Project: FreakyDRSSTC MK1

[GrantV]

Hi all,

Having previously successfully built a test bench SSTC, I am now building my first DRSSTC, so I am going to use this space to document my progress and ask questions :-)

To kick it off, I have posted my JavaTC output and ask you to please give it a quick sanity check and make any comments you feel may be helpful?

Note: My secondary is already done so any tweaks I would need to make will have to be in the primary circuit!

I will initially be using the CloudLeopard driver from Ali. I do have a Loneoceans UD2.7c board and will build this up and add a suitable interrupter as things progress.

For the bridge, I will be using my own version of the Loneoceans 80mm Fullbridge which uses 4x FGH60N60SMD IGBT's and will be running at around 200-250Apk

My MMC is a 4x4 stack of 0.15uF CD caps (942C20P15K-F)

[GrantV]

My pcb's have finally arrived after going on a bit of a world tour with Fedex ...

Anyway, just glad they are here now 

I have always wanted to learn how to use pcb software and get my own pcb's printed, so I figured this was a great opportunity ...

Having had great success using Loneoceans 80mm Full Bridge with my SSTC, I loaded up EasyEDA and got to work creating my own version of his board.

I know it's not perfect and has a couple of issues but it's mine and I am happy with my first attempt 

[Mads Barnkob]

Overall you seem to have a solid plan and all numbers are within a regular DRSSTC build. You did your home work and should have no significant problems from parts and dimensions chosen.

200-250Apk does however seem a little modest for a 40 cm long secondary coil, not that I would advise you to push it further, but the bridge might be a tad under-dimensioned to utilize the full potential of the secondary coil.

Looking forward to see the sparks

[GrantV]

Overall you seem to have a solid plan and all numbers are within a regular DRSSTC build. You did your home work and should have no significant problems from parts and dimensions chosen.

200-250Apk does however seem a little modest for a 40 cm long secondary coil, not that I would advise you to push it further, but the bridge might be a tad under-dimensioned to utilize the full potential of the secondary coil.

Looking forward to see the sparks

Thank you Mads, I appreciate your response :-)

I am aware that I should be able to get considerably more performance out of my secondary. I prefer to build up (in power levels) in stages so that I learn as much as possible along the way :-)

[GrantV]

Hi there,

I completed the build on my bridge and am happy to report that it is tested and working ... (on the test bench) :-)



I have done some initial scoping and would appreciate your comments :-)

These scope shots are based on a very simple test configuration :-
Load : No load other than the 10K MMC Bleeder Resistor
Bus : 16V Lipo Battery
Gate drive : 24V
GDT cores are Epcos 26.6mm N30 with 12 turns cat 5 (dual primary/single secondaries)
Gate circuit : 15ohm gate drive resistor with reverse Schottky and a 26V G-E TVS
IGBT's are FGH60N60SMD
Driver is given a 130Khz signal to simulate feedback from the primary


excuse the duty cycle anomaly ... it had me scratching my head for a while lol




And now just making sure that the interrupter is working properly:

OK .... I give up .... How does a beat interval of 800uS = 10Hz/bps

[davekni]

The "10Hz" display should be the frequency of trigger events.  Perhaps the scope software gets confused by multiple closely-spaced triggers followed by longer 800us gaps.

Gate waveforms have enough overshoot to be clamped by the TVS diodes on every falling transition.  There appears to be much more parasitic inductance in gate drive than I would expect from your GDTs.  GDTs appear to be nicely wound.  What does the wiring from driver to GDT primary terminals look like?  Is it short and using twisted pairs?  It might be fine as-is.  The TVS diodes can likely handle the clamping power.  I'd try a high-duty-cycle run in this unloaded test setup to check that TVS diodes don't get too hot.

[GrantV]

The "10Hz" display should be the frequency of trigger events.  Perhaps the scope software gets confused by multiple closely-spaced triggers followed by longer 800us gaps.

Gate waveforms have enough overshoot to be clamped by the TVS diodes on every falling transition.  There appears to be much more parasitic inductance in gate drive than I would expect from your GDTs.  GDTs appear to be nicely wound.  What does the wiring from driver to GDT primary terminals look like?  Is it short and using twisted pairs?  It might be fine as-is.  The TVS diodes can likely handle the clamping power.  I'd try a high-duty-cycle run in this unloaded test setup to check that TVS diodes don't get too hot.

Hi Dave,

Thanks for your response.

The wiring from the driver to the GDT's is twisted but certainly not the best as this was just meant to be a quick lash-up for initial testing.

Please may I ask you to explain where and how you see the 'parasitic inductance' you are talking about in the waveforms? These are just my second ever GDT's so I am still trying to get used to what the waveform 'should' look like :-)

Pushing the on-time and/or the bps higher, I can hear the GDT's 'whining' ? and over 200uS 200bps the gate resistors do start to warm up but do not get overly hot. I have not noticed any heating of the TVS diodes but I will check that today.

EDIT: Unloaded, the TVS's gets up to around 60°C at 50% duty cycle and tops out at about 85°C at extreme duty cycles. The gate resistor does about the same.

Could the parasitic inductance be caused by the way I have run the leads from the GDT into the board?

[davekni]

Since the TVS diodes are not overheating even at extreme duty cycles, there is nothing to be concerned about.  For estimating inductance, I'm looking at the ~8% overshoot on rising Vge waveforms with your 15-ohm gate resistors.  Extrapolating down to low drain voltage, gate charge for FGH60N60SMD is about 150nC at 15V, or 10nF average capacitance.  Roughly 1.4uH in series with 15 ohms produces 8% overshoot with a square-wave input.  (Ran an LTSpice simulation to get this.  My quick estimate yesterday was off a bit.)  So, total gate drive inductance must be about 1.4uH.  An individual cat5 twisted pair is about 0.66uH/meter.  So 1.4uH implies just over 2m twisted pair length.  However, inductance on the GDT primaries counts double, since it is feeding two gates.  0.5m of twisted pair from driver to GDT could account for half of the 2m total.

[GrantV]

Since the TVS diodes are not overheating even at extreme duty cycles, there is nothing to be concerned about.  For estimating inductance, I'm looking at the ~8% overshoot on rising Vge waveforms with your 15-ohm gate resistors.  Extrapolating down to low drain voltage, gate charge for FGH60N60SMD is about 150nC at 15V, or 10nF average capacitance.  Roughly 1.4uH in series with 15 ohms produces 8% overshoot with a square-wave input.  (Ran an LTSpice simulation to get this.  My quick estimate yesterday was off a bit.)  So, total gate drive inductance must be about 1.4uH.  An individual cat5 twisted pair is about 0.66uH/meter.  So 1.4uH implies just over 2m twisted pair length.  However, inductance on the GDT primaries counts double, since it is feeding two gates.  0.5m of twisted pair from driver to GDT could account for half of the 2m total.

