BrOdin coil - High power Big sparks!

[fh89]

Specs:

• Full bridge cm600ha-24h
  
• Liquid cooled primary
  
• 1000A primary current limit now 2500A
  
• 12.75x48" secondary 23awg now 22awg
  
• .57uF now 1.14uF mmc
  
• 6800uf 16,800uF 26,800uF bus cap
  
• 36x10 spun aluminum toroid
  
• EVR/Philip Slawinski UD+ freewheeling driver
  
• EVR polyphonic midi interrupter / Custom Philip Slawinski long pulse interrupter
  
• All polycarbonate/aluminum enclosure
  
• 30kW+ Pulling 180+ amps at 220V.Now 57kW draws 350A+ on 220V
• up to 15ms on-times 17%duty cycle
  

Photo dump of all of my construction photos (some of these photos show mistakes that were later corrected):
https://1drv.ms/u/s!Ap7oF5iAXsqR_Q4Ll1D90VbZBSRI?e=13UIZi


Human for scale.


















15' from tip of break out point to strike target


Clamp Ammeter showing 351 Amps






Updated photos showing doubled MMC, added bus caps, extra fans, etc.

[acobaugh]

Looks good. Does any of the hardware in the frame get warm at all? What kind of power draw are you looking at?

[fh89]

Looks good. Does any of the hardware in the frame get warm at all? What kind of power draw are you looking at?

Thanks! I have not noticed the frame hardware getting warm, but I also have not touch checked anything directly after a run. I haven't noticed any plastic discoloration or melting on the primary supports or disc yet though. Currently the max power draw is 8-9kW. I expect this will rise dramatically as I push up the on-times.

[acobaugh]

I take it you're not hitting your 2kA yet? Even at 360uS on time, your primary impedance might be too high to really push some current through. My DR82 coil uses a CM300 full bridge, OCD set to 1000A, peak current is around 1100A. Power draw is about 9kW, and I easily hit 11 feet.

I say you need to lower your tank impedance so you can really push some current.

[fh89]

I take it you're not hitting your 2kA yet? Even at 360uS on time, your primary impedance might be too high to really push some current through. My DR82 coil uses a CM300 full bridge, OCD set to 1000A, peak current is around 1100A. Power draw is about 9kW, and I easily hit 11 feet..

Haven't made any attempt to determine primary current, but I would guess you are correct. Very nice performing coil BTW!

I say you need to lower your tank impedance so you can really push some current

Well, I am going to have to build another MMC because the current one was designed and built before I knew anything about running long on-times in freewheel mode, and the Arms is too low. What would be an ideal value for a low impedance? I'll admit my knowledge of low-impedance tank circuit design is lacking.
 
I am going to push more current by increasing on-time to 1.12ms and see how it goes first. Assuming a 2kA limit, 1.12ms (my interrupter's current maximum value, possibly to be increased later via firmware update) on-time at 48Hz (my interrupter's minimum value) gives a 5.39% duty cycle or 316Arms- way over my MMC's 70Arms rating.

I am going to rebuild a MMC for much higher rms current rating, but I am not sure what the ideal C value is. The current .57uF puts me at 5.75 primary turns.

[acobaugh]

My CM300 coil runs at 38kHz with a 0.6uF MMC (5s20p array of 942C - not recommended if you're starting from scratch, but I already had most of these caps from the previous spark gap model). My CM600 coil will be between 1.25uF and 1.56uF, based on experimentation. For a coil your size, I would expect at least 0.8uF, probably closer to 1uF to be a good value, based on other similar designs. There are a bunch of tradeoffs involved, and I'm curious to see how your testing with long on times works out for spark formation. Just watch your temps. You might want to lower your OCD threshold to a known lower value just to see where your peak is at in case you don't have a means of measuring it in realtime. I don't think the UD+ will freewheel until you hit OCD. As others have said, you'll see more heat dissipation per unit time until you start freewheeling.

Check out Mads' MMC calculator here: http://kaizerpowerelectronics.dk/calculators/mmc-calculator/
and the corresponding mmc design guide here: https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/mmc-tank-capacitor/

Is this your first DR coil?

