manifesto

I am not holding my breath. GaN was announced as the ultimate cat's meouw for switch mode power conversion many years ago, with lots of pomp, parades and whatnot. Many moons later they are still a niche technology and there is still only one manufacturer.

From the above PDF, this is not so cool: GaN on SiC will remain the purview of low-volume, niche applications due to the inherent cost structure of substrate material. Fundamentally, at a physics level, SiC boules grow 200X to 300X slower than silicon. The cost of producing substrates - notably capital depreciation and energy consumption during material growth - scales proportionally to production time. Thus, GaN on SiC will remain perpetually higher cost and thus prohibitive for mainstream commercial use.

Translation: Macom is not interested in the commodity and mass market devices and will do their best to keep prices high. I would not expect cheaper power FET's from Macom.

Lots of people are making GaN fets now. Nitronex, Cree, Macom, EPC, Toshiba, Eudyna, RFMD, and I think a few others.

We did some work with the Nitronex GaN-on-Si parts some years back. They had astounding power/capacitance ratios and low gate swings on/off, but they weren't reliable. They claim that they've fixed them.

If epitaxial GaN on silicon gets made in volume, we're going to get some killer parts. I want a GaN version of the 2N7002, sot23, but 20x as fast.

Hey, your EMI consulting will pick way up when everybody is making 30 MHz switching supplies.

His point was that GaN on silicon will get cheap; it's getting cheap already. Silicon conducts heat better than SiC anyhow.

GAN Systems comes in once a year or so, trying to sell their stuff. It's too expensive but not really by all that much. One of the controllers I've used claims GAN compatibility but I have no idea if it really works. GAN is a win for switching regulators but not enough, at least yet. It looks more promising over 100V.

SiC will take more.

I am not so sure about that.

After all you need much smaller inductances and capacitances and transformer core areas than at 30 kHz. Thus, filtering the inputs and outputs and putting the whole mess into a sealed metallic box is easy, without too much material cost or weight penalty.

Oops. I didn't notice that the author had switched from discussing GaN on Si to discussing GaN on SiC.

IME, stomping out radiated emissions are a lot easier than conducted. As you point out, the filters are a lot smaller at high frequency, too. Ferrite beads don't work so well at 30kHz. ;-)

ifesto.pdf

class-D amps

Read it again, it is a comparison between GaN on SiC versus GaN on Si for volume production (power conversion) to leverage that technology for RF and microwave. Macom is betting on GaN on Si.

?-)

Well, it WOULD be nicer to have more contrast between the background and the text; say light blue at worst - and maybe no images to interfere