Hi Dave and thanks again for your response.

Thanks for your most detailed explanation, I think I understand what you were seeing in my waveform :-)

As I build, I'll keep an eye on that overshoot and make sure that I do not make it worse!

[GrantV]

Ok, well it took longer than expected and there were some problems with the setup of the laser cutter at the Maker Shop ... BUT .... my new DRSSTC themed 3D Puzzle is finally here 

[GrantV]

Hi all,

I am using shielded 4 core DMX cable for my GDT and CT lines and was planning on grounding the shield/s, but, as there is going to be HF noise in the ground circuit, I am now second guessing myself as to wether this is a good idea or if I should just leave the shielding unconnected.

Your thoughts?

[Da_Stier]

Hi,

I don't really have any experience in the field of tesla coils, however I do have some expience with the topic itself.
If your GND potential has a lot of superimposed noise on it, it might not be the best idea to directly connect your shielding to it.
However leaving it completely floating might also be a bad idea since it can float at pretty high potential, if the field around it is strong enough.
A common approach that is used for example with ethernet cables is to couple the shielding to GND via a high value resistor and a small capacitance in parallel.
Common values are 100k and 1nF.

Another possiblility is to connect the shielding to any other static potential in your system which has less noise on it.
From an RF point of view it doesn't really matter if your shield is at GND or +5V or whatever, since powerplanes will always be connected to GND via decoupling caps.
This makes them basically shorted to GND anyway ... from an RF point of view of course.

I hope this might be useful,
Michael

[GrantV]

Hi,

I don't really have any experience in the field of tesla coils, however I do have some expience with the topic itself.
If your GND potential has a lot of superimposed noise on it, it might not be the best idea to directly connect your shielding to it.
However leaving it completely floating might also be a bad idea since it can float at pretty high potential, if the field around it is strong enough.
A common approach that is used for example with ethernet cables is to couple the shielding to GND via a high value resistor and a small capacitance in parallel.
Common values are 100k and 1nF.

Another possiblility is to connect the shielding to any other static potential in your system which has less noise on it.
From an RF point of view it doesn't really matter if your shield is at GND or +5V or whatever, since powerplanes will always be connected to GND via decoupling caps.
This makes them basically shorted to GND anyway ... from an RF point of view of course.

I hope this might be useful,
Michael

Hi Michael, thanks for your response and very useful info!

I agree that leaving it floating would probably be a bad idea so I think I'll go with your suggestion of connecting it through a decoupling circuit (cap/resistor) as you suggested 

[prabhatkumar]

Hey there,
I was browsing through the forum and found that your PCBs look great. I actually required the PCBs from loneoceans and also contacted him for the same and he did reply to me about the shipping prices and everything. I accepted it and told him I am ready to pay for the boards if he ships it. But his reply is awaited for more than a month now.
I actually required this board for a college research project( Its a phase shifted power supply). I had also built a simple SSTC with the help of the fellow members of this forum last year and would now Like to upgrade to a full bridge using this board.
Please tell me if you are willing to post the board files in the open domain :)or if its paid.
I am myself leaning PCB design and got familiar with the single sided designs but double sided and that also for power circuits will take some time and hence I am relying on other options.
Thank you

[GrantV]

Hey there,
I was browsing through the forum and found that your PCBs look great. I actually required the PCBs from loneoceans and also contacted him for the same and he did reply to me about the shipping prices and everything. I accepted it and told him I am ready to pay for the boards if he ships it. But his reply is awaited for more than a month now.
I actually required this board for a college research project( Its a phase shifted power supply). I had also built a simple SSTC with the help of the fellow members of this forum last year and would now Like to upgrade to a full bridge using this board.
Please tell me if you are willing to post the board files in the open domain :)or if its paid.
I am myself leaning PCB design and got familiar with the single sided designs but double sided and that also for power circuits will take some time and hence I am relying on other options.
Thank you

Hi there,

Thank you for your kind words :-)

I wouldn't have a problem with sharing these but am yet to get permission from Gao, so for the moment I would prefer not to.

Also, my board is far from optimal, so I think you would be better off with the bridge design from Dan (Profdc9) :-)

https://highvoltageforum.net/index.php?topic=353.0

https://github.com/profdc9/DRSSTC-PCB-Pack/tree/master/full-bridge-transistor

[prabhatkumar]

Thanks for the prompt reply !! I understand that you have mentioned that in the board, there is a lot to improve. I had also seen profdc9 boards which is also great. But for my requirement, I needed the gate drive transformer, the bulk capacitor, rectifier and the IGBTS/MOSFETS all in the same board which unfortunately is not there in profdc9's board( not a con, his boards are great given we usually have the gate drive section and logic section on the same board for the tesla coil).
And as far as permission from Gao, we all always acknowledge his immense contribution for teaching the theory of tesla coils and also how to build one. And your board is an inspiration from his work and given he has also made the schematics available to everyone I don't think it should be problem. I will also add the gate drive section( The FDD8424H on the same board and make it available if it turns out to be good:)
Still it would be nice if someone like Gao or Mads himself comment on this .
Thanks

[Mads Barnkob]

Still it would be nice if someone like Gao or Mads himself comment on this .

I am a bit unsure what you are asking me about?

[prabhatkumar]

Hello Mads,
 I was asking about the release the of PCBS made by GrantV which are inspired by Gao's 80 mm full bridge plus. He is unsure whether he can release his version of PCBs over here as I required them urgently for a college research project. I will make sure his name and Gao's name are there in the final documentation.
Thanks

[Mads Barnkob]

Hello Mads,
 I was asking about the release the of PCBS made by GrantV which are inspired by Gao's 80 mm full bridge plus. He is unsure whether he can release his version of PCBs over here as I required them urgently for a college research project. I will make sure his name and Gao's name are there in the final documentation.
Thanks

I can not answer on behalf of Gao, contact him directly through his website/email at loneoceans.com if you want to publish his designs. If he wanted them public, he would properly have published them himself on his website. As he do sell some of his PCBs, both empty and populated, I would guess that it would be a no.

Modify profdc9's PCBs instead, would be much faster to just use something that is already public available instead of asking for complete solutions. If you require something urgently, you better just pay for the service or get working yourself. You are on the verge of breaking forum rules on asking directly for full instructions, designs etc.

[prabhatkumar]

Thanks for the reply Mads. I wanted to redesign the board made by profdc9 but the thing is the design is in kicad and I am only familiar to eagle some what:(
Anyways thanks for the help. I think problems only shape the way to new solutions. I will sit and try to build the entire board on eagle itself when I have some free time.

