PID Control on FPGA - LabVIEW 8.6.1

[srm@viewpointusa.com]

What is the response of your system under control?  how fast does it react to a step change in output.  What is your required response bandwidth?  It would seem that if you want it stable, use a low P and tune mainly with I.  leave D 0.

[trouinky]

It's tuned, but the device sits in an airflow stream.  Sometimes it needs to be heated in a zero flow condition and sometimes in a high flow condition (when the PID really needs to kick).  I've got it where the PID works in both cases (D set to 0 for sure), but eventually I'll put a flow sensor in there and change the PID gain settings based on whether there is flow or not.  Now that will be something!

These aren't really RTD's that I'm heating, but they are close to those . . . thermal time constant (open loop, no PID) is about 40-60 ms (response to step input of say, 5 V).  With the PID tuned really well at zero flow I get that down to single digit ms, and under flow maybe 15-20 ms . . . we need to heat these as fast as possible, hence the need for the PID.  Cooling is not an issue, although we are designing actively cooled devices as an extension to the project.