PID integral implementation

[Reverant]

The current code reads:

Code: Select all

Ki = ((((long) error * looptime) / (long)pid_divider) * (long)set_Ki);

The problem with this implementation is that it's essentially a P term (as I understand it, and it certainly acts that way), whereas an I would be more like:

Code: Select all

overall_error += error;

if (overall_error > integral_limit)
    overall_error = integral_limit;

Ki = overall_error * set_Ki;

Thoughts?

[muythaibxr]

Incorrect. I have answered this question over and over again. It is done this way because the whole PID equation is being summed with the current output every time through the loop, effectively integrating the whole thing. So you have to take the derivative of the whole thing to make it work.

It does not act like a P term in that over time if there is error it continues to correct.

Depending on what code you look at we use type B or C.

http://bestune.50megs.com/typeABC.htm