BrOdin coil - High power Big sparks!

[Hydron]

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.

Software, schematic PDF, Altium SCH/PCB source and (I think) gerbers are all on the page I linked (sources hosted on github so you can just clone the repos). No idea if the EVR version uses <51R for feedback, I'd expect it to already be a smoking crater if it's not smaller than 51!

[fh89]

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.

Oh the feedback resistor- my bad. Yes, it's a 1024:1 ct and the resistor is a 51 ohm:


Looks like possibly a 2w resistor? The schematic Hydron mentioned is the closest thing I know of.

[davekni]

Software, schematic PDF, Altium SCH/PCB source and (I think) gerbers are all on the page I linked (sources hosted on github so you can just clone the repos). No idea if the EVR version uses <51R for feedback, I'd expect it to already be a smoking crater if it's not smaller than 51!

Poked around that link more and found PDF schematics buried a ways down in a zip file.  Thank you!  However, these don't appear to be the EVR version, as OCD reference is divided down from 5V, with ~4.1V maximum setting.  Also, the schematic shows R10 as the feedback CT burden 51-ohm 2W resistor.  fh89's image shows "R6" for the 2W 51-ohm resistor.

It is amazing that R6 hasn't burned or charred the ECB under it.  I can think of two possible reasons.  One is that the EVR version includes TVS diodes to clamp feedback voltage.  The other possibility that comes to mind is CT saturation.  Either of these possibilities might cause phase-lead to increase at high current.  Definitely warrants some exploration to see if this UD+ can correctly handle 2.9A feedback current at 17% duty cycle.  Schematics for this EVR version would be a great place to start.  Did any come with the board?  Would EVR share them electronically? 

[fh89]

Nothing came with the board other than printed pages from Phil's webpage on the UD+ that Hydron linked. I reached out to Phil Slawinski (the designer of EVR's version of UD+) for a schematic. He suggested modding my CT. I think 48 turns should get me to where I need to be - 32*48=1536:1 = 3000/1536=1.95a

[davekni]

If you can get a schematic, I'd much appreciate seeing it.  Unless there is some other difference, even 1.95A peak at 17% duty cycle is much more than the 2W resistor rating.  16W per my calculations:  0.17 * 51 * 1.95 * 1.95 * .5 = 16.48W.  It will take a little time to ramp up to 1.95A skip-pulse threshold, but I think you are spending the majority of your 17% time at the limit based on line power draw.

[fh89]

If you can get a schematic, I'd much appreciate seeing it.  Unless there is some other difference, even 1.95A peak at 17% duty cycle is much more than the 2W resistor rating.  16W per my calculations:  0.17 * 51 * 1.95 * 1.95 * .5 = 16.48W.  It will take a little time to ramp up to 1.95A skip-pulse threshold, but I think you are spending the majority of your 17% time at the limit based on line power draw.

I will see if he's cool with sharing it. I think there must be something else we're missing because I ran the coil this whole time with a 1024:1 CT and the FB resistor was fine when I did the current limit resistor mod on the board. That photo of the resistor was taken after I smoked the secondary the second time and had the controller opened up to mod it, so it had tons of run-time at 1.95A from the CT.

[Hydron]

If EVR hasn't touched that part, then the original schematic is, like I said, on the page I linked. The actual PDF is here: http://classictesla.com/pslawinski/UD2%20CPLD.pdf

Would be very interesting to see the primary current waveform - that resistor does NOT look like it has dissipated anywhere near the power that it in theory it has (I'm amazed it's still in existence!).

[fh89]

If EVR hasn't touched that part, then the original schematic is, like I said, on the page I linked. The actual PDF is here: http://classictesla.com/pslawinski/UD2%20CPLD.pdf

Would be very interesting to see the primary current waveform - that resistor does NOT look like it has dissipated anywhere near the power that it in theory it has (I'm amazed it's still in existence!).

I can confirm the schematic is the same.

