Felix's DRSSTC

[Felix B.]

Hello

I'm a 16 y.o. from Austria (sorry if my english isn't perfect ) and decided to build a DRSSTC a few months ago. It turned out pretty good so I decided to post the specifications and results for reference for people who wanna build one too or are just interested. Anyway here are the specs:


 - The overall size is 50x50x150cm. The F_res of the secondary is 82kHz unloaded and 73kHz with 2m streamer simulated with a wire. F_res of the primary at turn 5,65 is 73kHz. My bus voltage is 565V from rectified 3 phase 230V. OCD is currently set to 1130A and my on time is 70us. My DRSSTC can produce streamers easily exceeding 2m with most of the DRSSTC staying cold. (after 3min of continous opertation: Inverter ~40°C, MMC ~30°C but the primary gets to over 70°C)

 - Driver: UD2.7 with phase lead and two 1:900 current transformers. The GDT was wound with 10 turns CAT 5 cable on a B64290L0082X830 core

 - Inverter: FF450R12KE4P Fullbridge with two 450V 5600uF electrolytic capacitors in series and 68nf 2kV bypass cap to ground and 10kΩ loading resistor

 - MMC: 8kV 323nF made from 76 (4 series and 19 parallel) Wima MKP10 68nF 2kV with 470kΩ balancing resistors

 - Primary: 5,65 turns 10x1mm copper tubing with 10mm space between turns (edge to edge)

 - Secondary: 160mm PVC tubing wound with 0,33mm (AWG 28) enameled copper wire to 70cm (around 2100 turns). The whole pipe is 80cm long.

 - Topload: beautiful spun aluminium topload 500x125mm

Here are a few pictures of the DRSSTC and its components:






Here are some oscilloscope shots:
Gate Signal: 10V/div
Inverter Output: 250V/div
Primary current: 200A/div
Bus voltage at 570V with 70us on time and 10us/div:


I'd say that is some nice ring free switching. Here are the final results with my self-programmed MIDI interrupter:

Felix

[davekni]

Nice build - your waveforms look great!  If you ever run it hard enough to get the IGBTs hot, you may want slightly more phase-lead to compensate for the slower IGBT switching when IGBTs are hot.

[Mads Barnkob]

Hi Felix and welcome to HVF.

Checklist of awesome-ness.

• 16 year old
• Builds flawless DRSSTC following all best practices
• Programs own MIDI interrupter
• Posts thread that is nicely structured, a lot of information and videos

I hope to see many more projects from you, what you have done so early in life is way further than many of us

A couple of questions though...

- From the primary coil and MMC size, it "seems" like a higher impedance coil and thus I find the 70uS on-time for 1130Apeak a bit off, I did however not calculate this, its just a hunch from looking at the numbers It could suggest that its not tuned properly. The sweet spot is usually within 3-5cm, so moving the tap back and forth a 1/4 turn is too much for fine-tuning.
- Your MIDI interrupter is limited to what on-time? It seems like it is quite a stressful input to a DRSSTC, not much time between notes, if any

Do not excuse your English, had you not mentioned it, I would never have thought about it

[GrantV]

Wow, a perfectly EXCELLENT coil! Well done Felix :-)

[futurist]

Felix excellent coil and welcome to hvf

As mads pointed out thread is nicely written with lots of information
The only thing I notice that might cause problems on you coil are acrylic primary supports, they like to melt on longer runs, especially on inner turns

[Hydron]

Very nice coil, at 16 I was fooling around with neon sign transformers and flyback drivers, not a DRSSTC (though they weren't really a thing at that point).

A couple of questions though...

- From the primary coil and MMC size, it "seems" like a higher impedance coil and thus I find the 70uS on-time for 1130Apeak a bit off, I did however not calculate this, its just a hunch from looking at the numbers It could suggest that its not tuned properly. The sweet spot is usually within 3-5cm, so moving the tap back and forth a 1/4 turn is too much for fine-tuning.
- Your MIDI interrupter is limited to what on-time? It seems like it is quite a stressful input to a DRSSTC, not much time between notes, if any

Is it hitting the OCD at 70uS though, or is that just the setpoint? My similarly specified coil does about the same spark length at a similar or slightly lower OCD setting (can't recall if I need to push beyond 70uS though when in upper pole tune though). I also didn't find too much of a sweet spot, didn't get a dramatic difference with 1/4 turn tap changes.

