A few questions

[davekni]

Thanks man!

I definitely wouldn't already be where I am now without all of your support!!! I really appreciate that!

You are welcome!  Glad to be able to help.

So far, I went up to 150V displayed on my variac before I ran the coil, as the display isn't of any usage when the coil runs (just jumps around and shows garbage).

Analog meters are quite useful around Tesla coils   Or digital meters with shielding around them and filters in lines to meter.

I haven't measured the bus voltage during the runs, but as I currently have a voltage doubling config with the two caps over the bridge, I guess at 150V input I already should be over more than 230V right?

Presuming 150V is RMS AC, peak is sqrt(2) higher, 212V.  Voltage doubler will produce 424Vdc (slightly less due to diode Vf) unloaded.  Of course, voltage drops due to loading.  230Vac has 325V peaks.  Bridge rectifier (not doubler) would generate just under 325V unloaded.

But then I am wondering how I connect the primary to the bridge as it is currently connected to the center point between the two caps in series.

Connect half of MMC to Vbus- and half to Vbus+ (2S1P to each of the incoming DC supply rails, which I'm calling Vbus).  Connect primary coil between half-bridge output and both halves of now-split MMC.  In other words, primary coil and MMC need to trade places before splitting MMC into two halves.  (I'm inferring from other statements here that half-bridge output currently connects to MMC which then connects to Vbus center.)

I've built the coil like the OneTesla with its 0.068uF capacitance and also used the same capacitor, but what I did differently is I have used 4 of those capacitors in a 2S2P config... don't know why... does that make sense to do so? Or can I save me 3 caps and use them otherwise.

2S2P provides margin for higher voltage and current and duty cycle, but may be more than you need.

which brings me to the next question, can I use those as snubber caps?

Yes, but not best choice.  Snubber caps (caps from Vbus+ to Vbus- immediately adjacent IGBT connections) are usually higher capacitance and physically smaller.  Short leads and small allow mounting closer to IGBT connections to reduce parasitic inductance.  Multiple parallel small caps can further reduce parasitic inductance of snubber.

Those two huge bus caps are 450V 5600uF each... Are those values more than I need for this size of coil? Can i go for lower values and smaller sized caps?

Likely more than you need.  As with MMC, depends on current and duty cycle.

Ahh... another one: In regards of measuring primary voltage and current during the runs, everyone is speaking of measuring the bridge output? What exactly is the bridge output in that terms? Is it before the resonant capacitor or behind?

Voltage on bridge output (low-side IGBT collector and high-side IGBT emitter) to probe tip.  Probe ground (with scope ground floating) to either Vbus- or to Vbus center tap (center between bulk caps).  Vbus- for "ground" allows probing close to IGBTs which is most accurate.

I tried to scope the primary current and output voltage with the help of an CT with a 1R burden resistor attached to it which seems to work fine I guess (that's actually what you see on the scope in the video) and a probe attached to the bridge output (before the cap) and then tried to adjust phase lead with an self wound inductor which I can adjust from ~2mH up to ~17mH according to my LCR meter, but I couldn't see anything significantly changing in the waveforms. Is there any tutorial for dummies how to do the scoping (During the primary voltage scoping, I had disconnected ground lead from my scope power cord)

There are many posts on the forum about how to scope and about phase adjustments.  Can't think of any tutorials off-hand.  Scoping both the 3-400V half-bridge output and lower voltage CT output requires come care.  Ideally have coax from CT burden resistor to scope input to avoid picking up noise.  Scope probe and ground clip should form minimal loop area and connect close to IGBTs and away from primary coil wiring.

Scope images are much easier to interpret here if you post separate still images, from camera or scope's internal capture capability.  Both overview as in video and zoomed in to see just two or three cycles is helpful.  To interpret current, need to know both burden resistor value (which you provided as 1 ohm) and CT ratio (which I don't see listed).

Does "2mH to 17mH", refer to milihenries or microhenries?  I'm used to "uH" for microhenries and mH for milihenries.

[Mads Barnkob]

which brings me to the next question, can I use those as snubber caps?

Yes, but not best choice.  Snubber caps (caps from Vbus+ to Vbus- immediately adjacent IGBT connections) are usually higher capacitance and physically smaller.  Short leads and small allow mounting closer to IGBT connections to reduce parasitic inductance.  Multiple parallel small caps can further reduce parasitic inductance of snubber.

