HV transistor heatsinking / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

HV transistor heatsinking

1 2 3 next

Move Thread

LAN_403

Henry H

Tue Nov 17 2009, 08:14PM

Registered Member #2298 Joined: Sat Aug 15 2009, 08:16PM
Location: ex UK, now Santa Cruz, CA
Posts: 35

I am trying to build a 5kV to 800V stepdown DC-DC converter using an H-bridge of seriesed 1200V silicon carbide JFETs in "cascode" configuration. These parts come in non-isolated TO-247 packages, and the top one in the chain will be at 5kV relative to ground. I'd like to put them all on the same heat sink. Clearly I need a standoff insulator of some kind. The system is operating at several kilowatts and each JFET will probably have at least 60W of dissipation so I also need thermal impedance of better than 3W/K between the transistor case and the heatsink.

The traditional way to do this is to sandwich a sil-pad or mica or other insulator between the transistor and the heatsink. The transistor can be bolted on with a steel/other conducting screw because there is an isolated area around the hole in the transistor (see attached pic). However, 5kV is higher than the TO-247 was really designed for. A creepage arc will form across this isolated area. A ceramic or nylon screw, besides the unfavourable mechanical properties, buys you only another mm or so of creepage distance (the thickness of the insulator) since it must still pass through a hole in the insulator.

I'm planning to use a 25x25x1mm square aluminium nitride insulator with no holes in. I'll glue the transistor to the insulator, and the insulator to the heatsink, with Arctic Silver thermal epoxy and plenty of clamping force during curing.

Has anyone tried anything similar? Is this a terrible idea?

Thanks
Henry

doctor electrons

Tue Nov 17 2009, 09:18PM

Registered Member #2390 Joined: Sat Sept 26 2009, 02:04PM
Location: Milwaukee Wisconsin
Posts: 381

The company i work for does lots of work with induction heaters. They had a similar problem! The way they fixed it was by not using metal heat sinks.
Now i don't know if you have the ability to do this or not. They switched to ceramic blocks (sorry not sure exactly what type but i could try and find out for you) That solved it. Ceramic blocks that were drilled? molded?, and used some sort of high temperature plastic pin to hold the part on. Tapered i am guessing. As for the glueing, never tried it! I do know you can call the friendly people at loctite, they probably have a glue designed specifically for that application! (kiddn, who knows) Hope that helps!

MinorityCarrier

Tue Nov 17 2009, 09:33PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

Beryllium Oxide ceramic insulators work the best. They are expensive, and poisonous. Do not abraid or grind the ceramic, the BeO dust is nearly as poisonous as cyanide.

There are articles on applying "mount down" clamping pressure on TO-247 packages pressing them against a heatsink, in lieu of using machine screw hardware mounting.

Are you using Cree SiC MOSFET's?

Henry H

Tue Nov 17 2009, 09:54PM

Registered Member #2298 Joined: Sat Aug 15 2009, 08:16PM
Location: ex UK, now Santa Cruz, CA
Posts: 35

BeO would be nice but I haven't been able to find any in the right shape. Getting a custom batch fired takes a long time and is 4-digits expensive. AlN is nearly as good (200W/mK versus 350W/mK, translating to a 3K drop vs a 2K drop, which is trivial compared to the drop over the thermal epoxy) and more obtainable / less hazardous.

I'll look up the "mount down" phrase, thanks. That has the advantage of being able to easily replace them when they blow, although it's more complicated than just gluing them.

Using SemiSouth SiC depletion-mode ("normally on") JFETs.

MinorityCarrier

Tue Nov 17 2009, 11:54PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

Kunze seems to suggest that they have TO-247 BeO insulators as standard available parts, up to 1.8mm thick.

Using depletion mode fets in an H-bridge could get very exciting. Are you using a Si-SiC Cascode configuration?

teravolt

Wed Nov 18 2009, 05:15AM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

you could use 24 sheets say 4*4" of 1/8 aluminuim and mount them together with insulating hardware say 1/4" to 1/2" spacing and put a big fan over the top of them. each sheat has a mosfet on it and then mount the hole assembly to a PCB for the other components. all of the sheets will act like big a heat sync and it would be easy to set up. the more compact this device the more exotic your heat synk will have to be to hold off the voltage.

Henry H

Wed Nov 18 2009, 05:22AM

Registered Member #2298 Joined: Sat Aug 15 2009, 08:16PM
Location: ex UK, now Santa Cruz, CA
Posts: 35

This will be mounted on a large multirotor hovering vehicle, so vibration and mechanical robustness is a big issue which means I'd rather keep them on a single heatsink. Also I think it will need more surface area than a 4x4" sheet, even with forced air. I'd budgeted a 14x9x1.5" heatsink for six transistors and freewheel diodes dissipating 500W total, with 10mph of forced air.

1mm of AlN or BeO should hold off about 20kV. I'm later intending to bump up the system voltage from 5 to 30kV so at that point I might need to separate into multiple heatsinks.

Patrick

Wed Nov 18 2009, 06:01AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Hmm, just my 2 cents but I dont know about your set up, especially at 30Kv for your future design. At 5 kv you can get away with the TO-247 but at 30 kv your design may have to change. I think maybe you should use multiple independent aluminum heat sinks in cooling/insulation oil. I would arrange your semi's carefully with concern for proper strike spacings. Your device can be made smaller and run cooler if you use insulating oil (if for flight that matters)

Feel free to ignore my crazy rantings.

-Patrick

Henry H

Wed Nov 18 2009, 06:29AM

Registered Member #2298 Joined: Sat Aug 15 2009, 08:16PM
Location: ex UK, now Santa Cruz, CA
Posts: 35

Could be. I was really hoping to keep it dry but it is possible that using oil would result in a system with an overall lower weight and better reliability. I will keep trying with dry heatsinks and forced air for now.

Patrick

Wed Nov 18 2009, 07:34AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Hey, Henry H, I can see your new like me, look over to your Private message box and reply to my message in regards to this thread, if you can please.

Also, why do you need to step down from 5Kv to a lower voltage, from where is your 5-30 kv being generated?

-Patrick

1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob