# Why does this step response overshoot?

1. May 17, 2014

### 215

I have this transfer function which i don't know why it is overshooting due to an step input.

http://snag.gy/Vh1SM.jpg

Could someone explain why it does that.

It's it a close loop transfer function, where i've designed a controller for it, but for some reason, eventhoug all my poles are placed correctly, it is not possible to get the response i want to.

2. May 17, 2014

### rude man

With the given transfer function there is no overshoot.
Wolfram alpha agrees!

3. May 17, 2014

### 215

4. May 17, 2014

### rude man

Oops, there is too a small overshoot. I couldn't see it on wolfram. It's due to the zero at s = -1.513. Sorry!
You can cancel it by inserting a pole at the same point (s = - 1.513).

5. May 17, 2014

### 215

But how come does zeroes apply overshoot? I thought it was only poles which affected the system

6. May 17, 2014

### milesyoung

You should check your entry. It does have overshoot.

The transient response of your system is not only determined by its poles. You have a LHP zero as well, which can have a significant impact on the overshoot of the step response of your system.

Quick Google search for 'poles zeros transient' turned up:
http://courses.engr.illinois.edu/ece486/documents/lecture_notes/effects_zero_pole.pdf

7. May 17, 2014

### 215

but what criteria should i apply for determining the zeroes, and how does it affect the system. I mean it is still a LHP zero??? so bad can't it be

8. May 18, 2014

### 215

How should i using a PID controlle be able to adjust the Zero location, while i am adjusting the location of the pole.

9. May 18, 2014

### rude man

You need to describe your problem more fully, not just talk about an isolated transfer function. What is your plant and what are the response criteria?

10. May 18, 2014

### 215

I tried to make a general problem, but seems to have forgotten where it began at.
I posted my problem on another forum.

http://snag.gy/OqhxH.jpg

I hope this helps.
My repsponse criterias is that I want an overshoot to be 0% and settling time be under 1 sec.
Using Matlab tuning application i managed to do with only the I part of the controller. The problem is though i've read alot of places that it is not recomended to use just I or D regulators.

Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted