Experiment No.

Aim: For step,ramp and parabolic input find steady state error for the unity
feedback digital control system.

Software: MATLAB

Theory:

We now examine the effect of sampling upon the steady-state error for digital
systems. Any general conclusion about the steady-state error is difficult because
of the dependence of those conclusions upon the placement of the sampler in the
loop. Remember that the position of the sampler could chang the open-loop
transfer function. In the discussion of analog systems, there was only one open-lo
transfer function, G(s) upon which the general theory of steady-state error was
based and from which came the standard definitions of static error constants. For
digital systems, however, the placement of the sampler changes the open-loop
transfer function and thus precludes any general conclusions. We assu the typical
placement of the sampler after the error and in the position of the cascade
controller, and we derive our conclusions accordingly about the steady-state error
of digital systems. Consider the digital system in Figure (a), where the digital
computer is represented by the sampler and zero-order hold. The transfer
function of the plant is represented by G 1(s) and the transfer function of the
z.o.h. by (1 - e ^ (- T_{1}))/s Letting G(s) equal the product of the z.o.h. and G 1(s)
and using the block diagram reduction techniques for sampled-data systems, we
can find the sampled error, E^ * (s) = E(z) . Adding synchronous samplasa the
input and the feedback, we obtain Figure (b). Pushing G(s) and its input sampler
to the right past the pickoff point yields Figure (c). Using Figure (a), we can
convert each block to its z-transform, resulting Figure (d).
a Digital feedback control system for evaluation of steady-state errors

b. phantom samplers added

c. pushing G(s) and its samplers to the right past the pickoff point,

d z-transform equivalent system

From this figure, E(z) = R(z) - E(z) * G(z) or

E(z) = (R(z))/(1 + G(z)) -- 1

The final value theorem for discrete signals states that,

e^ * (∞) = lim z -> 1 (1 - z ^ - 1) * E(z) -- 2

where e^ bullet (∞) is the final sampled value of e(t) or (alternatively) the final
value of e(kI)

Using the final value theorem on Eq. 2, we find that the sampled steady state
error, e^ * (∞) for unity negative feedback systems

e^ * (∞) = lim z -> 1 (1 - z ^ - 1) * E(z) = lim z -> 1 (1 - z ^ - 1) * (R(z))/(1 + 6(z)) -- 3

Equation 3 must now be evaluated for each input step, ramp, and parabola.

Unit Step Input

For a unit step input, R(s) = l / s

R(z) = z/(z - 1) -- 4

Substituting Eq. 4 into Eq. 3, we have

 e^ * (∞) = 1/(1 + lim z -> 1 sigma(z)) -- 5

Defining the static error constant, Kp as

K_{p} = lim z -> 1 G(z) -- 6

we rewrite Eq. 5 as

Unit Ramp Input

For a unit ramp input, R(z) = (Tz)/((z - 1) ^ 2) Following the procedure for the step
input, you can derive the fact that

G v ^ * (∞) = 1/k_{v} - 8

Where

K_ {v} = 1/T * lim z -> 1 (z - 1) * G(z) -- 9

Unit Parabolic Input

For a unit parabolic input R(z) = (T ^ 2 * z)/((z - 1) ^ 3) Similarly,

e^ * (∞) = 1/K_{a} -- 10
Where

K_{a} = 1/(T ^ 2) * lim z -> 1 (z - 1) ^ 2 * G(z) -- 11

The equations developed above for e^ bullet (∞) Kp, Kv, and Ka are similar to the
equations developed for analog systems. Whereas multiple pole placement at the
origin of the S-plane reduced steady-state errors to zero in the analog case, we
can see that multiple pole placement at z = 1 reduces the steady-state error to
zero for digital systems of the type discussed in this section. This conclusion
makes sense when one considers that s = 0 maps into z = 1 under 2 = e ^ (Ts) For
example, for a step input, we see that if in ) has one pole at the limit G(z) Eq .(?
will become infinite, and the steady-state error will reduce to zero. For a ramp
input, if G(z) in Eq. (9) has two poles at z = 1 , the limit will become infinite, and
the error will reduce to zero. Similar conclusions can be drawn for the parabolic
input and Eq. (11). Here, needs three poles at z = 1 in order for the steady-state
error to be zero. z = 1 G(z)

Procedure:

> Type the program in MATLAB editor that is in M-file.

>Scan the value of sampling time interval "T" by using input('text') command. >
The numerator polynomial of transfer function is represented as numgls.

> The denominator polynomial of transfer function is represented as dengls.

> Make the transfer function of the system by using tf(numgls,dengls) command.
>Form closed loop system by using feedback (sys,fb) command.
> Use c2d (sys,T,'zob') command to Convert from continuous to discrete time
models. >Find the location of closed loop poles.

> Use the following formula,

K_{p} = lim z -> 1 G(z) ,

K_{a} = 1/(T ^ 2) * lim z -> 1 (z - 1) ^ 2 * G(z),++++++

K_{V} = 1/tau * lim z -> 1 (z - 1) * G(z)

 Use the mineral(sys) command to cancel out common term in numerator &
denominator.

Use the dcgain(sys) command to find DC gain of the system.

 Find the steady state error of the system.

e(∞) = 1/(1 + K_{p})

e(∞) = 1/K_{0} ,

 Save and run the program.

, e(∞) = 1/K_{a}
Example:

For step, ramp and parabolic input find steady state error for the unity feedback
digital control system.

Gz=

404537z+0.04679

38

of 38

2-1.905 z+0.9048

Sample time: 0.1 seconds

Iz=
801837z+0.04679

z^2-1.856 z+0.9516

Sample time: 0.1 seconds

Closed loop Z-plane poles

r=0.9282+0.3000i

0.9282-0.3000i

M-0.9755

0.9755

Kp-Inf

Estep=0

K_{v} = 10
Eramp=0.1000

K_{a} = 0

Eparabola = Inf

Conclusion: For a digital control system, we have find out the steady state error.