1000V Electromagnet Driver

Hi everyone

Im attempting to build a high voltage (1000V) electromagnet pulse driver using a 1200V rated IGBT with a gate driver and an arduino for switching.
I have attached datasheets for the IGBT and the driver.
I have attached a schematic as well. It simulates fine, but obviously iv screwed up somewhere.

I simulated at 48V, 600V and 1000V using isis. According to the simulation the electromagnet draws 20A at 600V, well within the datasheets specs.

The resistance of the electromagnet is 30 Ohms.
I initially tested the setup with 48V DC from a Meanwell power supply to drive the electromagnets and a 12V battery to power the transistor gate driver.
All worked flawlessly! Opening and closing the gate of the IGBT :slight_smile:

I then substituted the 48V supply for a voltage multiplier and powered that with 220V AC, giving me 600V DC.

I powered up the driver, then the arduino, then the multiplier. Just as I had done before with the 48V supply.
This time it energised the electromagnet and it stayed energised. The 2 electromagnets are not meant to be driven for more than 10-50ms.
So they overheated and started to melt their bobbins.
I powered everything down quick quick! Then tried again, the transistor no longer switched, and was permanently on.

I felt the driver and it was too hot, the IGBT however was not hot at all.

I checked the resistance of the IGBTs emitter to gate and I get a dead short. Obviously I murdered that poor transistor.

I have to admit, Iv never been too good with high voltage stuff, I like my 5V logic signals! :wink:

Where did I screw up and why? Did I miss something obvious? Was it the back EMF from the coils? My lack of intelligence or my bad luck? :slight_smile:
Go easy on me :wink:

Thanks in advance!


PS. While typing out this post I think i figured out where I screwed up. In simulation I used a single 30 Ohm solenoid, it drew 20A at 600V.
In testing I used two 30 Ohm electromagnets, which obviously drew 40A, which is above the rated current for the transistor.
Am I correct?

IRG4PH40KD.pdf (347 KB)

TC4426-TC4427-TC4428 1.5A Dual High-Speed Power MOSFET Drivers.pdf (431 KB)

Well that's not going to help...

The problem is likely the large dV/dt swing pulling the gate well below 0V, popping the gate oxide and
frying the driver (btw I would only ever consider using opto-isolated drive at these voltages
with fast logic opto-couplers). 600V in 31ns (the on rise-time) is 19GV/s, which will pump 0.5A
directly into the gate via the reverse transfer capacitance of 26pF, so you must have a high current
driver connected directly to the gate with a low-inductance feed. The driver you used is rated for
0.5A reverse current though, but only if you used enough decoupling on it - 0.1uF + 10uF ceramic
right next to the chip would be essential to control the 12V line I think.

So another thing that might have happened is the large 0.5A current from the gate pulled the
output of the gate driver chip down towards negative voltages, this in turn pulled the +12V
rail negative locally, perhaps put the driver into CMOS-latch-up (reverse supply polarity will do
this), then the MOSFET gate voltage was ill-defined for a few milliseconds until the IGBT melted
(because the device wasn't hard on or hard off it dissipated a few kW).

However if you did have proper decoupling on the driver chip I am surprised it blew, 600V isn't excessive.

Did you check the 1N4007 survived?

The moral is you should have gone from 48V to 100V only and put a good 'scope on the
system and checked everything was switching correctly... Then to 200V, then to 400, then to 600.

Hi Mark!

Thank you so much for the help! I really appreciate it :slight_smile:

I have modified the circuit with your suggestions.
Also im using a different IGBT, one rated at 600V. I cant afford to pop a 1200V $10 chip everytime.

Iv added an optoisolator and decoupling caps to the VDD of the IGBT driver.
I tried reversing the supply polarity on the IGBT driver but proteus wont let me.
All simulates fine.

But will it work?



irg4bc40s.pdf (162 KB)

As I said proceed in steps with higher and higher voltages, observing with an oscilloscope
that the transient is exactly as expected. You need a variable high voltage supply and
oscilloscope and take no risks with these high voltages, they kill without compunction if
you take stupid risks.

You can limit the total energy by using a capacitor bank - you can then just look at
the first 10us of the transient knowing the available current will drop sharply thereafter.

And I should also have said wear eye-protection when doing high power electronics,
devices can explode violently.