DRSSTC II

[AstRii]

Hello!
Long time haven't i been here since i couldn't do anything with my DRSSTC I... My neighbors complained that it interferes with their TV so i can't run it at home and since everything (even school) is closed thanks to coronavirus, there is no way for me to test it. Nevertheless i decided to meanwhile begin the construction of my 2 meter high DRSSTC II. This time i would like not to make mistakes because i don't want to destroy IGBT bricks, so to be sure i would like to share with you my build with great detail.
So here are the specs of my coil:

Secondary - wounded about 2500 turns with 0.35mm dual isolation wire. Resonant frequency with toroid is approximately 60kHz.
Primary - total of 8 turns with 1cm copper pipe, currently resonating with the secondary at about 4.2turns at frequency of about 53-59kHz.
MMC: 660nF 12kV symetric system 2x 5S3P combination of MPAPB1K2W22J0M6 1200V 2.2uF snubber capacitors rated 1320A peak current. bought from https://cz.farnell.com/multicomp-pro/mpapb1k2w22j0m6/snubber-capacitor-2-2uf-1-2kv/dp/3106538
IGBTs: Full bridge of MG150Q2YS40, you can download the datasheet from here: http://www.datasheetcatalog.com/datasheets_pdf/M/G/1/5/MG150Q2YS40.shtml
DC bus capacitors: 2S2P combination of 10000uF 450V for total of 10000uF at 900V.
Driver: UD+ with pulseskip OCD from PhilsLab.
Topload: 3D printed 60-90cm rings covered with aluminium foil and tape.

Since i don't want to make this message very long, i attached all the images and waveforms as a pdf file on this post. DRSSTC II WAVEFORMS AND LAYOUT.pdf

So there are few things i'm worried about...
1.) The primary current right now is almost 150Apk at only 30V! Which seems crazy to me, with my 220kHz DRSSTC I, i had 150Apk at nearly 100V input. I want to power the coil from 3 phase rectified power which is 560V DC. If the current on the primary grows linearly, then i will have 2700A at full power, which would definitelly blow up my IGBTs. Which brings me to another problem...
2.) If you downloaded the datasheet of my IGBTs, you can see that they are only rated 300A peak current. But i think i can push them much higher, i don't belive that these bricks which are the same size as CM300 bricks could only push 300A of current at peak... What do you think? How many amps would you try to push at maximum from these bricks?

Thank you for all your help and suggestions. Hope you're safe during these hard times

[Mads Barnkob]

Very nice construction of your coil, there is not much to critique, this will all work just fine. Do not worry about cosmetic imperfections of your secondary or primary coil, they will do just fine if varnished properly and primary turns are not exactly so near they can arc to each other

First, low voltage testing is deceiving! Look at page 4 of the IGBT datasheet for the capacitance vs Vce graph and you can see that Coes is about 3000pF at 30VDC, but only 600pF at 100VDC, Coes is the output capacitance of the IGBT module and this will give you extremely high switching transients when looking at the inverter output voltage. So these switching transients will get lower at higher voltages!

Second, your gates are ringing way too high up, use a higher value gate resistance and make sure its not a wire wound resistor! Wire wound resistors have a high self-inductance that could only make things worse.

Now to your questions:

1) Current does not scale linear to the voltage, the power will be transferred to the arc and at some point your system will be in equlibrium. You do have OCD set for a maximum safe current dont you? There should not be a need to worry about that in 2020

2) Almost all IGBT brick specifications are for hard switching, when doing soft switching we can push them way harder (read and calculate if you want: http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/ ) so looking at the 300A Icm specification, I would say that since these are not THAT fast, some 800-1000A would properly be the maximum. Your coil is rather large and you should give it a large IGBT bridge at some point, you have afterall dimensioned everything else to use a bridge that could be pushed to 1500-2000A.

Good job and happy coiling

[davekni]

Concerning secondary imperfections, is one end better than the other?  I'd suggest putting the better end at the bottom.  Volts/turn will be higher at the bottom adjacent the primary, so any overlapping turns are more likely to arc though insulation there.  Good varnishing helps as Mads said.

Also as Mads said, ringing will improve at higher bus voltage.  However, you are running with significant phase lag.  Those Toshiba IGBT bricks have rather slow diodes, which cause current spikes if turned off rapidly, which happens without phase lead.

I'm more skeptical than Mads about pushing current with these old slow IGBTs.  They have relatively high Vce, and attempting to extrapolate the current vs. Vge plots suggests a need for high Vge to get high current.  Perhaps you could get 500A with >= +-20Vge.

