Arc quenching in ZVS driven TC

[AstRii]

Hello guys!
I've seen some videos of people driving their SGTCs with ZVS driver (like this guy:

/>And since ZVS driver is so easy to build and i have some spare MMC, i wanted to try the same thing.
I created 30kV 1A diode from 30 UF4007 diodes in series for capacitor charging.
I'm assuming people use this schematics:


When it oscillates, it works well, but the problem is.. it doesn't oscillate for a long time! Once the air in the spark gap is hot enough, an continuous arc is ignited.

Why does this not occur in MOT driven SGTCs and to other people which i've seen are driving their SGTC with ZVS driver?
And how can i prevent this?

Thank you,
Mark
 

[davekni]

Other people have had the same issue:
https://highvoltageforum.net/index.php?topic=822.msg5518#msg5518
High-speed air through the spark-gap will help blow out the arc, along with good heat-dissipation capability for the points.  A rotary spark gap is even better.

The difference with MOT or other AC sources is that the source reverses polarity periodically.  The momentary zero-current point allows the arc to quench.  For my rapid-fire Marx generator, I had to add circuitry to pause the DC charging circuitry for ~1ms after each spark-gap firing to allow the arcs to quench.  (Had to add air flow through the spark gaps as well.)

Your UF4007 string may die after a while.  UF4007 are not avalanche-rated.  At the end of each diode conduction period, the fastest diode will recover first.  It then must conduct the remaining diode's reverse-recovery charge in avalanche-breakdown.  Even avalanche-rated diodes struggle to handle this situation for extended periods.  I've had lots of experience with fried diode strings with my Marx generator charging circuit.

You may also find that T1 fries eventually.  Depends how it is wound.  Without a Terry-filter (some form of series impedance), the rapid discharges from spark-gap firing can fry HV transformer output windings.  The sudden voltage change will distribute voltage inside the transformer based on capacitive diving.  If parasitic capacitance isn't uniform enough, one area of the winding can be voltage-stressed into breakdown.  That same thread linked above includes some issues with TV flyback transformers frying that way.

[johnf]

As Dave Says
you need an inductor between your diode string and the spark gap and as it is DC you will need a very big cored one that will not saturate a MOT secondary would be nice. Also the plasma between the spark gap could do with a stiff air flow to get rid of plasma and ozone products that tend to help permanent gap firing.

[AstRii]

For now i only added a fan to blow air on the spark gap and i manage to keep the spark gap firing for more than few minutes this way. I will upgrade the spark gap so it's not just 2 sharp wires but more round electrodes.
I will try what Dave did with his Marx Generator, i will turn the ZVS on and off, to quench the spark if it ignites.
As a transformer in ZVS circuitry i use home wound transformer with decent insulation, i'll see if it's gonna breakdown, maybe it will survive..
It only produces like 10cm+ sparks, which is not that much, i'd expect more from 30cm secondary.
But that's maybe because of my low capacity MMC (16nF 30kV), i need to use 10 primary turns. In monday i should have more capacitors so i can use less primary turns for better output.

[AstRii]

Hello again!
I see i'm missing out on more SGTC theory than i thought..
The circuit with ZVS driver works and arcs do not ignite anymore thank so high speed fan blowing air on the spark gap. But now there is a different problem...

You can see on this video that each time a spark occurs, some flash happens inside some of the capacitors in the MMC.
I'm using the following circuit:


I was already warned that i need some inductor in series with the diodes not to break the insulation on the transformer, but i can't see where would this voltage come from.
Primary starts to oscillate at the MMC peak voltage, since then there is nothing which would add more energy into the circuit. Those oscillations should be damped oscillations and voltage on the circuit should not exceed
MMC voltage or transformer voltage, since the transformer is also designed to operate at ~30kV. Am I missing something? Because my MMC is okay when charged from the transformer but when there is a spark, there are some flashovers
which i assume are caused by some kind of overvoltage.

Just to be sure that there should not be any voltage spikes i created a very crude simulation of such a circuit:
https://i.ibb.co/7r7Gb3C/image.png
I tried to approximate the primary circuit well, but the secondary is just a very very crude approximation.

Simulation seems to agree with me, that the voltage never exceeds those initial ~30kV of the MMC.

Also, just a side question... I made my capacitor bank 3x bigger, therefore i assume the energy in the circuit will be 3x higher, therefore longer "ontimes" and better sparks?

Thank you for your answers!

[davekni]

Your home-wound transformer may be fine.  It depends on distribution of parasitic capacitance.  The instantaneous change from 30kV to 0V will distribute voltage across the windings based on capacitance, not on number of turns.  If capacitance is too non-uniform, then the section with low capacitance will see a sudden large voltage reversal.  I've destroyed a dozen automotive spark coils this way (running within their voltage rating) in my Marx generator before changing to long strings of small inductors in a physical arrangement to make capacitance uniform.  Others have destroyed TV flyback transformers in the same way.

The MMC arcing may be a similar issue, or just insufficient spacing for AC voltage.  What capacitors are you using in your MMC?  I'd bought some polystyrene capacitors that were listed as good for Marx generator use.  They weren't good for that or any AC use, as they had tiny air bubbles within the polystyrene at edges of the foil windings.  Every voltage change (AC cycle or step discharge) causes corona within the bubbles.  After enough cycles, corona damages insulation until cap failure.  Many caps have high DC voltage rating and MUCH lower AC rating.

[AstRii]

Hi Dave!
I'm using Tesla TC209 capacitors, as they are Czech and here in Czech Republic they are cheap Unfortunately i haven't found any useful datasheet. They are rated 220nF 1kV and i'm using 150 of them for 34nF 30kV MMC (30S5P).

[AstRii]

Oh okay i just found out that they are only rated 250Vrms, well that's a shame, now i need new MMC.
Anybody can recommend some nice capacitors for 30nF+ 30kV MMC?

[davekni]

I'm using Chinese induction cooker capacitors, 0.33uF 1200Vdc 630Vac, polypropylene film.  I did some abuse testing to +-2100V at 1% duty cycle, with a long run at +-1500V initially (several days at 1% duty cycle).  They are available on EBay etc. for a bit under $1 each.

[AstRii]

Thanks for the reccomendation, they do look as the cheapest option. Yet it will take a long time before they arrive from Ebay, therefore i'm putting this project aside meanwhile..