BrOdin coil - High power Big sparks!

[fh89]

If V/turn is really an issue then you could try spacing the turns with nylon monofilament fishing line (may need to reduce wire diameter to keep turn number appropriate when spaced out a bit).
Might not help a lot though if the problem is direct hits from streamers (which won't care about your insulation or spacing!).

In this case, I think it was streamer puncture.
 [ You are not allowed to view attachments ]

I've been playing around with the numbers in JAVATC including space-winding, but figuring out the best compromise between frequency, reactance at resonance, winding length and coupling is not easy.

 

Have you considered a larger top-load?  That would help make the field more uniform down the secondary, which should reduce strikes to the secondary.  It should also help with longer arcs.  The larger top-load capacitance supports longer arcs before resonant frequency drops too far below primary frequency.  Even if building a new nice-looking toroid is too hard/expensive, it would be fun to experiment with a cheap hack top-load addition on top of your existing toroid.  Something like a sheet of 4x8' styrofoam insulation sheet rounded with a hot-knife and covered with aluminum foil.  Would look odd, but might get you another couple feet of arc length.  That might be something to try along with your planned lower-inductance secondary.

This has crossed my mind too, but as silly as it sounds, I am hesitant to use anything other than the spun toroid after the ridiculous amount of money I spent on it. Maybe I would be ok with a ring-type toroid on top of the spun one. I'll have to give this some more thought.

Sad to hear that your secondary was damaged again. It is a really remarkable coil ranking near or at the top of in terms of power for QCW types. I hope you get it running again soon

Thanks, I appreciate that! Don't worry, I will get it fixed up and upgraded soon.

Any chance you can take high fps movies of it? It would be very interesting to see how much gain in arc length you get from these long on times.

Unfortunately the best camera I own currently is a gopro hero 8. It can shoot 120fps I think, but it isn't great in low light. I would love to get some high-quality slow-mo video. I will ask around when I'm back up running.

I think, that spacing the secondaries winding at the bottom will decrease coupling. Coupling depends much on the overlap of primary and secondary fields and winding spacing will decrease it. I haven't calculated the effect, though.

I found Mads remarks and links to the whiplash effect interesting. How much the voltage gradient along the secondary is affected depends much on the speed of voltage change of the top load during a ground strike. From Hydrons top load measurements a typical value would be maybe 100 kV/us. That voltage jump travels down the secondary at a speed of about 400 km/s, so the jump is about 40 cm long. That means a 100 kV jump along 40 cm length. Quite a lot but it doesn't last very long.

The jump travelling speed can be estimated, by calculating or measuring the secondary resonance frequency without a top load. The coil is then a quarter wave resonator, so that frequency and wavelength is known. From this the speed of the wave can be calculated.

Hydron has made some measurements of secondary bottom currents. Whiplashes could be nicely observed there. AFAIK, there weren't any ground strikes in his last data set.

To the extent that top-load and primary/strike-rail diameters are large compared to secondary length, there should be minimal whiplash.  Field up the primary will be roughly uniform, the field between two parallel plates of an air-dielectric capacitor.  Of course, in a real coil the diameters aren't semi-infinite.  Still, the whiplash voltage should be much less than would be experienced by a coil with no top-load.  I suspect that electric field non-uniformity up the secondary is driven more by magnetic coupling from the primary than from top-load voltage transients.

I'm not sure whiplash or flashover/racing sparks contributed to either of my secondary failures, so I think I am ok in that regard. I just need to keep the streamers away from the secondary!

I'm also considering modding my UD+ to run 3000A peak (along with new MMC and secondary).

For the new MMC, I've been looking at 4CAYUD3220AA3J https://content.kemet.com/datasheets/KEM_F3042_C4C_AXIAL.pdf (.22uF 3kV 12Arms) .  5 series 25 strings 125 caps 1.1uF 300Arms 15kV for $566.58. Any opinions on these caps?

[Uspring]

@davekni:
I think we are looking at different extremes of the same issue. I am considering secondary surface charges only and am neglecting top load charges. You are doing vice versa. Presumably both type of charges need to be added in their effect.

I completely agree on the strong effect of induced voltages from the primary. The induced voltage is V = M * dIpri/dt, with M being the mutual inductance. Numerically we have M=0.55mH, Ipri=2000A and f about 33kHz for fh89s coil. That amounts to 230kV. This voltage is generated around perhaps the lower third of the coil. On top of that voltage is that produced from the secondaries current. How these add up depends on the relative phase between primary and secondary current. This is hard to predict, since it depends on the tuning, which varies with arc load.

