Non-Isolated power supply for three phase inverter

[smanches]

I just went off a recommendation that I found someplace. They are 101k 1kv ceramics. Quite small, but I did look at the switching waveform with my scope. It made quite a difference, even being that small.

My load is only 2A at stall too, so much lower than yours. Although your test load was only around this much.

Do IGBTs have reverse diodes in them like MOSFETS?

 

[tgrandahl]

Yes, it looks as if the IGBT's i was using did have freewheeling diodes on the HI and LO side IGBT's.

I will be using some FSBB30CH60 smart power modules now, as I am out of the **broken link removed** and got some free samples of the FSBB30CH60 modules. They still do use IGBT's and have freewheeling diodes on the output, i am assuming that these diodes kick in at the max DC bus voltage.

I looked through an application note concerning these smart power modules, snubber capacitors are discussed on page 28 here . It looks like you just provide one as close as possible across V+ / V- and the same general 0.1uf value is recommended.

I am trying to understand why the freewheeling diodes and snubber capacitor are placed where they are. I was expecting them to be placed across the load, as a relay would. However it appears the voltage spike here is passed back into the dc buss where it is suppressed by the snubber cap? I am referring to the examples in the application note and IEEE article I referenced.

I am trying to understand how this could have lead to the massive failure that occurred the other night. From what i am understanding now, this circuit without a snubber cap but with the diodes was essentially acting as a pseudo boost circuit from the high frequency pwm.

Was this pseudo boost dumping enough charge into the DC bus to momentarily overcome the 600V collector emitter breakdown voltage?

In that case could this have damaged the IGBT gate so that I am seeing this 20ohm resistance from gate to emitter?

 

[smanches]

It may not have been the boost alone, but without snubber caps there could have easily been very high voltage spikes during the turn on ringing. I myself have seen a 2.5X increase on the biggest spikes. And I bet they can get a lot higher.

Yes, place the snubbers right next to the collector and emitter pins. You want to catch as much of (as close to) the ringing as you can. The inductance of the traces on the board will cause the snubbers to not be as effective otherwise.

The diodes are there to help with any reverse currents that might temporary be caused by the switching. They are not freewheeling diodes like what you would use on the motor. They're "facing" +V.