TO220--thin leads--high current--How?

Again an infinite heatsink driven by copious amounts of LN2.

inductance, and

specified

years.

rise/fall

they didn't

Class D or class S?

And I insist on not buying their parts. So everybody, even you, is happy.

John

Gp they use LN2? The appnote cited here said "nucleated boiling liquid."

John

That wouldn't work.

Tim

This is nothing new :-).

For decades power transistor maximum dissipation is specified at 25 C _case_ temperature. The heat sink has some internal thermal resistance, so it must be cooled with some cooler fluid. So in order to get 115 W dissipation on the 2N3055, the heatsink would in practice be cooled by 0 C flowing water.

Cooling a semiconductor connected to a large heat sink with LH2 will spread out the heat and the case temperature can be dropped below 0 C and hence the junction can be kept within acceptable values.

An infinite heat sink would have a large contact surface with the fluid, reducing the power density.

With proper orientation and a sufficient liquid flow across the heat sink, the created vapor will be blown away with the stream.

Where can I buy one of those infinite heat sinks? The scrap metal value alone would be ... calculates furiously ... infinite!

But where would I store it? I already have an almost-infinite heat sink

ftp://jjlarkin.lmi.net/Infinite_Sheet.jpg

and it's a nuisance already.

John

I think the quotes indicate that the writer didn't originate the text. But even cliches have meaning.

Tell us about your recent design experiences with power mosfets. Or even your ancient design experiences with power mosfets.

John

e, and

fied

ars.

all

y didn't

Really class S at 16MHz. The "D" ness was a duty cycle control to dial the power up and down a little.

I was experimenting with a servo regulated discharge lamp idea.

Ancient but it was fun at the time. Had to produce current pulses and ship them into a vacuum chamber and be able to prove their shape. The smaller current pulses were 20 and 30 A for 75 and 100 microseconds. Not too bad. The higher current pulses were 120 and 300 A for 900 and 300 nanoseconds respectively. Those were somewhat challenging back then (25+ years ago). Car batteries worked pretty well for the bulk source. Wound some hefty current transformers for that one too. Different fuses (the DUT) for each test case.

use

Tim, John, Read the relevant App Notes, etc. Look at the pictures. It is possible to do interesting things at 77 K. And vaporizing many SCF of LN2 per second transports a lot of heat.

Are you proposing that LN2 cannot boil in the normal (nucleate) way?

A black hole?

(retrieval could be a problem...)

No. I was just wondering which liquid IR uses to generate their absurd dishonest ratings.

John

Simulated liquid? :)

I worked for a company over 30 years ago, that took IR to court regarding over-rated SCRs that failed too often in the field.

This IR attitude seems well established.

Grant.

Their "ultimate current" ratings are apparently done in an "inert nucleated boiling liquid" at 23C, which doesn't sound much like LN2 to me.

And they seem to be testing DPAKs on a solid aluminum "pc board."

John

It could be, with the appropriate resistor inbetween. Kinda tough to stabilize the mess, but it only has to work long enough to make the measurement.

Speaking of which, does anyone have information on the JEDEC JEXD51-7 standards for high and low "effective thermal conductivity test boards"? Unfortunately we're not members and aren't likely to be.

LN2 is pretty thin stuff, namely low specific heat. Water is fabulous.

Maybe they used distilled water under enough vacuum to boil at 23C. Wouldn't put it past them.

John

junction

IMO.

lead.

absurd

JEDEC JESD51-3 and JESD51-7 look inexpensive enough, just register.

--=20 Transmitted with recycled bits. Damnly my frank, I don't give a dear

----------