DRSSTC trouble

[buchtawill]

Hi all,
I am a high school senior in the US and I started making a DRSSTC back in January 2020, taking inspiration from Mads' DRSSTC I. I made the oneTesla kit a number of years ago and decided to come back to the wonderful land of Tesla coils. I have also made a spark gap tesla coil, which can be seen off to the right in the video below. Working on and off during the coronavirus lockdown, I had my first successful test last month (200VDC bus from variac):

It blew right after I took this video, which I attribute to my poor bridge design: https://drive.google.com/file/d/1YcG2pWm8kf7eFIVq646JIEfXXsDc9pin/view?usp=sharing. Whole rat's nest of wires. I have since remade the bridge with IXXN110N65C IGBTs and busbar, found here: https://drive.google.com/file/d/1Ah39WuPqT2fYzhU6FvLrx1yl8nClExOo/view?usp=sharing. Since making the new bridge, I have not been able to get it running. New MMC is 3x7 CDE942C20P15 for 0.35uF @ 6KVDC, or around 900VAC derated @74kHz. Bus is 15,000uF 400V Kemets(2 x 7,500). Driver board is a variant of oneTesla's: https://drive.google.com/file/d/1pFHJJzeRBw8jN1Er48E7NNonjFVLksk1/view?usp=sharing. I use a 15v 1:1:1:1:1 GDT, something I plan to change in the future. I also added an OCD circuit after blowing the first set of IGBTs, which is not shown on the schematic. For an interrupter, I used oneTesla's SD interrupter. It's designed for much smaller coils, so I plan on modifying that in the future as well, but it works for now.

After firing up the new coil last night, no sparks flew. I have been stuck scratching my head. I was running with 130VDC from a variac. I tried flipping the feedback CT, but still no luck. I did notice, however, that the MMC gets charged, which is curious (After signaling the coil to fire and nothing happened, I shut everything down and measured a quickly declining voltage across the MMC). Any thoughts/comments/suggestions on what to try next? If you have any clarifying questions, please let me know and I'll try to get back ASAP.

[davekni]

Likely not your initial issue, but it's better to have CT (current transformer) on one of the two H-Bridge outputs, not after the MMC.  After the MMC is several kV, so can cause corona and insulation failure eventually.

The schematic you reference looks like a design for antenna feedback.  DRSSTCs need primary current feedback.

Do you have an oscilloscope?  It's very difficult to diagnose issues without one.

[buchtawill]

Hi davekni,
Thanks for the reply. The driver circuit does use primary feedback, H3 and H4 are connections to a 33:33:1 cascaded transformer. I borrowed it from oneTesla, which also employs primary feedback. That being said, I am open to redesigning it/moving towards a UD2 type controller. I will move the CTs for my next test. I do, thankfully, have an oscilloscope which I can use, but I am nervous about putting it near something like a Tesla coil while in operation. It would definitely be useful to see GDT output on top of primary current. Anything else I should measure for the next test?

[davekni]

H3 and H4 feed a circuit typical of antenna feedback.  It is not a suitable load for the CT.  Yes, clone the input part of UD2.7.  Or, here's a variation I prefer:


For schematics such as yours that don't need both comparator outputs, I prefer TLV3501 for its rail-to-rail input range.  Also, the above circuit is weakly self-oscillating.  That guarantees startup no matter what initial state the H-Bridge outputs may be in.  (With UD2.7, a bleed-resistor across the H-Bridge outputs or other way to initialize state is needed for reliable startup.)  The only nuisance with the above circuit is that C1 and C2 need to be scaled to roughly match your operating frequency.  Above values self-oscillate at roughly 130kHz.  Scale caps by 1/frequency.  Doesn't need to be precise.  Even if self-oscillation is at half of operating frequency it will work.

Another detail in your schematic:  The GDT coupling capacitor C10 doesn't have any damping (additional R and C) as in the UD2.7 schematic.  That may be OK, but can cause trouble with the resulting low-frequency series resonance of C10 and GDT inductance.

Yes, scoping IGBT gate waveforms is helpful, as is scoping H-Bridge outputs.  This is if you have an isolated power supply for the H-Bridge or a differential probe.  If neither is an option, then at least scope the driver output (input to GDT) and current feedback.

[buchtawill]

Hello once again,

Success! Or at least somewhat. I moved the OCD and feedback CTs to before the cap bank as you suggested. I ran a test with 190VDC bus (I think part of my problem the other day was a too low bus voltage) and got sparks so small I could only hear them, not see them. Please note that the resonance is ballpark tuned (10% lower than secondary rule), I still need to do fine adjustments. After this somewhat successful test, I tried playing with my OCD circuit. I raised the threshold a little and saw negligible effects, and proceeded to disconnect the OCD CT entirely. Here are my results:

So the pros: it works with decent sized arcs
Cons: There is something wrong with my OCD circuit
Any tips or tricks for OCD? I borrowed the OCD circuit chunk from https://www.stevehv.4hv.org/DRSSTC1/DRSSTC1OCDsch.JPG. My connection to the transformer is pretty long, there may be some issues with noise.

[davekni]

Use Steve's feedback circuit as well as OCD.  The feedback circuit you showed previously is for antenna input.  Better, use a UD2.7-style feedback which has phase-lead (or the variation I posted earlier).  Phase lead will reduce stress on IGBTs, making them more robust to poor bridge layout (parasitic wiring inductance).

The wrong feedback circuit w/o CT load is the most likely cause of OCD issues.

If capacitively-coupled noise is the issue, increase C15 from 1nF to something larger.  100nF should still respond plenty quickly, but provide much more noise immunity.