[volt-nuts] Temperature controller for ovenizing and temperature cycling

[Jan Fredriksson]

Maybe I should not be posting this on the VN list, as it is only indirectly
related to volts.

Still; I have been getting some messages and feedback, with no disrespect
to me, asking, if I know what I am doing. With no disrespect to the people
who replied, of or on list (on the contrary, I'm grateful for feedback!):

My background: I have been working for 15 years with thermal engineering
and measurement techniques.  I have calibrated in liquid baths,
ovens, against reference probes. I have calibrated reference probes in
water triple cells, and gallium melting pots, (both are ITS-90 temperature
scale reference points) calibrated reference probes etc. I know it its hard
to do measurements with 0.1C accuracy in real life, sometimes even 100C is
very difficult, ie at very high temperatures. Absolute uncertainties at mC
are beyond all but very qualified calibration labs.I also know that
measuring temperature differences in time and space CAN be very accurate,
if conditions are optimal. I frequently measured in stirred water baths
that had a stability of around 0.005Crms overnight (checked with PT100
reference probes, and actually measured with 0.001 resolution, but with
nothing near the same temperature uncertainty) with thermocouples.
The reference points where huge water-ice slurry Dewars, the thermocouple
measurements where done with Keithley Nanovolt meters and the PT100 where
measured with a reference bridge).

I am not an electrical engineer, but come from mechanics and thermal
engineering. My PID / control loop maths are now a bit rusty but I have
developed amplifiers for highly capacitive loads before. By searching new
methods and ways, I have also more than once developed practically usable
measurement techniques that people in advance told me where almost
impossible ;-)

The circuit I am working with, ONLY shows temperature of a sensor, heated
by resistors, taped together, under a shield. I am aware that this is a
serious limitation. Adding any mass to the circuit thermal feedback loop
will be seen as an increased capacitive load to the circuit, so it will of
course be harder to stabilize. Air convection, in and around the final
solution is unpredictable and can dramatically change the heat transfer and
temperature.

I am mainly thinking of a heater that can keep a small circuit at a
decently stable temperature, by which I mean in the order of 0.1C-0.01C. I
am not really aiming for mC stability in a real life application, it was
something I got, a bit to my surprise, for this circuit which is still on a
breadboard level. But I think that a first board / circuit stable to the
mC level is a decent start and I do think that kind of stability is
possible in a very small scale.

I think I will not be posting more on this until I have a complete working
solution which may take a month or four...