Optimal tuning

[flyingperson23]

Ok, so I got my drsstc working, and now have no clue how to tune it properly.
Specs:
BSM150 full bridge (with excessive cooling)
340V bus
.3uF mmc (2s6p cde940)
rather fat secondary, 3.33:1
ud2.9+ with pulse skip
almost exactly 50k secondary reactance at resonance (180kHz)
coupling of .131
attached is a picture of the coil

My questions is how do I optimize it for arc length and sound quality?
I've seen a few coils with conservative ocd and very long on times/duty cycles getting huge arcs with pulse skip, but my current impression is that that's only feasible for low notes, making it unsuitable for music.
So, how should I tune it? low current and high on times or vice versa, or is high current and on time feasible?
And how should I determine the ocd? By guessing based on thermal dissipation?
This seems like fairly basic information, but the tuning section of kaiser's guide is unfinished and I'm having trouble finding other resources.
Any help or links to resources would be appreciated.

edit: this is the best I've gotten so far: https://drive.google.com/file/d/1KTiZVL3Kdm18simvhhEKExdP0c7ETTkH/view?usp=sharing
This is using TMax's midi stick with ocd at 300A, Vbus~100v, max on time 150us, max duty cycle 8%
The ocd is the yellow light above the red one, and it's continually tripped here.
The ocd similarly remains almost continually tripped for lower notes, but they sound strange, quiet, hardly any arcs, and they produce weird harmonics.
I'm not sure what it is about the high notes that makes them and only them sound normal.

[davekni]

rather fat secondary, 3.33:1

coupling of .131

Fat secondaries are great.  The main advantage AFAIK is that coupling can be higher before getting problems with racing sparks.  I'd suggest raising height of primary or lowering height of secondary to increase coupling.  With my skinnier secondary, couldn't get above 0.141 without racing sparks.  I'm guessing you could go significantly higher, which will likely improve performance.

My questions is how do I optimize it for arc length and sound quality?

In my experience those are conflicting goals.  Longest arcs are achieved with lower secondary frequency, as low as possible while still managing to start an arc.  Sound is better with secondary frequency closer to primary frequency, around 10% below.

This is using TMax's midi stick with ocd at 300A, Vbus~100v, max on time 150uS, max duty cycle 8%
The ocd is the yellow light above the red one, and it's continually tripped here.

Higher coupling is likely to help here.  If that isn't enough, perhaps primary frequency is a bit too far below secondary.

[Manz]

I'm currently tuning my pulseskipping DRSSTC and from what I've seen so far MIDI files sound better when the primary isn't detuned aggressively, as davekni said.
It seems like your primary coil has a really low impedance that leads to rapid current rise, from my limited experience with pulseskipping the arcs sound much better and you get a better performance when the current rises slower so I would encourage you to also try different mmc configurations to get a lower capacitance (I started with 156nf with my small coil and it now behaves much better with only 97nf)

[flyingperson23]

Yep, thanks manz. That makes perfect sense now, noob mistake. The primary was tapped at 2.75 turns. I changed the mmc from 300 nF to 75, and the primary to 5.75 turns, which bumped the coupling up to .137, and got a drastic performance increase. However, because the coil is so compact and the primary radius so much larger, the frame is arcing on the loop making up the bottom. That won't be hard to fix, but makes me a little worried about the inverter being exposed to that strong of a field. The driver is enclosed in an aluminum box, but the gate drive PCB and IGBT gates are quite close to the primary now; should I be worried about them? I like the look of the compact coil, but I can raise the coils if necessary.

[Manz]

I'm happy you were able to get better results.
I think you don't have to worry about the inverter or shielded driver getting interferences. The only problem you might have is induction heating on the metal parts that are too close to the primary but I guess at this power level it won't happen. You can just run the coil for a while and then check if everything is nice and cold.

[flyingperson23]

yep, aside from the coils, everythings cool! by any chance, is there a way to make bass notes louder besides 'build a bigger coil'?

[Rafft]

midi mode works OK.

javaTC is used for calculations?. try setting PRI resonant freq lower than secodary freq. also try different mmc capacitor TYPE.

ocd 200Apk for test

150vbus is enough for testing also.

fullbridge (4 IGBT) = more arc length

coupling k 0.3 at least(but I guess this is too high for regular drsstc)

phase lead adjustment also


edit:
just for reference, I have here an old coil setup. its using only TO-247 IGBT in half-bridge. mains powered. 310vdc bus. Im unsure if it was in-tune either(just based of javaTC calculations only) and I had no scope that time. its using an old UD with no OCD & no phase lead adjustment. secondary feedback. no idea how much current it pulls.

/>

[Manz]

A low duty cycle limiter could help a bit by making high pitched sounds quieter.
The best option is probably to edit the MIDI files and lowering the high notes volume.

[davekni]

is there a way to make bass notes louder besides 'build a bigger coil'?