[davekni]

Yes, definitely explore behavior of your coil with longer on-times and more measurements using your existing MMC before changing.  Measuring primary current during operation would be ideal.  This can be done with an additional CT, or by scoping across the driver's feedback input resistor.  Knowing the operating frequency and current ramp-up and limiting (if any) will help in picking your next MMC capacitance.  If you are hitting 2000A for a reasonable portion of each enable pulse, then I wouldn't increase capacitance.

Does your controller have a single-shot mode?  I think most do.  I'd use that first for long-enable-pulse testing before running at 48Hz.  5% duty cycle will likely trip your breaker besides overheating the existing MMC.

Do you have a JavaTC dump for your coil?  Or a link to parameters from a similar EVR coil?  That would provide estimates of secondary inductance and frequency, coupling factor, etc.

[fh89]

My CM300 coil runs at 38kHz with a 0.6uF MMC (5s20p array of 942C - not recommended if you're starting from scratch, but I already had most of these caps from the previous spark gap model). My CM600 coil will be between 1.25uF and 1.56uF, based on experimentation. For a coil your size, I would expect at least 0.8uF, probably closer to 1uF to be a good value, based on other similar designs. There are a bunch of tradeoffs involved, and I'm curious to see how your testing with long on times works out for spark formation. Just watch your temps

1uF puts me at about 4.5 turns, 1.5uF puts me at about 3.5. Is there some minimum number of primary turns for optimum performance?

You might want to lower your OCD threshold to a known lower value just to see where your peak is at in case you don't have a means of measuring it in realtime. I don't think the UD+ will freewheel until you hit OCD. As others have said, you'll see more heat dissipation per unit time until you start freewheeling.

I'lll try it with 1000A current limit first and see how it goes. That should put me at 158Arms at 1.12ms at 48Hz 5.4%duty cycle. 

Is this your first DR coil?

This is actually my second one, but the last one was built way back in 2007! The technology and knowledge has changed quite a bit since then. I've also built 4 or 5 spark gap coils long ago.
 

Yes, definitely explore behavior of your coil with longer on-times and more measurements using your existing MMC before changing.

Will do some testing next weekend, won't be able to fire the coil up this weekend.

Measuring primary current during operation would be ideal.  This can be done with an additional CT, or by scoping across the driver's feedback input resistor.  Knowing the operating frequency and current ramp-up and limiting (if any) will help in picking your next MMC capacitance.  If you are hitting 2000A for a reasonable portion of each enable pulse, then I wouldn't increase capacitance.

Is this also true for running in freewheel mode?

I have a current monitor installed on the coil, just haven't had it connected to the scope while the coil was actually running. Is it safe to connect the scope with 12'+ streamers flying around? I think i have a 25' coax to connect it with.

Does your controller have a single-shot mode?  I think most do.  I'd use that first for long-enable-pulse testing before running at 48Hz.

Unfortunately no single-shot mode on this controller.

5% duty cycle will likely trip your breaker besides overheating the existing MMC.

I don't doubt this. I will be running on a 100A breaker, so I can turn it up at least a bit.

Do you have a JavaTC dump for your coil?  Or a link to parameters from a similar EVR coil?  That would provide estimates of secondary inductance and frequency, coupling factor, etc.

I don't but secondary inductance is 265mH. I can run the inputs into JavaTC if you're interested.

[acobaugh]

# of turns might affect your coupling. JavaTC will predict that. But, as @davekni suggested, try to explore the parameter space you have now without changing components. Being able to burst is really helpful when changing parameters. I usually use a super short burst from my interrupter after swapping parts and such to minimize the potential damage if I get something wrong. My interrupter is Gao's MIDI2 controller with a simple 555 circuit switching a BJT in front of the controller's output. Two different ranges, and I can vary the on time and off time separately. If this is just PWM control, then you could breadboard together Steve Ward's 555-based interrupter if you don't/can't modify your existing controller.

Yes, it is safe to use the CT with the coil running. Just make sure the coax is routed so it won't get hit by arcs and/or shield it with some metal screen, conduit, etc. I do this all the time to watch the current envelope with the coil running full power. I actually picked up one of those cheap import battery powered single channel scopes off of Amazon just for this purpose, "just in case." Also, make sure the CT is on one leg of the primary circuit right after the bridge. You don't want your CT between your MMC and your primary, or you'll be stress testing the insulation on your CT.
 

