SiC higher-voltage gate drive

Does the slower electron mobility, mu_n, of SiC semiconductors, contribute to the higher required gate voltage for SiC MOSFETs, e.g., as high as +20V,-5V?

--
 Thanks, 
    - Win
Reply to
Winfield Hill
Loading thread data ...

E-Field ?

Reply to
BibO

These SiC things are hard to drive, especially fast. There are the big Ixys RF-packaged drivers, but they are huge and expensive. We have to make our own gate drivers, which is a nuisance. Maybe we should pot them and sell them.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

If you accept a 5 to 8ns risetime, you can drive them with standard driver ICs. I've used UCC27511, UCC27538 IXDD609 TC4422A and UCC21520. If you need faster than that, yes, roll your own.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

The latter had 1.0 ohms to the gate, paralleled with an NPN PNP emitter followers: DNLS350Y and DPLS350Y 5-amp BJTs in SOT-89 packages.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

Unfortunately UCC21520 is too slow. 6ns typ, 16 max rise time and 20 ns min pulse width. I'm making a 10 ns wide HV pulse, so I need gate drives with a few ns rise and fall. With that many volts of gate drive, we are probably slamming the Cree on/off, through its transfer curve, pretty fast.

I expect there will be some faster SiC drivers eventually.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Nah, those numbers are for higher capacitance, plus the SiC switches faster than the driver risetime. It's the time it takes to go through the Miller plateau that counts. I made under 10ns-wide pulses (measured at half-height). I also reduced the swing down to -3 to +16V, to help it go through faster. Works fine.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

As for your original question about mobility, I will try to ask that question while at APEC next week in Anaheim, CA.

Good question.

I just drive my SiC FETs with regular 2.5 amp drivers and they work well. Some of these SiC FETs may have high internal gate resistance anyway I understand.

They sure work well though ! But I'm not running them at 100s of kHz.

boB

Reply to
boB

e to

5V?

Reduced SiC is only part of it. Apparently there is a lot going on with the gate SiO2 oxide layer at the channel SiC interface by way of electron trap s with energy levels comparable to the conduction band energy of the channe l electrons, which effect opposes the VGS enhancement of channel conductivi ty. End result is a sluggish drain I-V characteristic that is more like a V GS dependent resistance than ideal current source, with a fairly small forw ard transconductance, gm. The reduced gm is most responsible for the large VGS drive requirement. Their problems don't end there because the large req uired gate drive results in an extremely large electric field within the ga te oxide, which in turn causes defects to develop that move the gate thresh old even higher, in effect reducing the gate drive effect on channel conduc tance, increasing power dissipation and possibly leading to device failure. This note is a fairly self-contained account:

formatting link

Reply to
bloggs.fredbloggs.fred

That's a great paper; thanks.

Table 2 shows scary gate time constants. But they look wrong.

525 pF times 6.5 ohms sure ain't 221 ns.
--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

bute to

V,-5V?

the gate SiO2 oxide layer at the channel SiC interface by way of electron t raps with energy levels comparable to the conduction band energy of the cha nnel electrons, which effect opposes the VGS enhancement of channel conduct ivity. End result is a sluggish drain I-V characteristic that is more like a VGS dependent resistance than ideal current source, with a fairly small f orward transconductance, gm. The reduced gm is most responsible for the lar ge VGS drive requirement. Their problems don't end there because the large required gate drive results in an extremely large electric field within the gate oxide, which in turn causes defects to develop that move the gate thr eshold even higher, in effect reducing the gate drive effect on channel con ductance, increasing power dissipation and possibly leading to device failu re.

That's another thing about this device. There's a disconnect between the ga te voltage rise times and the turn-on time of the channel. The channel turn s on much faster. That doesn't mean it's on as much as you would like by wa y of minimal RDS,ON, but it's on enough to cause substantial current flow a nd switching loss.

Reply to
bloggs.fredbloggs.fred

The Cree Spice models seem to be pretty good, except that they got the substrate diodes wrong, and their thermal modeling hooks are a real PITA.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Hello Win. No. The Vt is not changed by mobility. Here is the formula for Vt in a link.

formatting link
Alan Folmsbee

Reply to
Alan Folmsbee

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.