2z )
(f) x(z) = 1 + z
−1
−2
+z
1 − z −1
5. Solve the following difference equations:
a. 2x(k) – 2x(k1) +x(k2) = u(k)
Where x(k) = 0 for k< 0
k = 0,1,2.....
1,
And u(k) =
k<0
0,
b. x(k+2) –x(k+1) + 0.25x(k) = u(k+2)
where, x(0) = 1, x(1) = 2
u(k) = 1 for k ≥ 0.
c. x(k2) – x(k1) + 0.25 x(k) = u(k2)
x(0) =1 x(k) = 0, for k<0.
u(k) = 1 for k ≥ 0
(6) Consider the difference equation .
X(k+2) = x(k+1) +x(k)
Downloaded from www.bhawesh.com.np /  26
Downloaded from www.bhawesh.com.np
Where x(0) = 0 and x(1) = 1.
Note that x(2) = 1 , x(3) = 2, x(4) = 3.
The series 0,1,2,3,5,8,13,………is known as Fibonacci series. Obtained in general solution
x(k) in closed form. Show that the liming value of x(k+1)/ x(k) as k →∞ is (1+√5)/2 or
approx 1.6180.
Date: 2065/5/15
Example 05: (c ) x(k2) –x(k1) + 0.25x(k) = 4 (k2)
X(0 ) = 1
X(k) = 0 for, K < 0 and u(k) = 1 for k ≥ 0
Solution:
X(k2)  x(l1) + 0.25x(k) = u(k2)
Z[x(k2)] – z[x(k1)] + 0.25 z [ x(k0] = z [ u(k1)]
2
1
2
Or, z x(z) – z x(z) + 0.25 x(z) = z
2 1
2
u(z) Or, x(z) [z –z + 0.25] = z . 1/(11
2
1 2 1
z ) X(z) = z /(1z )(z z +0.25)
2 2
Or x(z) = z.z /z ( z1)(12
z+0.25) = 4z/(z1)(z2)
2
2
X(z)/z = 4/(z1) (z2) = A/(z1) + B/(z2) + C/(z2)
By partial fraction method we get,
A = 4, B = 4, C =4.
2
Therefore, X(z)/z = 4 [ 1/(z1) – 1/(z2) + 1 (z2) ]
1
1
1
1 2
X(z) = 4 [1/(1z ) – 1/12z + z /(12z )
2
Or, X(z) = 4 [ z/z1 – z/(z2) + z/(z2) ]
Taking inverse z transform on both sides
k
k
k1
x(k) = 4[1 – 2 + 2 .k] , k = 0, 1, 2…….
k+2
k+1
=4–2
+ k.2
Example: 06
Given, x(k+2) = x(k+1) + x(k) , x(0) = 0 and x(1) = 1
To get x(k)
And to prove,
Lim k ∞ x(k+1) /x(k) = (1+ √5)/2 = 1.6180
We have ,
X(k+2) = x(k+1) + x(k)
Z[x(k+2)= z[z(k+1) +x(k) ]
2
2
Z x(z) – z x(0) – 2x(1) = z x(z) – 2x(0) + x(z)
2
2
Or, z x(z) – 2x(z) – x(z) = z x(0) + 2x(1) – 2x(0)
z
Or, x(z) = z 2 − z −1
x(z) =
z
2
z
z − z −1
2
Now, the roots of z z 1 are
2
− b ± b − 4ac
=
2a
Downloaded from www.bhawesh.com.np /  27
Downloaded from www.bhawesh.com.np
−1 ± 1 − 4.1(−1)
=
2.1
= 1±
52
= 1+ 5
2
= 1− 5
2
x(z) =
z
=
2
z
z − z −1
1
1+5
z −
=
z
−
2
+
A
1+5
1−5
z−
2
B
1−5
−
z
2
2
We get,
A = 1/√5 , B = 1/√5
1
x(z) =
z
5
z
x(z) =
−
1
1 + 5
2
1
z
1
5
−
1
−
1+
1 −
1
1− 5
2
−
5
1− 5
z
2
−1
1 −
z
2
−1
Taking inverse ztransform we get,
1 1+ 5 k
1 − 5k
x(k) =
…………….(i)
−
2
5
2
Similarly we can write,
1 1 + 5 k +1
x(k +1) =
From (i) and (ii)
1+ 5
2
1−
k +1
−
2
1+ 5
k +1
……….(ii)
2
5
x(k +1) =
x(k)
1− 5
−
k
5
k +1
2
1−
5
k
−
2
2
Downloaded from www.bhawesh.com.np /  28
Downloaded from www.bhawesh.com.np
Here,
1− 5
< 1 when k
2
∞
Lim k
1− 5
2
1−5
The terms
k
→0
k
1−5
k +1
can be neglected when
2
2
∞ , Thus the equation (iii) become,
k
and
1
lim k → ∞
x(k +1)
=
x(k)
+ 5 k +1
2
k
1+ 5
= 1.6180 proved
2
=
1 + 5
2
Impulse Sampling:
Let us consider an ideal sampler commonly called an impulse sampler. We assume
= 0 for t< 0. The sampler o/p is equal to the product of the continuous time i/p x(t)
train of unit impulse δT(t) which means the sampler may be considered as a modular
x(t )
and
with
x(t) as the modulating signal and δT(t) as the carrier as shown in fig1 below.
Modulator
x(t)
x*(t)=x(t)×δΤ(t)
δΤ(t)
Figure:1
*
*
Let us consider x (t) to represent the impulse sampled o/p. The sampled signal x (t), a train
*
of impulses, can
thus be represented by the
infinite summation, i.e x (t) =
∞
∑x(kT )
(t − kT ) ……..(i)
k =0
*
x (t) = x(0) (t) + x(T ) (t −T ) + x(2T ) (t − 2T ) +........  (ii)
Here, we are defining ,
∞
T (t) = ∑ (t − kT )
………….(iii)
k =0
Now taking the laplace transform of equation (iii) we get
*
*
[x (t)] = x (s) = x(0)
*
X (s) = x(0).1+x(T).e
∞
X (s) = ∑x(kT )e
*
−kTS
[δ(t)]+x(T) [δ(tT)]+ x(2T) [δ(t2T)]+…..
Ts
2Ts
+ x(2T).e
+…….
…………….(iv)
k =0
We define e
sT
= z ……….(v)
Downloaded from www.bhawesh.com.np /  29
Downloaded from www.bhawesh.com.np
Then equation (iv ) can be rewritten as
∞
X (s) = ∑x(kT )z
*
−k
= X(z)
k =0
i.e X * (s) s = 1 .ln z = X (z)
T
……………(vi)
Data hold:
zero order hold
x(t)
x(kT)
(ZOH)
h1(t)
Fig 2(a): Sampler and a zero order hold.
x(t)
X*(s)
δΤX*(s)
Gho(s)
h2(t)
H2(s)
Fig 2(b): Mathematical Model that consist of a sampler and a
Transfer function Gho(s) for ZOH.
Data hold is a process of generating a continuous time signal h(t) from a discrete time
sequence x(k). A hold ckt covert the sampled signal into a continuous time signal. A hold ckt
hold the amplitude of the sampled from one sampling instant to the next. Such a data hold is
called zero order hold or clamper or stair case generator. The o/p of the zero order hold is a
staircase function.
Assuming x(t) = 0 for t<0,
The ZOH circuit smoothens the sampled signal to produce the signal h(t) , which is constant
for the last sampled value until the next sample is available, i.e ,
h1(t+kT) = x(kT) for 0 ≤ t ≤ T …………..(i)
The o/p h1(t) may be assumed as
h1(t) = x(0) [ 1(t) 1(t1)]+x(T) [ 1(t1)1(t2T)] + x(2t)[1(T2t) 1(t3T)]+ ………….
∞
h1(t) = ∑x(kT )[1(t − kT ) −1(t − (k +1)T )] …………..(ii)
k =0
We know that
kTs
[1(tkT)] = e
/s
Therefore , the laplcae transform of equation (ii) be written as:
Downloaded from www.bhawesh.com.np /  30
Downloaded from www.bhawesh.com.np
∞
[h1(t)] = H1(s) =
∞
= ∑x(kT )
k =0
∞
= ∑x(kT )
H1(s) =
1−e
e
∑x(kT )
−kTS
−
s
e
−kTS
−TS
[1(tkT) 1(t(k+1)T)]
k =0
−(k +1)TS
−e
e
S
−(k +1)TS
∞
∑x(kT ).e−KTS ………..(iii)
s
k =0
But from fig (ii) b,
[h2(t)] = H2(s) = H1(s)
Thus,
k =0
−TS
∞
1−e
H2(s) =
∑x(kT ).e−KTS ……….(iv)
s
k= 0
From fig (ii)b, we can write , H2(s)
= Gho(s) × X*(s) ………(v)
But we know
∞
X*(s) = ∑x(kT ).e
−KTS
k =0
Thus from equation (iv) become,
−TS
1−e
H2(s) =
X*(s)
s
−TS
1−e
Gho(s) =
………..(vi)
s
ztransform by convolution integral method:
The convolution integral is defined by
1 c+ j∞
1
X * (s) =
X ( p)
dp ……….(i)
−T (s− p)
2 j c− j∞∫
1−e
Where, the integration is along the
line from cj∞ to c+j∞ and this
line is parallel to
1
imaginary axis in the pplane and separate the poles of X(P) from those of [1 − e−T (s− p) ]
Equation (i) also be written as
1
X ( p)
X * (s) = 2 j ∫ 1 − e −T (s− p) dp
The above integral is equal to the sum of residues of X(p) in the closed container i.e
X ( p)
X * (s) = ∑ residue of
at poles of X ( p)
−T (s− p)
1−e
But we know that
Downloaded from www.bhawesh.com.np /  31
Downloaded from www.bhawesh.com.np
Ts
e = z X*(s)
= X(z)
X ( p)z
X (z) = ∑ residue of
poles of X ( p)
Tp at
z−e
By changing the complex variable rotation from p to s obtained
X (s)z
X (z) = ∑ residue of
at poles of X (s)
z − es
Assume that X(s) has poles S1 , s0 ……sm. If a pole at s = sj is a simple pole , then residue.
X (s)z
lim
k j = s →s j (s − s j ).
z − e TS
If a pole at s = si is a multiple pole of
order ni , then the residue ki =
d
1
ni−1
X (s)z
lim
n
s →si
(s − s j ) .
ds
(ni − 1)!
