Sourcing a thermally conductive epoxy or glue

[Franz Zinn]

Hi

I am trying to DIY a Temperature Control experimentation kit like this one http://apmonitor.com/pdc/index.php/Main/ArduinoTemperatureControl. It is basically a TIP31C power BJT with a TMP36 temperature sensor glued to it. I am trying to find out what would be a good thermally conductive epoxy/glue for this purpose. The kit actually uses some thing that has a Thermochromic pigment that is gray but turns pink at 37C, though this is probably not an essential requirement for doing the process control experiments.

I would be grateful for your suggestions. I am looking to source something inexpensive in the EU/UK.

regards

[cdev]

I have some heat sink glue here and I'd be happy to test it for you if you thought of a good procedure for me with stuff that I have. I have a meter that has a thermocouple. I dont have anything else that adequately can be used in that manner (a direct reading thermal meeasurement)

I don't know how one would test this glue that I've had for years and which may have even dried out. The last time I used it was around a year ago, and it seemed to work. (it was used to stick a transistor and heat sink on to a power supply with assistance from some screws. )

The manufacturer is not marked. A long time ago I got it on dealextreme.com, a Chinese electronics and USB stuff site. It is a little tube that I think says Heat Sink Glue on it, which describes it well. Its supposed to be spread very thin and it dries hard.

Its unlike the other thermal interface materials I have in thatit is, unlike the others, supposed to dry out, the huge tube of silicone I have for heat sinks, bought at Frys in the bay area is a generic silicone that is supposed to be spread on thermal interfaces and I think it is not supposed to dry out. I have a large toothpaste size tube and indeed, it does stay tacky a long time. Its not a good glue.

On computer CPUS I usually use something like Arctic Silver, which stays tacky and is labeled as dangerous to ingest or breathe.

Hest sinks that use thermal pads, like 3M's thermal pad stuff, tend to move if they often get really hot. For example, inside a computer, thermal padded heatsinks I applied to a video capture card moved because they stayed hot for a long time. This made me fear that they might fall off, so I removed the card.

I also wasn't comfortable with this happening inside my Raspberry Pi, the heat sinks moving was not something I could live with, so I actually removed the heat sinks rather than have them moving around. I removed the overclock that made me apply the heat sink. The clock jitter is very low now which was the most important thing on that PC for me.

[Ian.M]

You'd only need a real thermal epoxy for metal to metal heatsink bonding.

The thermal conductivity of both parts epoxy encapsulation isn't that great, so the effect of the thermal resistance of the joint will be negligible as long as its thin and void-free.  Any hardware store Epoxy or even superglue would do the job.

Do you really need to glue it? I'd simply flip the TMP36 so its face to face with the TIP31C to get maximum contact area and assemble with a dab of ordinary heatsink compound.   Putting another TMP36 on the other side, in contact with the heatsink will be a useful exercise as you will see a significant difference in their response speed.

[cdev]

You might want to look at the web site for that company that makes LCD materials that change color with temperature. I forget their name right now but they are in England and their site should be easy to find. Its Fair-something. They might have something perfect for that kind of application.

The sheets of material they sell work extremely well for finding shorts on PCBs without having to touch potentially hot or electrified components.

[Franz Zinn]

Thanks cdev and Ian.M. I was under the impression one would like to avoid any voids, because the behavior with a void would be more complex. This was the reason I was thinking of thermal conductive epoxy. I have a hot glue gun with me. But the TIP31C can get pretty hot and that will certainly melt the hot glue.

Also I was thinking by covering temperature sensor with the epoxy bonded to the BJT will ensure maximal heat transfer so the sensor will respond faster to changes in BJT current making it a easier system to learn process control with :-) . But I am going to try just putting them in close contact and see how well it works (i am waiting on my tmp36 to arrive). The LCD material sounds like a nice idea.

regards

[jpanhalt]

If you want to improve heat conductivity of "epoxy" without a lot of mess, you can mix in boron nitride (BN).  It is an excellent thermal conductor and good electrical insulator.  It is also slippery like graphite without the messy black color.

Lots of studies about its use.  Here's one: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6723785/

I have it for that reason as well as a lubricant.  The powder I have is light tan.

[Franz Zinn]

Thanks. Boron NItride is actually available on ebay 10g pack as a lubricant. Nice. So I do have couple of options now to try with their sample lab code

- Just juxtaposition of sensor and bjt
- Using super glue
- Using thermal compound
- Using Boron Nitride mixed with superglue or some common epoxy.

Thanks.

[geggi1]

Loctite got some stuff that I used about 15 years ago