Discrete-Time Control Systems

Introduction
• Backgrounds necessary for the analysis and design of
discrete-time control systems in the z plane are
presented.
– The main advantage of the z transform method: it enables us to
apply conventional continuous-time design methods to discrete-
time systems.

• The chapter covers:

– Mathematical representation of the sampling operation
– The convolution integral method for obtaining the z transform
– The sampling theorem based on the fact that the Laplace
transform of the sampled signal is periodic
– Mathematical modeling of digital controllers in terms of pulse
transfer function
– Realization of digital controllers and digital filters
 Impulse Sampling and Data Hold
• Impulse Sampling
– A fictitious sampler
– The output of the sampler is a train of impulses.


x (t )   x(kT ) (t  kT )
*

k 0

x* (t )  x(0) (t )  x(T ) (t  T )    x(kT ) (t  kT )  

 Impulse Sampling and Data Hold
• Impulse Sampling (cont.)
– Let’s define a train of unit impulses:

 T (t )    (t  kT )
k 0

– The sampler may be considered a modulator with

• The modulating signal: the input x(t)
• The carrier : the train of unit impulses T(t)
 Impulse Sampling and Data Hold
• Impulse Sampling (cont.)
– The Laplace transform of x*(t)
 
X * ( s)  L x* (t )  x(0)L  (t )  x(T )L  (t  T )  x(2T )L  (t  2T )  
 x(0)  x(T )e Ts  x(2T )e  2Ts  

  x(kT )e  kTs
k 0
1
– If we define e  z or s 
Ts
ln z
T
1  
*
X ( s)  X  ln z    x(kT )z k  X ( z )
*
s (1/ T ) ln z
T  k 0

– Hence we may write

X * ( s)  X ( z)
s (1/ T ) ln z
 Impulse Sampling and Data Hold
• Impulse Sampling (cont.)
– Summary
• If the continuous-time signal x(t) is impulse sampled in a periodic
manner, the sampled signal may be represented by

x (t )   x(t ) (t  kT )
*

k 0

• The Laplace transform of the impulse-sampled signal x*(t) has been

shown to be the same as the z transform of signal x(t) if

eTs  z
 Impulse Sampling and Data Hold
• Data-Hold Circuits
– Data-hold: a process of generating a continuous-time signal h(t)
from a discrete-time sequence x(kT).
– A hold circuit approximately reproduces the signal applied to the
sampler.
h(kT   )  an n  an1 n1    a1  a0 where 0    T
Note that signal h(kT) must equal x(kT):
h(kT )  x(kT )
h(kT   )  an n  an1 n1    a1  x(kT ) n-th order hold

– The simplest data-hold: zero-order hold (clamper)

h(kT   )  x(kT )
 Impulse Sampling and Data Hold
• Zero-Order Hold
h(kT  t )  x(kT ), for 0  t  T
 Impulse Sampling and Data Hold
• Zero-Order Hold (cont.)
A real sampler and zero-order
hold

h1 (t )  x(0)1(t )  1(t  T )  x(T )1(t  T )  1(t  2T )  x(2T )1(t  2T )  1(t  3T )  


  x(kT )1(t  kT )  1(t  (k  1)T )
k 0

e  kTs
Since L 1(t  kT ) 
s

e  kTs  e  ( k 1)Ts 1  e Ts 
L h1 (t )  H1 ( s)   x(kT )   x(kT )e  kTs

k 0 s s k 0
 Impulse Sampling and Data Hold
• Zero-Order Hold (cont.)
Mathematical model: an impulse
sampler and transfer function

L h2 (t )  H 2 (s)  H1 (s)

1  e Ts 
H 2 ( s) 
s
 x(kT )e
k 0
 kTs

From the figure H 2 (s)  Gh 0 ( s) X * ( s)


Since X ( s)   x(kT )e kTs
*

k 0 1  e Ts
Gh 0 ( s) 
s
1  e Ts *
H 2 ( s)  X ( s)
s
 Impulse Sampling and Data Hold
• Transfer function of First-Order Hold
h(kT   )  an n  an1 n1    a1  x(kT ) n-th order hold

1-st order hold h(kT   )  a1  x(kT )

By applying the condition that h((k  1)T )  x((k  1)T )

h((k  1)T )  a1T  x(kT )  x((k  1)T )
x(kT )  x((k  1)T )
x(kT )  x((k  1)T ) h(kT   )  x(kT )  
a1  T
T
 Impulse Sampling and Data Hold
• Transfer function of First-Order Hold (cont.)
– Derivation of the transfer function

 t t T
h(t )  1  1(t )  1(t  T )  1(t  T )
 T T

Taking the Laplace transform

1 1  1 1
H ( s )    2   2 e Ts  e Ts
 s Ts  Ts s
1  e Ts 1  e Ts Ts Ts  1

s

Ts 2
1  e  Ts 2