AnsweredAssumed Answered

vrf PID autotune/advice

Question asked by VRFuser on Feb 22, 2008
> I have used VEE to design a temperature control system for a
> resistance heated furnace. I have written a program such that the
> temperature ramps up to a setpoint, T1, and then when the temperature
> is (approaching!) stable a cycling program kicks in and the
> temperature cycles between two setpoints, T1 and T2. I have used an
> adapted version of the PID controller available with the VEE software
> to attempt to smooth out the approach to each setpoint*.
> It is with the PID control that I have been experiencing problems: 
> not only is the trial and error method time consuming, the performance
> of the furnace varies with time (heating of
> electrodes) so PID values from early cycles often become redundant
> later on. I've been wondering about an autotune system, but I don't
> really know where to start. Can anyone offer me any advice?

This is not a simple problem. I would imagine that a furnace is slow enough that VEE could be used to set up a reasonably good controller. There are certain rules and recipies to set the parameters of a text-book PI(D) controller, which involves measuring the oscillation period if the gain is set too high on purpose and using this as time for the integration. Afterwards lower the gain until the oscillation ceases and half the gain. In my experience with temperature controllers any derivative action does not really help alot, so staying with a PI controller is in most of the times good enough.

I do not have any experience with fuzzy logic but I hear there are certain benefits for complex systems that are hard to describe and possibly time varrying. So maybe digging into fuzzy logic might give some interesting new ideas.


PS What sometimes works is to create an array of PI-values in the above described way for different target temperatures and modify the controller as the actual set point or actual temperature changes. This way you always have the "best" controller in place.

You are currently subscribed to vrf as: To subscribe please send an email to: "" with the word subscribe in the message body.
To unsubscribe send a blank email to "".
To send messages to this mailing list,  email "". 
If you need help with the mailing list send a message to "".
Search the "unofficial vrf archive" at "".
Search the Agilent vrf archive at "".