another price point

Well, that would be a rather ineffective use of such a material.

That's the high conductivity blend. It's also thick, and goopy. You'd be an idiot to try and squish it behind a TO-220, there'd be nothing left.

You can, however, get a fair amount of heat out of a DPAK just by smooshing it over one. If you absolutely must do that, you might consider buying it.

Same goes for this stuff, but sideways.

It's significantly worse than an equal cost of copper foil by area, but on the condition that you only have 100 microns overhead to work in, you might consider it.

Or you can cut a small square and levitate it on top of some NdFeB magnets, that's fun I guess.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

I want to insulate and heat sink a TO-247 with minimal capacitance. It will only be 10 watts or so, so AlN isn't practical... the numbers don't work very well. I'd basically need an aspirin sized insulator.

The higher-end sil-pad types claim 6 or even 8 w/m-k, and dielectric constants in the 3-4 range, which does the job for around 3 pF. The cheap sil-pads can be as bad as 0.05 w/m-k.

Problem is, the people who make this stuff are chronic liars. Spec'd thermal conductivity is seldom seen in real life. I'll have to get samples and test them.

I don't understand the conductive carbon pad stuff. Why add anything between things that you want to have in thermal contact?

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Nothing has a lower dielectric constant than air. 10W and a small tab heatsink and a blower would do.

Kind of noisy and requires moving parts, but hey.

I have no reason to doubt a product that costs $10/in^2. Someone must be buying it, and they wouldn't be buying it at that price for very long if it were crap.

The squishy pads are good material, even at (a very believable) 1 W/m/K. They tend to be self-adhesive, even without an actual coating of adhesive proper. The one-side-tacky kind still need to be clamped, but the clamping force is much less than the old firm sil-pads, and the conductivity is better (both in bulk and in use under a device).

I've removed sil-pads from equipment where the imprint on the device and heatsink was maybe 50% coverage. Way too textured and firm.

The modern squishy kinds are much too squishy for screw fasteners. You need to think about alternative methods, like a bracket or clip with a light spring holding things in place, or (gasp) just leaving it sit there under its own self-adhesion.

Typical laptops have a few squares of stuff like this, under the heat spreader for the CPU, GPU, system chip and RAM. The heat spreader is captive in a bracket which is screwed in place, applying light spring pressure. In turn, a heat pipe connects the spreader to the main heatsink, at the back or side of the enclosure, where a copper fin heatsink and low profile blower delivers refreshing ventilation.

Can't be too bad if it works for ~100W CPUs.

Because you can't get a 4-mil-thick heat pipe.

Check the datasheet, it's not isotropic!

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

Fat finger typo? I see a 150x150 sheet for $11

Cheers

Weird. I see a 640 x 320 mm sheet for $985.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

No you don't, at least not the same material and thickness:

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

Ok, yep I had the thickness wrong.

The best price for the 150x150mm 5.0mm sheet is $59

The page has lots of thermal sheets from differnet mfg's...

Cheers

Grin, yeah I bought some of that carbon stuff to test and was shocked to discover it was conductive... (I should have read the specs better but at the time I was just ordering a few of everything.)

Does anyone make SiC insulators? Mica? (or is that no good thermal conduction wise.)

Does someone make an insulated package?

George H.

Mica is one of the best natural insulators and conducts thermally well too. That is why it got utilized as both an insulator and thermal transfer pad on transistors.

On a hard anodized Al heat sink they are not needed, but not many trust the "non-conductive" claim of hard anodized Aluminum (unless level

3 mil spec).

The polymer ones out these days have poorer conduction figures, but still work as they have 'oils' in them which guarantee (practically) that there are no air spaces in the mated surfaces.

Those air spaces are the primary failure mode of a heat sink system. That and improper coplanar mating upon assembly.

SiC is a good heat conductor but in its common forms, not an insulator.

Mica was the classic transistor (and toaster) insulator. Its thermal conductivity is pretty bad, about 0.7 w/m-k, but it's thin and tough. Pushed againt the top of a chip-scale fet, it would be a thermal bottleneck.

AlN is about 200x better a thermal conductor than mica.

The carbon stuff might be useful as a lateral heat spreader. I should get some to play with.

There are some internally insulated mosfets and ICs. No exotica yet.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Duh, semiconductor, sorry.

Yeah mica's 2-D like graphite. I think it's birefringent too. I remember a prof. pulling apart layers to make a 1/4 waveplate.

So why not AlN? or AlO2? You need it so thin it cracks? maybe someone bonds it to a metal?

You mentioned three in your other thermal thread, I was thinking (maybe you already thought this), that for 2-D, three is a magic number. With three devices you could make a nice triangle heat spreader of some metal. (TeCu?) with a bolt in the middle to hold it tight... it'd be pushing on the pcb, so maybe that's not so good? A washer underneath?

George H.

AlN conducts heat about as well as aluminum alloy, and maybe 6 times better than alumina. AlN has a fabulous ratio of heat conduction over capacitance: 1 pF of AlN is about 0.5 K/w. Only BeO and diamond are better.

I want thick to minimize capacitance, but need thin to dump heat. High power and high speed get interesting.

I've considered just pushing an AlN washer down onto three devices, possibly one a dummy. A gap-pad could then transfer the heat to the PCB. AlN is relatively hard to get, but Sienna is sending me some sample washers to try. The down-force would need to be controlled.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Den tirsdag den 6. februar 2018 kl. 02.18.31 UTC+1 skrev John Larkin:

can't get the heat out via the pins and use some of these dothats

Thermal resistances add.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Den tirsdag den 6. februar 2018 kl. 02.56.24 UTC+1 skrev John Larkin:

I know :P

it should have been the other thread "EPC GaN fets"

But it looks like you're onto something. Those 'Q-bridges' are available in BeO and AlN, handy little chunks of useful materials.

Might be useful for something. Even the BeO should be fine as long as you don't Dremel it.

Cheers, James Arthur

Maybe you could seal up the whole product, evcuate and and part fill with something nice & volatile. :)

NT

I have a sheet of SIL-Pad 2000, 12'x12' 0.015"

Left over from a business closure. Still in original flat cardboard container. I'll let it go for $50, shipping included. (lower 48) Less than 1/2 Mouser price. I'll be gone next four days, but I'll get email.

Mikek

12 feet by 12 feet? I don't think so.