ENGR487 – Digital Control W2015

Lecture 4 – Pulse Transfer Function for Open-loop (Jan 15, 2016)

1. Pulse Transfer Function for Open-Loop System (or Cascaded Elements)

We will investigate open-loop sampled-data system of different configurations. Mainly,

configuration refers to the presence of the sampler and location the sampler.

Example 1: Obtain the pulse transfer function of the following system

∗
( ) ∗
( ) ( ) ( ) ( ) ∗
( )
( ) ( )

( )= ( ) ∗( ), ( )= ( ) ∗
( )

Taking starred Laplace transform of the above 2 equations, we have

∗( )= ∗( ) ∗( ), ∗( )= ∗( ) ∗
( )

Thus,
∗( )= ∗( ) ∗( ) ∗( )

In terms of -transform,

( )= ( ) ( ) ( )

∗ ∗
The pulse transfer function between ( ) and ( ) is

( )
= ( ) ( )
( )

@Dr. Yang Cao Page 1

 ENGR487 – Digital Control W2015

Example 2: Obtain the pulse transfer function of the following system

( ) ∗
( ) ( ) ∗
( )
( ) ( )

( )= ( ) ( ) ∗( )= ( ) ∗
( )

Note that ( ) = ( ) ( ). Taking starred Laplace transform of the above equation, we have

∗( )=[ ( )]∗ ∗
( )

Thus, in terms of -transform,

( )= ( ) ( )

∗ ∗
The pulse transfer function between ( ) and ( ) is

( )
= ( )= [ ( )]
( )

Note that, by comparing results from example 1 and 2, we can see that ( ) ( ) ≠ ( ).

@Dr. Yang Cao Page 2

 ENGR487 – Digital Control W2015

Example 3: Obtain the pulse transfer function of the following open-loop

system with a digital filter

Note that if no digital filter, this system would be the same as example 1 in lecture 3.

Digital filter solves a linear difference equation and can be represented by transfer function

( ). Such that

( )= ( ) ( )

D/A converter usually has an output data-hold register which gives the D/A characteristics of a

ZOH. Then,

1− ∗(
( )= )

Note that we only feed discrete signal to ZOH, not to any other continuous system. The
∗( ).
mathematical model for the discrete signal can be modeled as

Finally, the output ( ) is

1− ∗( ∗(
( )= ( ) ( )= ( ) )= ( ) )
( )

Taking starred Laplace transform, we have

∗( )= ∗( ) ∗( )

In terms of -tranform,

( )= ( ) ( )= ( ) ( ) ( )

@Dr. Yang Cao Page 3

 ENGR487 – Digital Control W2015

The pulse transfer function of the system is

( ) 1−
= ( ) ( )= ( )⋅ ( )
( )

An equivalent block diagram based on the pulse transfer function is shown on next page.

ZOH

( ) ( ) 1− ( )
( ) ( )

( )

Example 4: Determine step response of the system described in example 3.

Suppose that ( ) = and ( ) is described by
( ) = 2 ( ) − ( − 1)

( ) 2 −1
( )= =2− =
( )

1−
( )= ( )

1− 1
[ ( )] = ( ) = (1 − )
( + 1)

(1 − )
= (1 − )
( − 1)( − )

1−
=
−

The pulse transfer function is

( ) 1− 2 −1 (1 − )(2 − 1)
= ( ) ( )= =
( ) − ( − )

@Dr. Yang Cao Page 4

 ENGR487 – Digital Control W2015

Note that the system has two real distinct poles, = 0, = . Since 0 < < 1, the

system’s transient response is stable. However, as sampling time gets faster, or → 0,

→ 1. The system will appear to have a pole at 1. This means that the system’s transient

response (or impulse response) will be marginally stable. Also observe that system has a gain of

(1 − ). When → 0, or → 1 , this gain → 0. Thus we should expect very small

amplitude of impulse response.

For a step input, ( ) = , thus

(1 − ) (2 − 1)
( )=
− −1

By partial fraction expansion, we have

( ) (1 − )(2 − 1) (1 − ) 1 −2
= = + +
( − )( − 1) −1 −

( ) = (1 − )+ +( − 2)
−1 −

From transform table, we can see that

[1 − ] = (1 − ) ( ) = 1− , =0
0, ≠0

= 1, = 0, 1, 2, …
−1

( − 2) =( − 2) , = 0, 1, 2, …
−

Then for ( ), it is

( )=1+( − 2) , = 1,2,3, …

(0) = 1 − +1+ − 2 = 0, =0

As → ∞, (∞) = 1. Or from final value theorem,

@Dr. Yang Cao Page 5

 ENGR487 – Digital Control W2015

( )( )
(∞) = lim → [(1 − ) ( )] = lim → =1
( )

0.9

Matlab Code (assume = . ): 0.8

0.7

T=0.5; 0.6

num=[conv([1-exp(-T)], [2 -1])]; 0.5

den=[conv([1 0], [1 -exp(-T)])]; 0.4
%step response 0.3
x=ones(1,100); % unit step input
0.2
y_step=filter(num, den,x);
0.1
%stem(y);
0
0 5 10 15

%or equivalently using a function ‘stepz’ from filter toolbox

% stepz(num, den)

% stepz(num, den, N, f);

% computes the first N samples of the step response.
% separates the time samples by T = 1/f second

% [H,T]= stepz(num, den, N, f);

% stepz returns the response in column vector H and a vector of times (or intervals) in T

%impulse response
r=[1 zeros(1,20)]; %Kronecker delta input
y_impulse=filter(num, den,r);
stem(y_impulse);
%an equivalent function from filter toolbox ‘impz’
%impz(num,den, 20); compute 30 samples of impulse response

See ‘Lecture4_Example4.m’ which uses slight different method for simulation.

@Dr. Yang Cao Page 6

 ENGR487 – Digital Control W2015

 Concept of DC gain

For many control systems, the steady-state gain for a constant input is important. This gain

is called DC gain. DC gain is the ratio of the output of the system to its input (presumed

constant) after all transients have decayed.

If the input is a unit step input ( ), the steady-state value of the output is the DC gain,

which can be easily found using final value theorem.

 For continuous system

1

( )

1
DC gain = lim ( ) = lim ( )= (0)
→ →

 For discrete system

ZOH

( )= ( ) −1 1− ( )
( )

( )
Using final value theorem,

DC gain = (∞) = lim(1 − ) ( )

→

= lim(1 − ) ( )
→ −1

= lim ( ) = (1)
→

Hence for DC gain,

DC gain = lim ( ) = lim ( )

→ →

@Dr. Yang Cao Page 7

 ENGR487 – Digital Control W2015

This relationship gives us a relatively easy check on the calculation of ( ) because DC

gain of a system should not change whether it is continuous or discrete.

In example 4,

2 −1 1−
DC gain = ( ) ( ) = =1
−

Or

2 −1 1
DC gain = ( )| ( ) = =1
+1

Note that in Charles Phillips’ Book, the initial and the final value theorem is

(∞) = lim[( − 1) ( )]
→

which is equivalent to what’s given in Ogata’s book. The reason is that from Ogata’s expression

−1
lim(1 − ) ( ) = lim ( ) = lim( − 1) ( )
→ → →

The on denominator can be set to 1.

@Dr. Yang Cao Page 8