I work in an AMO Lab, and I have been tasked with creating an experiment enclosure that will keep the air (and instruments) stable within about 50 mili-Kelvin with a set point of about 20C-21C. Our lab is almost always at 23C+, so we will 99% of the time be cooling down the enclosure, not heating it up.

I have two main ideas that are taken from what some others have done before:

  1. Use a water chiller to overcool the water, then use a PID controlled water heater to quickly heat the water to the needed temperature. Then, pass that temperature-corrected water over a radiator with some fans and that should stabilize the box.

  2. Use Thermoelectric Coolers (TECs), which are cooled by a water chiller, and a PID circuit to control the current going through the TECs and therefore control the temperature.

These two systems do indeed work, so I am not asking for feedback on these designs.

I would like to know what other designs people have come up with for stabilizing their experiment enclosure temperature. Did you use similar designs, or something completely different?

Thanks for the help!

  • 1
    $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$
    – Community Bot
    Oct 18, 2021 at 20:33
  • 1
    $\begingroup$ How big is the enclosure? What sorts of instruments? Do they all need to be stabilized, or only some of them? $\endgroup$
    – Jon Custer
    Oct 18, 2021 at 21:05
  • $\begingroup$ I'd question whether you really have to have the setpoint just below room temp instead of just above - heating is easier than cooling. You're hopefully close enough to room temp to avoid condensation, but over-cooling will require something colder than the target temperature. $\endgroup$
    – Chris H
    Oct 19, 2021 at 8:07
  • 1
    $\begingroup$ Also: how much heat are you trying to get out - "instruments" implies power dissipation? $\endgroup$
    – Chris H
    Oct 19, 2021 at 8:07
  • 1
    $\begingroup$ And one more: how important is the temperature distribution within the enclosure? Will you be measuring at multiple points? $\endgroup$
    – Chris H
    Oct 19, 2021 at 8:09

1 Answer 1


Just a basic trick: Don't PI-control your cooling. Instead, over-cool and PI-control a heater that counter-acts your cooling. Cooling typically is difficult to control, whereas heating is trivial using ohmic (power-)resistors. So, much easier to do it that way around.

Also, don't forget your safety controls, malfunctioning heaters are a fire hazard. You may consider some form of an alert system. Think about what happens when the thermometer comes off and now has a worse thermal coupling to your heater, will your heater then heat all the time? That could be bad.

@jwenting in a comment correctly points out that redundancy may be important for your application. Thermal control failure can be an issue more readily than you think, for example, you may have a relay that will easily switch every few seconds. At $10^7$ seconds a year that may fail within months. So think about a maintenance plan, and/or a redundant setup, and/or invest in quality products.

  • $\begingroup$ and if it's that critical, have multiple backups for everything that automatically come online when the primary fails. $\endgroup$
    – jwenting
    Oct 19, 2021 at 7:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.