Infineon's CoolSic MOSFETs - a notch up.

You mean that my stash of TO-3 horizontal output NPNs with betas around

Cheers

Phil Hobbs

I have already completed a series of PSR flybacks based on the Cree part. It would have saved me so much trouble... You could have sent that a month ago, James... ;-(

Best regards, Piotr

Beta? I only used the c-b junctions.

That sounds beautiful. Gate driving the Crees was a real nuisance.

I could, seriously, use two of those in parallel in my Pockels Cell driver. The thermal situation would be much improved.

They're not obsolete, not at all! They're still excellent fishing weights, heaters, burnishing tools, double-barreled push pins for bulletin boards, foot-piercing devices, varactors, HV diodes... the uses are endless!

Cheers! James

I could've used them a month ago too!

I like the Cree parts quite a lot.

I've recently surveyed all the SiC offerings in my application's class, and Cree's was clearly best choice...until this new part.

I hope someone at Cree will take notice of this discussion and tweak their parts to compete with this new offering.

Cheers, James Arthur

It *was* a bit of a nuisance. Even backing down from 20 to 12Vgs is a driver-help; getting rid of the -4V would be even nicer.

I'm still suspicious of 0V drive. Infineon's app notes say it works, but they might not pushing it as hard as I would.

It should be emphasized that the discrete CoolSiC? MOSFET products can be safely operated with a designed turn-off voltage of 0V. Thus, the values in the guidelines do not have a negative impact on the performance. Furthermore, it even enables a less complex, unipolar gate-driver circuit design.

I looked at GaN, too. GaN's even easier to drive but none of the devices I saw could compete with SiC on Coss, and Vds(br) is still just 650V.

Cheers, James

With the right audiophile woo-words, I expect you could flog them on fleabay for obscene amounts of money. Audiophiles seem to buy anything.

This part looks incredible. I am very much obliged for pointing it out! I will include it in my next order from Mouser just to evaluate it.

Competition is always a good thing.

Best regards, Piotr

What makes you suspicious? The negative drive was there to buy sufficient noise margin, as the older SiCs have V_th of around 1.5V. This part is claimed to have 4.5V, very much silicon-ish...

Best regards, Piotr

Best regards, Piotr

Why did you use a SiC FET for a flyback? Optimizing a inherently bad efficiency design (flyback) seems counterintuitive. But maybe your case is special? High frequency, or very large step down ratio?

Cheers

Klaus

Simplicity of the final design was at a premium. And with a SiC at the input, synchronous rectifier at the output and a piece of 3C95 ferrite in between has resulted in a design which is not even warm at full load.

Indeed, overengineered like hell, but should be super-reliable.

Best regards, Piotr

I think what I was seeing with the Cree parts was Crss current, from a fast slew, producing voltage drop in the internal gate resistance, keeping them on, basically making a soft Miller plateau. I drove mine from about -5 to +18 at extreme slew rates, making narrow pulses for pulse-picking an 80 MHz free-running laser.

Part of the reason for extreme gate drive was thermal, to avoid losses and chip heating. Rds-on vs temperature tends to be high in SiC. The Infineon is mediocre in that respect.

The Cree is kind of silly, a tiny chip in a giant package.

Two of James' parts would spread the heat out, and the absurdly low Crss should improve the gate drive situation. I'd probably still do some inductive peaking in the gate, to crisp up the internal gate waveform. Negative gate bias may well not be needed, especially with a little inductive peaking. I'll try those parts when things calm down.

GaAs and GaN were mediocre (or downright awful) first couple of generations.

I boosted 48v to 1200 at pretty good efficiency, at 4 MHz, with an air-core inductor. Water cooled!

I just hadn't fully checked their figures. At high slew rates Miller could flow enough current across Rg to turn the part on, for example.

But doing the math, the Infineon is almost certainly good enough for nearly every common use, and enough better than my Cree in this respect that unipolar drive might be almost feasible for me, too.

Cheers, James Arthur

I'm flogging that same horse. If I could get the total gate drive energy low enough that would help the support considerations, even open alternative drive schemes.

Most of the makers seem focused on big amps and low Rds(on). I'd really appreciate unipolar drive, lower Crss (thank you Infineon!), and lower Rg.

Cheers, James

Somebody else's SiC fet, I think Japanese, had absurd Rds-on vs temperature, not to mention a huge internal gate resistance.

I tried the ST equivalent of the C2M0280120D, but it was about the same, so I stuck with Cree.

There's a lot of money going into SiC, so we can expect good stuff.

I can show you my GaN-based SiC gate driver, which is admittedly obscure. I'm swinging -5 to +18, all that I can without blowing out the Cree gate.

I theorize that a bit of inductance will fix that.

I did find that the Cree Spice model was pretty realistic for tuning gate drive.

Have you shoveled the Infineon into LT Spice?