Ts
z−e
ni−1
1
Example:01: Obtained the ztransform of X(s) = s (s +1) by the use of convolution integral
method.
Solution:
1
2
s (s
X(s) =
+1)
X (s)z
z
TS = 2
TS
z−e
s (s +1)(z − e )
Thus by convolution integral method,
z
X (z) = ∑ residue of 2
X (s)
TS at poles of
s (s +1)(z − e )
= k1 + k2
z
lim
Where,
k1 = s →−1 .(s
+1). 2
(s +1)(z − TS
s
e
)
z
−T
= 1 − (z − e )
z
2
K1 = (z −e
1
K2 =
−T
s →0
d
lim
(2 −1)!
=
)
s →0
lim
ds
d
.
s
z
2
.
2
s (s +1)(z − e
z
2
−Ts
)
Ts
ds s (s +1)(z − e )
2
− z + z + zT
2
(z −1)
Thus X(z) = k1 +k2
2
− z + z + zT
z
=
+
−T
(z −e )
(z −1) 2
=
Downloaded from www.bhawesh.com.np /  32
Downloaded from www.bhawesh.com.np
1
=
+ −1 + z
−1
+z
−1
T
−1 2
− z −1e−T
(1 − z )
G(s)
−1
X (s) = 1 − e−TS .G(s) . Prove that
Example:02:
Given
X (z) = (1 − z
)z
OR
s
s
consider the zero order Hold circuit succeeded by the plant
with transfer function G(s).
Explain how would you determine the ztransform of such combined system.
Soluotion:
1
Suppose the transfer function G(s) follows the zoH. Then, the product of the transfer function
of ZOH and G(s) becomes:
1−e
X (s) =
−TS
s
X(s)
=
.G(s)
(1 − e −TS
G(s)
)
s
TS
X(s) = (1 − e ).G (s)
1
TS
X(s) = G (s) − e − .G (s)
1
1
X(s) = G1 (s) − X 1 (s)
TS
Where X1(s) = e .G1(s)
1
1 TS
[X1(s)] =
[e G1(s)]
1 TS
1
=
[e ]*
[G1(s)]
1
[X1(s)] = g0(t) * g1(t)
x1(t) =
∫
t
g0 (t − ) * g1 ( )d
0
But,
1 TS
[e
] = δ (tT)
Thus , x1(t) =
∫t
(t −T − ).g1 ( )d
0
= g1(tT)
Hence by writing z[g1(t)] = G1(z)
Thus,
Z[x1(t) ] = z[g1(tT
T)] X1(z) = z .G1(z)
Thus , X(s) = G1 (s) –X1(s)
Or, X(z) = z[g1(t)]z[X1(t)]
—T
= G1(z) –z .G1(z)
T
= G1(z)[1z ]
T
= [1z ].z[G(s)/s]
[G1(s) = z[(G(s))/s]
T
X(z) = (1z ) z.(G(s)/s)
For T = 1,
1
X(z) = (1z ).z [G(s)/s]
proved
Downloaded from www.bhawesh.com.np /  33
Downloaded from www.bhawesh.com.np
Criteria For stability in Zdomain:
Example: 01: consider the closed loop control system shown in fig (i) below. Determine the
stability of system when k = 1.
Fig.
−s
G(s) = k.
1−e
1
.
s
s(s +1)
−s
1−e
. [since k = 1]
2
s (s +1)
G(z) = Z[G(s)]s
−
G(s) =
1−e
=z
2
s (s +1)
0.3679z + 0.2642
(z − 0.3679)(z −1)
Now the closed loop transfer function of G(z) is
C(z)/R(z) = G(z)/[1+G(z)]
Therefore , the characteristics equation is given by ,
1+G(z) = 0
=
1+ 0.3679z + 0.264z
(z − 0.3679)(z −1)
Or (z0.3679)(z1) +
2
0.3679z+0.2642 Z z+0.6321 = 0
From which we get,
2
Z –z+0.6321 = 0
Form which we get,
Z1 = 0.5+j0.6181 Z2
= 0.5 – j0.6181
2
2
z = z1 = z2 = √ (0.5 + (0.6181) ) =
0.7950 Since the z < 1 , the system is stable.
The jury stability test:
Assume,
n
n1
n2
P(z) = anz + an1z + an2z + ……. a1z+a0
……..(i) Where,
An > 0 , then for stable system the following condition should be specificed.
1. a0> an
2. P(z) z =1 > 0
3. p(z) z =1 > 0 if n = even
< 0 if n = odd
4. bn1 >bo
cn2 > co
.
.
q2 > qo
Downloaded from www.bhawesh.com.np /  34
Downloaded from www.bhawesh.com.np
Where,
a
bk = n
a
a
a
0
ck =
k +1
b
a
n −1
k = 0,1,2…….n2
n
b−2−k
b0
k +1
.
.
qk =
k = 0,1,2…….n1
n−1 −k
P
p3
2−k
k = 0,1,2
P
P
0
k +1
Date: 2065/5/22
The jury stability Test:
n
n1
n2
P(z) = a0z + a1z + a2z + ……. An1z+an
……..(i) Where,
a0 > 0 then, for a stable system the following conditions should be fulfilled.
1. an > a0
2. p(z)z =1 > 0
3. P(z)z= 1 > 0 n = even
< 0 n = odd
4. bn1 > b0
Cn2 > c0
q2 > qo
4
3
Example:01: construct the Jury stability table for the following equation. P(z) = a0z + a1z +
3
a2z +a3z+a4 ,Where a0 > 0. Write the stability conditions since n = 4 the stability conditions
are:
(i)
(ii)
(iii)
(iv)
a4 > a0
p(z)z =1 = a0+ a1+ a2 +a3+a4
P(z)z= 1 = a0 a1+ a2 a3+a4 > 0 [since n = 4 = even ]
b3 > b0
C3 > c0
The Jury stability table can be constructed for n = 4 as follows:
0
Row z
Z1
Z2
Z3
Z4
a4
a0
a3
a4
a1
a4
a0
a3
a2
a4
a0
a2
a3
a4
a0
a1
b3
b0
b0
b1
b1
b3
b0
b2
b2
b3
b0
=
b3
=
b2
=
b1
=
b0
=
c2
=
c1
=
c0
b1
Downloaded from www.bhawesh.com.np /  35
Downloaded from www.bhawesh.com.np
4
3
Example:02: Examine the stability of following characteristics equation. P(z) = z 1.2z +
3
0.07z +0.3z  0.08 = 0
Use Jury stability test.
Solution:
4
3
3
P(z) = z  1.2z + 0.07z +0.3z  0.08 =
0 Here,
a0 = 1 , a1 = 1.2 , a2 = 0.07 a3 = 0.3 , a4 = 0.08 The
conditions for stability in Jury tests are as follows.
1. a4 > a0
0.08 < 1 (which is true)
4
3
2
2. p(z)z =1 = 1  1.2×1 + 0.07×1 +0.3×1 0.08
= 0.09 > 0 (which is true)
3. Since n = 4 = even ,
p(z)z =1 = 1+ 1.2+ 0.07 0.3 0.08 = 1.89> 0 (true)
4. b3 > b0
c3 > c0
Row
z0
Z1
0.08
1
0.08
1
0.08
1
0.08
0.3
1
1.2
0.9936
0.204
0.9936
0.204
0.9936 1.176
0.204
0.0756
Z2
Z3
1.2
0.3
0.07
0.07
Z4
1 = 0.993
0.08
= 1.176
= 0.0756
= 0.204
0.204
0.0756
0.0756
1.176
= 0.9456
= 1.1838
= 0.315
b3 > b0
i.e 0.9936 > 0.0204
0.09936 > 0.204
(true)
Again c2 > c0
0.9456 > 0.3150 (ture)
3
2
Example: 03: A control system has the following characteristics equation P(z) = z – 1.3z
– 0.08z + 0.24 = 0
Determine the stability of the system by jury stability test.
3
Example:04: Examine the stability of the characteristics equation given by : P(z) = z – 1.1z
– 0.1z + 0.2 = 0 . Use jury method.
Downloaded from www.bhawesh.com.np /  36
2
Downloaded from www.bhawesh.com.np
Example: 05: Consider the discrete time unity feed back control system whose open loop
transfer function is given
K (0.3679z + 0.2642)
G(z) =
Determine the range of k for stability by the use of Jury
(z − 0.3679)(z −1)
method.
Solution:
K (0.3679z + 0.2642)
G(z) =
(z − 0.3679)(z −1)
The close loop transfer function is
T(z) = G(z)/ 1+G(z) of whose characteristics equation is given by
1+G(z) = 0
Or, 1+ K (0.3679z + 0.2642) = 0
(z − 0.3679)(z −1)
2
P(z) = z + ( 0.3679k – 1.3679)z + 0.3679+0.2642k =
0 Here ,
a0 = 1
a1 = (0.3679k – 1.379)
a2 = 0.3679 + 0.2642k
For the system to be stable the conditions are :
(i)
a2 < a0
i.e  0.3679+0.2642K< 1
0.2642K < 0.632
K < 2.3925
(ii) p(z)z =1 = 1+ 0.3679k 1.3679 + 0.3679+
0.262K K > 0
(iii) p(z)z =1 = (1 0.3679k+ 1.3679 + 0.3679+ 0.262K ) > 0
= 2.7358 – 0.1037k > 0
2.7358 > 0.1037k
K < (2.7358)/(0.1037)
K < 26.3818
Combining (i) (ii) and (iii)
conditions K < 2.3925
K>0
0 < K < 2.3925
Stability Analysis by use of the Bilinear transformation and Routh stability:
The bilinear Transformation is defined by:
z = w +1
w −1
w=
z + 1
z −1
We choose ‘w’ so
that W = +jω
3
2
Example:01: Consider the following characteristics equations P(z) = z – 1.3z – 0.08z
+0.24 = 0
Determine the stability of the system using Bilinear transformation and Routh stability
criterion
Downloaded from www.bhawesh.com.np /  37
Downloaded from www.bhawesh.com.np
Solution:
Here,
3
2
P(z) = z – 1.3z – 0.08z +0.24 =
0 Using Bilinear Transformation , i.e
z=
w + 1
w −1 , we get P(z) as,
3
2
P(z)z=(w+1)/(w1) = P(w) = [(w+1)/(w1)] – 1.3 [(w+1)/(w1)] – 0.08[(w+1)/(w1)] + 0.24
=0
3
2
Or w – 7.571w – 36.43w – 14.14 = 0
The Routh stability criteria is checked by routh arrow as follows:
3
W
2
W
1
W
W
1
7.75
38.30
0
36.43
14.14
0
14.14
The above equation and array show that there is a sign change in the following system is not
stable. It indicates that there is a pole in the right half of w plane which in turn indicates that
there is a pole outside the unit circle in zplane.