[Hydron]

That'll buff right out!

If it is as suspected a secondary strike then not a lot more I can suggest other than the longer breakout point, and possible putting more attractive streamer targets within range. Wind is certainly a killer as well - a decent breeze wrecks performance and makes the streamers go all over the place.

Secondary looks fixable again with a bit of surgery too, worth doing even if only as a spare.

What is the reason you're looking at a beefier MMC? Is the current rating of the existing one a bit suspect, or are you looking for a different capacitance/voltage rating? Could be that it can be pushed harder than expected, especially if you're looking at conservative ratings (e.g. for long life, at high ambient temperatures).

[fh89]

If it is as suspected a secondary strike then not a lot more I can suggest other than the longer breakout point, and possible putting more attractive streamer targets within range. Wind is certainly a killer as well - a decent breeze wrecks performance and makes the streamers go all over the place.

Yeah, I think I had issues when running raised up off the ground and lower on-times throwing it out of tune causing shorter streamers to hit the secondary. It's a real PITA having to retune every time I want to increase or decrease the on-time significantly.

 

Secondary looks fixable again with a bit of surgery too, worth doing even if only as a spare.

Yeah it is fixable, but the coilform cost $400 so I'm not really wanting to buy another one. I'm just going to re-use it and wind it from scratch.

What is the reason you're looking at a beefier MMC? Is the current rating of the existing one a bit suspect, or are you looking for a different capacitance/voltage rating? Could be that it can be pushed harder than expected, especially if you're looking at conservative ratings (e.g. for long life, at high ambient temperatures).

Yes, the RMS current rating of the MMC is only 70A, and at 2000A primary current limit, I'm pushing 547Arms at 15% duty cycle. If I bump the pri limit to 3000A, I'm looking at 822Arms, so I definitely will need a beefier MMC. I also wanted to try a lower impedance primary design with a higher capacitance value.

[davekni]

I'd guess that the pictured damage event was initiated by a strike from the top, but that most of the burning came from shorted-turn current after the initial strike compromised insulation and formed plasma there.  Or it could be an arc formed from ground or primary to that point after the initial strike from top.  This is just my guess.  Others may have more experience in this area.

Yes, re-tuning is a nuisance, but necessary.  (My re-tuning is by connecting or removing banks of MMC, as my primary is fixed 4 turns of litz wire.)

I like Hydron's idea of fixing.  Run more experiments with more possible faults before bothering to rewind.  However, the wire was likely cheap compared to the form, so perhaps rewinding makes sense.

At high currents, IGBT forward drop looks more resistive, so power increases as current squared.  3000A is at least doubling IGBT power.  You might consider 2500A first before making such a large jump.

Yes, you have a beautiful toroid!  Aesthetics are lacking in most of my projects.

[Mads Barnkob]

Thumbs up for chasing the outer limits of what is physically possible with DRSSTC   

As DaveKni mentions, a bigger toroid would suit the coil better. I know its a very nice and very large toroid you have. I am a sucker for matching secondary winding length to toroid major diameter and secondary coil diameter to toroid minor diameter.

Regarding the 4CAYUD3220AA3J capacitors, seems to be a little "worse" than the famous CDE 942C series capacitors. For a coil this size I would recommend looking at large GTO snupper capacitors in the 1-5uF range. With M5 threaded connections and "heat sink" plates inbetween them, you can dissipate a lot more power in those than a 100+ cap MMC.

I ran your capacitor through the Tesla coil MMC calculator and I do not think that your proposed MMC is worth the money.

[fh89]

At high currents, IGBT forward drop looks more resistive, so power increases as current squared.  3000A is at least doubling IGBT power.  You might consider 2500A first before making such a large jump.

I've seen other coils running cm600ha-24h at 3500 or even 4000A primary current, so I was actually thinking 3000 was somewhat conservative.

Thumbs up for chasing the outer limits of what is physically possible with DRSSTC   

Thanks, I'm just trying to have some fun with it! I'll probably get sick of blowing stuff up eventually and back it off just a tad when I find the limits of what is possible with my current setup. Hopefully one day I will have the time and $$ available to build something truly monstrous!

As DaveKni mentions, a bigger toroid would suit the coil better. I know its a very nice and very large toroid you have. I am a sucker for matching secondary winding length to toroid major diameter and secondary coil diameter to toroid minor diameter.

I might be able to live with a ring style toroid on top of the spun one. I'm giving the construction some thought.

