What's causing MOSFET driver failure in my coilgun circuit?

Question

I'm working on a small coilgun, and am running into an extremely annoying problem with my circuit.

Circuit schematic:

PCB Layout:

Key components:

• Flyback diode: BYC30X-600PSQ
• Flyback resistor: Ohmite 45 W 1 ohm
• Coil: 95 uH
• Switching MOSFET (Q1): ISC027N10NM6
• MOSFET driver (U2): MCP1406T-E/SN

I've chosen everything to run right at it's limits (probably a bad idea in hindsight).

The coil is 85 mohm, and my battery's internal resistance is 15 mohm. Fully charged it's 8.4 V, so I should be getting around 84 A of current at most. My flyback resistor is 1 ohm, so the flyback voltage peaks out at 92 V. ISC027N10NM6 has a breakdown of 100V. I've tested shorting the resistor though, and it still fails the same way so I don't think this is the cause.

When testing my circuit in a single isolated stage, it works fine. I've repeat fired it 9-10 times and nothing ever got hot and everything behaved as expected.

Testing single stage with direct wiring (Works)

But then when I got to assembling the stages together, and started testing again in that scenario, the gate driver kept blowing up on the firing test! I tested slight variations in each one, but they all resulted in the gate driver breaking (in the same way), and I'm now out of boards :(

I tried:

• shorting the flyback resistor (to rule out breakdown voltage issues)
• adding a gate discharge resistor from gate to ground (10k)

But none of these affected the circuit. The driver still broke in the same way (described further down).

Multiple coil assembly with single stage wired up (Fails!)

My gate driver (MCP1406T-E) and circuit worked perfectly fine when using a test LED, which is just a small LED and a resistor that I plug in instead of the coil's leads. This is how I adjust the comparator before plugging it into the coil.

The driver is inverting, so it provides no signal when it has an input voltage, and only turns on when the input is low. I've verified this functions correctly with the test LED and my oscilloscope.

Since everything looked good on the test LED, I moved onto testing with the coil plugged in, but it kills the gate driver when I insert the projectile! As soon as it turns on, the driver gets stuck on, which leaves the MOSFET on. I've unplugged it and tested with the LED again, and can see that regardless of the voltage comparator signal (which is still outputting correct values), the gate driver is stuck outputting 12 V.

I've also verified that the MOSFET itself is still fine by removing the driver, and poking the gate with the 12 V wire (and adding a discharge resistor).

Some things to note:

• In the future, I'll probably go for a Zener diode instead of a flyback resistor to cap the voltage to a predictably lower value than the MOSFET breakdown. But considering I ran 2 tests with the flyback shorted, I don't think the flyback voltage is the culprit.
• I've done some research and realized I should have added decoupling capacitors at all the ICs (comparator + driver), but that still doesn't really explain why it worked in the old setup but broke in the new one.
• Some people have mentioned the EM pulse could induce voltages in the photo diode leads, which might trick the comparator/driver to turn on. Unfortunately I didn't think to hook up the voltage comparator to my scope while firing, but again it worked in the old setup!

I'm completely at a loss as to what to do to fix this. If anyone has any ideas I will graciously accept anything. It literally feels like I've lost everything right at the finish line and I'm tearing my hair out at trying to figure out the issue.

Answer 1

So after much trial and error (and about 25 drivers blown up for testing), I've come to find a number of problems and their solutions. Thank you everyone for the comments, and I hope this helps someone who runs into the same issue:

1. (Biggest problem) I did not account for the flyback RLC voltage oscillations! While the flyback diode + resistor was chosen to clamp the voltage @ ~80-90V, the RLC oscillations on top of that was making it peak out past the 100V MOSFET limit. While the MOSFET was able to function fine afterwards, parasitic capacitance in the circuit was causing the voltage transients to ripple throughout the circuit, which at the MOSFET driver was occasionally hitting 24V (well above the 12V input). This is outside of the driver's 20V max supply range, destroying the driver. An RCD snubber network across the MOSFET removed the RLC oscillations, and I never saw it peak above the expected 80-90V range afterwards (even after 100+ firings!).

2. Decoupling capacitors around the driver were useful, not just for providing a local burst of current, but because it further damped the voltage transients that rippled through the system. This dropped the transients even lower, to 17V at most.

3. I should have picked a higher R1 resistor value (sensor resistor). I didn't realize how little current the diode would allow even with full light exposure, so I was only getting about 1/2 the voltage range to work with for sensing. Using a 1M ohm resistor made it better to tune and also minimized noisy switching coupled with the turn on/off transients, which caused erratic switching oscillations.

All 3 of these problems catalyzed each other. Now that they're all addressed, the system is very stable, with almost 0 unwanted switching oscillation! I was able to do a multi-stage test over 100 times, and not a single part got warm to the touch :)

Answer 2

If I understood correctly, it works with one stage with the projectile. The only major difference between one stage, and multiple stages is that the projectile is travelling faster.

It should be noted that the voltage induced by a magnetized piece of metal traveling through a coil is proportional to the speed. It's possible that with more stages you created enough kickback voltage to break down the isolation between the MOSFET gate and drain, which in turn damaged the driver.

To confirm this, with the power disconnected, I would measure between the MOSFET gate and drain, and also measure between the MOSFET drain and source. If either are short, then that's the likely cause. Also, measure the resistance between the driver output and the positive power supply (between pins 1 and 7) with the power off to confirm a short.

It may also be worthwhile to either use a higher voltage MOSFET or put a TVS across the MOSFET drain/source.