[GrantV]

Hi there,

My build is coming along nicely and I was able to do my initial power on tests today!



After having to correct the phasing of my feedback CT, it rang up nicely :-)


Vin is just 16V DC using my trusty Lipo battery :-)

I do have an issue that I am going to have to look at though, as the OCD CT is arcing under the CT to the track on the PCB .... not sure why, but I'll desolder it later and see what is going on 

Another issue that is concerning me is the BPS ........ with the interrupter set at minimum (supposedly 10 to 20BPS), my scope shows just 800uS between bursts which would be 1.2KHz?



Am I reading this correctly or have I missed something? I would hate to move on to higher power testing and blow my IGBT's due to the increased duty cycle........

[davekni]

Is your CT between MMC and primary coil?  That wire has high voltage, so isn't good for CT placement.  The other possibility is a bad connection or failed component in the OCD diode bridge or burden resistor.  An unloaded CT will generate high voltage spikes at each zero-crossing.  May be damaging the CT internally too.

Looks like your controller has a problem with 1.25kHz minimum frequency.  Definitely fix that before increasing power.

[GrantV]

Is your CT between MMC and primary coil?  That wire has high voltage, so isn't good for CT placement.  The other possibility is a bad connection or failed component in the OCD diode bridge or burden resistor.  An unloaded CT will generate high voltage spikes at each zero-crossing.  May be damaging the CT internally too.

Looks like your controller has a problem with 1.25kHz minimum frequency.  Definitely fix that before increasing power.

Hi David,

No, the CT is not between the MMC and primary, it is however on the opposite leg to the feedback CT, which I do not think should be a problem.

I think you nailed it however regarding the unloaded CT! Desoldering the CT showed a carbon arc trace between one of the pins and the opposite pins trace which I had inadvertantly (stoopidly) placed on the top of the board and routed it just a mm from the pin! Doing my first tests I unplugged the CT and used that line to push a signal into the driver from my sig gen ... the CT thus ran unloaded .....



Happily however, it doesn't seem like I have damaged the CT itself as I have now wired it up directly off-board and it is working fine. So I live and learn :-)

The interrupter issue is ongoing ....

[davekni]

Opposite sides of H-Bridge is no issue.

The 1mm gap isn't a problem either.  The problem is having no load on the CT secondary.  A burden resistor or shorting-strap or other load on the "CT 2 OUT" connector is necessary whenever running.  Or, if you want to make that board robust to unconnected CT load, add TVS diode(s) across the CT secondary.  (TVS diode capacitance is generally high, so that could affect phasing a bit if used on the feedback CT.)

That 1mm gap may have saved your CT from damage.  Sometimes intentional spark gaps are added to circuit boards as over-voltage protection.  The 1mm gap sparked over before voltage became high enough to spark inside the CT.

[GrantV]

Opposite sides of H-Bridge is no issue.

The 1mm gap isn't a problem either.  The problem is having no load on the CT secondary.  A burden resistor or shorting-strap or other load on the "CT 2 OUT" connector is necessary whenever running.  Or, if you want to make that board robust to unconnected CT load, add TVS diode(s) across the CT secondary.  (TVS diode capacitance is generally high, so that could affect phasing a bit if used on the feedback CT.)

That 1mm gap may have saved your CT from damage.  Sometimes intentional spark gaps are added to circuit boards as over-voltage protection.  The 1mm gap sparked over before voltage became high enough to spark inside the CT.

Hey David.

So what you are actually saying, is that I am soooo extremely clever that I subconciously added a safety feature to my board and then ran the CT unloaded JUST to test the efficacy of that safety feature? NICE! Cheers for that 

Seriously though, thanks for your reply and the info! Learning through mistakes is made a lot easier when someone takes the time to explain properly!

If or when I do do a rework of the board I think I would opt to add an intentional spark gap for each of the CT's on a visible section of the board rather than add additional components. The only reason it took me a while to figure out what was going on was because as much as I could hear the HF crackling, I couldn't see the arcing under the CT until I switched the lights off and went searching for the source of the offending sounds :-)

[davekni]

Yes, exactly, you added a safety feature

Even with an intentional spark gap, always connect the CT output to the driver or with a shorting strap.  The spark gap should be there to handle only mistakes such as a broken connector wire.

[GrantV]

Ok, so I have decided to throw the CloudLeopard Yunbao driver and interrupter in the spare parts bin as I am getting nowhere with it! (cannot get the BPS below 1250Hz)

I have gone ahead and ordered all the components I need to build up my Loneoceans UD2.7C driver, and those should be arriving in the next few days.

My budget is now however very sadly lacking, so as much as I would like to buy a OneTesla or EVR interrupter, I honestly just do not have the cash for that right now :-(

What is your favorite "lower cost" interrupter?

[Max]

It takes quite some effort to say this (joking) but I guess the best bang per buck interrupter is the MidiStick from TMaxElectronics: https://highvoltageforum.net/index.php?topic=1117.0
I'd normally recommend my own work (ofc  ) but since your budget is limited that wouldn't make any sense; the build costs start at about 80$.
There are of course more interrupters out there. Some are listed in this topic: https://highvoltageforum.net/index.php?topic=573.0

Aaand there are the simpler old school interrupters which are based on 2-3 NE555 chips like this one from Steve Ward: https://stevehv.4hv.org/drsstc_interrupter.htm No MIDI and not expensive


Kind regards,
Max

[Mads Barnkob]

Aaand there are the simpler old school interrupters which are based on 2-3 NE555 chips like this one from Steve Ward: https://stevehv.4hv.org/drsstc_interrupter.htm No MIDI and not expensive

In Profdc9's PCB pack there is a combined Steve Ward interrupter, audio interrupted and onetesla MIDI interrupted on one PCB: https://highvoltageforum.net/index.php?topic=353.msg2401#msg2401

Another cheap and old-school audio interrupter: https://kaizerpowerelectronics.dk/tesla-coils/musical-sstcdrsstc-interrupter/

[GrantV]

Hi again,

Max and Mads, thanks for your replies :-)

Max, I absolutely love your Syntherrupter but will leave that for later when I have built up my funds again a bit :-)

Tmax has offered me one of his MidiSticks so I think I am going to go that route for the Midi interface!

Since my post yesterday, I have resurrected an old project I had been working on and after a bit of tinkering, it is now working quite well :-)



It is a STM32 F4 Black Pill with an old LCD panel, a 3 way switch (mode switch), a momentary push button (burst or pause depending on mode), 2 rotary encoders (BPS and pulse width) and a transistor driven output stage for the IF-E96E.

