Sinking Heat From a Surface Mount Footprint to the Chassis

But that was not what was said.

It is for a switch mode power supply. The part will either full on, or full off, and will drive 15A maximum. This part will serve for both the low side and high side switch. It was chosen for is very low CRSS/CISS ratio which is essential for the low side MOSFET.

The MOSFETs will be driven by a MAX8523:

The switching frequency will be 10kHz.

The copper will be at least 2oz and more if practical.

The board will have at least four layers and possibly six.

I am not expecting it to dissipate much power when correctly operating. I am concerned about the heat that might be generated should there be any shoot through while dead time is adjusted. I also do not have a good way to model the heat that will be dissipated during the switch. I can only make a crude guess at it given its turn on/off time, the gate resistance, and the output resistance of the MAX8523, and gate capacitance.

I do not think the expense of machining a custom aluminum block can be justified for this. I was thinking the standoff would be a low cost way to do this. If I do use one it will be aluminum with an aluminum screw, for which I thank the responses in this thread.

Another thing I am looking at is using the AAVID 573100D00000G:

There are lots of things not in the periodic table. What's your point? As someone else pointed out there are more than one form of carbon crystal. Graphene is a 2D carbon crystal.

Rough diamonds? No one is talking about natural diamond... at least that I know of. The gentleman I saw referred to synthetic diamond and that was some 20 years ago.

Now I'm really confused. When I asked if you meant diamond, you said "yes". Now you are saying you mean *any* carbon crystal?

Yeah, or raised in *some* school system. Those crappy school systems... ;)

monds are not that expensive. We need to cut it into some kind of can/cont ainer, then vapor deposit Gallium Nitride on it.

If you need large and flat wafer, then you need synthetic diamond.

But i only need a small round, perhaps several mm in diameter.

"yes". Now you are saying you mean *any* carbon crystal?

Many carbon crystal will work. Synthetic carbon crystal is good in many fo rm. Diamond is good for small package, like power transistor. Carbon Nanot ube will conduct heat in one direction, but that's not a problem either. A ny tightly packed carbon crystal will work.

I've successfully used this over a field of SOT-23 packages. I just bolted the pcb down onto aluminum and let the material squish down over the packages. Not optimum, but worked.

Mouser 739-A15324-01

ChesterW

You are a hard person to discuss a topic with. When you make up your mind about what you are saying, let me know.

Okay, so how about this:

I conduct heat with many vias to copper on the bottom side of the board. On the chassis floor I have several aluminum hexagonal standoffs arranged in a tightly packed hexagonal pattern. Between the standoff's tops and the MOSFET footprint's bottom layer I would have a self adhesive thermal pad. To make room for this pad I put, under the other standoffs that are used to hold the board, washers or nuts to lift the board enough to fit the thermal pad.

Another way would be to put the MOSFET on the underside of the board, and put the thermal pad between the MOSFET and the standoffs. 15 A of current would have to be carried by vias to the MOSFET. How many and what size vias would be necessary to carry that current?

s. I

be less surface area in contact with the ali and less conduction. If you a dd solder 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 elsew here.

r?), and does not corrode or oxidize.

Gold is only needed where the environment is mildly hostile. I see it used so often where really it achieves nothing. Grease with zinc oxide is cheape r for heatsink use.

NT

Thank you. Normally, I don't engage in grammar/spelling/context disputes. But in this case, I am updating my proposal to:

We need GaN transistors with very low on resistant and very good thermo pac kaging. Carbon crystal (Diamond is a common form of carbon crystal) has ve ry good thermo conductivity and is the perfect packaging and heat sinking m aterial. We can mount existing GaN dice on the carbon crystal, or we can v apor deposit Gallium Nitride directly on the crystal. This will allow high er current flow for any given temperature, as well as efficient heat sink/c onduit to the external environment.

That works, but you could use a solid chunk of Al just as well. That saves all that "hex packing," hardware, and such.

You'll have to calculate it. I wrote a spreadsheet for such calculations.

As a starting point, an array of (10) x 0.012" vias with 3/4 oz. plating in .062" PCB have 115 micro ohms' resistance, and dissipate 26mW at 15 amps.

Unfilled, they have a net thermal resistance of 16K/W. Filled with solder,

HTH, James Arthur

I'd say put blocks of alu under the board where you need it, as if you had a heatsink but instead of fins it touches the enclosure

-Lasse

Without an estimate of the mosfet power dissipation, you can't do a meaningful thermal design.

Why not? It's a CSD17312Q5 30V 100A 0.0015O N-CH At 15A, it's 15A^2 * 0.0015 = 0.34W

Your circuit never switches? Why use the FET at all? A hunk of 14GA wire is cheaper. BTW, that 15mohm is a nominal number at 25C. The WC isn't spec'd.

As an aside, I like that series of parts. I've used them for a fairly good sized boost, in the past.

Why not? OP's circuit is fully on or fully off, just not PWM. Anyway, it's 1.5mOhm, very nice FET. Unfortunately. no matching P-CH.

I guess in reality it's good for 10A or so.

Best regards, Spehro Pefhany

Den tirsdag den 5. august 2014 01.21.22 UTC+2 skrev snipped-for-privacy@gmail.com:

how hot is ambient?

datasheet says it'll do 10 time that on 1 square inch copper so why all the fuzz?

-Lasse