PID-Tuning, rule of thumb?

Question

Dear Colleagues,does anybody remember the following rule-of-thumb for PID controllertuning?  :1. set all parameters kp, ki, kd to minimal influence (0 or 1,depending on the controller realization)2. increase kp until the control loop starts oscillating(damping behaviour of the differential component)3. increase kd until the control loop stops oscillation4. increase ki until there is no control error anymoreDoes that make any sense?Is that a common rule of thumb?Is there any book, where this is written in?Any comment is greatly appreciated,bestTilo-Tilo Gockelgockel2000 at ira dot uka dot de//remove the 2k

John Popelish · Accepted Answer

You have the general idea, except for the integral component. Any value for the integral gain below that which causes oscillation will, eventually, bring he system to zero average error.

I have written a nontechnical tutorial on how to tune PID controllers based on watching a trend line of the input and output. It may give you some insight on how to recognize too much or too little of each component of the PID control. It is available at:

formatting link

Tim Wescott · Answer

Steps 2 and three presume that differential control will be a benefit -- it isn't always. My preferred sequence is:

Set Kp to the minimum amount necessary to get things working in the correct range (none is good, if possible).
Increase Kd until the onset of oscillation, then back it off by a factor of two or three.
Increase Kp until the onset of oscillation, then back it off by a factor of two or three.
Increase Ki until the onset of oscillation, then back it off by a factor of two or three.

DON'T GO BACK FOR ANOTHER ROUND.

Go to

formatting link

for a pointer to an article on how to implement and tune embedded PID controllers.

PID-Tuning, rule of thumb?

Join the Discussion

Didn't find your answer?