Now that it is working (even though the coding is very dirty and needs some work), it is successfully giving me a beautifully clean output with absolute control over the pulse width and the BPS in auto mode and is also working very nicely in burst mode with a single burst at the set pulse width.

Although it does show me the duty cycle on the screen, I am yet to implement a max duty cycle limiter.

I'll carry on working on it and report back :-)

OH .... and my parcel has just arrived from DigiKey so I'll be able to start the build of my UD2.7C driver as well :-)

[GrantV]

OK, I need more help please :-)

I am trying to decide on what my max duty cycle should be and how to implement a limiter in the code.

Is there a simple rule-of-thumb for TO-247 IGBT coils that I could use?

OR

Do I need to set it up based on actual coil parameters?

I have spent the last 30minutes reading through Mads https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/ but I am probably more confused now than when I started 

Next:
To the coders out there, how have you implemented your active max duty cycle limiter? My thinking at this point is that :
Case 1: Duty cycle is set and the operator is busy increasing the BPS - when max duty cycle is reached the code activelt decreases the pulse width while alowing BPS to be increased further.
Case 2: BPS is set and the operator is busy increasing the pulse width - when max duty cycle is reached the code actively decreases the BPS while allowing pulse width to be increased further.

OR, do you just set a hard limit and the operator has to manually lower one or other parameter?

OR, am I overthinking it entirely?

[Max]

Most coils - no matter their size - have their duty cycle limit somewhere around 10-20%. In the early days people tended to push their IGBTs really hard (like 2kA-through-a-300A-IGBT-hard). This causes extreme temperature spikes on the IGBT die which aren't measurable from the outside. The rule of thumb that DRSSTCs should be run at <=10% duty cycle comes from those setups. Today however most people don't push their IGBTs that hard anymore (todays IGBTs are different, too) and there are many coils out there running happily at 20% or even more. I'd suggest you monitor the temperature of your setup. Parts that can get (too) warm include IGBTs/heatsink, MMC, wiring in the primary circuit, the primary itself (especially the inner turn) and the secondary. As long as nothing is overheating you should be able to increase the duty. Depending on your setup your limiting factor might even be the circuit breaker 

Now considering the code side... Since we're talking about a limiter I personally would prefer it to be just like a potentiometer that reached the end of the range, you can't turn it further (your third case). To keep the analogy, it would confuse me if hitting the end of a potentiometers range would start turning other pots. This is however a personal preference. After all, you're the one who will use it so you have to find out what you prefer. From a programming perspective there's not much difference between both, so you could try either one out.

I can see some use from what you describe as case 1 and in fact I implemented something similar in my Syntherrupter. Its Simple Mode (=simple interrupter like what you're doing here) has three sliders, ontime, BPS and duty cycle. And yes, this is redundant, so I need to decide what value gets adjusted when the user moves a slider. Instinctively I wouldn't expect the BPS to change if I mess with the duty or the ontime (which would be your case 2). Hence, if either the duty or the ontime slider changes, the other one is adjusted automatically. The more interesting scenario is a change of the BPS. In that case Syntherrupter keeps whatever value has been modified lastly. If the user sets an ontime, then moves the BPS slider, the duty cycle will adjust accordingly. And if he set the duty cycle before modifying the BPS, the ontime will adjust. I thought this was the most intuitive solution. On top of that, you have of course the ontime/bps/duty limitation, which as I said above only limits you without modifying anything. Once one of the sliders hits the end of its range no new value will be applied. To get a feeling of it, you can try it yourself; to run my UI you don't need to buy any hardware (Steps: 1. Download latest release, 2. Download Nextion Editor, 3. In the editor, hit "Debug" and select the Syntherrupter_Nextion_NX8048T050.tft file from the release zip file, 4. Write "click comOk,1" (without quotes) in the bottom left window and hit enter, 5. Have fun!).

Of course, if you don't like such hard limits, you could make your interrupter such that it always adjusts the oldest of the three values. Neither solution is harder than the other so I'd suggest you try a couple of them and find the one that suits you best.


Kind regards,
Max

[GrantV]

Most coils - no matter their size - have their duty cycle limit somewhere around 10-20%. In the early days people tended to push their IGBTs really hard (like 2kA-through-a-300A-IGBT-hard). This causes extreme temperature spikes on the IGBT die which aren't measurable from the outside. The rule of thumb that DRSSTCs should be run at <=10% duty cycle comes from those setups. Today however most people don't push their IGBTs that hard anymore (todays IGBTs are different, too) and there are many coils out there running happily at 20% or even more. I'd suggest you monitor the temperature of your setup. Parts that can get (too) warm include IGBTs/heatsink, MMC, wiring in the primary circuit, the primary itself (especially the inner turn) and the secondary. As long as nothing is overheating you should be able to increase the duty. Depending on your setup your limiting factor might even be the circuit breaker 

Now considering the code side... Since we're talking about a limiter I personally would prefer it to be just like a potentiometer that reached the end of the range, you can't turn it further (your third case). To keep the analogy, it would confuse me if hitting the end of a potentiometers range would start turning other pots. This is however a personal preference. After all, you're the one who will use it so you have to find out what you prefer. From a programming perspective there's not much difference between both, so you could try either one out.

I can see some use from what you describe as case 1 and in fact I implemented something similar in my Syntherrupter. Its Simple Mode (=simple interrupter like what you're doing here) has three sliders, ontime, BPS and duty cycle. And yes, this is redundant, so I need to decide what value gets adjusted when the user moves a slider. Instinctively I wouldn't expect the BPS to change if I mess with the duty or the ontime (which would be your case 2). Hence, if either the duty or the ontime slider changes, the other one is adjusted automatically. The more interesting scenario is a change of the BPS. In that case Syntherrupter keeps whatever value has been modified lastly. If the user sets an ontime, then moves the BPS slider, the duty cycle will adjust accordingly. And if he set the duty cycle before modifying the BPS, the ontime will adjust. I thought this was the most intuitive solution. On top of that, you have of course the ontime/bps/duty limitation, which as I said above only limits you without modifying anything. Once one of the sliders hits the end of its range no new value will be applied. To get a feeling of it, you can try it yourself; to run my UI you don't need to buy any hardware (Steps: 1. Download latest release, 2. Download Nextion Editor, 3. In the editor, hit "Debug" and select the Syntherrupter_Nextion_NX8048T050.tft file from the release zip file, 4. Write "click comOk,1" (without quotes) in the bottom left window and hit enter, 5. Have fun!).

Of course, if you don't like such hard limits, you could make your interrupter such that it always adjusts the oldest of the three values. Neither solution is harder than the other so I'd suggest you try a couple of them and find the one that suits you best.