Example: 02: Consider y(k) = 0.6y(k1) – 0.81y(k2)+0.6y(k3) – 0.12y(k4) = x(k) where,
x(k) is the input and y(k) is the output of the system. Check the stability of the system using
Jury stability method.
Solution:
1
2
3
4
Y(z) 0.6z y(z) – 0.81z y(z) + 0.6z y(z) 0.12z y(z) = X(z)
Y (z)
1
=
−1
−2
−3
−4
X (z) 1 − 0.6z − 0.81z + 0.6z − 0.12z
4
z
4
3
2
z − 0.6z − 0.81z + 0.6z − 0.12
For which the characteristics equation is
4
3
2
Z – 0.6z – 0.81z + 0.6z – 0.12 = 0
Date: 2065/6/5
Splane to zplane mapping:
Mapping of the left half of splane into zplane:
We know z and s are related by the equation,
TS
Z = e ………….(i)
This means that a pole in splane can be located in the zplane through transformation
TS
Z=e ,
Also we know that ,
S = +jω ………….(ii)
Thus equation (i) become ,
T( +jω)
Z=e
¡
T
jωT
=e .e
Downloaded from www.bhawesh.com.np /  38
Downloaded from www.bhawesh.com.np
Or in general,
+ 2 π k)
T
¡ ¥
Z=e
……….(iii)
We see that the poles and zeroes in splane , where frequencies differ in integral multiple of
the sampling frequency 2 π /T are mapped into the some locations in the zplane . This means
that there are infinite may values of s for each value of z.
)
¡¨¢¦
is –ve in the left half of the splane , the left arm of the splane corresponds to z = e
=e
T
¡
Since
<1
The jw axis in the splane corresponding to z = 1 (
=0) i.e the imaginary axis in the splane (the line =0) corresponding to the unit circle in the zplane and interior of the unit
circle corresponds to the left half of the splane.
jw
im
Complementary
strip
3
Primary
strip
Re
Re
Complementary
strip
zplane
splane
(i)
Mapping of constant –attenuation loci:
A constant attenuation line ( a line plotted as
into a circle of radius z = e
T
= constant) is the splane maps
centered at the origin in the zplane as shown below.
jw
im
eTσ2
1
σ1
0
σ2
σ
Re
eTσ1
splane
zplane
Downloaded from www.bhawesh.com.np /  39
Downloaded from www.bhawesh.com.np
jw
im
1
Re
σ1
splane
zplane
(ii)
Mapping of constant frequency Loci: A constant frequency loci w = w1 , in the splane is mapped into a radial line of constant angle Tw1 ( in radian) in the zplane as
shown in the fig. below.
jw
im
w1
w2
w2T
w1T
σ
Re
w1T
w1
s plane
z plane
(iii) Mapping of constant damping ratio. ( ) A constant damping ratio line (a radial line) in
the splane is mapped into a spiral in the zplane as shown in figure below.
ξ = ξ1
Re
s plane
z plane
2
In splane a constant damping ratio is related to it by the equation . s =  ξ wn + jwn √ (1+ξ )
=  ξ wn +
jwd Where,
Wn = natural frequency (rad/sec)
Wd = damped frequency (rad/sec)
2
= wn√ (1ξ )
Thus ,
Z = eTS = eT(ξwn+jwd) = e(ξ Twn+jTwd)
Downloaded from www.bhawesh.com.np /  40
Downloaded from www.bhawesh.com.np
Z = exp.
2
2
1−
wd
wd
+ j2
w
ws
s
Hence ,
−
z = exp
w
2
d
2
1−
w
s
wd
z=2
ws
Example:01: For a given region in splane draw the corresponding region in z – plane.
jw
jw1
e−σ1Τ
(2π −w2T)
w1T
e−σ2Τ
1
jw2
σ=−σ2
σ=−σ1
2065/6/5
Reconstruction of original signal from samples:
Instantaneous sampling (i.e when the duration of the sampling pulse τs –0 or delta function )is
referred to as ideal sampling . In this case the sampling function is the train of impulses,
i.e
∞
Sδ (t) = ∑ k=  ∞ δ (t kTs )
Therefore , the sampled signal for ideal sampling can be written as
Xδ(t) = x(t) x sδ (t)
∞
= x(t) x ∑ k= ∞ δ (tkTs)
The fourier transform of the above expression is the convolution
Of the fourier transformation of x(t) and sδ (t) , i.e
xδ’(f) = x(f) *sδ (f)
where,
∞
Xδ’(f) = f[sδ (t)] = fs ∑ n= ∞ δ (fnfs)
∞
xδ(f) = x(f) *fs ∑ n= ∞ δ (fnfs)
∞
xδ(f) = fs ∑ n= ∞ δ (fnfs)
Xδ (f) = fsX(f) assuming the condition fs ≥ 2fx is satisfied , let us
Pass the sampled signal through ideal LPF with the following parameters
HLPF (f) = K for fx ≤ f ≤ fs –fx
= 0 , otherwise
At the output of LPF , we ge ,
∞
Xδ(f) HLPF (f) = [ fs ∑ n=  ∞ x(f nfs)]
HLPF(f) Xδ (f) HLPF(f) = X(f) …………….(i)
Downloaded from www.bhawesh.com.np /  41
Downloaded from www.bhawesh.com.np
As HLPF(f) = 0 for the frequency components higher then fx , the spectrum at the output of
LPF for
K = 1/fs will be x(f) , the spectrum of the message signal .
Which is shown in equation (i)
In time domain
1
1
X(t) = f [x(f)] = f [xδ(f) HLPF(f)]
x(t) = xδ (t) * hLPF(t)
Where hLPF(t) is the impulse response of ideal LPF which is expressed as.
hLPF (t) = 2BTs sinc [
2Bt] Where,
Ts = 1/fs
B = fx , the higher frequency component.
Thus signal at the output of low pass filter for
∞
Xδ(t) = x(kTs)
∑ x(kTs) δ(t kTs)
at t = kTs
n=−∞
Will be,
∞
x(t) = x(kTs)
∑ x(kTs) δ(t kTs)*2BTs sinc [2Bt]
n=−∞
∞
x(t) = 2BTs
∑ x(kTs) sinc [2B(tkTs)] ……….(ii)
n=−∞
The above equation (ii) shows that the original message signal
Can be reconstructed from its sampled values x(kTs) if the sampling is done at Ts ≤ 1/2fx
Chapter: 3
Analysis of control system:
The pulse Transfer Function:
X(z)
G(z)
Y(z)
The transfer function for the continuous time system relates the Laplace Transform of the
continuous time output to that of continuous time input while the pulse transfer function
relates the ztransform of the output at the sampling instants to that of the sampled input.
The pulse transfer function is defined by
G(z) = Y(z)/X(z) …………..(i)
Of the discrete time system As seen from the above equation we write,
Y(z) = G(z) x(z) ……..(ii)
Equation (ii) may also be written as,
Y*(s) = G*(s) x*(s) ……………..(iii)
General procedure for obtaining Pulse transfer function:
x*(t)
x(t)
y(t)
G(s)
δΤ X*(z)
δΤ
y*(t)
Y(z)
Downloaded from www.bhawesh.com.np /  42
Downloaded from www.bhawesh.com.np
Figure (ii) (a)
x(t)
X(s)
y(t)
Y(s)
G(s)
Figure (ii) (b)
The pulse transfer function for the figure (ii) (a) will ve ,
G(z) = Y(z)/X(z)
Also , the transfer function for the system shown in fig (ii) (b) will be , G(s) = Y(s)/X(s)
The important fact to be noted is that the pulse transfer function for the system
in fig (ii) (b) is not ztransfer of z[G(s)], because of the absence of the instant sampler.
The presence or absence of the input sampler is crucial in determining the pulse transfer
function of the system. This difference is explained below.
Consider the case as shown in fig (ii) (a)
[y(t)] = Y(s) = G(s) X*(s)
Taking starred Laplace transform on both sides, we
get, Y*(s) = G*(s) X*(s)
Or, Y(z) = G(z) X(z) ……………(i)
Now let us consider the case for fig (ii) (b) ,
[y(t)] = Y(s) = G(s) X(s)
Taking Starred LT on both sides,
Y*(s) = [G(s)X(s)]*
Y*(s) = [GX(s)]*
Where, GX(s) = G(s)X(s)
Or Y(z) = z[GX(s)] …………………(ii)
Thus equation (i) and (ii) clearly describe the presence and absence of instant sampler.
Date:2065/6/12
Pulse Transfer function of cascaded elements:
x*(t)
G(s)
x(t) δΤ
u(t)
u*(t)
H(s)
y(t)
y*(t)
δΤ
δΤ
Fig. 3(a)
x(t)
x*(t)
δΤ
G(s)
H(s)
y(t)
y*(t)
Fig. 3(b)
Consider the system shown in fig (iii) (a) we can write that
G(s) = G(s).X*(s) ……………..(i)
Y(s) = H(s).U(*(s) ……….(ii)
Now taking starred laplace transform of equation (i) and
(ii) U*(s) = G*(s).X*(s) …………(iii)
Y*(s) = H*(s) U*(s) …………….(iv)
From equation (ii)
Y*(s) = H*(s)U*(s)
= H*(s).G*(s).X*(s)
Or, Y*(s)/X*(s) = G*(s). H*(s)
Taking ztransform of the above equation.
Y*(z)/X*(z) = G(z).H(z) …………….(v)
Which is the pulse transfer function between i/p x*(t) and o/p y*(t)
Downloaded from www.bhawesh.com.np /  43
Downloaded from www.bhawesh.com.np
Now consider the cascaded system shown in fig (iii)
(b) Y(s) = G(s)H(s).X*(s)
Or Y(s) = GH(s).X*(s)
Now taking the starred laplace transform of the above equation , we get ,
Y*(s) = [GH(s).X*(s)]*
Y*(s) = GH*(s).X*(s) Or
Y*(s)/X*(s) = GH*(s)
Now taking ztransform on both sides we get,
Y(z)/X(z) = GH(z) …………………….(vi)
Which is the required PTF for fig (iii) (b)
From equation (v) and (vi) , it is to be noted that,
G(z)H(z) ≠ GH(z)
Example:01: Consider the system shown in figs (iii) a & b obtained the pulse transfer
function (PTF) Y(z)/X(z) for each of these tow systems.
