Some questions about my first drsstc.

[davekni]

I tested the bridge with lipo in series for a total of about 110Vdc on the bus, here is what it gives :

Common inductance of scope leads and power connections to H-bridge output is significantly improved.  Makes viewing easier.

We agree that the max voltage Vce of a single IGBT is not the "peak to peak" value on the oscilloscope but the "max" value which is around 236v, in theory I can have 1200v peak to peak to have a max of 600v on each IGBT ?

Yes, presuming output waveform is symmetric, which it generally is.  More accurate to look separately at positive peak voltage and negative peak voltage.  Neither should exceed +-600V.

[Mathieu thm]

So I was able to test the drsstc in MIDI and here is the result! I tested it at around 110Vac:

/>

/>

IGBTs are cold, which is good news.
So I think I can test with 150Vac to see the behavior.
I really feel that the DRSSTC is in better "health"

[davekni]

IGBTs are cold, which is good news.
So I think I can test with 150Vac to see the behavior.
I really feel that the DRSSTC is in better "health"

Yes, looks good.  BTW, there are a few places in the videos, especially first one, where sparks and sound seem to fade out.  I presume that corresponds to quieter parts of the MIDI input and not some coil performance issue.

[Mathieu thm]

Yes, looks good.  BTW, there are a few places in the videos, especially first one, where sparks and sound seem to fade out.  I presume that corresponds to quieter parts of the MIDI input and not some coil performance issue.

In fact it's me who modifies the on-time to see how the sparks react.

[davekni]

So adjust the phase before the primary current, this corresponds to the igbt turn on a little before the primary feedback, and adjust the phase after the primary current, this corresponds to the igbt turn on a little after the primary feedback ?

I don't know if it translates like that but, I have either spikes that appear from rising edges or from falling edges, here is a diagram that illustrates what i mean:

Seeing actual scope captures will make it more clear, especially zoomed-in horizontally to see details of spikes.  The general answer is that IGBT switching should occur slightly before zero current times.  IGBT turn-off occurs first, then the short dead-time gap, then turn-on of opposite IGBTs, then current crossing zero to other polarity.  This is proper phase-lead adjustment.  Should minimize spikes and minimize IGBT switching power loss.

[Mathieu thm]

Seeing actual scope captures will make it more clear, especially zoomed-in horizontally to see details of spikes.  The general answer is that IGBT switching should occur slightly before zero current times.  IGBT turn-off occurs first, then the short dead-time gap, then turn-on of opposite IGBTs, then current crossing zero to other polarity.  This is proper phase-lead adjustment.  Should minimize spikes and minimize IGBT switching power loss.

Thanks, I'll look at the waveforms next time.

As everything worked well with 110Vac, I tested the DRSSTC at around 150 Vac, after having it running for quite a long time, the IGBTs were cold (with peaks at 900-1000w on the highest notes), so I think the phase lead is pretty well tune.
Here are two videos of the tests at around 150 Vac :
The videos are a little blurry, I apologize, I only noticed that after. It's quite hard to focus in the dark.

/>

/>
I also noticed that the OCD never turn on at 130 Vac even on heavy ground strike.
At about 150 Vac with an OCD of about 280A (I'm not sure, I need to check it) the OCD rarely triggers even on heavy ground strike.

[Mathieu thm]

I need to know something, it may be a bit of a silly question, but I would like to be sure. When I do my tests with a dummy load, at about 110Vac, with MIDI mode, I put the on time at 100us and I leave it constantly at 100us, so for about two minutes, there is 300A passing through the primary circuit and I draw around 7A on the mains on the highest notes (IGBTs are almost not lukewarm).
When I test with the secondary coil, at the same bus voltage, I draw around 6A on the highest notes, but on the highest notes I feel like it's going "heated" more compared to the test with a dummy load and I dare not pushing the on time too high.
I think it's all psychological, we agree that IGBTs cannot heat up significantly more with the secondary coil with the same primary current and the same (even a little less) mains current draw ?

[davekni]

I need to know something, it may be a bit of a silly question, but I would like to be sure. When I do my tests with a dummy load, at about 110Vac, with MIDI mode, I put the on time at 100us and I leave it constantly at 100us, so for about two minutes, there is 300A passing through the primary circuit and I draw around 7A on the mains on the highest notes.
When I test with the secondary coil, at the same bus voltage, I also draw around 6-7A on the highest notes, but on the highest notes I feel like it's going "heated" more compared tested with a dummy load.
I think it's all psychological, we agree that IGBTs cannot heat up significantly more with the secondary coil with the same primary current and the same mains current draw ?

If I understanding correctly, you are comparing equal mains current.  Equal mains current indicates equal total power input.  (Mains power factor is likely the same at the same current.)  Equal mains current does not imply equal primary coil current.  It may require significantly higher primary RMS current to cause similar power dissipation.  Depends on the Q of your dummy load compared to the system Q of your operating Tesla coil including arc loading.

There are other possible (but less likely) causes for H-bridge power dissipation (heat) to be higher in operation than with dummy load, including arcs slightly disrupting driver circuit operation (if not sufficiently shielded).

[Mathieu thm]

Seeing actual scope captures will make it more clear, especially zoomed-in horizontally to see details of spikes.

I made scope captures with a bus voltage of about 100v:




I also did a test with MIDI at around 140Vac, I set the on-time at 100us (I did not reduce the on-time on the high note like the last time) and I set the OCD at 300a :

/>
For the first time, I completely closed the case with the air vent plate.



I'm happy, the IGBTs are cold after this run, I also did a test with MIDI with slightly higher notes and the IGBTs don't even heat up !
I think the frequency of the primary circuit is a little bit high, I would try to adjust the primary for a frequency of about 5-10khz lower.
(currently the primary is at 190 khz and the secondary at 213 khz)
I also think I can increase the bus voltage a little more, I think to adjust my OCD for a final value of 325a.

[davekni]

I made scope captures with a bus voltage of about 100v:

Looks good.  Spikes near the beginning of bursts are common.  UD2.7 phase lead and duty cycle are less accurate at low current (low feedback input voltage).  Also, initial low current may not be enough to cause H-bridge output voltage transitions fast enough during the dead-time between one pair of IGBTs switching off and the other pair switching on. Your switching looks nice and clean after the normal start-of-burst low current spikes.

Enjoy your coil!  Hopefully no more IGBT failures.

[Mathieu thm]

Looks good.  Spikes near the beginning of bursts are common.  UD2.7 phase lead and duty cycle are less accurate at low current (low feedback input voltage).  Also, initial low current may not be enough to cause H-bridge output voltage transitions fast enough during the dead-time between one pair of IGBTs switching off and the other pair switching on. Your switching looks nice and clean after the normal start-of-burst low current spikes.

I don't understand why I couldn't remove the spikes at the beginning of the burst, thank you for explaining it.

Enjoy your coil!  Hopefully no more IGBT failures.

Thanks a lot! Now I reach a little over 70 cm of streamers, which I find rather impressive for a DRSSTC of this size, I also wanted to thank you, you made me understand a lot of things that I did not understand well 

[Mathieu thm]

Hello everyone,

Since my last post, I made some changes to my drsstc.

I detune the primary a bit more for about 8-9%, I added an ammeter on the front panel of the tesla coil and made a small protection on the left side to protect the primary circuit.



My final OCD setting is 315A, with a little more detune I was able to enlarge the arcs a bit.
I used it quite a lot, almost every weekend for about 1h-1h30 and I no longer have any IGBT problems!

/>