[davekni]

Yes, scoping is fine.  Do add a secondary shield of some sort as acobaugh said.  I usually run a strip of grounded aluminum foil or aluminum window screen over scope cables.  Since I'm focused on the electrical/physics side, I scope four channels every time my DRSSTC runs.  Usually that is primary current, secondary current, one H-Bridge output and one gate.  (Gate probing is with a crude home-built fiber-optic-isolated probe.)  Of course, run scope cables the opposite direction as the breakout point.

My quick estimate is 45pF capacitance on your 0.265H secondary for 46kHz unloaded.  That is likely dropping to ~38kHz with arc capacitance, so your primary may be tuned around 38kHz for good performance.

For long on-times, it is generally better to keep primary resonant impedance on the high side.  My guess is that 0.57uF may be about ideal for your build.  Testing will show.

With a 100A breaker, you will probably be fine testing a few seconds at a time at 5% duty cycle.  Worst-case current draw would be just hitting 2000A without much pulse skipping.  That would be 1273A * 5.4% = 68.7A DC from your voltage-doubled bus, so 137.5A average from the line.  Power factor will bring that to 200+A RMS.  But this is worst-case.  And a 100A breaker should allow a couple seconds at 200A.  Reducing OCD setpoint is still a good idea for initial long-enable-pulse testing.

Would love to see some scope captures of this nice coil when you get to testing!

[acobaugh]

More importantly, do you have video you can share?

[fh89]

Thanks for all the help and valuable feedback guys! I will keep you all updated with results!

More importantly, do you have video you can share?

Running in manual mode

running in midi mode

[Uspring]

1uF puts me at about 4.5 turns, 1.5uF puts me at about 3.5. Is there some minimum number of primary turns for optimum performance?

The short answer is yes. The long one is:
The optimum primary inductance depends on your bridges capabilities. If you have a high bus voltage and low max currents you need a higher Z (=impedance) primary tank than in the low voltage and high current case. If you use pulse skipping to limit primary current, you can get away with quite a low primary Z. The drawback is, that you are inputting energy only part of the time, so performance will suffer. It is better to choose primary Z in such a way, that you can input max voltage all of the time and also have a current just below your upper limit. This will maximize input power.

When you start a burst, the input energy will first go into the primary tank and then into the secondary. The arc will then draw power from the secondary and limit current in the primary. The secondary power draw depends on the coupling between primary and secondary, so the coupling value k will also have an effect on primary current. An estimate of the best primary Z can be obtained from the bridges voltage and current capabilities and the coupling.

Even for an optimum choice of primary Z, pulse skipping can still be beneficial, since the primary current initially can become too large due to the secondary being out of tune. That is because primary resonance frequency is often chosen somewhat below secondary resonance frequency. The arc lowers secondary resonance frequency, when it becomes big and so pulls the secondary better into tune. Then secondary power draw will increase, limiting primary current and removing the necessity for pulse skips.

To give a recommendation for the primary tank some more data about your coil is needed.

[fh89]

Here's some video running at 1.12ms PW. PRF 48-156Hz. Primary current limit set at 2kA. Very hot streamers lighting my grass on fire!

Drawing over 150 Amps at 220V from the wall.


At 48Hz duty cycle is 5.39%


At 156Hz duty cycle is 17.53% which is about 592Arms primary current.


Some phone pics:




The ground strikes are nearly continuous at higher frequencies limiting maximum spark length. I think I need to build a platform to raise the coil higher and let the streamers grow.

[davekni]

Probably depends on coil parameters, but my longest arcs were with a very short breakout on top, extending barely beyond the toroid OD.  I think that electric field from the toroid is helping push the arc up and away.

[fh89]

   Probably depends on coil parameters, but my longest arcs were with a very short breakout on top, extending barely beyond the toroid OD.  I think that electric field from the toroid is helping push the arc up and away.   

I'll give this a shot next time I run it.