I wrote about snubber sizing here https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/snubber-capacitor/

Those two huge bus caps are 450V 5600uF each... Are those values more than I need for this size of coil? Can i go for lower values and smaller sized caps?

Likely more than you need.  As with MMC, depends on current and duty cycle.

To calculate the ripple voltage on your DC link, I went through that here: https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/dc-bus-capacitor/

[ako]

Hi folks...

no more questions for a while... a good sign, right? I think yes, but one came up a day ago... but first of all, a progress status update. I'm quite happy with what I've built and I'm almost completely finished with the build (mostly worked on building a case and placing everything). I am working with 230V mains, OCD set to 300A and haven't blown my last two IGBTs for a while now.

Now heading to my question. The last technical change I did, was to change the topload with a slightly bigger one. As I moved the coil to another place in the meantime, I didn't have my oscilloscope to hand and used JavaTC to roughly get the new primary tapping point at ~6.5 turns. As my primary unfortunately only has 6 turns, I tapped it at the end and did a test run. Just to mention: I am using the OneTesla SD interrupter running on an Arduino Nano. As you may know, it boots up to Fixed mode. I started that and the first bangs were much quiter than with the smaller topload and my previous tapping point. I was even disappointed in the first moment. Then I went to the midis and started a song. While it was running I started to increase the volume (I guess the on-time) and the first few clicks increased the volume very slightly still giving much less output than before the change but - I don't know - the fourth or fifth click suddenly made the coil go into badass mode and I honestly got scared by the size and length of those new streamers it started to spit out. Then I moved on with increasing the volume and the remaining clicks until full volume felt like giving a more linear response compared to the first few clicks.

So my question basically is, what happens with the new topload and the new tapping point in combination with the SD interrupter and the configured on-times on its first clicks. Why do I get less output on the lower on-times than with my smaller topload and primary tapping point but then suddenly with slightly higher on-times much much more output?

Here are two videos to compare the output at full volume before and after I changed the topload and tapping point.

Before:
https://youtube.com/shorts/fG4AiAPHJRU

After: (By the way: Celebrating myself for having a strike rail installed, was thinking that it was unnecessary... until this point)

/>

[Mike]

The behaviour you're seeing now is consistent with a coil with the primary tuned below the secondary. At low on times the coil spend most of the on time slowly ramping up the voltage in the secondary because of the relatively poor energy transfer between the primary and secondary. Eventually enough energy transfers to the secondary for break out to occur and the arc starts growing, but then you end the on time so you get a smaller steamer.

If you extend the on time the steamer would grow which pulls down the resonant frequency of the secondary so it comes more and more in to tune with the primary. This increases the amount of energy that can be transferred between the primary and secondary so you can grow larger streamers. Eventually the steamer growth would pull the secondary frequency below the primary frequency and energy transfer starts reducing again. More commonly your on time elapses and you terminate the steamer growth that way.

If you had more primary left you could have tuned it lower for even more growth, but you do need to be a little careful because if your secondary isn't pulling energy out of your primary the current can ramp up very quickly, which can result in higher peak currents before the OCD kicks in and disables the coil.

As to why you end up with a lower primary frequency (or higher secondary frequency) than expected, despite JavaTC saying you needed more turns on your primary, I have no idea...

[ako]

Hm…

I can‘t tell (yet) what frequency I ended up with… if anything is lower or higher compared to what JavaTC calculated. All I can tell is that I had to tap the primary at the end of what I have which is 6 turns instead of the JavaTC calculated 6.5 turns. And if I‘m not wrong, the lower the primary turns the higher the resonant frequency which in my case would mean that my primary currently is tuned for a higher frequency than the resonant frequency of my secondary

Another quick test I can do in a few minutes is to modify the interrupter firmware and add another factor to the on-time multiplier...

EDIT:

...which also didn't seem to change the behavior... the first clicks while increasing the volume still give less output than before the change of topload and primary turns.

[davekni]

The behaviour you're seeing now is consistent with a coil with the primary tuned below the secondary.

I was thinking the same thing.  Seems opposite your actual tap being at 6 turns instead of calculated 6.5 turns.  Perhaps there is much more primary lead inductance than modeled by JavaTC.  (Did you include primary lead length in JavaTC input?)  May be worth trying less than 6 turns just in case primary truly is too far below secondary frequency.

