Things have improved quite a bit since my last post (partially thanks to a proper scope). Hopefully my experience will be useful for other people.

The RC snubber is very important. With a good scope, I used this spreadsheet to calculate the required values for R and C (instead of using random small values). After adding the calculated values to the circuit (for a damping factor of 1), the ringing stopped completely, and the current draw doubled, together with the spark length!

However I still couldn't get my interrupter to work. Or rather, it worked as long as the coil was off. When the coil was on, I just got a 0v output (due to interferences, I guess). I tried everything I could with no success, and eventually decided to go for a separate supply for the control circuit, hoping to solve the issue but also to later on use a higher voltage for the primary coil circuit.

The separate supply didn't change anything on the interrupter part (still a flat 0V signal), until I increased the voltage from 15V to 32V on the primary coil circuit. Somehow this made the interrupter work, I'm still unsure why. The control circuit runs at 12V.

So I now have quite reasonable sparks (I would say 4 to 5 cm), and I can play tones, which is nice. I got also a huge improvement in noise/interferences by grounding the circuit (using mains grounds): the sound is much clearer (and the sparks much sharper).

Next, I would like to try to change the coupling to get more current draw (and longer sparks of course), i.e. 1 turn only at the primary instead of 2, and try to place it at different heights. I'd like to also play with the PWM duty cycle to see if it changes anything (I'm using 50% duty cycle at the moment).

Thanks again for your great advices.

It only stops oscillating when I try to draw arcs. The feedback is direct feedback from the secondary base to the schmitt trigger input (with diodes to clamp it), as in the circuit schema in my first post. There are a lot of interferences in the air near the circuit; I plan to shield it in a grounded ATX power supply case when I'm done. And indeed, an interrupter will fix this issue, so I'm not too worried about it.

About toploads, I hope it will improve performance because the arcs I get are still very short compared to the ones in these videos, where the same design is used: and />
I can think of a few reasons why I get relatively small arcs with respect to these videos:
* No topload
* Running in continuous mode (without an interrupter)
* Running at 15V (the coils in the videos probably run at at least 30V, maybe more).
* Number of turns of the primary (I have 5 turns vs 1 turn in the second video)

I plan to work on all of these issues, but I was assuming (maybe incorrectly) that the lack of a topload was the main culprit. I'll also post pictures.

If I am reading the oscilloscope waveform correctly, the mosfet gate waveform scale is 1V/div? It seems that you do not have enough voltage on the MOSFET gate. That waveform is only ~5V peak and you want a higher voltage to turn it on, something like 12V. You can also see that the MOSFET is not turning on fully by the fact that the drain voltage on the mosfet never gets close to zero. If the mosfet is fully on it should be a low resistance and there should be relatively low voltage across it.

You're reading it right, indeed the MOSFET gate is turned on by the 5V circuitry. I didn't think this would be a problem since the datasheet of the FQPF9N50C that I'm currently using states that the gate threshold voltage is 4V maximum. But I neglected to look at the graph of drain current at a Vgs of 5V, and indeed, it's around 1A. That must be the issue, thanks a lot!

You said the control circuit voltage is 5V, does that include the push pull stage driving the gate? The push pull stage needs a supply voltage equal to what you want for your gate voltage (~12 - 15V). Additionally, the push pull stage does not provide any voltage gain, only current gain, so the input signal needs to be 12V and right now you are using 5V logic. If you want to keep using 5V logic you could add a common emitter amplifier (which would invert the signal but also provide gain).

Alternatively you could use a gate drive IC and do away with the totem pole. Something like this could be directly driven by the logic signal and drive the MOSFET gate faster than the push pull stage: https://www.digikey.com/product-detail/en/ixys-integrated-circuits-division/IXDN630CI/CLA374-ND/2623260

Thanks, I'm gonna look at different options; I think I'll get a gate driver IC, but I'll try to see if I have a schmitt trigger that supports 15V first. I'll update when it works (or doesn't).

Thanks so much for your answers.

I've been experimenting with the circuit and working on the PCB (which I'm all new to, so it's taken quite a bit of time). Now I need a good secondary...

Mads, I've read your guide, and decided to go for a 30 cm height, 8 cm diameter secondary and 800 to 1200 turns. Started looking for wire and then found a pre-wound secondary on ebay: https://www.ebay.com/itm/Teslaspule-32x7-5-cm-1000-Wdg-0-2mm-Tesla-Coil-Spule-TC-001-HVSHP/273279706078

Seems to be perfect for my usage, and to be honest I'd rather buy it pre-wound than do it myself. So I have a few more questions now

• Is the secondary in the link above able to produce good sparks in my use case?
• For the primary, I think I'll just use some insulated copper wire from an extension (rated 30 amps or so). Ok as well?
• The grounding: I'd like to just use the earth connection from an extension. I suppose it's ok as well?

Thanks, and I'll post photos when I have something working (which might take quite some more time...).