Gentlemen,
Taking the 2N3055 as an example (as I'm sure we're all familiar with this once-ubiquitous decades-old device) say if you were to rig up some apparatus which enabled you to flood its case with a stream of say liquid helium, how much power could you wring out of it? I don't mean theoretically, I mean *practically* taking into account its real-world limitations. Any idea?
Thanks,
CD
It would probably be limited by die cracking due to differential thermal expansion at the high voltage end and bond wires blowing off at the high current end.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Wow, Phil, you broke the speed of light with that reply!! :-D
Bored, bored, bored. ;)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Bipolar transistors don't work at liquid helium temps! Better try liquid nitrogen. Or cold water.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
You just have to turn the current on before the helium, and do the measurement fast before you run out of money. ;)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
LTSpice includes thermal modelling blocks, you could probably use it to model liquid cooling loops, too.
Not that it makes much sense in most applications to liquid-cool BJTs. Maybe in avionics bays in aircraft or on spaceships something.
Silly thing I did, put a ~70 watt TDP processor in this teeny lil aluminum box. Thermal epoxied the cooler blocks to the enclosure and fashioned the right-angle headers from brass plumbing T-joints, solder and a Dremel. External PSU and coolant pump, just deionized water with the pump speed under control of the processor temp sensor. 40mm fan at ~700 rpm on the final block.
Box got very toasty but the processor stayed within limits under full load.
Okay, the liquid helium bit is unworkable for various reasons. Let's be realistic and say forced water cooling. Mount the transistor on a finned copper heatskink and force water through the fins sufficiently fast to carry away the excess heat. Under those circs., how many Watts could be screwed out of it? The normal max dissipation for this device is 115W and there's a datasheet here:
-- Walk away, Boris!
The transistor is a lump of silicon bonded to the metal base of the transis tor case. There a thermal resistance from junction to case, more thermal re sistance through metal that forms the bottom of the case, and if you clamp the case hard down on the top of a nice thick heat-sink you need a smear of zinc-oxide loaded thermal grease to fill in the air-gaps between the two.
lists the junction to case thermal resistance as 0.584 C/W.
If I remember rightly, the bottom of the case is fairly thick lump of coppe r, and I used one once to save a design where the original designer had mes sed up his heat-dissipation calculations.
For the 2N3055 it is 1.52 C/W
The liquid helium would be outside the case. The junction temperature is li mited to 200C. Liquid helium boils at -268.75C, so you could get 308W out o f a 2N3055, or 801 W out of 2N5685.
You'd be boiling off a great deal of liquid helium - it's heat of vaporisat ion is 0.0854 kJ/mole, so 801 W is about 10,000 moles per seconds, or 200,0
00 litres of gas at STP. 4.25K is lot lower than standard temperature, so i t's 3030 litres of helium gas at -268.75C.It may not be impracticable - you probably could squirt a high pressure str eam of liquid helium at the base of the heat sink and pump away about 3 cub ic metres of gas every second - but it certainly wouldn't be worth doing.
The 2N3055 is nominally rated at 115W and the 2N5685 at 300W (and you aren' t going to come close to either figure in a practical set-up).
-- Bill Sloman, Sydney
Based on that data sheet, 115 watts is the max.
But based on the other numbers, if you cool the case to absolute zero, Tj hits 200C at 311 watts.
If you blast ice cold water directly onto the bottom of the can (no heat sink needed) it looks like 131 watts to me.
Some 2N5055s had a tiny chip, Fairchild I think, and blew up fast.
We use one fet that's claimed to be able to dissipate 1000 watts.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Liquid helium isn't particularly cheap either, it's a non-renewable resource, or at least not in large volumes at reasonable cost.
Use liquid hydrogen!
Actually given that it's CD asking I expect it's some hare-brained scheme to use power transistors to blow all the world's helium reserves into space to "prove" there's actually an infinite supply.
I saw a design once where they carefully drilled a hole in the top of the TO-3 case, and soldered in a pipe with a nozzle pointed right at the face of the die, and another pipe was soldered to an outlet hole somewhere at the edge of the TO-3 can. I think it was a German project but can't recall.
By spraying or squirting an inert, insulating coolant (fluorinert or maybe transformer oil, or a low-boiling-point liquid e.g. butane or freon) right at the surface of the die, it ought to be possible to ignore the junction-to-case thermal resistance and also to exceed the usual current capacity of the bondwires.
I am dubious of the long-term reliability of such a scheme, and I'm not aware of any magic device in TO-3 that couldn't be replaced by multiple parts in a different package on a water-cooled heatsink, but if you had to do it, I expect it would work for a while. If you could do it with some fancy RF power amplifier part it might be worth it, but they generally don't come in TO-3.
I guess that's the killer here: the junction temp.
Does it?
You recirculate it in any practical set-up - recondense it and pump it back
Fine until it leaks - mixture of hydrogen gas and air are explosive over a very wide range of concentrations.
Cursitor Doom isn't hare-brained - that would be unkind to hares. He's just a rather nasty half-wit who cuts and pastes a lot of nonsense that he doesn't actually understand.
-- Bill Sloman, Sydney
===========================================
** So did Motorola, when back they used Aluminium paks. Same devices had a short thermal cycle life and by the mid 80s they returned to using steel paks with the chip sitting chip on an alloy slug.Most other 3055s had far larger chips, eg RCA, while the small chips boasted a higher Ft.
Because of the JEDEC numbering system and rules, any TO3 device which met or exceeded JEDEC specs of a 2N3055 could be labelled as one.
.... Phil
=======================
** Q, "wring out of it " = what ?The makers rating is for dissipation, nothing else.
As a linear PSU regulator, it might realistically deliver 8 amps at 60 volts = 480W of useful DC.
Use a pair in a push pull class B stage and you can have 100W at 4 ohms with about 40W dissipation per device at worst case sine wave.
.... Phil .... Phil
I agree.
The limit could be inside the chip but ON Semi says with thermal greased bare metal contact you can get the junction to case thermal resistance down to 0.1 W/'C if you can manage rated torque or make a contact pressure > 200 PSI to heatsink.
I've used crushed dry ice with a muffin fan inside a picnic box to test products down to -45'C but contact pressure and out gassing won't give you great thermal resistance to the dry ice. So I used forced air inside while generating 3W and used thermocouples on the D.U.T.
On the old days the story was that when a manufacturer tried to make some high quality power transistors, but some chips did not meet the original parameters (such as h_FE or f_T) but exceeded the very relaxed 2N3055 requirements, these were sold as 2N3055 transistors. This is usually OK, but some manufacturers labeled some VHF power transistors as 2N3055. However, when installed into a typical LF circuit layout, the transistor might oscillate wildly somewhere in VHF or UHF :-).
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.