Discrete control Systems

In contrast to the continuous –time system whose operation is described or modeled by a

set of differential equations, the discrete –time system is one whose operation is modeled
by a set of difference equation. The transform method employed in the analysis of linear-
time –invariant continuous –time system is the Laplace transform, in similar manner, the
transform used in the analysis of linear-time invariant discrete-time system the is Z-
transform. The modeling of discrete –time system by both transfer function and state-
space equations. Discrete –time system are subdivided into:

1- Digital control system: refers to the use of digital computer or controller in the systems
so that the signals are digitally coded, such as in binary code

2-Sampled-datasystem: The signals are in the form of pulse data. The error signal can be
supplied only in the form of pulses (no digital controller in the system)

Advantages of digital control system

1-Size and flexibility 2-Price and reliability 3- High accuracy

4-Reduced sensitivity to signal noise. 4-Avoid distortion in sampled-data system

5- Capability to easily reconfigure the control algorithm in software

6-Many computer are able to receive several input

Ideal sampler: Is basically a switch that close every T Second for one instant of time.

Zero-order hold (ZOH): The sampling operation causes a loss of information since the
sampled signal is not known at all instants of time. In an effort to reduce this loss of
information, a data reconstruction device ,called a data-hold, is inserted into the system
directly following the sampling operation.

1
The T.F of zero –order hold is

L.T

Z=

2
3
Example:

4
Solution:

Example:

5
Example : consider the sampled-data system shown in
figure below . Find the closed-loop T.F ,ζ ,ωn and τ.

Solution:

6
ζ=0.25 , ω n=0.919 rad/sec , time constant τ=4.36sec

Selection of sampling peroid(Ts)

It is worthy to note that the single most important impact of implementing a
control system digitally is the delay associated with the hold.

A delay in any feedback system degrades the stability and damping of the system.
The reconstructed signal e(t) lag e(Kt) by Ts /2. The effect of a delay can be
analyzed using frequency response because a time delay of T s /2 translates in to
phase decrease φ = - ω1Ts /2 ω1: gain crossover frequency

In order to reconstruct a continuous signal from the sampled sequence, the sampling
frequency must verify the condition (Nyquist theorem): fs > 2 fmax

The rule used to choose the sampling frequency in control systems is

6 f BCL  f s  25 f BCL
CL
fs: sampling frequency, f B : closed loop system bandwidth
1St order system:
1
In this case the system G( s) 
bandwidth is 1  T0 s
T0
f B  f0 
1  Ts  T0
2T0 4 T0 : Time constant
2nt order system:

7
 n2
G( s)  2
s  2 n s  n2
0.25 1.5
 Ts  for 0.7    1.0
n n

Alternative method for selection sampling frequency (fs)

ωs = (20 – 35)ωn ωn: closed-loop natural frequency

ωs = (20 – 30)Bw Bw: closed-loop bandwidth , Ts = 1/fs

For first time constant: Ts =(0.25 – 0.5) T0 , ωs =2πfs

Example1: consider the closed T.F

ωn = 1 , ωs=20X1 =20 , ωs=35X1 =35

Ts = 1/fs=2π/ ωs 0.179 < Ts < 0.314

Example2: consider the closed T.F

T0= 0.5 sec , Ts =(0.25 – 0.5) T0 , 0.125 < Ts < 0.25

The smaller the Ts , the more computation it requires. Using a
small Ts may also introduce computational problems.

Example3:plot the discrete frequency response

corresponding to the plant T.F
1 H(S)=1
G( s) 
s( s  1)
Ts with ZOH at 0.2 , 1 ,2 sec and compare with continuous
system.

Solution: the phase margin P.M=52Deg for continuous system

P.M =47Deg for Ts=0.2sec , φ = - ω1Ts /2 = - 4.5Deg

8
P.M =30Deg for Ts=1sec , φ = - ω1Ts /2 = - 22Deg

P.M =13Deg for Ts=2sec , φ = - ω1Ts /2 = - 42Deg

For Ts =0.2 sec the bode plot close to its continuous system

9
Example: Consider the following second order
Characteristic equation
Q (z) = z2 + z + 0.25 = 0

11
Applying the conditions, we have

Q (1) = 2.25 > 0

Q (-1) = 0.25>0

/ao/ =0.25 < a2 =1

Thus the conditions are all satisfied. The two roots are all
inside the unit circle, and the system is stable

Example:

11
Example

The characteristic equation of the closed loop system

transfer function is:
Q( z )  z 2  (0.368K  1.368) z  (0.368  0.264K )  0

Test the stability of the system using Jury's Stability

Criteria

Solution:

 Condition: Q(1)>0
1  (0.368K  1.368)  (0.368  0.264K ) >0
Which yields to K>0

 Condition: (-1)2Q(-1)>0
1  (0.368K  1.368)  (0.368  0.264K ) >0
Which yields to? K  2.736  26.3
0.104
 Condition: a0  a2

(0.368  0.264K )  1
0.632
Which yields to? K  2.39
0.264
Thus the system is stable if 0<K<2.39
Digital Implementation of the PID Controller
Unlike analog electronic (resistor , capacitor and operational amplifier)
,digital computer cannot integrate .Therefore in order to solve a
differentail equation in a computer,the equation must be
approximated by reducing it to an algebraic equation involving sums

12
and products only . These approximation techniques are often referred
to as numerical integration.

The PID controller transfer function U(S)/E(S) in the continuous control

system is described by

The proportional component Kp is implemented digitally by a constant

gain Kp . U(Z)/E(Z)= Kp

The time- derivative of afunction e(t) at t=KT can be approximated by

the backward-difference rule(Eulers Method) using the values of e(t)
measured at t =KT and (K-1)T that is

U(s) / E(s) = kD S ,The difference equation is

u(KT)= (e(KT) – e[(K-1)T])/ Ts Z-transform on both sides

U(Z)/E(Z) =Z-1/ TsZ

Gd(Z)= kD (Z-1)/ (TsZ)

There are a number integration rules that can be used to digitally
approximate the integral controller Ki /s .The three basic methods of
approximating the area of a function numerically are :

1- Forward Rectangular Integration

U(s) / E(s) = Ki /s ,The difference equation is

u(KT)= u[(k-1)T] + Ts e(KT) Z-transform on both sides

U(Z)/E(Z) = TsZ / Z-1

Gi (Z)= Ki TsZ /( Z-1)

13
2- Backward Rectangular Integration
U(s) / E(s) = Ki /s ,The difference equation is

u(KT)= u[(k-1)T] + Ts e[(K-1) T] Z-transform on both sides

U(Z)/E(Z) = Ts / Z-1

Gi (Z)= Ki Ts /( Z-1)

3-Trapezoidal integration (Bilinear or Tustin transform)

U(s) / E(s) = Ki /s ,The difference equation is

u(KT)= u[(k-1)T] + Ts /2{e(KT)+ e[(K-1) T]} Z-transform

U(Z)/E(Z) = 0.5 Ts (Z+1) /( Z-1)

Gi (Z)= 0.5Ki Ts (Z+1) /( Z-1)

The z-domain transfer function of the PID controller is:

Gc (Z)= U(Z)/E(Z)= Kp+ Ki TsZ /( Z-1)+ kD (Z-1)/ (TsZ)

The complete difference equation algorithm that provides the

PID controller is obtained the adding the three term to obtain

u(k)= [Kp+Ki Ts+ kD/Ts]e(k) – (k D/Ts)e(k-1)+ Ki u(k-1)

The above equation can be implemented using a digital
computer, microprocessor or microcontroller. Of
course, we can obtain a PI or PD controller by setting
an appropriate gain equal to zero.

14
15