G(s) = 1/(s+a)
H(s) = 1/(s+b)
Y(z)/X(z) = G(z)H(z)
= z[G(s)] z[H(s) ]
= Z[1/(s+a)] z[1/(s+b)]
aT 1
bT 1
Y(z)/x(z) = 1/(1e z ). 1/(1e z ) for fig (iii)
(a) For fig (iii) (b)
Y(z)/X(z) = GH(z)
= z[GH(s)]
=z[G(s)H(s)]
=z[1/(s+a).1/(s+b)]
A
B
+
=z
s+
b
s +a
1
−1
=z b − a + b − a
s+a
s +b
1
1
−
.
b−a s+a
1
1
1
=z
Y(z)/X(z) =
.
−
−
AT
−
s+b
1
−
1
BT
for fig (iii) (b)
−
1
b−a 1 −e z
1−e z
Thus we conclude that G(z).H(z) is not equal to function GH(z).
PTF of a closed loop system:
R(s)
E(s)
+
E*(s)
G(s)
C(s)
δΤ
H(s)
Fig(iv) closed loop system.
From fig. (iv)
E(s) =R(s) – C(s)H(s) ………….(i)
Where,
Downloaded from www.bhawesh.com.np /  44
Downloaded from www.bhawesh.com.np
C(s) = G(s) E*(s) ……….(ii)
Thus equation (i) become
E(s) = R(s) –G(s)H(s)E*(s) …………….(iii)
Taking starred Laplace transform
E*(s) = R*(s) – [ G(s)H(s) E*(s)]*
E*(s) = R*(s) – GH*(s)E*(s) ……………..(iv)
Or, E*(s)[1+GH*(s)] = R*(s) ………..(v)
E*(s) = R*(s)/(1+GH*(s)) ………….(vi)
Now taking Starred laplace transform of equation (ii) ,
C*(s) = G*(s) E*(s)
= G*(s).R*(s)/(1+GH*(s))
C*(s)/R*(s) = G*(s)/1+GH*(s) …………..(vii)
Now taking ztransform of equation of (vii) we get,
C(z)/R(z) = G(z) / (1+GH(z)) …………..(viii)
which is the required PTF for closed loop
system.
Assignment # 3:
(1) Prove the given PTF for the closed loop system shown in Table 3.2 (page 112) form
k. Ogata book .
Closed Loop PTF of a digital control system:
R(s)
E*(s)
E(s)
+
δΤ
G*(s) m*(s)
D
1eTS
s
U(s)
G (s)
C(s)
p
Fig. (v)
From fig (v) , let ,
Ts
1e /s. Gp(s) = G(s)
……..(i) Also,
C(s) = G(s) GD*(s) E*(s)
Or C(s) = G*(s) GD*(s) E*(s) ………(ii)
In term of ztransform, equation (iii) can be written as:
C(z) = R(z) – C(z) ………(iv)
Thus , equation (iii) becomes
C(z) = G(z) GD(z)[R(z)C(z)]
C(z) = [ 1+G(z)GD(s)] = G(z) GD(z) R(z)
Or C(z)/R(z) = G(z)GD(z) / 1+G(Z)GD(z)
C(z)/R(z) = GD(z)G(z)/1+GD(z)G(z) ………(v)
Which the required closed loop PTF of a Digital control system.
PTF of a digital PID controller:
The PID control action in analog controller is
Downloaded from www.bhawesh.com.np /  45
Downloaded from www.bhawesh.com.np
m(t) = kp e(t) +
1
T
de(t)
t
∫e(t)dt
+ Td
………….(i)
dt
i 0
where e(t) is the i/p to the controller , m(t) is the o/p of the controller, k is the proportionality
gain , Ti is the integral time and Td is the derivative time.
To obtain the PTF for the digital PID controller , we first discritise equation (i) .
Then , by approximating the integral term by the Trapezoidal summation rule and the
derivative term by twopoint difference formula. Thus equation (i) becomes at t = kT ,
1 e(0) + e(T ) e(T ) + e(2T )
e(k −1) + e(kT )
+
e(KT )
m (kT) = kp
+
T
T
+ ......
2
e(kT ) − e(k −1)T
+ Td .
+ ........
2
i
2
T
k
=kp e(KT ) + T . e((h −1)T ) + e(hT ) +
T
i
Let,
k
∑
d
[e ( kT ) − e(k −1)T ] ….(ii)
∑
Ti
2
h=1
e((h −1)T ) + e(hT ) = f (hT ) where, f(o) = 0
2
k
=∑ f (hT ) ………(iii)
e(h −1)T + e(hT )
2
h=1
h=1
The ztransform of equation (iii) will be,
k
=
z.
e((h −1)T ) + e(hT )
∑
2
h =1
=
k
z.
∑
f (hT )
h=1
1
= 1/(1z ). [F(z)f(0)]
1
= 1/(1z ). F(z) …….(v)
Also ,
F(z) =z[f(hT)]= z
e((h −1)T ) + e(hT )
2
1
F(z) = (1z )/2.E(z)
Thus from equation (iv) becomes,
k
−1
e((h −1)T ) + e(hT )
1+z
z.
∑
=
−1 E(z) ……….(V)
2
=
2(1 − z )
Hence equation (ii) becomes by taking ztransform on both sides,
−1
1−z
Td
T
−1
M(z) = z[m(kT)] = Kp E(z) +
.
E(z) +
(1− z
)E(z)
T 2(1 − z −1 )
T
h1
i
1
1
1
M(z) = kp [1+T/2Ti +(1+z )/(1z )+Td/T .(1z )].E(z)
1 + Td .(1 − z −1 ) E(z)
M(z) = kp 1 − T + T .
−1
2T T 1 − z
T
i
i
Downloaded from www.bhawesh.com.np /  46
2
Downloaded from www.bhawesh.com.np
Ki
M(z) = kp'+
+ k D .(1 − z −1 ) E(z)
−1
……….(vi)
1−z
Where, kp’ =
T
k
p
2T
1−
=
proportional gain
i
T
Ki = k
p T
= Integral gain
i
Kd = k p
T
d = Derivative gain
T
Form equation (vi) we can write that
Ki
GD(z) = M(z)/E(z) =
kp'+
−1
+kD
.(1 − z −1 ) ………(vii)
1− z
Where, GD(z) is the closed loop PIF for digital PID controller equation (vii) is also know as
the proportional form of PID control scheme.
Downloaded from www.bhawesh.com.np /  47
Downloaded from www.bhawesh.com.np
Realization of Digital controllers
We know,
G(z) = Y (z) = bo + b1 z −1 + B2 z − 2.......bm z − m _____________(i)
X (z) 1 + a z −1 + a z − 2 +.......a z −n
1
2
n
Where, ais and bis are real coefficient and n # m. The above transfer function can
be represented by various structures (below) using the following methods:
(i) Direct programming
(ii) Standard programming
(iii) Ladder programming
(iv) Series programming
(v) Parallel programming
(i)Direct programming:Equation (i) can be represented as
1
2
Y(z) = boX(z) +b1X(z)z +b2X(z)z +……….
2
bmXzm a1Y(z)z1a2Y(z)z n
………..anX(z)z
(ii) Standard programming:Y(z) = Y(z) = H(z)
X(z) H(z) X(z)
1
= bo + b1z
1
2
+b2z
2
m
+…………..+bmz
n
1+ a1z +a2z + …………..+anz
Y(z)
H(z)
And
1
2
m
= bo +b1z +b2z +……………+bmz
_________(ii)
Downloaded from www.bhawesh.com.np /  48
Downloaded from www.bhawesh.com.np
H (z) =
−1
X (z) Ha z + a
1
1
z −2
2
_______________(iii)
+.........a
z −n
n
From equation (ii)
1
2
m
Y(z) = bo H(z) +b1H(z)z +b2H(z)z +…….bmH(z)z
_____________(iv) From equations (iii),
1
2
n
H(z) = X(z) – a1H(z)z –a2H(z)z …………………..anH(z)z ____________(v)
Fig for equation (iv)
Thus, the final realization will be.
Downloaded from www.bhawesh.com.np /  49
Downloaded from www.bhawesh.com.np
(iii) Ladder programming: In this case the PTF G(z) is represented in continued
fraction as follows:
1
G(z)=Ao =
1
B 1z +
1
A1 +
1B2Z +
1
.
.
.
.
1
A+
DZ+ 1
n−1
n
An
The programming method based on this scheme is known as ladder programming.
Let, us defines:1
,i =1,2.............n −1
Biz + Gi (A)(A)(z)
( A)
1
Gi (Z ) =
,i =1,2.............n −1
Gi
(b)
(Z ) =
Aiz + Gi+1 ( B) z
1
( B)
Gn (Z ) =
1 ,i =1,2.............n −1
BnZ +
A
n
For simplicity, let us assume,
n=2
then,
1
G(z) = Ao =
1
B1 z +
1
A1 +
1B2 Z +
1
.
.
.
.
A +
1
DZ+ 1
n−1
n
By the use of the function, Gi
(B)
(A)
(Z) G
An
(B)
(Z), and G2 (Z); the G(z) may be rewritten as:
Downloaded from www.bhawesh.com.np /  50
Downloaded from www.bhawesh.com.np
G(z) =Ao +
1
1
B1z+ A +G(B)(Z)
1
2
1
= Ao + B z + G ( A) (z)
1
1
(B)
G(z) = Ao +G1 (z)
Also , it is to be noted that ,
Gi
( B)
(z) =
Yi (z)
=
X i(z)
Or, Xi(z) = Bizyi(z)
1
Bi z + Gi
( A)
(z)
(A)
+Gi zYi(z)
(A)
Or, Xi(z) = Gi (z) Xi(z) = BizYi(z)
1
(A)
Or, Yi(z) =1/Bi[Xi(z)Gi (z)Y(x)]Z
Now for Gi
Gi
( A)
(A)
(z)
(z) =
Yi (Z )
1
=A
+G
X i (z)
i
i+1
( B)
(z)
(A)
+Gi zYi(z)
Or, Xi(z) = BizYi(z)
(A)
Or, Xi(z) = Gi (z)Yi(z) =BizY(z)
(A)
1
Or, Yi(z) =1/Bi[Xi(z)Gi (z)Y(z)]Z
For n = 2, the various blocks are as follows:
Downloaded from www.bhawesh.com.np /  51
Downloaded from www.bhawesh.com.np
`
Fig: for n= 2, representation of G(z) using ladder programming..
