I'd like to know what is the best thermal adhesive in terms of bonding and thermal conductivity regardless of electric conduction.I don't care if it conducts electricity, I just need it to be as strong bonding and thermally conductive as possible.
Does it bond ? I need it to be a strong adhesive too...I was first thinking of two siringe epoxy for metal...as transistors cases are made of epoxy resin ...
I could find this though: https://www.amazon.com/MG-Chemicals...ilver+thermal+adhesive&qid=1651000150&sr=8-13
Some years ago (ok, more than a decade), I often used Fischer Elektronik WLK (e.g. WLK 30) at work:
https://www.fischerelektronik.de/we...$catalogue/fischerData/PR/WLK30_/search.xhtml
This is an epoxy with metal oxide filler (non-conductive).
I found the 10:1 mix a bit difficult to manage in small quantities. Viscosity of the epoxy varied with my ability to get the ratio right.
Once hardened (and it always hardened well regardless whether I mixed it correctly), the joints made a pretty strong impression.
A bit pricey though IMO.
https://www.fischerelektronik.de/we...$catalogue/fischerData/PR/WLK30_/search.xhtml
This is an epoxy with metal oxide filler (non-conductive).
I found the 10:1 mix a bit difficult to manage in small quantities. Viscosity of the epoxy varied with my ability to get the ratio right.
Once hardened (and it always hardened well regardless whether I mixed it correctly), the joints made a pretty strong impression.
A bit pricey though IMO.
Interesting product.I wonder NOW if there's ever a chance to get the transistors off the heatsinks after bonding them with such materials 🙂
Not sure WHY would you want to Epoxy/permanently bond power transistors to heat sink, what´s the point?
There´s two completely different jobs to be fulfilled there:
1) Thermal paste is there to fill minuscule imperfections causing air gaps, and as such it needs to be fluid and flow under pressure, no rigidity advisable or allowed whatsoever.
To boot, power transistors suffer constant thermal cycling by definition, a fluid paste will keep doing its job for a long time (until it dries up, go figure) , but a rigid unyielding component? ... not that much,
2) fixing/keeping constant pressure/minimizing paste thickness are best done by mechanical clamping, at least by a screw head but also some clamp pressing transistor body against heatsink.
3) what will you do when that transistor fails and must be replaced?
To boot, old (classic) thermal paste can be wiped away with a paper towel, worst case you wet it with a drop of alcohol or some other solvent ... what will you do to fully remove hardened epoxy?
You might even scratch (damage) softish aluminum surface when trying to.
There´s two completely different jobs to be fulfilled there:
1) Thermal paste is there to fill minuscule imperfections causing air gaps, and as such it needs to be fluid and flow under pressure, no rigidity advisable or allowed whatsoever.
To boot, power transistors suffer constant thermal cycling by definition, a fluid paste will keep doing its job for a long time (until it dries up, go figure) , but a rigid unyielding component? ... not that much,
2) fixing/keeping constant pressure/minimizing paste thickness are best done by mechanical clamping, at least by a screw head but also some clamp pressing transistor body against heatsink.
3) what will you do when that transistor fails and must be replaced?
To boot, old (classic) thermal paste can be wiped away with a paper towel, worst case you wet it with a drop of alcohol or some other solvent ... what will you do to fully remove hardened epoxy?
You might even scratch (damage) softish aluminum surface when trying to.
Maybe solder the transistor to a clean copper heat sink or a copper spreading plate connected to aluminium heat sink?
Thank you very much for that! I already watched all his thermal inquiries videos and a few more.The mirror polishing surface intro on one of his videos made me think of an absurd coupling with nothing but pure atmospheric pressure , but i think I will use a mechanical method of clamping the transistor to the cooler just because i might need to replace a few transistors.
After looking all over the places I realised that I might better look for a graphite based grease for ballbearing use and there are some varieties that work up to 600 degrees Celsius , but I just found on the shelf of a car parts shop a copper grease that is aimed as an anti seize grease with a max temp of 1150 degrees Celsius .Its only other component disclosed being mentioned as thickener was bentonite, but I thought that it can't be bad if its stable and non corosive or reactive up to that temp.
I thought of asking your opinion on it as it's delivered in quite an impressive quantity for about 10 bucks.I think ot has some graphite in it too as its colour is dark brown...to black: https://www.commaoil.com/productsguide/view/6/220
I thought of asking your opinion on it as it's delivered in quite an impressive quantity for about 10 bucks.I think ot has some graphite in it too as its colour is dark brown...to black: https://www.commaoil.com/productsguide/view/6/220
Last edited:
Of course I so.I just tried to take the 1150 degrees mark as an indicative for better thermal transfer although googling I found that the other component , the bentonite is actually cooled volcanic ash and while very resilient to high temperature its conductivity which is not fantastic depends on how much water or oils it can "swallow"...so I can only help that the other undiclosed oils of the full recipee are good conductive materials cause it has only about 12% copper in it. For the moment I only bought some regular 130 degrees celsius graphite based grease and I'll try mix it with the fine copper powder I'll be grinding off the heatsink in the process of flattening it.
No questions thermal epoxy can make for a good heat transfer.
The die itself of a transistor is epoxied to the package already.
But I would argue that you won´t gain anything epoxying that (pretty flat) package again to some heatsink.
That´s where thermal paste or super thin gap pads or phase-change materials are simply better and much easier to work with.
If you really need to epoxy due to some reason you will have to watch thermal expansion coefficients of the two materials,
otherwise the connection can "break" easily. If you want to do it right (which you should), you will have to follow the recommended curing process which often is curing at temperatures >100°C and times ranging from 10min, to hours.
For high power applications use a paste that doesn´t dry out so quickly. The typical cheap white stuff does dry out in no time in high power applications.
The bigger the transistor package is, the more important it is to make sure the heatsink is flat too in those places where you push it against.
Sand the heatsink flat if it isn´t.
I personally like to use gap pads of ~100µm. Less messy than paste and pretty durable.
Plastic packages can warp, bend, even break and the front side might not be too flat either
which again would worsen the effective contact.
And finally and most importantly at least roughly calculate the junction temperature.
A transistor/MMIC with a junction temperature of 150° as an example won´t live very long.
The die itself of a transistor is epoxied to the package already.
But I would argue that you won´t gain anything epoxying that (pretty flat) package again to some heatsink.
Exactly. This and only the air gaps. (something lots of people don´t get right and apply paste liberally so that it literally lies on a bed of paste)
That´s where thermal paste or super thin gap pads or phase-change materials are simply better and much easier to work with.
If you really need to epoxy due to some reason you will have to watch thermal expansion coefficients of the two materials,
otherwise the connection can "break" easily. If you want to do it right (which you should), you will have to follow the recommended curing process which often is curing at temperatures >100°C and times ranging from 10min, to hours.
For high power applications use a paste that doesn´t dry out so quickly. The typical cheap white stuff does dry out in no time in high power applications.
The bigger the transistor package is, the more important it is to make sure the heatsink is flat too in those places where you push it against.
Sand the heatsink flat if it isn´t.
I personally like to use gap pads of ~100µm. Less messy than paste and pretty durable.
Again 100% right. Just don´t apply too much pressure. Again lots of people overdo this.
Plastic packages can warp, bend, even break and the front side might not be too flat either
which again would worsen the effective contact.
And finally and most importantly at least roughly calculate the junction temperature.
A transistor/MMIC with a junction temperature of 150° as an example won´t live very long.