Kind regards,
Max

Hi Max,

Thank you very much for your reply, it gives me a lot to think about and work with :-)

I'll most definitely have a play with your UI to see how you have implemented things!

I appreciate your comments on turning a pot or raising a slider till it hits it's maximum, however, as I have gone with digital encoders for my controls, there is no mechanical hard limit, which means that from a firmware point of view, the range for my encoders for BPS and on-time is from 1 to 4294967295 (uint32)  I can of course set max parameters, but the point is that I do not have any mechanical limts.

Using encoders is great because it gives me very fine control of the parameters but also has the down side in that it takes many many turns to set higher values. I am sure it will be fun initially, but in the long run it may end up just being annoying, and I'll swop them out for pots LOL

[Max]

Hi again,


I think you may have misunderstood my potentiometer analogy. I perfectly know that you can turn those things as far and as long as you like. However I personally wouldn‘t want it to start changing other values at some point in the endless scale. I would rather prefer nothing to change anymore if I try to turn past a limit. Of course I can continue to physically turn the knob but the effect is that I simply hit the end of the virtual range and cant go further.

Good point with having to turn a lot. I guess you have two options: If you want to have the full resolution across the entire range you need to implement some sort of velocity control. This means that you detect the turning velocity and increase the step size with the velocity (turn the encoder fast => sweep even faster through the range). From commercial products I know that this can result in a very intuitive control across the full range - no matter how many orders of magnitude the range covers. I also know that it can be annoying as f*** if it‘s not well done. So again, something you probably need to experiment with. I‘m pretty sure there are libraries out there that can do this, too.
The easier solution might be to implement something like a log potentiometer. So your step size simply depends on the range. That‘s what I have done with Syntherrupters sliders: 0-49: step size 1, 50-99: step size 5, 100-499: step size 10, 500-1000: step size 50, ...
this should be easier to implement while giving enough resolution (worst case is a step size of 10% of the current value). Of course, if you want to be able to control whether your ontime is 314 or 315 microseconds, this won‘t work for you.

Neither solution gives you a „continuous“ increase/decrease of the signal. If you‘d like to have a smooth transition from one value to another one I can recommend something like a moving average filter. My filter code (which is a bit more than only moving average) can be found here: https://github.com/MMMZZZZ/Syntherrupter/blob/dev/Syntherrupter_Tiva/Filter.cpp
For the ontime I use a filter with factor=2.0f, constant=30.0f, for the BPS I use factor=1.8f, constant=5.0f. Yet again, values to play with.


Kind regards,
Max

[GrantV]

Hey Max,

LOL sorry, no, I didn't misunderstand your pot analogy, I probably just did not answer correctly. It is early days for me as I have, as yet, never actually run or controlled a DRSSTC, so I am sure that my thinking and preferences will change once I get to testing in reality rather than just working with what is in my head *chuckle*

I have seen code snippets which implement velocity control, I have just not gone that route yet as my coding skills are rudimentary at best and it has all been a very steep learning curve!

I do currently have very basic step sizes implemented to lower the resolution but I have generally kept these at very low values during testing. Again, once I get the coil up and running I'll get a far better idea of what settings I prefer.

Your moving average filter sounds interesting, but to be honest, goes way over my head LOL

[TMaxElectronics]

Personally I consider your case 1 the best option, but primarily because I usually build musical interrupters where the note frequency is of course most important

The way I (and probably most others) have implemented is like this:

• Compute the current duty cycle (frequency * on-time in sec) of each voice (one for you if I saw correctly)
• Get the ratio between the maximum duty cycle and the current one (currDuty / maxDuty)
• And finally multiply the actual on-time with that factor if it is < 1

As for the velocity control, you can just measure the time between each encoder pulse and use that to scale your counting speed (so like maxSpeed/time). Make sure you debounce the switches properly though, as it could otherwise just shoot up like crazy even when moving slowly
And then you could of course also just dynamically calculate the maximums so you don't go over your maximum duty cycle. Basically just do a check every time the encoder increments to see if you have increased the value too far, and reset it if it is so.

[GrantV]

I am GUTTED!!!

Finally managed to get my UD2.7C all soldered up ...



BUT ... got the 9V regulator in the wrong way so when I tested, the 9V components got a few seconds of 24V!

I was hoping that nothing had been damaged but that does not seem to be the case.

After reorienting the 9V reg and reconnecting, the drive sends output to the GDT drivers wether or not I have signal from the interrupter and UVLO also does not cut the output.

So ... I now have some fault finding to do ....... Oh well, that's Monday's problem ...

OH .... and What does the jumper at JP1 do? Is it simply to enable/disable UVLO?

[John123]

Oh dear oh dear, how much did that mistake cost?

[davekni]

Is the GDT output a clean 24V square wave?  If so, I'm guessing that your 9V regulator mistake didn't fry anything.  There is probably some other unrelated issue causing the driver to be always-enabled.  Measure 5V and 9V supplies to make sure they are both good now.

Yes, it looks like JP1 enables UVLO.

With the driver disabled (interrupter input low or UVLO output low), measure voltage through the enable path.  IC6-1 (clear-n) should be low, so IC6-5 (Q) should be low.  That should make IC6-10 high and IC6-13 low, which makes IC6-9 low (other Q).  Measure these voltages to see where something goes wrong.  If these are all good, then continue measuring through IC5.

[TMaxElectronics]

Sorry to hear that!

But as davekni said aswell I think that if there is a clean output coming out of the GDTs most parts are likely fine and the issue is either limited to borken parts in the interrupter input path or even just a bad solder joint there

[GrantV]

Is the GDT output a clean 24V square wave?  If so, I'm guessing that your 9V regulator mistake didn't fry anything.  There is probably some other unrelated issue causing the driver to be always-enabled.  Measure 5V and 9V supplies to make sure they are both good now.

Yes, it looks like JP1 enables UVLO.

With the driver disabled (interrupter input low or UVLO output low), measure voltage through the enable path.  IC6-1 (clear-n) should be low, so IC6-5 (Q) should be low.  That should make IC6-10 high and IC6-13 low, which makes IC6-9 low (other Q).  Measure these voltages to see where something goes wrong.  If these are all good, then continue measuring through IC5.

Hi again,

Thanks all for your input :-)

After a quick check, I think I may have fried the LM311 Comparator because when the driver is disabled, IC6-1 is high making IC6-5 (Q) high thus IC6-10 is low and IC6-13 and 9 are high.

I'll do a bit more scoping but I think I may have hurt one or both of the mosfet drivers (UCC2743) as the GDT outputs, although square wave, are unbalanced (and floating in and out of balance over time) ....