Downloaded from www.bhawesh.com.np /  52
Downloaded from www.bhawesh.com.np
Example:01
` Obtain the block diagram for the following PTF by,
(i) direct programming
(ii) standard programming
(iii)ladder programming.
G(z) = Y(z) = 2 – 0.6z1
X(z)
1+0.5z1
(i) Direct programming:We know that,
1
X(z) = Y(z) = 2 – 0.6z
1
X(z) 1+0.5z
1
1
Or, Y(z) +0.5z Y(z) = 2X(z) 0.6z X(z)
1
Or, Y(z) = 2X(z) 0.6z1X(z) 0.5z Y(z)
Fig: Direct method structure for G(z)
(ii) Standard programming:Y(z) = 2 0.6z1 = Y(z) . H(z)
1 +0.5z1
Where,
H(z) X(z)
1
Y(z) = 2 0.6z
H(z)
H(z) =
1
X(z)
1+0.5z1
Now,
Y(z) = 2H(z) – 0.6z1H(z)
___________(i) X(z) = H(z) +0.5z1 H(z)
H(z) = X(z) – 0.5z1H(z) _______________(ii)
Downloaded from www.bhawesh.com.np /  53
Downloaded from www.bhawesh.com.np
For equation (ii),
Fig: standard structure for G(z)
For equation (ii),
(iii)Ladder Programming:G(z) = 2 – 0.6z1 = 2z 0.6
1+ 0.5z1
z + 0.5
Now, using continued fraction method,
Z+0.5) 2x – 0.6( 2
2z ± 1
1.6)z + 0.5( 1z/1/6
z
0.5)1.6(3.2
1.6
X
G(z) = 2 +
1
1z +1
1.6
3.2
Ao = 2 , B1 = 1/1.6, A1 = 1/3.2
Downloaded from www.bhawesh.com.np /  54
Downloaded from www.bhawesh.com.np
1
2
G(z) = 2 + 2.2 z +0.2z
Examples: 02
1
1 + 0.4z – 0.122
z1
(i) D
(ii) S
(iv) L
Transient and steady state Response
The transient response refers to that portion of the response due to the close lop
poles of the system and the steady state response refers to that portion of the response due to
the poles of the input or forcing function. Discrete time control system are frequently
analyzed with standard i/p such as, step input, Ramp inputs, or sinusoidal input.
Transient response specification:
Fig(i) unit step response curve showing transient response specification.
Delay time (td): the delay time is the time required for the response to reach half the final
value vary first time.
(2) Rice time : the rise time is the time required for the response to rise from 10% to
90% to 95% or 0 to 100% of it’s final value of ending on the response.
(3)
Peak time: peak time is the time required for the response to reach the first peak
of the over shoot.
Downloaded from www.bhawesh.com.np /  55
Downloaded from www.bhawesh.com.np
(4)
Maximum over shoot (Mp): It is the maximum peak value of the response
measure from the unity. If the final steady state value of the response differs from
the unity. Then it is common to use maximum percent overshoot which is defined
by
Maximum % overshoot = C(tp) – C(∞) x 100%
C(∞)
Where,
C(tp) = Amplitude at time tp
C(∞) = Final steady state value.
(5)
Setting time (ts): It is the time required for they response curve to reach an stay
within a range about the final value of a size specified as an absolute % of the
final value.
Steady state error analysis
Consider, the general block diagram of DCS as shown in fig(ii)
Let, us assume that the system is stable so that we can easily apply the final value
theorm in order to obtain the steady state value.
From fig (ii), we can write,
e(t) = r(t) – b(t)
or, E(z) = R(z) – B(z) _________(i)
Now, Let us assume, the sampling instant at t = KT and the final value theorem the o/p
or the steady state response will be:Lim e(KT) = Lim [(1 –z1) (E(z)] __________(ii)
K→ ∞
From figure,
TS
G(z) = Z [ (1 – e ) . Gp(s)]
S
Where,
TS
1–e
= PTF of
ZOH S
1
Thus, G(z) = (1 –z ) z [Gp(s)]
S
Downloaded from www.bhawesh.com.np /  56
Downloaded from www.bhawesh.com.np
And, GH(z) = Z[GH(s)]
= Z[G(s)H(s)]
1
GH(z) = (1 –z )z[GP\p(s).H(s)] __________(iii)
S
Hence, finally, the transfer function of the fig (ii) will be,
C(z) = G(z) _____________(iv)
R(z) 1+G(z)H(z)
Again, from equations (i)
E() = R(z) – B(z)
= R(z) – E(z) GH(z)
Or, E(x) [1 +GHD)] = R(z)
1
Or, E(z) =
.R(z)
1+GH(z)
Thus, by substituting the value of E(z) in equation (ii),
1
ess = lim e(KT) = Lim [(1 – Z )E(z)]
K→ ∞
z→ ∞
1
ess = Lim (1 – Z ). 1
R(z)___________(vi)
z→ ∞
1+GH(z)
which the steady state error value.
We now analyze the above equation (vi) with the various i/p, say r(t)
(i)
Static position error constant (Kp):If r(t) = 1 , then
R(z) =
1
1
1z
ess = lim [1/a+GH(z)]
Z→1
We now define the static error constant as,
Kp = lim Gh(z)
z→1
g
(ii)
ess = 1 1+Kp
Static velocity error constant (Kv):If r(t) = t, then
R(z) = Tz1
1 2
(1 –z )
Then,
1
ess = lim (1z )
1. .
1+GH(z)
z→1
Tz1
1 2
(1z )
Downloaded from www.bhawesh.com.np /  57
Downloaded from www.bhawesh.com.np
ess = lim
z→1
T
1
(1z )(1+GH(z))
now,
static velocity error constant Kv is defined by,
Kv = Lim
(1z1)(1 + GH(z)
z→1
T
ess = 1/Kv
g
(iii) Static acceleration error constant (Ka)
2
If r(t) = ½ t
Design Based on Root locus method:General Introduction
We know,
1 + GH(z) = 0
1 + G(z) H(z) = 0
In general ,
1 + F(z) = 0
Where,
F(z) =
GH(z) Or, G(z)
H(z) Again ,
F(z) =  1
Since, F(z) is complex quantity, the angle and magnitude of F(z) are, <F(z)
0
c
= 180 (2K + 1), k = 0, 1, 2, ………….. angle condition .
/F(x)/ = 1.
The value of ‘z’ that satisfy both the angle and magnitude conditions are the
roots of the characters equations, or closed loop poles.
A plot of the points in the complex plane satisfying the angle condition above is the
root locus.
Self study (page 207, methods).
Downloaded from www.bhawesh.com.np /  58
Downloaded from www.bhawesh.com.np
Examples:
Draw a root locus diagram for the system of the sampling period T = 0.5 sec. Also,
determine the critical value of k. finally locate the closed loop poles corresponding to
k = 2.
The General DCS of Fig (i) is of integral type for which.
GD(z) = Ki
Ki = integral gain.
Now, for closed loop poles, we provide as follows,
Z[Gn(S)Gp(s)] = Z 1
− e −TS .
S
1
−1
= (1 − z
1
−
)Z
s +1
s
= (1 − z
1
(s +1)
1
−1
)
−1
−
1−
1−z
1
Or, Z[Gn(s).Gp(S)] = 1  1Z 11 1
e z
1 1
=1 e z 1+
1
T 1
z 1e z
1
e−T
−1
Z
T
= z1(1e )
T 1
(1e z )
T
=1e
T
ze
The open loop poles will be,
T
Ki.
1
1z
1e
T
ze
¤
T
= G(z)
Or, Kiz(1e )
T
(z1)(ze )
The closed loop poles are given
by, 1 + G(z) = 0
T
¤
Or, Kiz(1e ) = 0
0
(z 1) (ze .s)
Now, for T = 0.5sec.
Downloaded from www.bhawesh.com.np /  59
Downloaded from www.bhawesh.com.np
G(z) = Kiz(1e0.5)
(z1)(ze0.5)
G(z) = Kiz(0.393)
(z1)(ze0.606)
G(z) =
0.393kz
=0
(z1)(z0.606)
g
In order to draw the root locus diagram, we need to first determine, we need to first
determine the break in points or break array point which is evaluated as below,
We know,
1 + G(z) = 0
Or, 1 + K B(z) = 0
A(z)
Where,
G(z) = K B(z)
A(z)
And , K = A(z)
B(z)
Then, the breakaway and break in point is determined as,
Dk = A’(z)B’(z) +A(z)B’(z) = 0
2
B (z)
Dz
If the value of k corresponding to a root z =zo of , is +ve , point z = zo is an actual
breakaway or break in points.
The given characteristic equations is
1+
ki 0.393z = 0
(z 1)(z 0.606)
Here, We assume,
Ki = K
The, 1 +
Kz 0.393
(z1)(z0.606)
Here, A(z) = (z – 1) (z – 0.606)
B(z) = KZ 0.393
g
K = A(z) = (z 1) ( z – 0.606)
z 0.393
B(z)
dk
Dz
= (z2 – 1.6062 + 0.606)
z 0.393
= z
 1.606z + 0.606
0.393
0.393z
z 0.393
Downloaded from www.bhawesh.com.np /  60
Downloaded from www.bhawesh.com.np
=
1  0  0.606
2
z 0.393
0.393
2
z  0.606
=
=0
2
Z – 0.393
Z = 0.7788 ( and also z = 0.7788)
For which,
Ki = A(z)
B(z)
= (z1)(z0.606)
Z 0.393
2
=  Z – 1.606z + 0.606)
Z 0.393
2
=  ((0.7788) – 1.606 x 0.7788 + 0.606)
0.7788 x 0.393
For which,
K1 = 8.048
for,
z =  0.7788
K2 = 0.124
for,
z = 0.7788
In order to obtain the critical value of K we use the magnitude condition
i.e.
