Ping Tim Wescott

What's the input? Open loop step/ pulse response?

I made a pid loop pcb with terminal blocks for the R's.

George H.

The input is a control voltage from 0 to 10V, corresponding to 0 to 50 C.

I usually do the compensation using the delay plus integrator model--you stick a current step into the TEC and watch the response at the temperature readout. At early times it looks like a time delay followed by a ramp.

The model is

H_plant(f) = A exp(j 2 pi f t_delay) / (j 2 pi f tau)

where tau is the integrator time constant.

Then you apply a PI loop, generally just an op amp integrator with a resistor in series with the integration cap:

H_amp(f) = (1 + j 2 pi f tau_zero) /j 2 pi f tau_int

You can solve this analytically to a reasonable approximation, given the desired phase margin, but it isn't clear that the customer can easily handle that even if I give them the expression--they're smart people, but mainly MEs and biomeds.

Besides, we haven't heard from Tim Wescott in months, so it was a good excuse. ;)

Cheers

Phil Hobbs