Regarding the 4CAYUD3220AA3J capacitors, seems to be a little "worse" than the famous CDE 942C series capacitors. For a coil this size I would recommend looking at large GTO snupper capacitors in the 1-5uF range. With M5 threaded connections and "heat sink" plates inbetween them, you can dissipate a lot more power in those than a 100+ cap MMC.

I appreciate the feedback.

I would love to get my hands on some doorknob caps, I have been looking for the last month or so but I cannot find any (at least in significant quantity at a reasonable price of a suitable rating.) If anybody has a source, please let me know!

Something along the lines of AVX/TPC FPG series (FPG86X0205J seems ideal); Kemet C4DRYAQ series.

[davekni]

I've seen other coils running cm600ha-24h at 3500 or even 4000A primary current, so I was actually thinking 3000 was somewhat conservative.

Are these coils running close to your 15ms on-times?  Are they pulse-skipping?  If so, then go for 3000A or more.  Or start with lower on-times again.

If you don't have other plans for your MMC capacitors, you could run them until failure.  Do you know how warm they get now?  IR thermometers are great for checking temperatures, and an IR camera even better.  They can't see through plastic, however, so you would need to measure the caps from some angle with a clear view path.

[fh89]

Are these coils running close to your 15ms on-times?  Are they pulse-skipping?

Good point, probably not. I will definitely start out with lower on-times. I'm not really sure I saw much length gain over 3ms, and I kind of prefer the more continuous look of higher break rates, so I might not go over that.

 

If you don't have other plans for your MMC capacitors, you could run them until failure.

I'm leaning towards using them now, but doubling the size/ Arms rating because I cannot find any doorknob caps even if I wanted to pay full price and spend $1500 bucks for new ones.

IR thermometers are great for checking temperatures, and an IR camera even better.  They can't see through plastic, however, so you would need to measure the caps from some angle with a clear view path.

I don't know how hot they get, but I pulled the bank out and they look pristine- no outward signs of overheating. I know this really doesn't completely correlate to how they are internally, but they've held up so far, and doubling the Arms rating should help. An IR camera would be awesome, but I don't have one. I will pick up a IR thermometer.

Quick question about the current limit burden resistor on the UD+
Given that the present current limit spec'd for the driver is 2000A, and the burden resistor is 5.1 Ohms, and my CT is 1024:1, that should give 2000/1024=1.95A*5.1r = 9.95v at the upper limit.
So If I want to change the limit to 3000A, I can parallel a 10 ohm resistor on top of the 5.1 equaling 3.4r, so that 3000/1024=2.93A*3.4r = 9.96v.

Is my thinking/math correct here? Is my understanding of how the current limit circuit works correct?

[davekni]

Quick question about the current limit burden resistor on the UD+
Given that the present current limit spec'd for the driver is 2000A, and the burden resistor is 5.1 Ohms, and my CT is 1024:1, that should give 2000/1024=1.95A*5.1r = 9.95v at the upper limit.
So If I want to change the limit to 3000A, I can parallel a 10 ohm resistor on top of the 5.1 equaling 3.4r, so that 3000/1024=2.93A*3.4r = 9.96v.

Is my thinking/math correct here?

Yes, math looks good.

The basic Seek thermal cameras run about $300.  That is what I have and found extremely useful.  I think smart-phone attachments may be available for a bit less, but I don't have a smart-phone.

[fh89]

The basic Seek thermal cameras run about $300.  That is what I have and found extremely useful.  I think smart-phone attachments may be available for a bit less, but I don't have a smart-phone.

Whoa, those are neat! $250 on amazon for the smartphone attachment. Very tempting!

Yes, math looks good.

Ok cool, but is my understanding of how the current limiting circuit works correct- that I want to end up with the same voltage at the current limit?

[davekni]

Ok cool, but is my understanding of how the current limiting circuit works correct- that I want to end up with the same voltage at the current limit?

Yes, current sense is based on the input voltage to the LM311 comparitor.  I think you were using UD2.9 or something similar.  For UD2.9, the 10k potentiometer RV1 generates the compare voltage for current-sense.  When turned all the way up, it theoretically would be 12V * (10k / (1.8k + 10k)) = 10.17V.  Your board could easily be 9.95V due to component value tolerances.  When that voltage is reached across the sense resistor, resistor current is V/R, so primary current is 1024 * V / R for your 1024:1 current transformer.

[fh89]

Yes, current sense is based on the input voltage to the LM311 comparitor.  I think you were using UD2.9 or something similar.  For UD2.9, the 10k potentiometer RV1 generates the compare voltage for current-sense.  When turned all the way up, it theoretically would be 12V * (10k / (1.8k + 10k)) = 10.17V.  Your board could easily be 9.95V due to component value tolerances.  When that voltage is reached across the sense resistor, resistor current is V/R, so primary current is 1024 * V / R for your 1024:1 current transformer.

