good high temperature epoxy

I received some switching power modules which I want to build into some projects. Some of then may need to deliver a couple of amps, at 5 v., dropped from an input of about 30 v. I know that switchers don't generate as much heat as linears, but I do want to attach some heat sinks in some cases. These came with small heat sinks with some stickum, but it isn't very good stickum.

So I want to use a good epoxy to attach the heat sinks, which means it

  1. should be able to handle up to100 Deg. C,
  2. be a fairly good conductor of heat.

Anybody have any experience along these lines?

Thanks in advance for any ideas.

I use two part "J B Weld Steel-Reinforced Epoxy" for glueing my heat sinks. 500'F

Any two part epoxy adhesive will stand 100 C. Use a thin one to get good heat transmission. I use West System epoxy as I have plenty left over from boat repairs but any brand will do. Make sure the surfaces are grease free. A quick wipe over with acetone will suffice.

Russell.

You can buy thermal adhesives that are designed for the purpose. "Arctic Silver" and "Arctic Alumina" come highly recommended, as does "Loctite 3875".

There's a bit of info here:- Attaching Heatsinks

Thanks,

I regard JB Weld as a miracle product for nearly everything, and I always keep some around the house. I’ll go with that. It kept my daughter’s car radiator operational for about three months after if split.

If you are talking about epoxy that needs to conduct heat (ie, between the heatsink and what is being cooled), you cannot use normal epoxy. It is not a good thermal conductor.

I have used both Arctic Silver and Arctic Alumina (note - be sure to get the epoxy kind, not the normal thermal grease - both are much more common as thermal grease than epoxy) - they work pretty well.

jrdoner: I regard JB Weld as a miracle product for nearly everything, and I always keep some around the house. I'll go with that. It kept my daughter's car radiator operational for about three months after if split.

Same here. I used it once on an old Bronco I had to hold an anti-backfire valve together; I couldn't find the valve anywhere, so I JB'd that sucker, and bolted it back onto the exhaust manifold - it held together no problem for the 10+ years I owned the vehicle.

My brother-in-law, though, probably has the record. He had this old Ford 14 wheel dump truck that he ran constantly - he drove it more than any other vehicle he owned, I think. Anyhow, one day I went over to his house and he had what appeared to be half the engine apart. Well - apparently he had a crack in the blower housing. He showed it to me; I asked "what's that other stuff"? He told me that the housing was made of aluminum, and he didn't have a way to weld the cracks, so he JB'd the cracks.

But get this: The cracks he had to fix were new cracks elsewhere on the housing - those JB'd cracks had held up under those grueling engine conditions over I have no idea how many years (maybe decades) and miles. He finally got rid of the truck when the cylinder sleeves slipped and he didn't want to put any more money into it. As far as I know, though, all the JB work was still perfectly functional.

So yeah - you could say I'm sold on the product. I've found that to get the best hold, though, you have to have proper surface prep, and a really good half-n-half mix - and to mix it really well (also, a hot dry day is best - we have plenty of those here in Phoenix).

Hi,

[soapbox] The important thing with fixing heatsinks to components is the medium you use between them.

If you use nothing, you have the best thermal contact, IF both surfaces are perfectly flat and in full contact.

However the world is real and most of the time you will have gaps, air pockets, these act as thermal insulation. So we have thermal grease, it fills in the gasp, NOTE it is not as good as metal to metal contact but better than still air. But people smother the surfaces, so most of the time the entire contact area is covered in grease, ending up with probaby a greater thermal resistance than no grease at all. You are supposed to smear it on as thinly as possible, I use a blade to get a very thin coating, I push the surfaces together, then pull them apart and check the coverage.

Gluing heatsinks is probably the worst thing you can do, the minimum thickness will be more than grease, its thermal transfer characteristics are questionable. Even the cheap chinese knockoffs don't use glue.

A solution is to clamp the heatsink on with minimum of grease, even if you have to make a metal clip over the heatsink and component and then glue the ends of the clip to the PCB. [/soapbox]

Tom..... :)

TomGeorge:
Hi,

[soapbox]
The important thing with fixing heatsinks to components is the medium you use between them.

If you use nothing, you have the best thermal contact, IF both surfaces are perfectly flat and in full contact.

However the world is real and most of the time you will have gaps, air pockets, these act as thermal insulation.
So we have thermal grease, it fills in the gasp, NOTE it is not as good as metal to metal contact but better than still air.
But people smother the surfaces, so most of the time the entire contact area is covered in grease, ending up with probaby a greater thermal resistance than no grease at all.
You are supposed to smear it on as thinly as possible, I use a blade to get a very thin coating, I push the surfaces together, then pull them apart and check the coverage.

Gluing heatsinks is probably the worst thing you can do, the minimum thickness will be more than grease, its thermal transfer characteristics are questionable. Even the cheap chinese knockoffs don’t use glue.

