Thread Subject: sisotool

My question is about feedback control. The open-loop process
I'm working with is G = tf(5000,[1 1020 20000 0]). I added
this to the "sisotool" as "G" and used a PD controller (i.e.
added a zero), and stabilised it.

Then I added a design constraint; percentage overshoot < 5%.
And this is where the problem comes: even when all the poles
are within the constraint, I'm still getting an overshoot
largely in excess of 5% and I can't understand why.

Could someone please try it and tell me what the problem is?
I would be very grateful.

Anyone?

The 5% overshoot requirement on an rootlocus plot is exact only if the loop
is a second order system. If the system is second order dominant the
requirement will also be accurate but in your case the controller zero (D
part of your PD controller) is interacting with the dominant close loop
poles and causing additional derivative action in the response - resulting
in the extra overshoot.

-Alec

"Waleed Najy" <waldo90009@hotmail.com> wrote in message
news:g02l2f$98n$1@fred.mathworks.com...
> My question is about feedback control. The open-loop process
> I'm working with is G = tf(5000,[1 1020 20000 0]). I added
> this to the "sisotool" as "G" and used a PD controller (i.e.
> added a zero), and stabilised it.
>
> Then I added a design constraint; percentage overshoot < 5%.
> And this is where the problem comes: even when all the poles
> are within the constraint, I'm still getting an overshoot
> largely in excess of 5% and I can't understand why.
>
> Could someone please try it and tell me what the problem is?
> I would be very grateful.

Hi Waleed,

The overshoot design constraint in the SISOTOOL is based on a second order
system with no zeros
1/(s^2+2*zeta*wn+wn^2). It uses the relationship between damping ratio and
overshoot for this system.
So the constraint is rigorous for this class of systems but is an
approximation for higher order systems.

-craig

"Waleed Najy" <waldo90009@hotmail.com> wrote in message
news:g02l2f$98n$1@fred.mathworks.com...
> My question is about feedback control. The open-loop process
> I'm working with is G = tf(5000,[1 1020 20000 0]). I added
> this to the "sisotool" as "G" and used a PD controller (i.e.
> added a zero), and stabilised it.
>
> Then I added a design constraint; percentage overshoot < 5%.
> And this is where the problem comes: even when all the poles
> are within the constraint, I'm still getting an overshoot
> largely in excess of 5% and I can't understand why.
>
> Could someone please try it and tell me what the problem is?
> I would be very grateful.

Thank you Alec and Craig. You've been very helpful. I've
been banging my head against the wall over this for some
time now.