High voltage full bridge driver

[s2k]

Hi,
a couple of months ago I started my own little nixie clock project based on Dave’s TPIC6595 design.
For the hour-minute separator I went with two neon lamps which I need to drive bidirectionally to get both electrodes to light up.
I designed that part as a full bridge (I now realize that I could probably have gotten away with a half bridge or even a single MOSFET solution + a cap but that’s beside the point).
Sadly, when testing it, I blew my microcontroller and I am uncertain as to what exactly happened/where my design flaw lies.


HVB1 and HVB2 go directly into the GPIOs of my MCU, VPP is +160V.

It is worth noting, that I accidentally shorted HVB2 to ground during soldering, while testing it like that, the other half of the bridge seemed to work just fine. After fixing that short and testing the full bridge my microcontroller blew up at what I assume is the switch from HVB1 -> HVB2 (I drove it by setting both controls low, delaying 1 ms and then setting the desired control high).

I assume one of three things happened:
  a)   I did not include any gate resistors so when switching the MOSFET, the gate capacitance shorted into the GPIO which blew up the MCU. However, considering one half of the bridge did work for some time I find that questionable.
  b)   The Gate-Source (or Gate-Drain) capacitance of the MOSFETs passed a high voltage spike when I plugged in the HV connector which killed the MCU.
  c)   The Gate-Source (or Gate-Drain) capacitance passed a high voltage spike during switching operation.
I already tried googling for high voltage MOSFET drive circuits and the likes but most of the results tried to sell me driver ICs which I think are a little over the top for this application.

I would be really happy about your feedback on my circuit before I try building an improved version so I don't instantly kill the next MCU.

[fourtytwo42]

You could be right about the cause BUT all MCU's are different so it would help if you specified it.

[s2k]

Sorry, completely forgot about that, the MCU is an ESP32.

[fourtytwo42]

I can't say I have any experience of those but the data sheet doesn't even bother mentioning ESD protection so you have to assume it's very weak on io protection of any kind. I would use a shottky from the io pin to mcu 3V3 (AND ensure 3V3 is protected from excessive voltages by a zener etc). In addition to that use a series resistor in the io to limit current. Depending upon your layout you might also have a ground undershoot problem on your io in which case you need another schottky to the mcu ground pin. I notice you have nothing to define the gate voltage at mcu power up, this may also bite you.

[mikerj]

Did the MOSFETs in the H bridge survive when the micro was killed?  High voltage spikes passing through the gate do not make MOSFETS happy, if this happened I'd be expecting at least one dead transistor.

[s2k]

Thank you so much for your replies, I will definitely try the protection with schottky + zener + gate resistor next time.
At least none of the MOSFETs failed short, I did not get around to desoldering and testing them more thoroughly though.

[David Hess]

Coupling through the gate-to-drain capacitance could have done it.  This becomes important at high voltages where the change in high voltage at the drain effectively multiplies the capacitance.