After a long chat with Phil, it turns out that I am probably only running at 1000A primary current because of the turns ratio on my CT. The max reference voltage on the comparator is 5v, and I was running 1024:1 which would give ~1A at 1000 primary amps across 5.1 ohms burden resistor - 5v triggering the OCD.

I'm going to rewind the CT so that I can actually run at 2000A. 64:32 = 2048:1 ratio and 2000/2048=.98A * 5.1ohm =4.98V.

[davekni]

If the schematic is actually the same, there is a 2.2k resistor from +5V to the OCD potentiometer (10k).  That gives 4.1V maximum (depending on part tolerances).  You could measure to verify.

Sounds like you have lots of room to push your coil yet harder!

[fh89]

If the schematic is actually the same, there is a 2.2k resistor from +5V to the OCD potentiometer (10k).  That gives 4.1V maximum (depending on part tolerances).  You could measure to verify.

Sounds like you have lots of room to push your coil yet harder!

I will measure it when I take it out to rewind the CT.  If that's the case, then with my modded 3.4r resistor, current limit should be closer to 2500A - 2500/2048= 1.22A * 3.4r = 4.15v.

@Kizmo - What was your primary current limit on your biggerDR coil? It was running cm600's correct?

[fh89]

Currrent transformer is rewound to 64:32 giving 2048:1


@Davekni I measured the voltage with the pot set to max and it was 4.13 as you said. I adjusted the pot to give 3.3v on the comparator to set current limit at 2000A. 2000A/2048 =  .98A * 3.4r = 3.33v

I will still have room to crank it up to 2500A if things aren't getting extremely hot extremely quickly.

FB resistor still looks ok, and should still be ok given that the current into it should remain the same right? 

[davekni]

Yes, the feedback 51-ohm resistor shouldn't run any warmer than it did in your last run (the run after lowering OCD resistor to 3.38 ohms), and even cooler with OCD set for 3.3V instead of 4.13V.  (I'm presuming OCD was set to maximum 4.13V for your last run.)

Feedback 51-ohm resistor power would have still been somewhat high in your last run.  4.13V / 3.38 ohms = 1.22A peak from CTs (1250A primary current).  Power in 51-ohm resistor at 17% duty cycle would be 0.17 * 51 * 1.22 * 1.22 * 0.5 = 6.45W.  If you were running 17% duty cycle for only a minute or two, 6W may not have been enough to scorch anything.  Or perhaps this last run wasn't hitting OCD and pulse-skipping as much as I'd estimated.

[fh89]

So I got a chance to run the coil tonight with the rewound CT and current limit set to 2000A. Nothing else changed from the last run. As soon as I turn up the PW, I get massive racing sparks, long before any large streamers even form.

I turned down the current limit pot about an 1/8 turn and it runs a lot better. I did not get a chance to experiment very much, and I have not yet been able to remove the controller and measure the voltage at the current limit comparator to see what it is set at now. I thought I saw some smoke inside the base and cut the runs short, but after looking everything over, I think it was just smoke from the frame fasteners being vaporized again. Haven't seen this issue since I made the mods to unshort the top turn and bond every frame rail together.

@davekni - I'm hoping you can explain why exceeding a certain current limit results in almost immediate racing sparks! Is there anything I can do mitigate this? Or is my best bet at this point to simply slowly turn up the current limit then back it off a touch when I get the racing sparks?

Another pic for scale. My buddy holding the tape measure is 6'2". The ladder is at 17' but did not get an arc to it this time.

[Mads Barnkob]

The racing sparks is properly a product of too high coupling to the detuning, it simply can not get to a point where its pulled into tune before the volt/turn product on the secondary coil is exceeded.

No matter what, its still amazing results

[fh89]

The racing sparks is properly a product of too high coupling to the detuning, it simply can not get to a point where its pulled into tune before the volt/turn product on the secondary coil is exceeded.