Perhaps only bass notes could be played with double (or triple) pulses to increase volume.  This post mentions double pulses with time-delay modulation for synthesizer effects.  Could likely work for bass volume too:
https://highvoltageforum.net/index.php?topic=1117.msg15752#msg15752

[flyingperson23]

Alright, I tuned it relatively well and bass boosted tracks more or less to my liking. I got max 30" arcs from non-doubled (us) 120v mains input. I designed it for 240v, but I finally tried 240v input and got lots of small streamers from my topload (made from https://www.homedepot.com/p/Everbilt-4-in-x-6-ft-Semi-Rigid-Duct-with-Collars-EVER007/313119823)and two medium sized ones, one from the breakout point (solid 12 awg copper wire) and a slightly larger one from the bottom of the topload continuously hitting the ground ring. Any suggestions? I don't have access to a 3d printer or any form of cnc or machine tool, so it'd have to be something I can make by hand. Attached is the javatc input/output if it's useful. Thanks!

Edit: remade the top load, fixed the problem. Works fine on 240vac in.
I think I'll stick with manually boosting the bass on the midi and get TMax's new midi stick (box?) when it comes out
Thanks for all your help!

[Max]

Concerning bass notes: there's a rather easy fix possible, I implemented and tested it in my own interrupter (Link). Play low notes at constant duty instead of constant ontime. Constant duty does not work universally, because the same duty cycle, that's sufficient to make low notes loud, is completely insufficient for high notes: the same duty cycle would result in much shorter ontimes, that are probably insufficient for decent arcs. Therefore constant ontime remains the preferred method for high notes. If this sounds confusing to you, please take a second with a pen and paper and go through some example numbers, like 5% duty at 200BPS and 5% duty at 2000BPS (calculate the resulting ontime). Or 50us ontime at at 200BPS vs 50us ontime at 2000BPS (calculate the resulting duty cycle).
Deciding whether to use constant duty or constant ontime is pretty easy; just take whatever results in the highest ontime. For high notes, this is constant ontime and for low notes this is constant duty - exactly what we want.

I don't know your interrupter and whether or not it's possible to implement this easily but here's how it looks like in it's easiest form:

Code: [Select]

float dutyBasedOntime = desiredDuty * 1.0f / bps;  // Ontime in seconds. Replace 1.0f by 1e6f if you want ontime in microseconds.
float finalOntime = max(desiredOntime, dutyBasedOntime);
For the example given above:

Code: [Select]

// desiredDuty = 5% = 0.05
// desiredOntime = 50us

// Case 200Hz
dutyBasedOntime = 0.05 * 1000000 / 200Hz = 0.05 * 5000 = 250us;
finalOntime = max(50us, 250us) = 250us

// Case 2000Hz
dutyBasedOntime = 0.05 * 1000000 / 2000Hz = 0.05 * 500 = 25us;
finalOntime = max(50us, 25us) = 50us

You can hear the effect of just increasing the desired duty cycle without increasing the ontime in the following video. Note how suddenly the lower voice gets louder, while the higher one remains the same.

/>

Kind regards,
Max

[flyingperson23]

heres the videos of the coil if anyone's interested: https://www.youtube.com/@benb32768/videos
I got 50" arcs and shortly thereafter blew the igbts

[flyingperson23]

I have a question about coupling -
I'm increasing the coupling from .131 to .142 by lowering the secondary. I'm not sure how high is too high and I really don't want to hurt my secondary. Can the primary tap wire, not counted in javatc's calculations, increase the coupling enough to take into account?

[davekni]

I'm not sure how high is too high and I really don't want to hurt my secondary.

In general, too-high coupling causes racing sparks along the lower part of secondary.  However, I believe the threshold that causes trouble depends on secondary aspect ratio (among other factors).  Low secondary aspect ratios (short compared to diameter) tolerates higher coupling factor without racing sparks.  AFAIK that is the key reason for large secondary diameter, in spite of the resulting increased weight and cost.  I reduced coupling from 0.151 to 0.141 due to racing spark issues.

Can the primary tap wire, not counted in javatc's calculations, increase the coupling enough to take into account?

Depends on wire routing.  If the wire continues around coil close to primary turn, it acts as a completion of that turn.  If headed straight down and/or radially, wire doesn't couple to secondary or primary, so just adds a bit of wiring inductance.  Primary lead length can be added to JavaTC.  JavaTC uses it as part of primary frequency calculation, but does not include it's generally-slight affect on coupling.  Lead inductance reduces effective coupling.

[flyingperson23]

I don't seem to have any issues with k=.145 excluding the primary tap. Would it be unwise to try .15? If I do get racing sparks, how likely is it to destroy the secondary? I can't seem to find much information on racing sparks besides the br0din coil eating secondaries at .141.

[davekni]

I don't seem to have any issues with k=.145 excluding the primary tap. Would it be unwise to try .15? If I do get racing sparks, how likely is it to destroy the secondary? I can't seem to find much information on racing sparks besides the br0din coil eating secondaries at .141.

As I'd mentioned, there are other factors besides secondary aspect ratio and coupling that affect racing sparks.  I think br0din's issues were exacerbated by top load size being a bit small for secondary height.
When I had racing sparks, I also had some sparks from my ground-strike ring around primary to part way up secondary.  Those made small pits in secondary coating.  I ended up adding more layers of epoxy to secondary along with reducing coupling and making a smoother ground-strike ring.  Not sure which combination of those changes was necessary.