0.393 k z
=1
(z1)(z0.606)
Since, z = 2, corresponds to , K = Kc at this
point
K = Kc = 8.165
Now,
Again, for K = 2, the closed loop poles can be find out as:
g
Z1 = 0.41 +j 0.66
Z2 = 0.41 – j 0.66
Downloaded from www.bhawesh.com.np /  61
Downloaded from www.bhawesh.com.np
( for T = 1 & T = 2 sec try it)
Recall that,
S = r Wn + jWn 1 − r
2
For,
Z = eTS
= eT(rWn +jWn 1−r
) From which, we get,
/z/ = e
Tr
2
Wn
And <Z = TWn 1−r
2
But, Wd = Wn 1−r
2
<Z = θ = TWd
¤
# For Z = 0.4098 + j 0.6623 and T = 0.5 sα, Find out the damping ratio. We know for
damping ratio r.
/Z/ = eTrWn __________(i)
2
And < Z = TWx 1−r __________(ii)
But, from the given condition,
/Z/ =
(0 .40987)2 + (0.6623)2
= 0.7788 _________(iii)
0.6623 = 58.25 0 =1.0166rad _________(iv)
0.4098
Form equation (i) & (iii)
eTrWn = 0.7788
1
and <Z = tan
or, =TrWn = 0.25
rWn = 0.5_________(v)
Similarly, from equation (ii) and (iv),
1.0166 = TWn 1−r
2
2
Or,Wn 1−r = 2.0332_________(vi)
Dividing equation (vi) and (v) ,
1− r
2
= 2.0332
r
0.5
r = 0.254
¤
Also, 1.0166 = TWn
Wn = 1.0166
T 1− r
2
1− r 2
= 1.0166
= 1.968 rad/sec
0.5 1 − 0.254
Downloaded from www.bhawesh.com.np /  62
Downloaded from www.bhawesh.com.np
Wd = Wn 1 r 2
g
g
= 1968 1
(0.254)2
= 1.903 rad/sec.
Question: Find out the static velocity error constant for G(z) for K = 2.
Kv = lim G(z) (1z1)
Z→1
T
For T = 0.5 sec, we had,
G(z) = 0.393 Kz
(z3)(z0.606)
= 0.393 x 2 x 2
(z3)(z0.606)
= 0.786z
Kv = Lim
z3)(z0.606)
1
0.786z
(1z )
Z→1 (z3)(z0.606) = 3.99
Try for T = 1 and T = 2 sec
.
Question: Design a digital controller for the system shown below. Given r = 0.5 and
setting time of 2sec. The sampling period is assumed to be 0.2 sec. obtained the response of
designed digital control system to a unit step o/p. Also obtained static velocity error
constant of the system.
Hint: ts = 4/rWn
n
Sol : first of all, let us find the Z transfer
TS
of GzoH(S) = 1e
S
Gp(S) = 1/s(s+3)
G(z) = Z[GzoH(S).Gp(s)]
Downloaded from www.bhawesh.com.np /  63
g
Downloaded from www.bhawesh.com.np
= Z[1eTS. 1 S
S(S+)
=(1z1)Z 1
2
S (S+2)
For which ,
G(z) = 0.01758 (Z+0.8760)
(z1)(z0.6703)
Here,
=Z
1
S2(S+2)
=Z A+
B+
S
A=d
Ds
2
B=S .
C
2
S
S+2
S. 1
S@(S+@
S =0
1
2
S (S+2) S = 0
C = (S+2).
1
2
S .(S+2). S = 2
1
G(z) = 1 – Z ) Z 0.25
+ 0.5 + 0.25
2
S
S+2
S
1
= (1Z1) 0.25 0.25 TZ
0 .25
1 2
1
 1
1e2TZ
(1z ) (1z )
But,
T = 0.2sec
G(z) = 0.01758 (Z + 0.8760)
(z1)(z0.6703)
Also, given
R = 0.5
Ts = 4 = 2 rWn
or, rWn = 2
or, Wn = 2/0.5 = 4
Wd = Wn 1−r
2
=4 1−
g
2
0.5 Wd = 3.464
Also, sampling frequency.
Ws = 2π
T
= 2 = 31.4 rad/sec
0.2
Downloaded from www.bhawesh.com.np /  64
Downloaded from www.bhawesh.com.np
Continu………………..
Now, using damping ration r = 0.5 and T 0.2 sec, the dominate closed loop pole is
obtained as:/Z/ = eTrWn = e0.02 x 0.5 x 4 = 0.6703
& <Z = Twd = 0.2 x 3.464 = 39.69
0

Z = /Z/ <Z = 0.6703 < 39.69
= 0.5158 + j 0.4281
Now, we locate the poles and zeroes on Z plane if point ‘P’ is the desired location on
g
0
the upper half of the Z plane, then, the total sum of angles at point ‘P’ must be ‘180 ’
Now, therefore, The net angle contribution at point ‘P’ will be,
< Zeroes  < poles
= 17.10 – 138.52 – 109.84
0
= 231.26
The angle deficiency will be:

0
231.26 +180 =  51.26
0
0
Since, There is an deficiency of  51.26 , We used to have a controller which must
0
0
provide +51.26 in order to contribute 180 at point P.
Now, let us assume the form of digital controller as,
GB(z) = K. Z+α
Z+β
Where, K is gain (or constant ) of the controller.
If we decide to cancel the pole at Z = 0.6 ≠3 by the zero of the controller at z = α
Then, the pole of the controller can be determined (from the condition that the controller
0
must provide +51.26 as a point at z = β
Example:2
Downloaded from www.bhawesh.com.np /  65
Downloaded from www.bhawesh.com.np
Design a digital controller shown in the above figure which has dead time of 2sec.
The samping peiod is 1sec. The digital controller must be PI controller. Given r = 0.5 and
Wd/Ws =1/10.obtained the step response and Kv.
(no of samples per cycle of damped sinusoidal oscillation is 10.).
From figure (i)
TS
G(s) = 1 – e
. E2s S
(S+1)
G(z) = Z[G(s)] = Z 1 eTS. E2s
S
S+1
1
= (1z ) Z e2s
S(S+1)
1 2
== (1z )z Z 1
S(S+1)
1
2
= (1z ) Z Z 1 _ 1
1
2
= (1z ) Z
S
1 _
1
1Z
(S+1)
1
T 1
1e Z
= 0.6321
2
Z (Z 0.3679)
Let, us consider the PI controller be of the form.
GD(z) = Kp + Ki 1
1
1Z
= Kp + Ki Z
Z 1
= Kp(Z1) +KiZ
(Z1)
= (Kp + Ki) Z = Kp
Z1
GD(z) = (Kp + Ki) (Z  Kp
(Kp+Ki)
(Z – 1)
Downloaded from www.bhawesh.com.np /  66
g
Downloaded from www.bhawesh.com.np
¤
The open loop transfer function will be
G(z) GD(z) = 0.6321 (Kp + Ki) (Z – Ki) Z  Kp
(Kp + Ki)
2
Z (Z – 1) ( Z – 0.3679)
Now, In order to specify the position of the PI controller We go as follows:From given,
R = 0.5
Wd = 1
Ws
10
T=1
Td = 2 sec.
From magnitude and angle condition,
−2 r
/Z/ = eTrWn = exp.
Wd
.
2
1−r
Ws
[ Since, T = 2 π/Ws & Wd = WN 1−r 2 ]
− 2Cos(0.5) 1
= exp.
2
1−r
10
= e0.3628
/Z/ = 0.6958
0
And <Z = Twd 2 Wd = 2 . Wd = 2 . 1 = 36
Ws
Ws
10
Z = /Z/ <Z
0
¤
= 0.6958< 36
Z = 0.5629 + j 0.4090
If point ‘P’ is the location of dominant closed loop, the plot for the upper half of the zplane will be:
Downloaded from www.bhawesh.com.np /  67
Downloaded from www.bhawesh.com.np
Total angle contribution = <zeroes  <poles
0
= 0 – (36 + 36+64.50 +126.901) +180
0
=  93.4
0
Therefore, the PI controller must provide +93.4 at Kp
Kp+Ki
0
tan(180 – 93.4) = 0.4090 X1 –
0.5629
X = 0.0242 = 0.5871
X = 0.5629 +
0.0242 = 0.5821
0
Thus, the controller must provide +93.4 at the point
Kp
Kp + Ki
Kp
= 0.5872
= 0.5872
Kp + Ki
g
g
GD(z) = (Kp + Ki) (Z – 0.5872)
(Z – 1)
Let, Kp + Ki = K
GD(z) = K ( Z – 0.5872)
g
(Z – 1)
We apply the magnitude condition at point P, ie.
GD(Z) G(Z)
Z = 0.5629 + j 0.4090 = 1
Or, K (Z – 0.5872 ) . (0.6321)
( Z – 1)
Z2 (Z – 0.3679) Z = 0.5629 + j 0.4090
K = 0.5070
i.e Kp + Ki = 0.5070 ___________(ii)
from equation (i) and (ii)
Kp = 0.5872 ( Kp + Ki) =
0.5872 x 0.5070
Kp = 0.2977
f
And Ki = 0.2092
g
The PI controller is of type:
GD(Z) = 0.5070 (Z – 0.5872)
(Z – 1)
Downloaded from www.bhawesh.com.np /  68
Downloaded from www.bhawesh.com.np
Again,
If r(t) = u(t)
The R(z) =
1
1Z
1
For this case,
C(Z) = ?
We know,
C(Z = G D(Z) G(Z)
R(Z)
HGD(Z) G(Z)
C(Z) = GD(Z) G(Z). R(Z)
1 + GD(Z) G(Z)
C(Z) = 0.5070 ( Z – 0.5872) . 0.6321
2
Z (Z0.3679)
(Z1)
1 + 0.5070(Z – 0.5872) . 0.6321
gg
(Z1)
2
Z (Z – 0.3679)
= ……………
Finally, we know,
Kv = Lim (1 – Z1) . GD(Z) G(Z)
Z →1 T
= ………………..
Downloaded from www.bhawesh.com.np /  69
Downloaded from www.bhawesh.com.np
Discrete time state equations:Introduction:For linear time varying discrete time system, the state equations and output
equations may be written as:X(x + 1) = G(K)x(K) + H®u(K)
Y(k) = C(k) x(k) + D(K) x(K)
Where,
X(k) = n – vector (state vector) Y(k)
= m – vector( output vector) U(k) = r
– vector( input vector) G(k) = m x n
matrix( state matrix ) H(k) = n x r
matrix ( input matrix) C(k) = m x n
matrix (output matrix)
D(k) = m x r matrix( direct transition matrix)
For linear time invariant system, equation (i) and (ii)
becomes: X(k+1) = Gx(K) + Hu(k) ___________(iii)
Y(k) = Cx(k) + Du(k) ________(iv)
Where , G,H, C & D are in matrix form.
