M-Labs TEC with 10K RTD #32
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Does M-Labs have a test setup in the lab for testing Thermostat? A setup consisting of the following would be helpful at checking out device performance and perhaps building simple human-in-the-loop tests.
Yes. Why?
A toy setup would be helpful at firmware validation. Basically, helps at finding bugs before end-users do. Examples...
does documented dt permit application of Ziegler Nichols approach at PID tuning? #11 (comment)
swapped RTD inputs -- resolved in #12
I sometimes see offset between target temperature and actual temperature (pid engaged). Haven't posted bug report because its not yet reproducible on my end.
would get developer using pytec; may reveal certain ease-of-use issues
it's fun!
Here is a test setup for the thermostat. A small block of aluminium is stacked on top of 3 TEC modules, only the one right underneath the aluminium block is wired to the thermostat, the other 2 TEC modules are wired to a constant current source set to 1A. The 10k thermistor is secured inside a hole drilled into the block being cooled. A aluminium box houses the above mentioned components. A large heatsink is adhered to the aluminium box, with a fan cooling the entire setup.
With a few small changes to the current firmware, the setup is able to control the aluminium block temperature and hold it to +-10mK within the set temperature once settled, according to the device temperature readout. PID loop is tuned mostly by intuition and following a few rules of thumb.
I've attatched a few photos showing the test setup, control loop response to a 40c -> 10c -> 40c setpoint change, as well as a graph showing temperature stability over a few minutes when trying to hold 40c (y-axis is deviation from 40c). Ambient temperature in lab is around 24c.
The changes to the software will be cleaned up and submitted for a pull request on sat.
Looks like a nice setup. And that's a good result for stability. Definitely better than what I've seen in my own setup. I typically see 100 mK noise and several 100 mK temperature offset from set point. Looking forward to taking a look at your patches when the post on Saturday. Great results!
@topquark12 Did you merge the changes you mentioned?
@jbqubit Yes, changes already merged into master. See
7c013ff4a4ande94601f54f