Thanks, I appreciate the sanity check! Seems I came to my conclusion working backwards, but still ended up with the correct value needed for the burden resistor. 9.96v is only the value I came up with assuming the Ilim was 2000A, it probably was a little bit more (the actual Ilim)  in reality if the comparator was cutting off at 10.17. I just had the pot cranked all the way up without any sort of measurement based on the stated spec of the UD+ controller.

[fh89]

Brodin coil is running again!

New secondary coil 22 awg. Not sure on how many turns/fres/q/ because JAVATC predicted that I would need 9lbs to wind the entire secondary length, but I ran out of wire and used up a whole 10lb spool before the end of the secondary form length. Coupling is cranked to .185 with the first turn starting at the bottom edge of the primary tubing.



UD+ controller modified for 3000A limit


MMC doubled in size to 1.14uF



MMC cooling fans doubled



Not pushing anything too hard here, just getting retuned and seeing where we're at. Running at 1.12ms pw and 156hz prf 17.5% duty cycle

Hit 40kVA and 30kW now:

[Kizmo]

Such a lovely coil!

Your secondary coil failures look like what I was experiencing with the BiggerDR.

At these power levels flashovers are not always recoverable accidents 

[Hydron]

Upgrade looks pretty nice!

Have you tried longer pulse widths (like 10+ms) and lower PRFs? Could change the character of the streamers quite a bit, and it seems like the coil is likely up to it (unless it pushes IGBT die delta-T too high) if you can run 17%+ duty cycle already.

[davekni]

Love watching that video in slow-motion, watching how the ground arcs develop and change shape over time!

Does anyone have a UD+ schematic or know what the feedback current capability is?  If it is like UD2.7 with a 51-ohm burden resistor, that resistor will be overheating at 17% duty cycle and 2.9A peak input (3000A / 1024:1 CT ratio).

[Hydron]

UD+ info is here:
http://classictesla.com/pslawinski/udplus.html

Looks like it is indeed 51R, if that's correct then it's running way past it's ratings, good catch!

[fh89]

Such a lovely coil!

Your secondary coil failures look like what I was experiencing with the BiggerDR.

At these power levels flashovers are not always recoverable accidents 

Thanks, I appreciate it! Yes, the flashovers can be catastrophic running at high power! Part of the price to play!

Upgrade looks pretty nice!

Have you tried longer pulse widths (like 10+ms) and lower PRFs? Could change the character of the streamers quite a bit, and it seems like the coil is likely up to it (unless it pushes IGBT die delta-T too high) if you can run 17%+ duty cycle already.

Thanks! Yes, I've run this coil up to 15ms at 10Hz - video on youtube/previous page.

Love watching that video in slow-motion, watching how the ground arcs develop and change shape over time!

Does anyone have a UD+ schematic or know what the feedback current capability is?  If it is like UD2.7 with a 51-ohm burden resistor, that resistor will be overheating at 17% duty cycle and 2.9A peak input (3000A / 1024:1 CT ratio).

UD+ info is here:
http://classictesla.com/pslawinski/udplus.html

Looks like it is indeed 51R, if that's correct then it's running way past it's ratings, good catch!

This is the EVR version of the UD+ designed to run up to 2000A. The original burdern resistor was 5.1 ohms. I paralleled a 10 ohm 3w resistor on top of the 5.1 equaling 3.4r, so that 3000/1024=2.93A*3.4r = 9.96v. So 9.96v^2/r = 99v/3.4r = 29w*.175 duty = ~5w, so maybe not too crazy over the rating?

[davekni]

Are there any PDF format (or image format such s JPG) files for the UD+ schematic?  EVR version would be great for your specific coil, or any UD+ version presuming it is close.  It would be handy to have that filed away in case other questions come up.

You'd previously discussed paralleling 10 ohms to the existing 5.1ohm resistor for over-current.  I don't know the power rating for those resistors, so don't know how much they may be past rating.  My latest comment was about feedback, not OCD.  Are you using the same 1024:1 ratio for feedback CT?  Perhaps the EVR version already has a lower-ohm feedback burden resistor to handle 2000A.  Or perhaps UD+ is expecting a higher CT ratio for 2000A?  Either way, presuming there is a separate CT input for feedback, it could be a concern for power capability inside UD+ with increasing to 3000A.