Fig. (i) Block diagram of linear time invariant discrete control system.
State space Representation of DCS:We know the discrete time system can be described by:Y(k) + a1y(x1) +a2y(k2) +……… +any(n) = b2u(k) +b1u(k1) +………. bnu(kn).
Now, taking z transform on both sides,
1
2
n
Y(z) +a1z y(z) +a2z y(z) +…….anz y(z), =
1
2
n
bou(z) +b1z u(z) +b2z u(z) + ….bnz u(z)
The PTF of the above equation will be,
1
2
n
Y(z) = bo +b 1z +b2z + ….bnz
U(z)
1
2
n
1+ a1z +a2z +…….anz y
Downloaded from www.bhawesh.com.np /  70
Downloaded from www.bhawesh.com.np
Or, Y(z) =
U(z)
n
bo z +b 1z
n
z + a 1z
n1
n1
+b2z
+a2z
n2
n2
+ ….bn
+…….an
]
There are various ways to relies the state space equations. The major 4 forms are:(i)
controllable canonical form.
(ii)
Observable “
(iii)
(iv)
(i)
Diagonal “
“
Jordon
“ “
Controllable Conical form:
“
The controllable canonical form is given by:X!(K +1)
X2(K +1)
X3(K +1)
Xn1(K +1)
01
00
= 01
01
Xn(K +1)
0 ………….0
1 ………….0
0 ………….0
0 ………….0
0 1 0 ………….0
X1(k)
X2(k)
X3(k)
Xn1(k)
0
0
0
0
Xn(k)
1
Y!(K)
X1(k)
Y2(K)
Y3(K) = [bn – anbn : bn1 –an1b0 ………..b1 – a1b0]
X2(k)
X3(k)
Yn1(K)
Xn1(k)
Yn(K)
Xn(K)
(ii)
Observable canonical form:In this case, the state space equation is represented by:X!(K +1)
X2(K +1)
.
.
01
10
+
.
an1
.
.
X1(k)
X2(k).
.
.
bn  anbn
.bn1anbo
+.
.
Xn1(K +1)
0 0 ………….0
a2
Xn1(k)
b2a2bo
Xn(K +1)
0 0 ………….0
a1
Xn(K)
b1a1bo
………….0
………….0
.
a1
The O/P is given by,
Y!(K)
X1(k)
Y2(K)
X2(k)
.
= [0, 0 ,0 …………………………..0 1]
.
.
Downloaded from www.bhawesh.com.np /  71
.
Downloaded from www.bhawesh.com.np
Yn1(K)
Xn1(k)
Yn(K)
Xn(K)
(iii)
Diagonal Canonical form:If poles of the PTF are distinct, the state
space equation in matrix form can be expressed as:
X!(K +1)
P1 0………….0
0
X1(k)
1
X2(K +1)
.
.
0 P2
0
.
.
X2(k).
+.
.
.
.1
.
Xn1(K +1)
0 0 ………….Pn1
0
Xn1(k)
1
Xn(K +1)
0 0 ………….0
ln
Xn(K)
1
.
……….0
.
U(K)
And the o/p matrix will be,
Y(K) = [ C1 C2…………………....Cn]
X1(K)
.
+ b0U(K)
.
.
Xn(K)
Where, P1, P2, …………..Pn are all distinct poles and C1, C2, ………..Cn are all
constant coefficients of the respective poles.
(iv)
Jordan canonical form: If the PTF given by equations (i), (ii) and (iii) involves
and multiple pole of order m at z = P1 and all other poles are distinct, then the
state space equation in matrix form may be represented as:X!(K +1)
P1 1 0………….0
0 ………..0
X1(k)
0
X2(K +1)
.
.
0 P1 1
.
.
……….0 0 ………..0
.
.
X2(k).
.
.
0
.
+ . U(K)
Xm(K +1)
Xm+1(K +1)
.
.
0 0 0…………..P1 0………..0
0 0 0 ………….0 0……….0
. ..
..
.
. . .
.
.
Xn(K+!)
0 0 0
and the o/p matrix is given by:
0
0
Xm(k)
Xm1(K)
.
.
Pn
1
1
.
.
Xn(K)
1
Downloaded from www.bhawesh.com.np /  72
Downloaded from www.bhawesh.com.np
Y(K) = [C1
C2.
.
.
.Cn] X1(K)
X2(K)
+ boU(K)
.
.
Xn(K)
Examples: 01 Consider the following systems:Y(z) =
Z+1
2
Z + 1.32 + 0.4
U(z)
Represent the above PTF in
(i)
Controllable
(ii)
Observable
(iii) Diagonal
Soln:The given PTF is
2
Z+1
= boZ + b1Z + b2
Y(z) =
2
2
Z + 1.32 + 0.4
Where,
Z + a1Z +a2
b0 = 0
b1 = 1
and
b2 = 1
a1 = 1.3
a2 = 0.4
Let, Z + 1
2
Z + 1.32 +0.4
= C1
+
Z+ 0.5
= Z+1
(Z+ 0.5) (Z+0.8)
= 5/3 + 2/3
Z+0.5 Z+0.8
g
C1 = 5/3,
C2
Z+ 0.8
P1 = 0.5
C2 = 2/5, P2 = 0.8
(i)
Controllable:X1(K+1)
X2(K+1
For o/p,
Y1(K)
=
0 1
0.4 1.3
=0
Y2(K)
1
X1(K) +
X2(K)
0
1
X1(K)
g
U(z)
X2(K)
U(K)
b0 = 0
bn – anb0
Downloaded from www.bhawesh.com.np /  73
Downloaded from www.bhawesh.com.np
=b2 – a2b0
= 1 – 0.4 x 0
(ii)
Observable:X1(K+1)
= 0 0.4
1 1.3
X1(K) +
=0
X1(K)
X2(K+1
And o/p,
X2(K)
g
= 0.5 0
0 0.8
X1(K) +
X2(K)
1
1
= 5/3 2/3
X1(K)
g
Y1(K)
1
Y2(K)
(iii)
X2(K)
1
1
U(K)
b0 = 0
Diagonal:
X1(K+1)
X2(K+1
For o/p,
Y1(K)
X2(K)
Y2(K)
U(K)
b0 = 0
Example: 01
1
2
Y(z) = Z + 2Z
U(z)
1
2
1+ 4Z +3Z
Example: 02
2
3
Y(z) =
Z + 4Z
U(z)
1
2
3
1+ 6Z +11Z +6Z
Solving Discrete Time state space equations:The LTI discrete time state equations is
X(k+1) = Gx(k) +Hu(k) ___________(i)
We need to find solution of equations (i) by the following
process. Equations can be rewritten as , for k = 0,1,2………..
X(1) = Gx(0) + Hu(0)
X(z) = Gx(1) +Hu(1)
= G[Gx(0) +Hu(0) ] +Hu(1)]
2
+GHu(0)
=G
[G x(0)
+hu(1) X(3) = Gx(z) +HH(z)
2
= G[G x(0) +GHu(0) +Hu(1)] +Hu(2)
3
= G x(0) +G2Hu(0) + GHu(1) +Hu(2)
Downloaded from www.bhawesh.com.np /  74
Downloaded from www.bhawesh.com.np
By repeating this procedure we obtained,
X(K) = G2 x(0) + Σ j=0
−1
J1
GK
+ Hu(j) K= 1, 2,3….__________(ii)
Equations (ii) shows that x(k) consists of two parts:(i)
(ii)
One representing the contribution of initial conditionm x(0).
nd
2 representing the contribution of input U(j), J= 0,.1,2, 3………….(k1) The o/p y(k) is given by
Y(K) = CGK X(0) +C ∑
k −1
j= 0
Gx+j1 Hu(J) + Du(K) ___________(iii)
State transition Matrix:We define,
ψ(k) = Gk ______________(iv)
Where the nique matrix ψ(k) is called the state transition matrix.
Therefore, in terms of ψ(x) equation (ii) can be rewritten as:X(K) = ψ(K) X(0) + ∑
k −1
j= 0
= ψ(K) X(o) + ∑
(Kj1) HU(j)
k −1
j= 0
ψ(J)HU(KJ1)
Now, we go toward the z transform approach to the solution of state space
equation
We know:X(K+1) = Gx(K) + Hu(K)
Or, ZX(Z) – Zx(0) = ZX(0) + HU(z)
Or, X(Z) (ZI – G) = ZX (0) + HU(Z)
1
1
Or, X(Z) = (ZIG) ZX (0) + ZIG)
HU(Z) Taking inverse z parameters.
1
1
1
1
1
X(K) = Z [Z (ZIG) Z]X(0) +Z [ZIG) HU(Z)]](vi) Now, comparing equation (v) and (vi)
1
1
ψ(k) = Gk= Z [(ZIG) Z]__________(vii)
∑kj=−10 GkJ1 HU(J) = Z1 [(ZIG)1 HU(Z)] ________(viii)
Where, k = 1,2, 3, ……………
Examples: 02
Obtained the state transition matrix of the following discrete time system:X(K+1) = GX(K) + HU(K)
And Y(K) = CX(K)
Where, G = 0
1
H=1 ,
C = [1, 0]
0.16
1
1
Downloaded from www.bhawesh.com.np /  75
Downloaded from www.bhawesh.com.np
Then, obtained the state space X(K) and the o/p Y(K) when the i/p U(K) = 1 for K=
0, 1, 2………. Assume that the initial state is given by.