[Felix B.]

Hello!

Thanks a lot for all your nice replies! I'll try to answer all of your questions:

Nice build - your waveforms look great!  If you ever run it hard enough to get the IGBTs hot, you may want slightly more phase-lead to compensate for the slower IGBT switching when IGBTs are hot.

I don't think I'm gonna run it much harder and if I do I still have power level 2 on my cooling fan but thanks for the tip!

- From the primary coil and MMC size, it "seems" like a higher impedance coil and thus I find the 70uS on-time for 1130Apeak a bit off, I did however not calculate this, its just a hunch from looking at the numbers It could suggest that its not tuned properly. The sweet spot is usually within 3-5cm, so moving the tap back and forth a 1/4 turn is too much for fine-tuning.

I think the fast current rise is because of my high bus voltage and I'm pretty sure that it's tuned properly but I will experiment a bit with the tuning on the weekend and see if I can run higher on times and hopefully get even longer arcs. At 70us my 1130A OCD only trips at heavy groundstrikes (in the videos you can see the blue led strip turning purple for a brief moment) and at 73us it trips contiously. My IGBTs could probably handle more than 1500A but my MMC is only rated at 8kV and on the verge of giving up (it stays pretty cold but more current and the 8kV will be exceeded). Anyway I'm gonna try to tune it better and see if it gets better.

- Your MIDI interrupter is limited to what on-time? It seems like it is quite a stressful input to a DRSSTC, not much time between notes, if any

My interrupter is controllable from the MIDI file so a volume of 10% with 70us on time set on the interrupter gives me 7us on time and 100% would be 70us. This works great and gives a really nice controllable arc length. Because of the polyphony 2 interrupts could overlap which would give me twice the on time. I compensate that with a 300us cooldown before the next interrupt. It doesn't really change the pitch in any way and fixes the problem. For high frequencies however I haven't programmed a protection feature in yet except for a cut off at above 880Hz. It basically relies on my MIDI editing skills to prevent overheating and in "the hall of the moutain king" I pushed it very hard with high BPS, polyphony and 70us on time. The coil can tolarate my abuse pretty well but because of the primary heating up I wouldn't push it any harder in terms of BPS. Also I will programm in a duty cycle limit hopefully protecting it more.

The only thing I notice that might cause problems on you coil are acrylic primary supports, they like to melt on longer runs, especially on inner turns

My old 1/4" primary I had before my 10mm one started melting my arcylic supports but you normally start smelling it before it melts and with the 10mm I haven't had any problems yet.

Again thanks for all your nice replies!
Felix

[TMaxElectronics]

I compensate that with a 300us cooldown before the next interrupt

That seems quite long to me. From my experiments deadtimes above 30us start to become noticeable as noise on the sound. What I'd recommend you do in order to set the deadtime is to scope your primary current and set it to the time it takes to decay after an interrupter pulse. Then there is no risk of hard switching at the start of the interrupter pulse. Then do the power limiting with a duty limiter.

All you need to do for that is to multiply the pulsewidth with the frequency to get the dutycycle. Add it up from all voices and divide the maximum dutycycle by that, then check if thats <1 and if so multiply all ontimes with that number and that should be it.
Thats the way I'm doing it in my midistick and so far I've not had any issues with it at all (and it has even saved my coil when I did something stupid like increase OT so much the dutycycle would be >100%).

[Felix B.]

Hello!

I don't hear any difference and didn't see any problems with the deadtime on the scope but I'm gonna reduce the deadtime and see if it makes a difference. I planned the duty cycle limiting pretty much as you explained it and I'm gonna programm it in tomorrow. Thanks for your tip with the shorter deadtime. I made it this long because I thought two pulses pretty close together will heat the coil up quite a bit but I guess it doesn't really matter that much if the interrupts are very close together or more appart.

