3
\$\begingroup\$

I designed the circuit below for an FSAE electric vehicle, for which the rules require an HV-powered LED to be lit when the HV bus is above 60 V. In simulation, the circuit works very well.

In real life, when 400 V was applied (smoothly thru a precharge circuit), it blew a fuse, the MOSFET failed (post-mortem RGS = 39 Ω, RDS = 4.1 kΩ), the 10 V Zener diode exploded, and its negative trace burned to death.

Is there a fundamental problem with this circuit that I haven't caught on?

My hypothesis is that the gate somehow broke down and it must have sank current straight from the drain. Would such a Zener arrangement be too slow to kick in, thus causing VGS to go beyond absolute maximum?

What could I do to fix the issue? If the problem is the Zener being too slow to kick in, I was thinking about adding 1 MΩ and 0.1 uF both in parallel with the Zener to better protect VGS.

MOSFET datasheet

Circuit simulation

Circuit

(HV and LV grounds are separate in real life)

Safe operating area

\$\endgroup\$
10
  • \$\begingroup\$ Did you definitely have source and drain connected correctly? \$\endgroup\$
    – Andy aka
    May 2, 2023 at 17:18
  • \$\begingroup\$ @Andyaka Yes, pin 2 tab (drain) is connected to the 51 V Zener, and pin 3 (source) goes to the LED. \$\endgroup\$
    – AgentRev
    May 2, 2023 at 17:29
  • \$\begingroup\$ What did you use for the 4Meg resistor? was it a through hole part? \$\endgroup\$
    – Voltage Spike
    May 2, 2023 at 17:31
  • \$\begingroup\$ @VoltageSpike SMD 2512 we had lying around in our inventory, I don't have the exact part number. All resistor measurements looked good after the fact. \$\endgroup\$
    – AgentRev
    May 2, 2023 at 17:39
  • 1
    \$\begingroup\$ At any time, did the \$V_{GS}\$ voltage exceed maximum rated +/-20 Volts? Might have to 'scope it to see (remember, scope ground is typically Earth ground; two 100x probes and math-mode subtraction is a common solution to differential HV measurements.) \$\endgroup\$
    – rdtsc
    May 2, 2023 at 17:43

1 Answer 1

3
\$\begingroup\$

There is a lot of capacitance from Drain to Source because that is a beefy MOSFET.

When the voltage is applied with a fast ramp, that capacitance will raise up the Source voltage well above Vgs(max). The gate is constrained to be 10V above ground so you get a huge negative voltage on the gate relative to the source. The zener should be be from gate to source to protect it and you may also need some series resistance (in series with the input, not the zener) that is resistant to failure from massive (but brief) power dissipation during the start-up pulses. Your LED and Optocoupler LEDs may be seeing potentially destructive peak currents as well as stressing the 1.5kΩ and 10kΩ resistors.

Since failure of the LED or the resistors would result in a false "safe" indication, I would expect you would want to do a very careful analysis and ensure that none of the parts are particularly stressed (and even that may not be sufficient).

\$\endgroup\$
6
  • \$\begingroup\$ Just to add on to this: the datasheet shows a dV/dt rating of 50V/ns. Was the rise time of the 400V less than 8ns? \$\endgroup\$
    – Eeyn
    May 2, 2023 at 19:51
  • \$\begingroup\$ @Eeyn OP says 2000V/s which isn't all that fast, but if part of the ramp is faster it could be bad. \$\endgroup\$ May 2, 2023 at 20:01
  • \$\begingroup\$ Okay, that sorta makes sense. I do actually need the Zener between gate and ground to ensure voltage regulation. Since regulation is roughly 7 V, I could simply add a second 10 V Zener between source and ground, and throw in a 10k at the gate. Would that suffice? Including the precharge circuit, this would be the result: tinyurl.com/2fw6h88s \$\endgroup\$
    – AgentRev
    May 2, 2023 at 22:12
  • 1
    \$\begingroup\$ You might consider a TVS rather than a zener for the source. \$\endgroup\$ May 2, 2023 at 22:20
  • 1
    \$\begingroup\$ Implemented the fix with a fresh MOSFET, cycled a dozen times and everything looks good so far. Thanks for the help! \$\endgroup\$
    – AgentRev
    May 7, 2023 at 9:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.