X(0) = X1(0)
X2(0)
=
1
1
Soln:The state transition matrix is given by,
ψ(x) = GK
But, GK = Z1 [(ZI – G) 1]
We go in the following manner
ZI = Z
0
0
Z
G= 0
1
0.16 1
ZI – G =
Z
1
Z+1
For, cofactor matrix of (ZIG )is
1
(Z+1)
0.16
2
The transfer of cofactor matrix,
2+1 1
0.16 2
The determinant,
(ZIG)1 = adj of (ZI – G)
/ZI – G /
Z+1
1
2
2
Z +Z+ 0.16
Z +Z+ 0.16
0.16
Z
=
2
=
=
2
Z +Z+0.16 Z +Z+0.16
Z+1
1
(Z+0.8) (Z+0.2)
(Z+0.8)(Z+0.2)
0.16
Z
(Z+0.8)(Z+0.2)
(Z+0.8)(Z+0.2)
4/3
1/3
5/3
+
Z+0.2
Z+0.2
Z+0.2
0.8/3 +
0.8/3
1/3
+
Z+0.2
Z+0.8
Z+0.2
1
5/3
Z+0.8
4/3
Z+0.8
1
ψ(k) = GK = Z [(GI –G) Z]
Downloaded from www.bhawesh.com.np /  76
Downloaded from www.bhawesh.com.np
d
Or,
1/3
Z+0.2
0.8/3
Z+0.8
0.8/3 +
Z+0.2
5/3
Z+0.2
(K) = Z1 4/3
Z+0.2
1/3
Z+0.2
= 4/3 (0.2)K1/3 (0.8)K
0.8/3 (0.2)k +0.8/3 (0.8)k
+
5/3
Z+0.8
+
4/3
Z+0.8
5/3(0.2)k5/3(0.8)k
1/3(0.2)k+4/3(0.8)k
Now, it is given that,
U(k) = 1 for K= 0, 1, 2, ………………..
U(z) = Z
Z1
It is to be noted that,
X(Z) = Z(X(K)] = (ZIG)1 [ZX(0) + HU(z)]
Now,
ZX(0) = Z 1 = Z 1 Z
HU(Z) = 1
Z
Z1
= Z
Z1
Z
Z1
ZX(0) + HU(z) = Z +
g
Z +
2
Z
Z =
Z1
Z
Z1
2
Z + 2Z
Z1
1
(ZIG)
Z1
ZX(0) + HU(z)]
=
2
(Z +2)Z
(Z+0.2)(Z+0.8)(Z1)
g
2
Z + 1.84Z)Z
(Z+0.2)(Z+0.8)(Z1)
X(z) = 17/6z
Z+0.2
34Z/6
Z+0.2
g
X(k) = Z1 [X(Z)] = 17(0.2)
6
+
22z/9 +
+
Z+0.8
Z1
17.6z/9 + 7/18z
Z+0.8
Z1
k
+
25z/18
k
22(0.8)
9
+
25.1
18
Downloaded from www.bhawesh.com.np /  77
k
Downloaded from www.bhawesh.com.np
k
24(0.2)
6
=
k
24(0.2)
6
17(0.2)
6
k
17.6(0.8)
9
+
k
(1 0) 17(0.2)
6
=
k

22(0.8)
9

17.6(0.8)
9
+
22(0.8)
9
k
7. 1
18
+
k
k
k
+
25.1
18
+
7. 1
18
+
25.1
18
k
k
k
Examples: 03
Find the inverse matrix ( 2KI – G) for
G = 0.1
0.3
0
0.1
0.1
0
0
0.2
0.3
Pulse Transfer function Matrix:A single –input –single –output discrete time system may be modeled by a
pulse transfer function s. Extension of pulse transfer function concept to a multiple –
input – multiple output discrete time system gives us the pulse transfer function matrix.
The state – space representation of a nth order LTI discrete time system with ‘r’ input
and ‘m’ output can be given by,
X(k+1) = Gx(k) +HU(k) __________(i)
Y(k) = (x(k)) +DU(k)_________(ii)
Where,
X(k) is an nvector
U(k) is an rvector
Y(k) is an mvector
G is an n x n matrix
C is an m x n matrix
D is an m x r matrix
Now, taking the ztransform of equation (i) and (ii)
ZX(Z) – ZX(0) = GX(Z) + HU(Z)
Y(Z) = CX(Z) +DU(Z)
Noting that we assume,
X(0) = 0
Then, we obtain
1
X(Z) = (ZI –G) HU(Z)
Downloaded from www.bhawesh.com.np /  78
Downloaded from www.bhawesh.com.np
1
Y(Z) =[C(ZIG) H+D]U(Z)
Or, Y(Z) =F(Z) U(Z)
Where,
1
F(Z) = C(ZI  G) H + D
___________(iii) F(Z) is called the pulse transfer
function matrix. Also,
Equation (iii) can be rewritten as:F(Z) = C adj(ZI – G) H +D __________(iv)
/ZI  G/
Equations (iv) show that the poles of F(Z) are at,
/ZI – G/ = 0
Discrimination of continuous time state space equations:In digital control system we need to convert continuous time state space equation into
discrete time state space equation. Such conversion can be done by introducing
instantaneous sampler and holding devices into continuous time system.
Consider the continuous time state equation and output equations:X = AX + BU ________(i)
Y = CX + DU___________(ii)
To clarify more in the following procedure we use (K+1) instead of (k+1) and KT instead
of K.
For which the state space equation in discrete time becomes,
X(()K+1)+T) = GX(KT) + HU(KT)
We also assume the constants factors G and H depends on the sampling period T,
thus, the above equation becomes,
X((K+1) T) = G(T) X (KT) +
H(T)U(KT) Where,
G(T) = e
AT
__________(iii)
H(T) =
Where, λ = T – t
Finally, the o/p equation is,
Y(kT) = CX(KT) + ∆U(KT) which is as it is
Example: 01
Consider the continuous time system given by :G(S) = Y(S) = 1
U(S)
S+!
Downloaded from www.bhawesh.com.np /  79
Downloaded from www.bhawesh.com.np
Obtained the continuous time state space representatikn of the system. Then
discretise the state equation and o/p equation and obtain the discrete time statespace
representation of the system. Also obtain the pulse transfer function for the system.
n
Sol :Given, G(s) = 1
S+a
st
It is of 1 order type where, the state space representation for equation (i)
in controllable canonical form will be,
0
X = aX + 1V
0
X = aX +U __________(ii)
& Y = X __________(iii)
Now,
In order to discretise, the above equations (ii) and (iii) we need to find out G(T) &
H(T). We know,
G(T) = eAT,
But, A = a
G(T) = eaT
H(T) =
∫
T
T
e
A
A
e
d B,
d
¦
¤
∫0
And, H(T)
0
but, A = a & B = 1
e
=
−a
d
−a
T
1−e
−aT
=
a
¤
0
Thus, the discretised version of equation (ii) and (iii) are:X((K+T)T) = G(T) X(KT) + H(T) U(KT)
X(K+1)T = e
aT
X (KT) + (1e
a
aT
)U (KT) __________(iv)
¤
and ,
Y(KT) = X(KT) ___________(v)
Finally, the pulse transfer function in this case is,
F(Z) = [C(ZIG)1H(T)
+D] Here, D = 0 [ Y = CX +
DV]
C=1
¤
aT
H(T) = 1e
a
aT
G(T) = e
Downloaded from www.bhawesh.com.np /  80
Downloaded from www.bhawesh.com.np
Or, F(Z) = (Z e
F(Z) = 1 – e
aT 1
aT
) . (1e
..
g
a
aT
)
A
1
(Z e
aT
)
Example: 2
Obtained the discrete time state and output equations and pulse transfer function
(when sampling period T= 1) of the following continuous time system:
G(S) = Y(S) = 1
U(S)
S(S+2)
Which may be represented in state space by the equation s:0
X
1
0
1
X1
X2
0
2
X2
Y=
[1
0]
X1
+
0
0
1
u
X2
Liapunov stability Analysis:Liapunov stability Analysis of LTI continuous time
system:Theorem:Consider the system described
by X = AX ________(i)
Where x is a state vector and A is an n x n constant nonsingular matrix. A necessary and
sufficient condition for the equilibrium state X = 0 to be asymptotically stable in the large is
that, given any positive definite Hermitian matrix or any +ve definte real symmetric matrix
Q1 there exist a +ve definite Hermitian / or any real symmetric ) matrix p such that,
A*p + PA =  Q ___________(ii)
The liapuov fraction for this system is given by,
V(X) = X x PX ______________(iii)
Here,
A* = AT
X* = XT
If
P = P11
P21
P31
P12
P22
P32
P13
P23
P33
Downloaded from www.bhawesh.com.np /  81
Downloaded from www.bhawesh.com.np
The P is called +ve definite
if /P11 />0
=
P11
P21
P11
P21
&
P12
P22
>0
P12
P22
P13
P23
>0
P31 P32
P33
We generally choose,
Q=I
Where, I is Identity matrix.
Example: 01
Determine the stability of the equilibrium state of the following system.
o
X1 = X1 –X2
o
X2 = X1 – 4X2
Also find out the liapunov function.
Soln: The given system can be represented as:X = AX
Where,
A = 1
2
1
4
Now for the system to be stable, we have
A*P + PA =  Q_________(i)
We choose,
Q=I=1
0
0
1
Also,
P=
P11
P=
P11
P12
P21
P12
P22
But, P12 = P21
g
P21 P22
Thus, equation (i) becomes,
1
1
2
4
P11
P21
P12 + P11
P22
P21
P12
P22
1
1
2 = 1
4
0
P11 + P12
P11+P12 +
P11 +P12
2P114 P12
P11
P11 P12
P11+P12
2P114 P22
+ P12
=
0
1
1
0
0
1
Downloaded from www.bhawesh.com.np /  82
Downloaded from www.bhawesh.com.np
2P11 +2P12
2P11  5P12+P22
=
1
0
2P11  5P12+P22
4P12 8P22
0
1
Comparing and equating the corresponding elements of the above system:
2P11 + 2P12 = 1
 2P11 – 5P12 + P22 = 0
P12 – 8P22 = 1
Solving the above set of equations (ii), we get,
P11 = 23/60
P12 = 7/60
P22 = 11/60
Hence, P =
23/60
7/60
7/60
11/60
st
The 1 matrix,
/P11/>0
The 2
nd
minor,
23/60
7/60 =
204/3600 > 0
7/60
11/60
Since P is +ve definite, The system is stable,
Now, the liapunov function is given by
V(X) = X*PX
=[X1 X2]
7/60 X1
7/60
11/60 X2
Liapunov Stability Analysis of Discrete Time system:Consider the discrete time system.
X(K+!) = GX(K) ________(i)
In this case,
G*PG –P = Q
Here,
G is constant nonsingular matrix
Q is identity matrix.
P is Hermitian matrix
And the system will be stable if P is +ve definite.
23/60
Example: 01
Consider the following system,
X1(K+1)
=
0
1
X1(0)
X2(K+1)
X2(0)
0.5 1
Determine the stability of the system and also determine the liapunov function (i.e
V(x) = X*PX)
Downloaded from www.bhawesh.com.np /  83