GaN

Nitronex, RFMD, and I think Toshiba have been doing GaN fets for a few years now. We've played with some Nitronex parts. The voltage, transconductance, and capacitance numbers are stunning compared to mosfets or gaas parts. This is going to be really hot for RF and power apps when they get the processes under control.

There are maybe a dozen serious players in GaN now.

We did blow up about $600 worth of the Nitronex parts one afternoon.

John

On a sunny day (Wed, 28 Jan 2009 21:20:10 -0500) it happened Boris Mohar wrote in :

Nice, 60MHz switchers.

So GaN is in about the same state today that GaAs was in 25 years ago when TriQuint, Link-a-Bit, and all the other GHz GaAs MSI vendors were new.

I was just reading one of my old Uni texts, and apparently (ancient history even back then) there was a major $$ effort to make a high- speed computer using GaAs tunnel diodes. Unfortunately, they were not stable so the entire effort was scrapped. Anyone know about that? Sometime in the early sixties, apparently.

Ge TDs were stable, and were the fastest devices around (risetimes down to about 25 ps) for a long time. But the fab process was ghastly, and doing logic with 2-terminal switches was harder than it first appeared. And no memory technology at the time justified screaming-fast CPUs.

The GaAs TDs did have a current-time degradation effect, sort of like LEDs now, but much worse. TDs were fast partly because they worked at extreme current densities.

RCA made some Ge TDs with peak point currents of 220 amps.

John

Well, a lot of what we have of useful theory and practice now is because some people built and tested extreme cases.

For what possible use? Radar transmitters or something?

They probably did them because they could.

One suggested application was for low-voltage dc/dc or dc/ac converters, like from solar cells or thermocouples.

One TD and one transformer - no other parts - made a nice 0.2 volt DC to AC converter. TDs were cool.

John