Thanks for all your tips! I'll keep you updated on further improvements.
Felix

[Felix B.]

Hello!

I tuned my DRSSTC yesterday and the arcs got about 30cm longer which is pretty good but the current rises pretty much as fast as it did before (altough the OCD is tripping less and the coil stays colder). The dudy limiter works too. Thanks again for all your help.

Felix

[Mads Barnkob]

Hello!

I tuned my DRSSTC yesterday and the arcs got about 30cm longer which is pretty good but the current rises pretty much as fast as it did before (altough the OCD is tripping less and the coil stays colder). The dudy limiter works too. Thanks again for all your help.

Felix

Less OCD tripping is a very good visual indication that you are transferring more energy into the secondary circuit, even if sparks was not getting longer, getting less OCD trips is still an improvement.

[JCF]

super ! excellent job

[Zipdox]

What microcontroller did you use? Also, would you share your code?

[Felix B.]

super ! excellent job

thanks!

What microcontroller did you use? Also, would you share your code?

the µc is a Arduino Mega 2560. I don't really wanna share my code because I worked on it very long. Sorry.

[Zipdox]

the µc is a Arduino Mega 2560.

I made interrupter with a BluePill STM32, It's a very powerful controller and dirt cheap. I suggest you check it out if you have a couple bucks to spare.

I don't really wanna share my code because I worked on it very long. Sorry.

Are you familiar with free and open source software? Here are a few reasons to share source code:

• People can give you feedback and suggestions
• People can learn from your code
• People can improve your code and contribute back
• Your work has greater value if people use it (I mean you're not gonna sell it are you)

You can license your code under so people will be required to share any modified versions they make of your code.

But of course it's up to you.

[TMaxElectronics]

Code: [Select]

I mean you're not gonna sell it are youI wouldn't even that an argument against open source. Many commercial pieces of software are open source. I mean heck I'm selling my Interrupter and the code and hardware files are out there.

As for the Licence... there are some that don't permit commercial use, some that don't permit use in closed source projects and so many more. Open source is really the way to go (unless you are programming the code for the next ICBM )

I made it this long because I thought two pulses pretty close together will heat the coil up quite a bit 

I highly doubt that two pulses after another would risk any damage unless the IGBTs are on the verge of dying already in which case better heatsinking might be required.
Especially without pulse skipping and waiting until the current decays to 0.

[Felix B.]

Maybe I'm gonna make it open source but not yet. It also has QCWDRSSTC ramp output build in but I worked on it for so long and don't want someone to claim it's their code. I hope you understand that.

[futurist]

It's off topic but I think it's important we discuss this.

I understand you but I'm sure nobody in hv community would claim it's their code.
As Max and Zipdox wrote, there are many advantages with sharing your projects with others. Remember, other people shared their projects which enabled you to build the coil in the first place. Imagine if Steve Ward kept his universal driver private, and others followed, it would be much more difficult to build the coils. Just think about it 

[TMaxElectronics]

don't want someone to claim it's their code

By posting your code under your name you actually make it less likely. If you keep your code private and somebody somehow copied it anyway you would have 0 chance of prooving that it actually is yours. But if you post it to github you can show proof that its yours and (if you really want) probably even take legal action.

[Zipdox]

Maybe I'm gonna make it open source but not yet. It also has QCWDRSSTC ramp output build in but I worked on it for so long and don't want someone to claim it's their code. I hope you understand that.

If you post it to GitHub you have proof that it's your code and you're the first one to post it. You'll also be able to enforce a license which can legally restrict people from using your code in certain ways.

[Max]

@Felix B,

Open source is great and yadi yadi yada (don’t get me wrong, I‘m a fan, too). But ultimately it‘s your work and no matter how great other people find open source, it‘s your and only your decision whether you want to share your work or not. You shouldn‘t feel embarrassed or the need to justify yourself if you want to keep it for yourself.

Regards,
Max