Elastomers, thermoplastics, heatsink Isolation?

[HellTriX]

I'm trying to find some electrically insulating but thermally conductive material to mate two heatsink surfaces together.

For instance a think flat copper bar with heatsinks on it carrying a positive voltage, that I would like to stick onto water blocks or another larger heat sink but have the two electronically isolated.

I know there are little mica type mosfet insulators but that will not work in this case and would be very expensive if I used a lot of them for the heat transfer.

I have seen a rubber material between cpu heatsinks before and some other switch mode supply mosfets. Its a gray rubber elastomer, but I have no idea where I can buy sheets of this stuff.

Any ideas on where to get something like this? ty..

TriX

 

[Sceadwian]

I'll have to watch this thread, I have some little sheets of that stuff I got from a power supply was wondering how it compares to heat sink compound. Thermal transfer equations are a bit of black voodoo to me.

 

[jpanhalt]

What surface area do you need to cover? Master-Carr has 3"X5" sheets of mica for a little over $2USD (#8802K16). McMaster-Carr is often more expensive than what you can find shopping around.

John

 

[dknguyen]

Is the elastomer thermal adhesive? Or phase change compound? If it is either, do not depend on them for electrical insulation even though they may be insulative, since they are only designed to fill in gaps and let the material interfaces touch whever they can.

What about thermally conductive epoxy?

 

[HellTriX]

What surface area do you need to cover? Master-Carr has 3"X5" sheets of mica for a little over $2USD (#8802K16). McMaster-Carr is often more expensive than what you can find shopping around.

John

This would probably work, I could cut it in 1.5" strips. and use 4 of them for a 1.5"x10" seconds times 2.
I'm glad to see the $2 usd. was afraid it was gonna go above $10 for that size of sheeting.

 

[jpanhalt]

@dknguyen: The elastomeric heat sink I have seen has fiber (fiberglass?) reinforcement, presumably to increase thermal conductivity and address squeeze out. The Loctite 383/384 series are cyanoacrylic adhesives. I am not sure what the difference in terms of performance would be between them and epoxy. The epoxy might even have a lower Tg than the allowable temp for the Loctite material, which is 150°.

I have seen strips/rolls of mica. McMaster-Carr was just given as an example. It has several sizes available.

John

 

[jpanhalt]

An added thought. I got some very thin PCB material on ebay a few years ago. It measures 0.006" thick, including 1/2 oz copper (I think it is 1/2 oz, not sure). The copper could be against the heat sink for great thermal conduction, once the heat got through the FG. John

 

[HellTriX]

TBH, I considered some transformer paper soaked with some oil, but I haven't done any testing to see if this would transfer enough heat and be electrically isolated.

My mosfets are not gonna get extreamly hot or dissipate much power. There will be large peaks but its in very short durations under a second or two so even a marginal thermal conduction to a good heat sink would be acceptable.

Whats the best search term to use when looking for sheets like this?
I find tons of other mica things, like capacitors, tiles, you name it.

 

[jpanhalt]

I don't think I would use oil, as it won't stay put. Thermal grease is an option. You might consider something like 4 oz FG cloth or 6 oz satin weave FG with the thermal grease. I have never tried anything like that myself. I suspect it might be messy.

For mica, I just searched on mica sheet or mica film. John

 

[Sceadwian]

Sounds like you should have just bought isolated FETs, they're only marginally more expensive than non-issolated. The fiber glass cloth with thermal grease sounds like a good idea though, if messy.

 

[HellTriX]

Sounds like you should have just bought isolated FETs, they're only marginally more expensive than non-issolated. The fiber glass cloth with thermal grease sounds like a good idea though, if messy.

Sounds like something I might try. Just gotta make sure the thermal grease is non-conductive. I have all the supplies. Might be slightly messy but everything will be bolted up with insulated bolts so it should stay put good.

And I was really limited in my choice of mosfets, so adding another variable like insulated mosfets would have been harder. Not to mention I'm doing these in parallel so I'm going to use the copper stock to bolt the mosfets on conductively. Then cool the copper rod with an insulated connection to a larger heatsink. The reason is this is for very high current and the copper rod is going to be the current flow as well as the main heat transfer. But I also want to isolate the case from the Positive voltage for protection from the 250 volts.

 

[dknguyen]

Why do you need to isolate them again? If they are all in parallel working together, then you shouldn't need to isolate them since the tab connections are the same for all of them in the package and the electrical connections? Or do you have multiple phases all mounted on the same chunk of copper?

THe info in your OP doesn't say definitively whether or not you actually need to isolate them.

Are you planning to use a metal case or something? I would think a plastic case would be much safer if you are worried about safety. Or IS your case your larger heatsink? Because you were just talking about water cooled blocks a second ago which doesn't make a lot of sense if your case is your heatsink.

 

[HellTriX]

This is a large case, gotta be durable as it will be exposed to some elements. It will be 1ft by 2ft long and maybe half foot high. If I use water blocks for cooling I don't want the water explosed to 250V because of electrolysis. But since 20 mosfets will be electrically connected via the copper bar 1/4"x1.5" by 2 feet long. I need to cool the copper bar. So I'm going to use a heatsink OR
water cooling. In either case I want whatever cooling device to be isolated from everything else. Only the copper bar and the mosfets will see this voltage..

To me it doesn't sound hard. Was just looking for some isolation material.
And the case will almost definitely be metal for durability

 

[jpanhalt]

Why are you using water for your coolant? Ethylene glycol is well established and less corrosive, and there are others as well.

Are you planning on soldering the tabs to the copper or using screws?

John

 

[HellTriX]

was thinking screws. Not sure I could solder them without damaging them?
Copper that thick may not get hot enough fast enough with my 100watt gun.

 

[dknguyen]

Use a toaster oven. More even heating with less peak heat, and better contact.

 

[HellTriX]

Thought that would be a bad idea. But then I looked at the datasheets.
Guess if its done right they are not exposed to too much heat for too long.

 

[Sceadwian]

An electric toaster oven is almost exactly the same as an IR reflow oven, the only difference is the type of element, and that might not even be different. If you look at reflow and wave soldering, they're stupid simple. Reflow is following a temperature profile in a controlled chamber using infrared heaters, and wave soldering is nothing more than dipping a board in a molten solder bath. Aluminum foil masking for the 'spike' part of the reflow heating cycle which is what does the work may be enough to keep sensative components from being damaged.