GDT1-1 and GDT2-1 traces (with respect to GND)

Then again ... I have just seen that although I am getting a good 5V square wave from IC5A (3), IC5B (6) is only sending a 0.2V square wave which is probably causing or partially causing the GDT output imbalances? So IC5 is probably also shot ... which would explain why I am not getting a full swing but I am not sure that that would explain the floating imbalance between GDT1 and GDT2?

Addendum: Hmmmm ok so on initial power up ... with no signal in the FB input, the signal LED is off until I hit the IF95D receiver with a few light pulses, then the signal LED switches on and off as expected. The moment I add signal in the FB circuit, the signal LED comes on and stays on. The same happens after a few minutes even with no input into the FB.

[davekni]

My mistake on the measurement path:  That was the over-current path I suggested measuring.  Of course it will be enabled (no over-current yet).

So, the path to measure is opto-output to IC4-1 and IC4-3 through D12 and D13 to IC5-10.  Are you using the 2412T opto?  If so, at least on the UD2.7 schematic I have, the IFD95 output needs to be wired low.  Otherwise a missing IFD95 will leave IC4-3 pulled high by R11.  If R11 isn't stuffed either, then IC4-3 will be floating, so could cause your random enable behavior.

With only 0.2V on IC5-6, it does seem likely that one or both of the UCC27423 chips has fried and is loading down its input.  Less likely that IC5 fried, although not impossible.  Try remove the phase jumpers to see if IC5-6 then has normal voltage.  If IC5-6 is still only 0.2V and it's inputs are good, then replace IC5.  If IC5 outputs look good with phase jumpers removed, the at least one of the UCC27423 chips must be fried.  Perhaps the fried one will be warmer, or have worse output signals.  Nice to guess right before needing to remove both chips.

[GrantV]

My mistake on the measurement path:  That was the over-current path I suggested measuring.  Of course it will be enabled (no over-current yet).

So, the path to measure is opto-output to IC4-1 and IC4-3 through D12 and D13 to IC5-10.  Are you using the 2412T opto?  If so, at least on the UD2.7 schematic I have, the IFD95 output needs to be wired low.  Otherwise a missing IFD95 will leave IC4-3 pulled high by R11.  If R11 isn't stuffed either, then IC4-3 will be floating, so could cause your random enable behavior.

With only 0.2V on IC5-6, it does seem likely that one or both of the UCC27423 chips has fried and is loading down its input.  Less likely that IC5 fried, although not impossible.  Try remove the phase jumpers to see if IC5-6 then has normal voltage.  If IC5-6 is still only 0.2V and it's inputs are good, then replace IC5.  If IC5 outputs look good with phase jumpers removed, the at least one of the UCC27423 chips must be fried.  Perhaps the fried one will be warmer, or have worse output signals.  Nice to guess right before needing to remove both chips.

Hi Dave, (skip ahead to the addendum below)

It looks to me like IC5 is the offender although it is not in the 9V path so I am a little stumped as to why this chip would have gone bad .... *shrug*

I am using the IF950C receiver and do not have the 2412T installed. I have tested and all is as it should be here.

IC4-1 is high thus IC4-2 is low and stays low regardless due to D12 so no problem there.

IF950C is an open collector device thus is low with no input. IC4-6 is thus low as expected with no input, and stays that way until I activate it with a light and then IC4-6 goes high and LED-2 lights to show that there is a signal.

THEN ... either over some random amount of time or when I switch on the signal into the FB circuit, LED-2 lights and stays lit even with no activation of the IFD950C. Testing the circuit shows 2.88V on IC5-10 (AND gate input) and correspondingly on IC6-3 (CLK).

At first I thought that it might be IC6 causing the problem because IF I cause IC6A to switch with a UVLO condition (Q goes low) then the signal LED extinguishes, but feel strongly that this is more a sign that IC5 is shot and is what is causing the 2.88V leakage between pins 9 and 10 (IC5C)

Moving further in, and after removing the phase jumpers as you suggested. Although IC5A seems to be working perfectly, IC5B has no output on pin 6 even though 4 is high and pin 5 is getting a good signal from IC8-7.

SO, all in all, at this stage I am pretty convinced that I need to replace IC5, unless you have some other thoughts? My biggest hesitation right now is caused by the fact that I cannot see any way that the 9V line going to 24V could possibly have caused any damage to the quad and gate as it seems way clear of the 9V circuit???

On the plus side, using only the active/working side of the phase jumpers, I have tested both of the UCC's and I am getting strong outs on OUTA and OUTB respectively on both chips, so I do not think that these were damaged.


ADDENDUM:

Oh good grief!!!!!! ... 

Was just on Digikey about to order another IC5 (and a couple of the others for good measure) when I noticed that something was wrong .......

I based my parts order on the list created by Octopus1 (https://highvoltageforum.net/index.php?topic=1299.msg9552#msg9552) and I did not check each and every component myself ....

It looks like he replaced 74HC08DR2G (obsolete) with MC14081BDG


instead of MM74HC08M


SOOOOOOOOO ... I am pretty sure IC5 is not damaged at all ..... It's just got the WRONG DARN PINOUT!!!!

[GrantV]

Great news! The MM74HC08M arrived, has been installed and everything seems to be working PERFECTLY 

Now to get on with the rest of the build!

[GrantV]

It has been a while since my last post ... life seems to have a way of getting in the way of having fun :-)

I have however managed to reach my next milestone and completed (?) my Interrupter!


It is based on the STM32F401CC "Black Pill" dev board using standard Arduino coding.

The LCD display shows me the mode, current duty cycle percentage, the max duty cycle, pulse width and BPS.

In Auto mode, I can adjust the pulse width and BPS independantly from min to their set maximums. While doing so, the duty cycle is updated on the LCD and once max duty cycle is reached, either the pulse width or the BPS is automatically lowered so as not to exceed the max duty cycle. The rocker switch in between the 2 rotary encoders allows me to increase or decrease the max duty cycle on the fly. I can also pause and restart the output of the program using the pause pushbutton.

In One-Shot mode, I can set the Pulse width using the encoder and then activate the output using the pushbutton.

User mode is currently just doing a sweep of the BPS from low to high and back again while alowing me to adjust the pulse width on the fly. This will probably be changed to an audio mode or something else ...

For power, I have used a buck step down converter and also have battery voltage monitoring in my coding so can use anything to power the interrupter between 6 and 12V. At the moment I am using a 2 cell 1500mAH LiPo battery.

I have not yet implemented velocity sensing on the encoders and will leave this until after I have it working properly with my coil and I get a better idea of how fine or coarse I want the controls.

There is enough space in the box for my MidiStick and I have already added a 2nd fibre transmitter for it. I'll get to adding this later after I have tested everything individually.

