Anecdotal Advice

[buchtawill]

Hi All,

I've made a lot of progress on my DRSSTC, and I finally got sparks longer than 1 m!

As you can see, I let out the magic smoke. The most likely explanation is that my OCD was set too high and it did not trip during a ground strike (it may be hard to see, but the final arc hit the ground ring). I have 4x IXXN110N65. The datasheet (attached) says it has a peak current of 470A @ 1 ms, but I know this does not help much when dealing with 75kHz DRSSTCs. That being said, does anyone have any anecdotal evidence of the maximum current an SOT-227 IGBT can handle? Also, could TVS diodes be an alternative to Zeners on the gates?

[TMaxElectronics]

I don't think that was caused by the OCD not triggering. It doesn't seem to go off as the coil is running too often anyway so I'm guessing the current would stay within a reasonable margin even if it failed to trip. My guess would be arc over somewhere or maybe inadequate overvoltage protection on your bridge.

I have used a whole bunch of MOVs on the output of my bridge and that has survived multiple strikes (my controller didn't survive the last lol). Not sure if they are designed to trigger and go high enough resistance again fast enough but MOVs are used for lightning strike protection so I think it should be fine. I just used a whole bunch so even if I discharged my entire bus capacitance into that they would have no issues whatsoever.

[buchtawill]

Hello,

Thanks for the response. I will consider adding MOVs. What ones do you use / what part numbers? One thing I forgot to mention is that you can see the OCD trip (signified by the red LED on the bottom left) in the same frame as the ground strike https://drive.google.com/file/d/1SDe58H46HlGJ7jnieNfvnLSN-CbcEb55/view?usp=sharing.

[TMaxElectronics]

I used B72220P3511K101, but again I have no idea if they are actually the right thing for the job, and can only say they worked for me

I looked for some that would not conduct during the maximum bridge output voltage plus some safety margin and took the largest one digikey had in stock. That is not to say that that is the way best way to go though... there might well be a cheaper alternative

But I just realised you said ground strike and not primary strike, my bad. That might mean that the failure has nothing to do with bridge output overvoltage protection. Even though the very sudden death kind of makes me wonder if it may have been some flashover thing after all

[davekni]

The ground strike is hitting the second ring in from the outside.  Is that also a ground ring, or is it an outer primary turn?  A primary hit can be problematic.  MOVs may help there.  If a ground strike, is there a solid path from that ground ring to ground and/or counterpoise, not a wire that runs right by circuitry?

High primary current from a ground strike is a reasonable possibility too.  That is the first ground-strike in the video, and while playing a low note that likely has longer on-times.  Many factors affect how much current any given IGBT can handle, including how clean gate waveforms are and if phase lead is adjusted properly.  If H-Bridge parasitic inductance is high and phase is lagging, high current may cause a failure due to resulting switching spikes exceeding IGBT voltage rating.

MOVs can have significant capacitance, although the B72220P3511K101j part Max used is fairly low at 600pF.  It looks good for 1200V IGBTs.  Clamping voltage is a bit high at 1355V.  MOVs would be most helpful with short leads to minimize parasitic inductance.  MOVs can help clamp switching spikes if that is your issue.

[buchtawill]

Thanks again for the responses.

I looked for some that would not conduct during the maximum bridge output voltage plus some safety margin and took the largest one digikey had in stock. That is not to say that that is the way best way to go though... there might well be a cheaper alternative

I went the same route and found myself V250LA4P‎. Maximum clamping voltage is 650v (perfect for 650v IGBTs), and I don't plan to run my bridge at over 200v anytime soon(staying within typical varistor limits). Datasheet: https://www.littelfuse.com/~/media/electronics/datasheets/varistors/littelfuse_varistor_la_datasheet.pdf.pdf

The ground strike is hitting the second ring in from the outside.  Is that also a ground ring, or is it an outer primary turn? If a ground strike, is there a solid path from that ground ring to ground and/or counterpoise, not a wire that runs right by circuitry?

High primary current from a ground strike is a reasonable possibility too.  That is the first ground-strike in the video, and while playing a low note that likely has longer on-times.  Many factors affect how much current any given IGBT can handle, including how clean gate waveforms are and if phase lead is adjusted properly.  If H-Bridge parasitic inductance is high and phase is lagging, high current may cause a failure due to resulting switching spikes exceeding IGBT voltage rating.

The ring it hit was the ground ring. Not only was it the first ground strike in the video, it was the first ground strike ever for this coil (arcs are finally long enough  ). I did take some time to make good ground connections; I have a small terminal bar and a good RF ground. Regardless of whether or not the IGBTs died from overcurrent or overvoltage by virtue of overcurrent, I will add MOVs for extra protection and lower my OCD threshold.

One more thing that I think might help: The arc hit the sharp part of the hose clamp I used to connect the ground ring to ground, so I will work on making a smoother connection as well. ALSO I took a leaf out of Dave's book and tuned the coil by adding a string of caps to my MMC.

[davekni]

One more thing that I think might help: The arc hit the sharp part of the hose clamp I used to connect the ground ring to ground, so I will work on making a smoother connection as well.

That makes sense.  I had a similar issue with my first crude ground ring.  Initially thought that a rough ground ring would help make arcs hit the ring instead of the primary.  However, in my case, it caused arcs from mid-secondary to ground ring points.  I now have three concentric ground rings of smooth copper pipe.

Looks like a good varistor part, rated for up to 330Vdc.  A few in parallel individually wired to bridge terminals would reduce inductance and clamp higher current.