Thermally conducting potting compound

I fitted an electric fan in place of the mechanical one and fitted a pair of temp sensors in the top hoses to trigger it. After more than 10 years those sensors have failed and I want to replace them. They were plumbed in as short aluminium tubes and have a small box which housed the sensor and switch controller. I just need to install a new sensor, the control relay will be elsewhere. The sensor needs to be potted into the small box but needs to be able to transfer heat from the aluminium tube. Will regular epoxy resin work, or is there some special compound I should use? When digging out the old controller the potting compound they used was slightly rubbery.

You do need to be a bit careful what you use to pot electronic components. It is important that they are not stressed by the epoxy curing, or from thermal cycling within the epoxy. And it must not be conductive or absorb or retain moisture.

https://www.digikey.com/products/en/fans-thermal-management/thermal-adhesives-epoxies-greases-pastes/220?k=potting%20compound

If you follow the link you will find examples of available compounds that are designed specifically for what you want. There are other manufacturers available, 3M and Loctite spring to mind.

The sensor is a standard NTC thermistor, it is already encased in a metal can and waterproofed. Some searching around says standard epoxy is not going to be good for heat transfer. I did find this: RS PRO White Chemical Resistance, Water Resistance Epoxy Potting Compound, which looks promising. But there is no offset adjustment in the temperature controller, so whatever I use, there is going to be some transfer loss - I’m unlikely to get a true reading on its digital display anyway.

So, rereading what you have posted, the NTC thermister is NOT directly interfacing to the coolant? I take it you have it touching the metal pipe and relying on thermal transfer and thermal inertia.

You could probably get away with using araldite and gluing it to the metal, then potting the whole thing. I wouldn’t trust the potting compound to be thermally conductive enough. You can get thermally conductive compound, but it is intended only to help pull heat away from components in power supplies while maintaining electrical isolation. I have never observed any that work particularly well. This is the reason why cheap power supplies for your indoor lighting fail so rapidly. The guts run too hot.

There are NTC thermistors with a eyelet mount.image or even clip on: image

Here’s what I have already installed, it’s an aluminium tube that you splice into the radiator hose. On the right is the A bank top hose which still has a working sensor. You adjust the pot to set the turn on temperature. On the left is the failed unit in B bank which I have dug out, and put in place the new thermistor, it is very similar to the metal tube type in your photo.


Internally the old system used an AD22015 programmable thermostatic switch on a little circuit board, potted into the rectangular housing. Setting the temperature was all relative so it wouldn’t matter too much what the heat transfer was like, just that it was consistent.

The new sensor will sit in the same space, as close to the aluminium pipe as I can get it.The controller circuitry is more much more sophisticated and allows monitoring of the actual temperature and setting specific start/stop temperatures for each bank.


It would be nice if the temperature displayed accurately reflected the true temperature, but I guess I can live with manually applying enough offset in the start/stop temperatures to compensate. Otherwise I can get a more modern version of the hose adapter that has a screw thread for an immersive NTC, which should work better.