[JCF]

Hello Grant,
could you tell me what are the characteristics of the box:
Fmin, Fmax, duty cycle min, max.

[GrantV]

Hello Grant,
could you tell me what are the characteristics of the box:
Fmin, Fmax, duty cycle min, max.

Hey JCF,

Because I am using a programmable PWM output from the micro, those values can be almost anything at all :-)

ie. It can go from an fmin of 1Hz to 2kHz (or more if ever needed) and the same with the pulse width. At the moment I have my min pulse width at 10microsec and max of 500microseconds but there is no reason I couldn't set that all the way up to some milliseconds if ever needed.

The duty cycle is just calculated from the other two so min is 0.01% and max is currently 10% which can then be adjusted using the rocker switch up to say 20% or down to whatever value makes sense for a specific coil.

[GrantV]

Hi all,

OK ... So I finally got hold of an isolation transformer and set my coil up to adjust my phase lead, and have hit 2 problems that I need your help with PLEASE!!!

The first problem is that I cannot adjust my phase lead ... no matter what I do, nothing changes while adjusting my phase lead inductor! In fact, with the jumper jumpered to INCLUDE the inductor into the feedback circuit, the current leads by just over 400ns, with the jumper shorting the inductor the currents leads by just 210ns ...?

I am using a UD2.7C driver and have turned my own tunable inductors using 0.4mm (AWG26) speaker wire. I do not have an inductance meter which reads very low values so the best I have been able to measure is the largest of the 3 inductors (40 turns) is tunable between 10uH and 20uH. I have not been able to measure the 20 turn and 10 turn inductors.


As I am driving TO-247 IGBT's (FGH60N60SMD), 10 to 20uH should be correct but as I said, I am seeing no visible change in the EC signal while adjusting the inductor, so would appreciate your words of wisdom please?
210ns without the phase lead inductor
400+ns with the phase lead inductor


OK ... now on to my second problem:
I am using Triag Magnetics CST206-3A 300:1 CT's on my inverter for OC and feedback. I have initially set my OC to a voltage of 2.5V (using 5 Ohm burden) which, at 300:1 should equate to 150Amps OC trigger.


I have also turned my own 256:1 cascaded CT (16 turn 16 turn) and have that loaded down with a 51 Ohm power resistor (on heatsink). Assuming that my calculations are correct, 150Amps should produce a voltage of 29.88V


When running my coil using a large pot with water as a dummy load sitting above my primary coil, the OC TRIPS at just 28uSec on-time?

My bus voltage is just 100VDC and the current draw from the mains is less than 500mA at time of the OC tripping!!!



The scope shot above shows the 28uSec burst with the OC trip. The yelllow trace is the current signal from my cascaded 256:1 CT and the blue trace is across R5




PLEASE HELP ... this just does not make sense to me!!!

MMC Cap is 0.15uF and Primary Inductance is around 10.8uH

[davekni]

Odd behavior on your phase lead.  I suggest probing both sides of the inductor while in-circuit.  Do that either sequentially or with your two probes.  The signal on the lower side (across 51-ohm burden resistor) should represent current.  The signal on the other side should be earlier in phase (and perhaps slightly larger due to inductor winding resistance).

28us to 150A is about right for your design.  125kHz scope traces match your listed 0.15uf and 10.8uH, 8.5 ohms impedance each at resonance.  You could try a simple SPICE simulation of that series-resonant circuit.  If no losses, each half cycle adds twice the full-bridge bus voltage, so 200V/half-cycle.  After the shown 7 half-cycles, voltage should be about 1400V, which would be 165A.  Given losses, 150A is quite realistic.

Given the above, your primary design impedance is low for your H-Bridge voltage and current capability.  More primary turns and lower MMC capacitance (and higher MMC voltage capability) would fit your H-Bridge better.

BTW, thank you for the very-clear problem description and scope images showing your exact questions.

[GrantV]

Odd behavior on your phase lead.  I suggest probing both sides of the inductor while in-circuit.  Do that either sequentially or with your two probes.  The signal on the lower side (across 51-ohm burden resistor) should represent current.  The signal on the other side should be earlier in phase (and perhaps slightly larger due to inductor winding resistance).

28us to 150A is about right for your design.  125kHz scope traces match your listed 0.15uf and 10.8uH, 8.5 ohms impedance each at resonance.  You could try a simple SPICE simulation of that series-resonant circuit.  If no losses, each half cycle adds twice the full-bridge bus voltage, so 200V/half-cycle.  After the shown 7 half-cycles, voltage should be about 1400V, which would be 165A.  Given losses, 150A is quite realistic.

Given the above, your primary design impedance is low for your H-Bridge voltage and current capability.  More primary turns and lower MMC capacitance (and higher MMC voltage capability) would fit your H-Bridge better.

BTW, thank you for the very-clear problem description and scope images showing your exact questions.

Hey David,

Thanks for your input :-)

I will certainly go scope out the inductor as suggested.

Regarding my OC problem (and yes, your answer was what I was worried you were going to say)......... OK, so assuming that I have already built everything and cannot now add turns to my primary, is there any other way for me to "de-rate" my primary circuit?



I was just soooooo happy to have received design signoff right at the beginning of this thread ......... UGH ......


EDIT: Just thinking here .... I currently have a large topload on my secondary which drops my fres by quite a lot! I could lose the large topload and use a much smaller one to raise the fres. This would allow me to drop my MMC to 0.075uF while keeping the 7 or 8 turns on the primary. Would this possibly work? or just shift the problem to a higher frequency? :-(

With 2 strings of 4 caps (instead of my current 4x4):

• The Primary Impedance would go up from 8.52 to 12.05 Ohms
• My MMC reactance and impedance would go up from 6.2445 to 12.5 Ohms
• My MMC peak voltage would go up from 1248 to 2497 Volts

Or would that not be enough of an improvement? If I could be so bold, what would your recommended figures be for my bridge?

This would be my coil with the small topload and would bring my fres up to about 175Khz rather than the 128Khz it is currently.

[davekni]

Looking back at your first post with JavaTC results, energy transfer time is 24us.  You may be fine with secondary in place, if arcs remove enough energy before OCD.  Also looked back at your IGBTs (FGH60N60SMD).  Those should be good for 250A, perhaps even 300A, especially once phase-lead is working correctly.  I recommend pushing your H-Bridge harder rather than changing anything with your nicely-constructed coil.

[GrantV]

Hey David,

Thanks for calling it a "nicely-constructed coil" 

OK, so ... it turns out that the inductors I had turned were just too small for me to notice any difference while adjusting them.

I turned a 70 turn this morning and all of a sudden, there was a big difference :-)

I have settled on a 55 turn that seems to be doing the job quite well (on very low bus voltage tests).