A solution is to clamp the heatsink on with minimum of grease, even if you have to make a metal clip over the heatsink and component and then glue the ends of the clip to the PCB.
[/soapbox]

Tom… :slight_smile:

That is interesting - can you cite any references or calculations to back up that claim? If this is true, it would demand significant changes in several planned projects.

Hi,

That is interesting - can you cite any references or calculations to back up that claim?

30 years experience and being taught that way.

https://www.youtube.com/watch?v=rjF5jabXRCY

https://www.youtube.com/watch?v=nfDsDLlj_Uo

Thermal Conductivity Table

A Spec Sheet

Spec sheet shows thermal conductivity of Super Thermal Grease = 3.0 W/(m.K) Table Air = 0.024 W/(m.K) Aluminium = 205 W/(m.K) JB-Weld = 0.59 W/(m.K)

http://www.farnell.com/datasheets/218722.pdf

The thermal greases are now where near as good as the heatsink material, but when you keep the thinkness waaayyy down to a minimum is when it performs.

Remember the grease is only there to fill in the air gaps.

Tom.... :)

Interesting. Looks like I could lap the two pieces together with some really fine grit alox to the point they stick and call it done. Better mating surfaces and the alox has a much better rating than the paste.

Hmm . . . I wonder if a thin liquid cyanoacrylate adhesive might actually be better than a thicker thermal grease for good flat surfaces?

Russell.

outofoptions: Interesting. Looks like I could lap the two pieces together with some really fine grit alox to the point they stick and call it done. Better mating surfaces and the alox has a much better rating than the paste.

Quite true, IF you can get the surfaces clean from the alox. The reason for the heat conducting compound is a business cannot take the time to ensure both surfaces are really flat and will mate with full contact. IF you have the time and patience, your way is much more efficient, but you can't make a commercial product that way.

The commercial way is to use a large fly cutter in a mill to make the surfaces flat, but with small grooves from the cutter. Then fill in the grooves with heat sink compound and finish building the product and get it sold!

Paul

russellz: Hmm . . . I wonder if a thin liquid cyanoacrylate adhesive might actually be better than a thicker thermal grease for good flat surfaces?

Russell.

In my experience its not very secure compared to epoxy. cyanoacrylate is great for softer materials, but metal joins are brittle to knocks.

Possibly the iideal material would be diamond grit, clamped - this would dig into the metal surfaces and create a very low resistance path for heat (diamond beats any metal). Expensive though!

Soft metal foils are either expensive (gold) or poisonous (lead, cadmium), so aluminium or copper-filled epoxy would be the next best thing, but you still want a very thing layer clamped tightly to get individual particles of metal bridge the gap as much as possible.

Paul_KD7HB:
Quite true, IF you can get the surfaces clean from the alox. The reason for the heat conducting compound is a business cannot take the time to ensure both surfaces are really flat and will mate with full contact. IF you have the time and patience, your way is much more efficient, but you can’t make a commercial product that way.

The commercial way is to use a large fly cutter in a mill to make the surfaces flat, but with small grooves from the cutter. Then fill in the grooves with heat sink compound and finish building the product and get it sold!

Paul

According to the linked reference the alox was MUCH better than thermal paste. Why clean it off? That was part of my point. I used to do telescope mirrors so I have plenty lying about.

outofoptions: According to the linked reference the alox was MUCH better than thermal paste. Why clean it off? That was part of my point. I used to do telescope mirrors so I have plenty lying about.

Because the alox (aluminum oxide) particles are embedded in the aluminum and keep the two pieces from making an intimate contact for heat transfer. I assume the telescope mirrors are glass with aluminum sputtered on the front surface. Glass is hard and will not allow the alox to be embedded.

Heat sink aluminum is really soft.

Paul

MarkT: In my experience its not very secure compared to epoxy. cyanoacrylate is great for softer materials, but metal joins are brittle to knocks.

It does require careful preparation of the surfaces but I have used it for holding small brass parts onto a faceplate for machining in the lathe. A more modern approach than the traditional wax chuck used in clock and watch making.

The high strength grades of Loctite retainer (urethane methacrylate) might be better though.

Russell.

I've read all these comments, and it gives me an idea. I'm going to use alox over most of the mated surface area, and then glue the heat sink to the regulator over just a small bit of the surface. I'll try it first on some other regulator chip, like a 7805, and drive it up to relatively high heat to see how it works.

Hi,

jrdoner:
I’ve read all these comments, and it gives me an idea. I’m going to use alox over most of the mated surface area, and then glue the heat sink to the regulator over just a small bit of the surface. I’ll try it first on some other regulator chip, like a 7805, and drive it up to relatively high heat to see how it works.

Good idea, time to try it in the real world, but KISS.
The higher the temp of the heat source , the larger the resultant differential temperature across the interfaces, the clearer to indication of thermal conductivity.
Tom… :slight_smile: