Choice of controllers

[garimella]

I was reading about on/off, P,I,D and lead,lag compensators, questions are

1)Are there any example of a system which can work appreciably well without controllers.
2) How do I decide that I need a particular type of controller as listed above

 

[CataM]

2) How do I decide that I need a particular type of controller as listed above

You figure out how it works without the controller and figure out what the system is missing (or has very little).
Based on that, you choose the controller that helps you improve the thing that is missing or has very little of it.

- - - Updated - - -

1)Are there any example of a system which can work appreciably well without controllers.

Well is relative. A lot of systems "work" (i.e. do not blow) without controllers, but they are very poor in performance.

 

[garimella]

When to choose a proportional controller

 

[CataM]

Ask yourself what a proportional controller provides, then, when you are missing something a proportional controller can provide, you use it.
https://en.wikipedia.org/wiki/PID_controller#Proportional_term

 

[garimella]

Hi CataM,

By adding a P to say a second order system , I find change in the transient response. But my question is what takes you from no controller to choosing a P controller, if quicker transient response is not the goal.

 

[FvM]

But my question is what takes you from no controller to choosing a P controller

You'll probably go back to the very first chapter of your control system literature about motivation for feedback controllers.

Stabilizing a process variable at a set point in the presence of disturbance values.

 

[LvW]

Hi CataM,
By adding a P to say a second order system , I find change in the transient response. But my question is what takes you from no controller to choosing a P controller, if quicker transient response is not the goal.

What, then, is your goal?

 

[garimella]

Yeh, I got the point. Control system's goal are three fold, Viz. Transient, steady state and stability. So With the feedback (with no controllers) if the goals are not satisfactory, then we go about different combinations to achieve the target. So this is clear.
This brings me to next level of clarification about the proof for PID controller math(particularly why P+I+D and why not P*I*D or any other combo)

 

[KlausST]

Hi,

why not P*I*D

If P is zero --> the output will be zero
If I is zero --> the output will be zero
If D is zero --> the output will be zero

At steady state D is zero .... how to get an output, then?

You may simulate PxIxD ... to see what happens. Play around to get a feel of it.

Klaus