integrator wind up in a controller

Hi
New to PI controllers and wondering why wind up occurs using PI controllers. Googling through could not provide me comprehension. Why the outputs of integrator increase indefinitely ? Many op-amp based PI controllers saturate at +/-15V. Simple example will be of great help which I can check in matlab/python etc

Hi,

Many op-amp based PI controllers saturate at +/-15V.

Click to expand...

...Only because the output voltage range is limited by the supply voltage...

Why the outputs of integrator increase indefinitely

Click to expand...

It depends on the input voltage.
A feedbacked system won´t increase infinitely.
It increases until the error becomes zero (this is why you use an integrator) and as soon as the error is zero it won´t stop increasing anymore.

Your question lets me assume you didn´t understand yet the usuage and meaning of the "integrating part" of a regulator.

Indeed analog and digital integrators are different in the bahaviour of limiting.
* An analog integrator is limited by the supply rails. It automatically stops.
* a digital integrator is limited by the range of the used variable. But the problem with digital integrators is that they will overflow and start with the most negative (or better say inverted) value. This may cause big problems, thus your software should prevent this.

But common to both regulators is that they should be limitd to meaningful values: As soon as the output of the whole regulator is at 100% it makes no sense to increase the inegrator anymore.

All this should be written and better explained in regulator tutorials. I recommend to read through some good ones.

Klaus

HI klausST

Could you explain windup with respect to speed controllers. Input set point is 0-10V (for speed 0-100 rpm). Feedback is through encoder. I assume that if feedback fails, PI controllers could go to +/-15V ( corresponding to 150 RPM which is beyond the rated speed of the motor). Is that what you call windup?

Hi,

I don´t like to reinvent the wheel by writing the same as others did before.

So please let us focus on:
* the document you already have read to design your regulator (give link, or upload document) and write what you don´t understand
* and/or your regulation system you talk about. (speed controller) draw a sketch of your system..

Klaus

Hi

I am just trying to understand wind up in the context of speed control system of a dc motor . I do not have any problem or design in my mind.

Hi,

1) Setup_speed > motor_speed
2) thus error > 0
3) thus integrator increases

Klaus

gary36 said:

HI klausST

Could you explain windup with respect to speed controllers. Input set point is 0-10V (for speed 0-100 rpm). Feedback is through encoder. I assume that if feedback fails, PI controllers could go to +/-15V ( corresponding to 150 RPM which is beyond the rated speed of the motor). Is that what you call windup?

Click to expand...

Yes this is exactly what windup is.

In analog controllers a zener type clamp (10V in your case) from output to amplifier inverting input can prevent it. In the digital world you have more options.

Hi asdf44
What would happen if my set speed is 20 RPM and due to feedback failure, speed jumps to 100 RPM. How to prevent that?

Hi,

The same, but in opposite direction.

Here, too, you need to limit the integrator.

Klaus

gary36 said:

Hi asdf44
What would happen if my set speed is 20 RPM and due to feedback failure, speed jumps to 100 RPM. How to prevent that?

Click to expand...

If this is an analog system then you'll need to clamp the voltage at the level that corresponds to 20 RPM.

Hi

The question is different. Based on the requirement, speed is settable from 0-100 RPM. 100 RPM is max speed of the motor. let us say we set the speed for a particular process to be 20 RPM. For change of the process (flow) requirement, operator is allowed to increase the set speed from 20-100 RPM. Under these circumstances what will happen if the feedback fails, and the speed jumps up to 100 RPM from 20 RPM. Can we dynamically limit or clamp the output voltage for a particular set speed?

Referring to the original question, in a usual drive application, the power supply is chosen so that that the motor can't be operated beyond safe limits. "Windup" means that the controller tries to demand more motor voltage than available. The main problem involved with windup is the delayed reaction when the saturation situation ends. Windup prevention is achieved by clamping the integrator near the saturation limit of the motor driver.

Feedback failure is a basically different problem. A simple solution could be to limit the motor voltage to a lower safe value. Or estimate actual speed from motor voltage and current, causing a shutdown if it's exceeded.

Hi,

I think the answer is already given in post#2:
You now ask "if feedback fails"
Post#2 says:

A feedbacked system won´t increase infinitely.
It increases until the error becomes zero (this is why you use an integrator) and as soon as the error is zero it won´t stop increasing anymore.

Click to expand...

If feedback fails --> there is no (proper) feedback.
Without feedback there will be any error causing the integrator to run..

Can we dynamically limit or clamp the output voltage for a particular set speed?

Click to expand...

Post#2 says:

But common to both regulators is that they should be limitd to meaningful values: As soon as the output of the whole regulator is at 100% it makes no sense to increase the inegrator anymore.

Click to expand...

Post#7 says:

In the digital world you have more options.

Click to expand...

You are free to write your own "dynamically" reacting software.
But what do you gain against:

As soon as the output of the whole regulator is at 100% it makes no sense to increase the inegrator anymore.

Click to expand...

This will work in either situation.

What else do you want to know?

Please read through some tutorial, do simulations, (even Excel would do), take a piece of paper and a oencil.
Show some effort of your own.

Klaus

gary36 said:

Hi asdf44
What would happen if my set speed is 20 RPM and due to feedback failure, speed jumps to 100 RPM. How to prevent that?

Click to expand...

This is a fundamental problem. If you’re worried about feedback failure then you need a redundant feedback of some sort.

(Funny story but we had a system where the feedback was failing probably due to esd or something causing the output to rail so the marketing guy demanded we use positive feedback instead of negative feedback).

Another general alternative that may or not be possible is to use a feed forward and then limit the maximum impact of the feedback loop. This only works if you have reasonably accurate open loop response and only need the feedback loop for fine tuning.

gary36 said:

what will happen if the feedback fails, and the speed jumps up to 100 RPM from 20 RPM. Can we dynamically limit or clamp the output voltage for a particular set speed?

Click to expand...

You REALLY need to think about what you're asking.

1 Your operating range is 0 to 100.
2 Your set point is 20.
3 Feedback fails but you want some magical method to limit your output to 20. Something like feedback???