[davekni]

Was your 151A 25kW reading at 17% duty cycle?  If so, I'm calculating about 80% of pulses are being skipped.  Quite high, but not too surprising given the heavy ground strikes.  Once an arc strikes ground it forms a low-impedance "short", limiting the further energy absorbed by the secondary.  Without much secondary load, primary current rises quickly, holding at your 2000A limit.  As you manage to get longer arcs rather than heavy ground strikes, line power at 17% duty cycle will increase as fewer pulses are skipped.

Did your MMC get hot?  Your high air flow may keep it cool enough even at 592A RMS.

[fh89]

Was your 151A 25kW reading at 17% duty cycle?

Yes, that photo of the power meter is a video still of the peak reading of a short 10 second run with all settings cranked all the way up as in the second scope shot.

If so, I'm calculating about 80% of pulses are being skipped.  Quite high, but not too surprising given the heavy ground strikes.  Once an arc strikes ground it forms a low-impedance "short", limiting the further energy absorbed by the secondary.  Without much secondary load, primary current rises quickly, holding at your 2000A limit.

Very interesting information, I appreciate you being able to make calculations like that.
It makes sense that so many pulses were skipped- the red LED on the controller signifying OCD was illuminated then entire video (you can see it on in the right side of the photo)

I may try to modify the controller for a higher current limit (I've seen people run 4kA on cm600's) but first I will experiment with longer on-times, raising the coil and the shorter break out point. I also have 10000uF worth of caps on the way to add to the bus.

As you manage to get longer arcs rather than heavy ground strikes, line power at 17% duty cycle will increase as fewer pulses are skipped.

Wow, already at 150A, wonder how high I can go for short runs before tripping the breaker? May have to look into getting a 200A breaker.

Did your MMC get hot?  Your high air flow may keep it cool enough even at 592A RMS.

All I can say for sure is that it was cool to the touch 30 seconds after a run. I wasn't brave (or foolish? I don't have any bleeders on the MMC) enough to stick my hand in there sooner. The primary was warm but not hot. Secondary was cool.
I don't currently have a way to actually measure the temp of the MMC or IGBTs, but I will pick up an IR thermometer and see if I can get a reading. I kept all runs <1 min and waited at least 2-3 minutes between runs.

[davekni]

BTW, 75% power factor is higher than I would have guessed.  For my curiosity, I made an LTSpice simulation of just line impedance and voltage doubler.  Adjusted parameters to get a fit with your meter readings based on unloaded line voltage being 240V RMS.  Setting total line resistance to 0.175 ohms and line inductance to 60uH made numbers match well.  These numbers seem reasonable since line impedance includes your power cord all the way back to winding resistance and leakage inductance of the transformer feeding your breaker box.

If this is accurate, power line resistance and inductance is providing enough filtering to keep power factor reasonably high.  That keeps power factor from being sensitive to bulk capacitance.  I was originally going to reply with a concern that increasing bus capacitance would reduce power factor.  In my simulation, adding 10mF (20mF per half of the doubler) made an insignificant reduction in power factor.

[fh89]

BTW, 75% power factor is higher than I would have guessed.  For my curiosity, I made an LTSpice simulation of just line impedance and voltage doubler.  Adjusted parameters to get a fit with your meter readings based on unloaded line voltage being 240V RMS.  Setting total line resistance to 0.175 ohms and line inductance to 60uH made numbers match well.  These numbers seem reasonable since line impedance includes your power cord all the way back to winding resistance and leakage inductance of the transformer feeding your breaker box.

If this is accurate, power line resistance and inductance is providing enough filtering to keep power factor reasonably high.  That keeps power factor from being sensitive to bulk capacitance.  I was originally going to reply with a concern that increasing bus capacitance would reduce power factor.  In my simulation, adding 10mF (20mF per half of the doubler) made an insignificant reduction in power factor.

Once again, I appreciate the analysis. I briefly looked into adding active PFC, but quickly gave up on the idea after realizing the cost and complexity of a boost converter with the necessary ratings. Good to know the extra capacitance shouldn't hurt PF too much.

I went ahead and uploaded the power meter video because I thought you might find it interesting. The PF drifts around a bit of course, but it seems to stay over 70% while the coil is running.