Sinking Heat From a Surface Mount Footprint to the Chassis

I doubt you'll have a corrosion problem in an ordinary environment.

How much heat do you have to dump? That arrangement won't have a very good thermal resistance.

That's a really tiny fet, so spreading out the heat will be hard. Can you use a DPAK or something bigger?

Will this be a multilayer board? Interleaved copper pours in a multilayer board would be better than the small mica, as a thermally conductive insulator.

A good way to transfer heat is to move it from top to bottom with big copper pours and lots of vias, then move it to the chassis through a gap-pad silicone thingie, like from Bergquist or Laird, as thin as possible. Even better, use a slab of anodized aluminum between the PCB and the chassis, with thermal grease.

It helps to do a little math, or a thermal breadboard, before laying out a board.

That's stainless steel. Nice looking, but a rotten thermal conductor.

The heat path through the #4 machine screw and the standoff is pretty poor--#4 machine screws are horrible heat conductors. A thick Al standoff with flat surfaces might help.

Whether it's "good enough" depends on how much power your FET has to dissipate.

If it's a lot of power I'd be thinking in terms of thermal vias from the FET to the bottom of the board, and a solid chunk of hard-anodized aluminum between there and the chassis, plus grease.

Cheers, James Arthur

I did something similar once. I built a 1000 watt heater that heated a 15 g allon tank of hydrochloric acid to 37C. The controller sat in a sealed alum inum box above the tank. There couldn't be any holes in the box for air flo w because of the fumes from the acid. I used aluminum screws and an aluminu m block to bring the heat from some TO220 packages on the board to a heatsi nk outside the box. I used standard TO220 silpads and plastic grommets to e lectrically isolate the TO220 parts. The PCBs were tinned. I don't recall d oing anything about about galvanic corrosion between the tinned PCB and the aluminum oxide of the aluminum parts. I guess I could have gotten the bloc ks anodized. The trick was the aluminum screws. Zinc screws didn't work.

gallon tank of hydrochloric acid to 37C. The controller sat in a sealed al uminum box above the tank. There couldn't be any holes in the box for air f low because of the fumes from the acid. I used aluminum screws and an alumi num block to bring the heat from some TO220 packages on the board to a heat sink outside the box. I used standard TO220 silpads and plastic grommets to electrically isolate the TO220 parts. The PCBs were tinned. I don't recall doing anything about about galvanic corrosion between the tinned PCB and t he aluminum oxide of the aluminum parts. I guess I could have gotten the bl ocks anodized. The trick was the aluminum screws. Zinc screws didn't work.

I am thinking about something similarly. Perhaps with carbon crystal. Car bon crystal has twice the thermo conductivity of copper and 4x that of alum inum. But making carbon crystal screw might not be easy or cheap.

By carbon crystal do you mean diamond?

Yes, obviously.

Carbon crystal has twice the thermo conductivity of copper and 4x that of aluminum.

not necessarily,

-Lasse

If you solder coat the pcb where it touches the ali block, there will be le ss surface area in contact with the ali and less conduction. If you add sol der lines radiating out where it wont interfere with touching the ali, you' ll get more heat spread and thus conducted to the ali block, or elsewhere.

NT

less surface area in contact with the ali and less conduction. If you add s older lines radiating out where it wont interfere with touching the ali, yo u'll get more heat spread and thus conducted to the ali block, or elsewhere .

Plated it with Gold. Gold is better thermo than copper or zinc (solder?), and does not corrode or oxidize.

Not sure why you say obviously and I'm not sure why you didn't just say diamond.

I remember years ago seeing a presentation by someone who had been doing research with wafer scale integration and proposed stacking wafers with diamond slabs in between to bring the heat out. The diamond wafers sticking out of the stack were cooled by a spray of coolant which cooled by evaporation. Impressive and could provide for some *very* dense electronics. Just not too sure how practical it was for a number of reasons even for military use.

Gold will give corrosion resistance but it only adds to the thermal resistively. Anything you add, even diamond, will only increase the thermal resistively. You're unlikely to add enough gold to get any appreciable spreading. Use 2oz. copper, if that's your goal.

I think the point is that you need to coat the copper with something to prevent it from oxidizing too much. Gold would be so thin that it would add appreciably zero resistance.

Almost all of our PCBs are gold plated these days. The gold is something like 10 micro-inches thick over a barrier nickel layer. I doubt that the thermal resistance changes much.

The big gold pour is used to transfer heat from the bottom of the board to a baseplate or enclosure.

Because we were discussing it in the "Google inverter" thread, and carbon is the scientific component. Diamond is not in the periodic table.

It would be practical enough for the $1M inverter prototype. Rough diamonds are not that expensive. We need to cut it into some kind of can/container, then vapor deposit Gallium Nitride on it.

There are many allotropes of carbon, and therefore many "carbon crystals". By attempting to be contrarily specific and thus "sound smart", you are in fact generalizing beyond what your state of knowledge is familiar with, and therefore "sound stupid". I would recommend talking about things the same way everyone else talks about them, so that you stand the least chance of being misunderstood (and thought stupid).

Tim

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

Eww, those ugly square-edged copper pours. Doesn't anyone working in PADS have style?

Then again, they don't even look very useful, I don't see many vias in them, and several are well cut up by intersecting traces. There's also a spot under that DIP8(?) that's bare, cut off by a trace would be my guess. Doubtful it does anything at all, but it looks ugly.

Nay, I probably shouldn't be so judgemental though. The software actively stifles any desire for creativity. It must've been written by people raised in the American Public School System. ;-)

Tim

But i do mean carbon crystals, in which diamonds and nanotube belong. Diamond is one form of crystal, but other form could work also. There are carbon crystals harder than diamond as well.

Yeah, it's pretty bad. All it does is make money.