Cooling transistors under subzero temperature / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Cooling transistors under subzero temperature

1 2 next

Move Thread

LAN_403

Tonic

Wed Apr 02 2008, 10:45PM

Registered Member #528 Joined: Fri Feb 16 2007, 10:32PM
Location: Warsaw, Poland
Posts: 166

Well, for example, currents in DRSSTCs are limited by low duty cycle and interrupter, otherwise transistors would overheat and blow up, right? Do cooling them under 0*C would be a interesting idea? Or, let's say that the without load cooler's temperature is subzero, but with load is abit higher than zero. I've readed somewhere that cooling resistors decreases resistance, but with semicondurcors (AFAIK, classic ones - diodes), it's opposite.

Logan Kennedy

Wed Apr 02 2008, 10:55PM

Registered Member #1103 Joined: Mon Nov 05 2007, 06:02PM
Location: Houston
Posts: 80

Use a peltier or buy a water cooling kit (made for computers).

Tonic

Wed Apr 02 2008, 11:29PM

Registered Member #528 Joined: Fri Feb 16 2007, 10:32PM
Location: Warsaw, Poland
Posts: 166

Logan Kennedy wrote ...

Use a peltier or buy a water cooling kit (made for computers).

I wasn't asking how I can get subzero, but what possibilities and disadvantages it would give.

Getting cooler shouldn't be problem, since I've been playing with water cooling and phase change systems for years :)

Proud Mary

Wed Apr 02 2008, 11:40PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

It would certainly be cheaper to buy larger, industrial size transistors than arrange costly cooling for smaller ones.

Cooling below zero has it uses in ultra low noise circuits - such as the RF amplifiers of a radio telescope - but its difficult to see any advantage of chilling switching transistors.

HV Enthusiast

Wed Apr 02 2008, 11:46PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

You don't want to chill the transistors that much. Although the effects of cooling are beneficial, once you get too cold you have to deal with CTE mismatches within the die, case, etc... especially with larger transistor cases. I've seen many crack right apart during thermal tests in low temperatures, even down the manufacturers min temps.

Tesladownunder

Thu Apr 03 2008, 12:54AM

Registered Member #10 Joined: Thu Feb 02 2006, 09:45AM
Location: Bunbury, Australia
Posts: 1424

It's also interesting watching LED's fade out when dunked in LN2 then recover when they are warmed.

TDU

HV Enthusiast

Thu Apr 03 2008, 02:46AM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

If you really want to improve thermal performance, look at the thermal interface between component (die) and heatsink. Thats where the biggest factor for loss in performance is. If you have a high thermal impedance there, doesn't really matter how cold you get your heatsink.

Tonic

Thu Apr 03 2008, 07:03AM

Registered Member #528 Joined: Fri Feb 16 2007, 10:32PM
Location: Warsaw, Poland
Posts: 166

Improving thermal perfomance isn't the only thing I would want to get - what I want to know, is what are benefits from improving this factor. Let's say we double DRSST's on time. With traditional cooling methods, like as radiators or water cooling, it would get hot (purely theoretical question, since I'm not experienced with this type of SSTC). With subzero cooling, even if temperature without load is lower than 0*C, with load it can be higher than 0*C - or around ambient temperature. Hey, don't think I'm going to be maniac and cool in ultra-low temperature such as -20*C to -60*C

I was thinking about reasonable range, -10*C to +20*C or so.
Then, shouldn't transistors work at "human" temperatures and give boost up for overall perfomance of Tesla coil comparing to setup with traditional radiator and two times shorter on time?

What is CTE?

Harry - how much it will cost, it depends. For example, I have one small refrigeration unit that was suppossed to cool CPU and to finish it, I only need to fill with gas. If it wouldn't be enough, I can build another more powerful and I wouldn't build it only for transistors - just building, charging and calibrating is a quite fun, and challenge too.

Steve Conner

Thu Apr 03 2008, 10:03AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I guess CTE is Coefficient of Thermal Expansion.

I read some papers a while ago about chilling power transistors with liquid nitrogen. Apparently MOSFETs get a performance boost (their Rds(on) gets smaller) but IGBTs don't like it and actually perform a bit worse than at room temperature.

However, performance is a separate issue to just getting rid of heat. If you can chill a transistor on a cold plate, and keep it cool under load, it'll be able to dissipate a lot more heat before it overheats and breaks down.

This is never done in industry, as Harry mentioned, when all the capital and running costs are taken into account, it's cheaper to use more/bigger transistors and more heatsinks. But if you're a hobbyist, and you've got a powerful chiller and some transistors just sitting there, I say go ahead!

I've always looked forward to the day when overclockers and Tesla coilers would unite to bring us an OMG extreme water cooled SSTC. Phase change cooling would be even more l33t

HV Enthusiast

Thu Apr 03 2008, 02:48PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

CTE is as Steve stated.
Probably is if you cool a device, it tends to shrink more than what its attached to or its mounting providions. I've seen semiconductors (SOT-227 devices like the ones used in DRSSTCs) simply crack their tops even when attached to cold plates of 30degF, with the rest of assembly at ambient temperature.

1 2 next

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