Your primary impedance is similar to my DRSSTC, but you are targeting a higher bus voltage and lower current.  I think you will get better performance using more of your 8 primary turns and reducing MMC capacitance, perhaps 7S2P on each half.  My current limit is set to 2600A, with intention of 3600A eventually.  Performance is reasonable at 1000-1500A and 300Vbus, but that's not within reason for the IGBTs you have now.  Higher primary impedance would allow similar performance at 560V and 500-750A, still a stretch for those IGBTs.  Your nice large coil will require better IGBT bricks to drive its real capability.

[Mads Barnkob]

Concerning secondary imperfections, is one end better than the other?  I'd suggest putting the better end at the bottom.  Volts/turn will be higher at the bottom adjacent the primary, so any overlapping turns are more likely to arc though insulation there.  Good varnishing helps as Mads said.

Also as Mads said, ringing will improve at higher bus voltage.  However, you are running with significant phase lag.  Those Toshiba IGBT bricks have rather slow diodes, which cause current spikes if turned off rapidly, which happens without phase lead.

I'm more skeptical than Mads about pushing current with these old slow IGBTs.  They have relatively high Vce, and attempting to extrapolate the current vs. Vge plots suggests a need for high Vge to get high current.  Perhaps you could get 500A with >= +-20Vge.

Your primary impedance is similar to my DRSSTC, but you are targeting a higher bus voltage and lower current.  I think you will get better performance using more of your 8 primary turns and reducing MMC capacitance, perhaps 7S2P on each half.  My current limit is set to 2600A, with intention of 3600A eventually.  Performance is reasonable at 1000-1500A and 300Vbus, but that's not within reason for the IGBTs you have now.  Higher primary impedance would allow similar performance at 560V and 500-750A, still a stretch for those IGBTs.  Your nice large coil will require better IGBT bricks to drive its real capability.

Would adding faster external diodes help on the internal slow diode problem?

He should do the Fmax2 calculations to get it sorted out how fast these can run, and at how high current. You are right about the Current vs. Vge at 125 degrees Celsius is bending downwards quite a bit, but the 25/40 curves are straight and going for the moon.

My point here is really that the IGBT bricks are too small for this big coil, pushing them to the limit and maybe blowing them up is not a great loss as bigger IGBTs should be a first priority upgrade anyway

Alternatively I agree on your suggestion of a higher impedance primary circuit to run longer on-times at lower current to get the same spark output.

[AstRii]

Well, i tried to calculate Fmax2, and i ended up losing more power on the transistors than i can dissipate. It's attached here if you want to have a look: Fmax2.pdf
Now i completely agree with you that the coil deserves better IGBTs. I wanted to use these, since i bought them very cheap on local market (about 10€ for both bricks) and i didn't have any other bricks to use.
I was hoping i could push them to at least 700-800A, now i'm also skeptical about this. But i will try it nevertheless, we'll see...

Yes Mads, the driver does have freewheeling (pulseskip) OCD. And i agree with both of you, i will decrease the capacitance of my MMC to use more primary turns.
My goal is to get some ground strikes, which means at least 2 meter long streamers. If the IGBTs will survive this goal, i will keep the coil as it is. If they will blow up, i will buy new better IGBTs.

I set up the OCD and it works great, now i'm going to solve the high voltage ringing on the gates and i still have to set up the phase lead for the IGBTs to switch in phase with current (This will take a lot of time for me, since here in Czech Republic nobody is selling adjustable inductors, so i have to compromise:) ). After that i'll try some power tests.
Thank you guys for your suggestions

[AstRii]

So i just added the gate resistors. The gate resistance went from 0R to 1R and the voltage spikes reduced by 2V from 38V to 36V. I was hoping for a better improvement but i also realized that i have no zener diodes on the gates. I'm going to buy some 27V zener diodes to protect the gates.

[davekni]

 You'll likely need more than 1 ohm gate resistance, unless GDT leakage inductance can be reduced dramatically.  People often use 2-5 ohms for the large bricks, so 5-10 ohms on each gate might be appropriate here, perhaps with 2-3 ohms in series with the parallel diodes to speed up turn-off without excess undershoot.

Good point about pushing these bricks - nothing to loose since the alternative is to replace them anyway.  You do have high enough gate voltage, so perhaps 700-800A will work.  Higher primary resonant impedance will help get performance at 700A.

Fast parallel diodes could help with the slow internal diodes, but it may be hard to find any with enough lower forward voltage to conduct the majority of current.  Phase lead will avoid diode issues without adding parts.  The ringing on the bridge output waveforms will mostly go away once you have sufficient phase lead.  The ringing is likely caused by diode reverse recovery current.

[Mads Barnkob]

So i just added the gate resistors. The gate resistance went from 0R to 1R and the voltage spikes reduced by 2V from 38V to 36V. I was hoping for a better improvement but i also realized that i have no zener diodes on the gates. I'm going to buy some 27V zener diodes to protect the gates.

I always start out with 4R7 and then move up or down according to the measured waveform. A single 5% overshoot without ringing is perfect.