As soon as I can get the isolation transformer again, I will do higher bus voltage tests and final adjustments and hopefully then be able to post pics of SPARKS!!!! 

Oh and thanks also for re-checking my specs! I have now set my OCD up to 250A and will see how that goes with the secondary in place!

[davekni]

It looks like the scope traces are of current and gate-drive, with roughly matched phase (no lead or lag).  If so, you'll want yet more phase lead.  The idea is to have the H-Bridge output switch slightly before current zero-crossing.  That includes delay through the IGBT, plus a bit more phase lead so that the H-Bridge output transition occurs due to remaining current, before the opposite IGBT turns on.

[Mads Barnkob]

It is a great built! Looks good and has nice cut parts, you deserve the applause for a good job on design and construction

If you can get your coil to a high enough impedance to get a steady energy transfer, you will properly see that while you reach maximum current fast, you are also transferring it all through secondary into sparks and you could reach equilibrium where turning up on-time will grow the spark length.

Low voltage testing can be a bit misleading, some 100VDC+ on the DC bus is much more true to how waveforms look at full voltage.

[GrantV]

It looks like the scope traces are of current and gate-drive, with roughly matched phase (no lead or lag).  If so, you'll want yet more phase lead.  The idea is to have the H-Bridge output switch slightly before current zero-crossing.  That includes delay through the IGBT, plus a bit more phase lead so that the H-Bridge output transition occurs due to remaining current, before the opposite IGBT turns on.

Hey David,

EDIT: I just re-read your post and no, those traces were of current and CE (not gate drive)?

I added in a bit of extra phase lead (120nS). Your comments on these traces would be most welcome :-) (these again show current and CE)

[GrantV]

Thanks for the kind words Mads 

OK, so the news update of the year is that I finally fired up my coil last night (on my birthday) Whoooohoooo 

fres: 128KHz
Bus voltage: 260VDC approx
Pulse width: 48uSec
BPS: 800Hz
Duty cycle: 4%
Streamer length: 80cm
IGBT heatsink temp after 2min run: 'cold to the touch'

OK, so as happy as I am to have gotten first light, I need to ask for suggestions on how I can improve it's performance??

The problem I have is that the specs above are the MAX I can run the coil at (although I have not pushed the BPS higher yet)!

Under 300-400BPS I just get a few cm of brush spark and cannot push my on-time much at all before the OCD trips.

I should be able to push the bus voltage to over 400VDC but even just 300VDC means that I trip the OCD at anything over 18-24uSec on-time and the same goes for my PW, at 260VDC, anything past 50uSec on-time trips the OCD!

Current draw from the mains is currently 7Amps at 170VAC

I have done primary tuning for streamer length but could possibly still do with fine tuning.

I will set my OCD a little higher (300A) and test again tomorrow but not sure how far I should really be pressing this coil.

By the way, what is the recommended upper limit for BPS? or is it purely down to the duty cycle?

[davekni]

Phase-lead looks perfect as shown.  You could increase it a bit more, since IGBTs (and other parts) slow down some as they heat up.  The zoomed-in image (400ns/div) shows Vce transition just before zero current, with a short 78ns time where the lower IGBT's diode is conducting (causing voltage to be a bit below ground).  As current reverses, the low-side IGBT goes into forward conduction.  The initial voltage bump up suggests that the low-side IGBT is not quite fully on (Vge isn't quite fully charged yet).  Vge is still high enough there to avoid the tripple-transition that sometimes happens when switching right at zero current.

My guess is that 300A will be fine now that phase-lead looks good.  You could try even higher if you have spare IGBTs to replace fried ones.  Higher current is the path to better performance for your coil.  (You could eventually parallel pairs of IGBTs, 8 total for the H-Bridge, to really push this coil.  (Then you may find other limits such as secondary racing arcs etc.)

Tuning your primary frequency slightly higher (closer to secondary frequency) will improve spark length at low BPS, but likely reduce the max length at high BPS.  That higher primary tuning will make music sound better (at the expense of max spark length).  Higher BPS is fine too until duty cycle is too high for either IGBT heating or line input current.  I've ran 2000BPS for short bursts.

I certainly agree with Mads:  Very nicely-constructed coil.  Much nicer than any project I've ever made or will make.

[GrantV]

Hi again David,

LOL thank you, I am not sure that that your last statement is true but I will take the accolades with pride and a grin on my face :-)

Thanks also for the amount of detail you provide in your answers and descriptions of what is going on in the circuitry! The learning curve is steep and every answer like yours helps progress my understanding just that little bit further every time!

As I had guests over, I did not get to push the coil any further yet BUT ...... I did get to test it with my MidiStick!!!! This is the very first music I have ever played through a DRSSTC and I am one EXTREMELY coiler 

[GrantV]

Hi all,

I have been doing some more tuning and testing and my coil is now running very nicely indeed 

Many many thanks to David and Mads for all the help along the way!

Here are a few video clips from today! There are a few more in my Youtube Channel if you are interested!
https://www.youtube.com/playlist?list=PLFOJCCNUmQKjUz0CD5Z2zGEFyJQWnwzjP

OH and many thanks to TMax for his MidiStick! I am having a lot of fun with it :-)

Intro Video showing construction:

Final Interrupter run today: (1m arc to target)

Musical - Zelda - using MidiStick

Musical - Popcorn - using MidiStick

Again, MANY MANY MANY MANY thanks 

[Mads Barnkob]

Congratulations on the great results!

Good looking build and good spark producer on MIDIs

You can elevate the quality of your sparks videos quite a bit with a few easy steps.
- Put your camera/microphone/smartphone inside a faraday cage, just a small one with a hole for the camera, this will take care of alot of the spikes and "hizz" in the sound recordings.
- Put some directed light on the Tesla coil itself, that does not light up the background at which the sparks appear.
- Keep the foil as spark target, gives off a nice effect from reflecting light
- Move that mirror that is behind the coil, I thought you had sparks flying out the back of the coil

Subscribed to your channel, looking forward to see more.

[GrantV]

Lol thank you Mads 

Thanks for the ideas, I will certaily give them a try!

The glass door, which looked like a mirror, to the side caught me out as well ...  I'll be sure to move it next time!

[costas_p]

GrantV,

I also designed a H-bridge based on "80mm" version of loneoceans, in your final product, how much copper oz you used?

[GrantV]

GrantV,

I also designed a H-bridge based on "80mm" version of loneoceans, in your final product, how much copper oz you used?

Hi there. I used standard 1oz copper and all seems to be working well.

[costas_p]

Thanks for the reply, I was thinking something like 4oz , but even china-made pcbs are expensive, so I will settle for 2oz just be a bit the safe side.