Sizing MOSFET Heat Sink

GOOD GRIEF! Thank you (!) (!) (!)

For the download.

One of the projects is for an extreme baffled wireless remote switch that is going to handle 200 amperes. This will allow me to O/C my battery bank from a 30 meter distance. Chop power to the inverter.

Engineers tend to design things for the bench. Where it's a comfortable 20C. This thing is probably going to see 50C ambient, and that does not give me much room to work with.

The heatsinks are extra thick anodized. Nylon bushings, screws and nuts to hold things together while the epoxy cures. I do not know if the infinitesimal mass of arctic silver would generate enough capacitance to affect the MOSFET. This is why I am using isolated heat sinks and the anodizing should offer enough insulation to isolate the MOSFET, totally.

Next is a search for conformal coating that is superior to the stuff I used to use when dinosaurs roamed the earth. The heatsinks and even the opto-couplers of course have to remain clean but underneath the PCB gets coated. Same for wire junctions.

I am going to use 3ns hyperfast freewheeling diodes, plus TVS for protection. I also found a source for .2 watt (oil immersed sealed ball bearing) fans.

Things have to be engineered rather overkill I'm afraid. There is no "oops, good golly, well, time to run down to the Radio Shack Store" (9 days travel).

I've ordered up a variety of materials. The above is the most extreme duty.

Going off-grid and employing wireless operation, for everything but the A/C.

Mex, I don't think the Arctic Silver is a good idea with an uninsulated case. Everything I read says that good old fashioned Thermalloy silicone with a mica sheet is about as good. I've also used TIM's, some of them better quality, and the heat transfer difference is negligible.

The sinks you have depicted above look like a good solution.

If you go to my dropbox you can download a small program that will calculate the heat sink thermal resistance needed.
Heat Sink Calculator

The main design parameter is fet junction temperature. No need to go to the extreme. It gains you nothing. You're just throwing money away.

Keep typical temperature below 80C and max below 120C and you'll be OK.

Before you blow anything up, it's best to show a schematic what you're planning on doing.

Actually I am going to be using 169 ampere National Semiconductor MOSFETS, and driving them with an enhanced (newer model) opto coupler with a 2000 ma output @ 9 volts.

MOSFET mounting tab is not isolated. Thinking about mounting to an anodized heatsink with extra slow cure Arctic Silver epoxy.

Here is an image of the heat sink(s) I am considering...

Perhaps too much sun right now in the Gorge.

Minor correction. P = I^2 * R; not I*R

Better parallel 3 MOSFETs. 50A/3 = 17A. They won't share equally. Assume one will conduct 22A.

Pmax = 22A^2 * 4.5 mohnm = 2.17W

Thermal resistance of heatsink must be less than:

theata = 45C/2.17W = 20.7 C/W

Now use the same heatsink as mentioned before. Only difference 3 fets & 3 heatsinks. Use thermal paste on the heatsinks.

Thanks for the correction.

LScamper wrote:
Long day Salvo?

Are you switching the high side or low side? N-channel or P-channel?
Got a schematic you can post?

Long day Salvo?

"Max power at 50A is: P_max = 4.5 mohm * 50A = 2.25 W"

P_max = 4.5 mOhm * 50A * 50A

P-max = 11.25W

Have a great day!

MrWizard wrote:
Chris
I think Mexi is switching shore power, . Not batteries...

Do'hh- I should have realized that because of the MOSFETs.
For that application I would use a remote controlled circuit breaker like this Square D model- they use small 24 volt motors to switch on and off, and consume zero power.

Do you want to gang those MOSFET's, have them closely located, or does each Fet get it's own sink?