thermal baggie things

ambient. One

The PCB

ambient.

about

better,

board.

reasonable

That's my opinion. But then el-customero said that ambient could go to 40C and they want more margin. So I did some research - emails to Germany and such - and discovered that there's a heat exchanger blowing cold air directly on my box.

Now, given the choice of getting -15C easily Bergquist pad) , or -17C with a lot of difficulty (Laird aluminum sandwich), they want me to do... the -17 thing!

Seems logical.

It's a 6

I scanned the board up close. The area between the ADC and the FPGA runs maybe

design

measuring

The IR shows the hot spot temp on the top of the package, probably close to die temp but not guaranteed.

It's a Cyclone 3.

The hard drives will like the cooler and better controlled temperature as well. The only stuff you have to move out of the cooler is DVD drives and the like (includes tape drives). Keyboard, pointing devices, and displays should also be outside for convenience.

?-)

Why don't you use some thermal epoxy to attach a copper heat spreader to the top of the chip? That should even out the temperature and help move it to the PC board. It would be a really simple to apply mod. Gets more heat out of the die.

ambient. One

The PCB

ambient.

about

better,

board.

reasonable

and

lot

Now whose fault was that... shouldn't have told them about that one!

ambient. One

The PCB

ambient.

that about

better,

board.

reasonable

and

and

lot

Yeah, I know. They want 16 piexed this week, and they are still arguing (on Tuesday!) about how to do it.

But customer aside, I learned some useful and non-intuitive stuff over this. Most thermal things turn out to be non-obvious to most people.

That's been tried. A pin-fin heat sink doesn't drop gross chip temperature measurably in still air, but its base slug may cool off the chip hot-spot, spread the heat laterally. But I can't demonstrate how much.

I drilled a 0.1" peephole in a pin-fin and IR imaged through that. When the sink was stuck down, the imaged hot spot dropped immediately, but climbed back up as the sink warmed up. Inconclusive.

Hey, glue a silver dollar to the top of it. It will improve the future scrap value. :)

and

lot

So with none of the needed or even useful other than as a PR effort, it seems logical to come up with an approach that could easily give them

-25°C, adding the sandwich directly to the hot chips! I've already told you how to do that, just flip the case so the solid metal is on the top.

Yes, that's what I thought you said...

I seriously doubt the PCB is within 2°C across the surface on either side. Metal is a good heat conductor, but 1 oz is what, 1.4 mil? That's still not much conduction.

die

I don't see any mention of a temperature measurement diode in the data sheet. So you will need to use some engineering to calculate your die temps.

and

and

lot

us.

maybe

That's what I measured with the FLIR. It's a 6 layer board with three copper planes. 1 oz copper is about 70 K/W per square, and the total power dissipation is about 6 watts or so.

If all the heat was on one edge of the board and the only heat sink was on the opposite edge, I'd expect roughly a 3 or 4 C temperature drop from one end to the other. So 2 C difference, with multiple heat sources and pretty even heat sinking, is perfectly reasonable. Or high, even. That's engineering!

Thermal measurements are never exact like electronic ones.

is

the

was,

Temperatures do that.

The image you showed us

Right, I can't presently measure actual die temp. Surface temp is as close as I cen get.

Then thank Goodness I don't have to work with you.