No matter what, its still amazing results

What I don't understand is why simply lowering the current limit a little bit allows it to run as intended, but raising the limit prevents any streamers from forming. I would expect changing the current limit to affect the way the coil runs at longer pw/higher prf or during ground strikes and hence higher primary current and I would understand racing sparks in those instances,  but not getting them before I can even crank the power up. Unless the primary current limit is being reached even before any long streamers are being generated?

[davekni]

Uspring, do you have any models that would help predict onset of racing sparks?

At high current, I'm presuming streamers don't form because you dial back pulse width once racing sparks form.  Or are you running 17% duty cycle at the higher current?  (I'm not suggesting any continued running with racing sparks, as I'd be concerned about damaging your secondary.)

As a simple model, primary coil voltage is proportional to primary current.  Higher current is higher volts/turn.  That translates to higher secondary voltage/turn near the bottom where most of the magnetic coupling is.  At some current threshold that secondary volts/turn creates enough local field to ionize air and start racing sparks.  This is a simple look, ignoring secondary current due to top load voltage.  I haven't tried to make any models for this - would be interesting to try some time.

This would also be a fun setup for controller experimenting, but I don't know of any with necessary capability.  Perhaps a current limit that starts each enable pulse at lower current and increases after an arc forms would allow higher final current without racing sparks.

BTW, as a guess, most trimmer potentiometers have about 3/4 turn of active range.  If so, your 1/8 turn is 1/6 of the range, or about 0.7 volts (4.2V / 6), so from 3.3V down to 2.6V, for 1576A OCD setting.

[TMaxElectronics]

This would also be a fun setup for controller experimenting, but I don't know of any with necessary capability.

Netzpfuscher and I have been planning to implement per-pulse ramped OCD into the UD3 for a while. It should be easy enough to do, as there is already the option to do this in QCW mode.
The question is if a ramped OCD would allow the streamer to grow from the breakout point better and thus limit voltage stress on the secondary.

But shouldn't it be easy enough to get the OCD on any other driver adjustable too? I'm thinking of a slow-ish rc LP behind the adjustment potentiometer, that also has a reset transistor across the capacitor. Maybe just an n-channel fet with the gate hooked directly to the output of the opto, so it is off when the interrupter is on. Then the OCD threshold starts at effectively 0. Obviously this would only work on pulse skip drivers as the others would just shut off immediately

[davekni]

Yes, an analog patch to UD+ could implement ramped OCD as you suggest.  Series resistor could make the starting current non-zero.  Adjusting ramp rate and starting and ending currents between experiments might be a bit tedious.  Your UD3 soft solution would be more conducive to experimenting.  Or, if someone gets around to modeling, there may be a closer starting point for an analog patch to UD+.  Without a model, I agree, we don't know for sure that more current later will help grow intended arcs without racing sparks.

[TMaxElectronics]

It might be interesting to see if there is a relationship between the secondary current and the voltage stress. In theory there should be right? after all that's only a resonant circuit too.
Then the third CT input on the UD3 (or an aux input) could be used to measure that and actively limit it to just below the maximum. Then the output power wouldn't be impacted as much.

[fh89]

So I ended up raising the secondary 1" to reduce coupling. That's as high as I could go short-term without major structural modifications. Javatc say this brings me from .164 to .146. I also turned the current limit back up to 2000A.
     

This reduction in coupling was enough to let me run at 2000A without racing sparks, but I smoked some silicon.

Something dies presumably at :13 seconds in the video without much drama (so hopeful it wasn't the IGBTs) and the breaker trips.  After resetting the breaker twice, something blows up at 1:10.

After getting it back inside and taking a look, it appears to be the diode bridge (Sanrex DF40AA160)
     

Hoping that's all it is, I don't see any damage to the bridge. If an IGBT dies because of overheating, would it normally just fail open without blowing up?  Is there a way to test the bridge as assembled or do I need to take it apart and test the IGBTs individually?