[ako]

Now that I seem to have successfully built my first DRSSTC including cases and stuff and haven‘t blown anymore IGBTs for weeks… the next questions come up:



Can I drive a full bridge with the UD2.9. I know it has only 1 GDT output and full bridges seem to all run with two separate but theoretically a full bridge could be driven with a GDT with 4 secondaries instead of two, right.

Though I‘m pretty sure that the amswer is going to be no, I would like to learn the reason why not.

[Manz]

You can run a fullbridge with a single GDT with a primary and 4 separate secondaries like you said.
For example my drsstc also uses a ud2.9 and has no problem driving a skm100 fullbridge at >100khz with a single GDT

[flyingperson23]

As far as I know, a single GDT with 4 outputs is fairly standard. Some drivers have 2 GDT outputs to enable one method of pulse skipping (not the one the 2.9 uses), or sometimes 2 identical outputs in parallel for very large bricks. You won't have any trouble with a single GDT; what usually works for me is using cat5 cable with 4 wires in parallel for the primary and the remaining 4 each as a secondary.

[ako]

OK, next queation came up:

What could be the issue?

I took out the GDT and the half bridge. I wound another GDT using the same type of core and a piece of CAT5 cable as per your recommendation or your tutorial. Then I placed in that and a full bridge with exactly the same parts and values for resistors diodes and caps as I have on my half bridge (both profdc9‘s PCBs from the PCB pack). I reconnected everything as it was with the nicely working half bridge and ran a test:

It is working, but I barely get some output. My „problem“ is, that my oscilloscope is somewhere else currently and that I can‘t look what exactly is going wrong. So maybe you could tell me if there are some obvious reasons for this behaviour when switching from half to full bridge and leaving everything else, as it is?

[flyingperson23]

the obvious one that I can think of is incorrect phasing; i.e. connecting wires backwards to the igbts

[ako]

Wires backwards to the IGBTs? Those are on a PCB… But just to be sure what you mean… a fix would be to turn the feedback transformer or in my case switch the two jumpers on the UD2.9, right?

[flyingperson23]

Yes, definitely try switching the jumpers on the ud2.9; the wrong setting will stop it from working. What I meant was if you mixed up any of the wires to/from the gdt, that may also prevent it from working.

[ako]

I mean… like I already said, it is working. I get output. But much less than with the half bridge board. Do I need to change something in the interrupter code (OneTesla SD) maybe?

Edit:

OK, like expected, switching the jumpers stopped it from working at all.

[davekni]

I mean… like I already said, it is working. I get output. But much less than with the half bridge board.

Lots of possibilities.  One likely issue is that only one side of H-bridge is functioning correctly.

For any significant change such as half-bridge to full-bridge, it is best to scope signals at low voltage before running anything at full bus voltage.  Ideally start with no bus voltage, feed UD2.9 from a signal generator, and scope all 4 Vge signals.  Then apply low Vbus (perhaps 12V) and scope both bridge outputs.  If your bench supply can go higher, by 30V or at least by 60V you should be able to get some coil operation using CT feedback instead of signal generator.  Run all this testing at low duty cycle.  Can all be done without secondary in place.

[ako]

Can a defective mosfet on the UD2.9 be the reason (for coil working but with low output)? Or would it then not be working at all? Because shortly after running the last test with switched phasing and then switching back and letting it run again for 3 4 bangs I stopped the interrupter signal but the system was still powered, so fan spinning and LEds on on the driver. Then all of the sudden lights went out und fan stopped. Status LED of the power supply also off. After a little search I touched a very hot mosfet. I removed it and can power everything normally again. Unfortunately my spare mosfets are somewhere else currently. Tomorrow I‘ll be able to replace it, but as I‘ve asked already, could that dying mosfet be the reason for the low output?

[davekni]

could that dying mosfet be the reason for the low output?

Yes.  As I said, lots of possible issues.  That's why it is good to scope at low power first after any significant change or repair.

[ako]

Indeed, both mosfets on one side of the UD2.9 were dead.

Replaced them, scoped the output to be sure to get proper signals, skipped the low power tests and just gave it a try.

Now I am getting the output I was expecting, but I guess I‘m still detuned currently. Will check that later.

I have some questions regarding the interrupter firmware/on-times/primary current but I‘ll bring those up later when I‘m on the laptop.

[Dominic@TWHV]

Quick suggestion (I did not scroll through everything):

be sure to heatsink your IGBTs, and enclose the driver in some ferromagnetic casing to reduce interferences.

Good luck making it!