suggested solutions to solve random gan fet fail when touching sparks about pcbtc HOT 2 OPEN

Hi niklas! how's the debugging going? I saw the 1.6 version and it looks cool so i wanted to ask if you tested it and if the random failing gan bug is still present. Maybe, if it isn't already fixed, you could provide us some oscilloscope plots to help us help you in solving this problem ;)

from pcbtc.

Hey Niklas,
thanks for your great suggestions and the time you spent on thoughts about this :)

1. I don't think an additional ground plate helps for this design. I think so for two reasons:
   First, GND of the bottom pcb is connected to the shielding of the USB type C cable, which is usually connected to an earthed computer or a laptop. For reference, I've experienced no difference in performance id I connected the ground cable of a scope probe. Second, I tried to avoid as much metal underneath the coil as possible, as it creates eddy currents. That's also the reason why I have a cutout in the middle of the pcb. I didn't mind that for the first revisions, and changing this made a huge difference. That is also the reason for the opto-coupler: I measure a voltage drop of 1.2V induced just from one side of the pcb to the other side. The opto-coupler helps isolating these domains.

2. That's a great hint, thanks! I did think about this before, but I never actually measured the actual voltage on the rail. The USBLC4LC6 does have an internal zener diode, but I don't know how well it performs.

Anyway, I wanted to try implementing primary current feedback for the next revision. This would also have the advantage that I can implement a phase lead circuit (google "DRSSTC phase lead") to compensate for the propagation time added by the deadtime logic. I was thinking of using a pcb-only current transformer as described here: http://sci-hub.tw/https://ieeexplore.ieee.org/document/7931056/
I wanted to do the next design in four layers anyway, so this might work out pretty well.

I also have some other ideas for this project. One of them is to implement a "QCW" mode by using the flyback transformer to ramp up the bridge voltage. I don't think this will work with USB, but with an external power supply this might give some pretty cool effects. Of you want to collaborate on this, feel free to keep giving suggestions :)

Best,
Niklas

from pcbtc.