BrOdin coil - High power Big sparks!

[fh89]

How about replacing one of the horizontal extrusions with something made out of non-conductive material, then you do not need to redo any corners.

This would work if I could find a suitable replacement material in the same t-slot design. The problem for me here is the strength needed. One of the great things about the 80/20 framing system is that the joints are rock solid and very strong. My secondary coil assembly weighs 77 lbs and the toroid is 29lbs, so it is supporting 106 lbs (48kg). I am able (with a helper) to lift the entire coil by the side handles without worry of something breaking in the frame.

You could also just cut out a 1" piece of a horizontal extrusion, as you suggested earlier and replace the missing piece with a machined counterpart of G10 knocked into each extrusion to give it some sturdiness or even mount some more pieces on the outside using just regular screws/nuts in the extrusion slots.

This is exactly what I was thinking until Davekni made some great alternative suggestions. I think isolating one end of each corner will get the job done without compromising the structural integrity of the frame.

You should also use a dedicated conductor from strike rail to RF ground, do not lead that through the anodized frame.

Terrific suggestion, I will make this change! Seems stupid in hindsight that I connected it this way to start!

Isolating both top rails at 2 opposing corners would be best (should eliminate the whole shorted turn path), though 1 should be a lot better than nothing (would leave half of the path there next to the primary and the other half via the bottom rails). Looks to me like some thin shims and isolating the screws from the triangular aluminium bits would do the job. If you're feeling lazy you could order some 0.8mm PCBs (without copper) with the correct shape and hole spacing (0.8mm was the same price as 1.6mm last time I ordered a thin PCB, going below this got pricey).

This is the way I'm going to go basically, but I'll go ahead and isolate one end of each of the four top rails.

As for the CT on the primary, looks like the rating is 700A RMS (and it is internally terminated) so you're probably safe there to leave it connected unless you start pushing a few 10s of % duty cycle, but as Mads said, put it between the bridge and one of the primary tank components (capacitor or primary coil), not between the cap and primary. It's likely possible to insulate it from the primary voltage seen in the current location, but is unnecessary effort vs just putting it somewhere safer.

I already removed it. There is no convenient place to mount it next to the inverter outputs so I'm just going to leave it out for now. I really only needed it to check the phase-lead was correct.

[Hydron]

Thinking about my previous suggestion, you could also just get the triangle bits made out of PCBs, that way no insulators are required for the bolts, just a little insulating shim at the end where the rail butts up against the upright. You'd want them thick rather than thin in this case, but stacking 2 or 3 could be an option given how cheap they are (also available in different colours!). Again would only need 2 corners to be fully insulated, or 4 with one rail insulated.

[davekni]

Hydron's version of fully-insulating two opposing corners saves a bit of work over my suggested version.  Only 6 corner pieces need modifying for shoulder washers rather than 8.  It functions the same electrically - one end of each 80/20 extrusion is insulated and another end is grounded to a vertical strut.  Only reason to do my version is if you like the symmetry of all four corners being identical.

[fh89]

Thinking about my previous suggestion, you could also just get the triangle bits made out of PCBs, that way no insulators are required for the bolts, just a little insulating shim at the end where the rail butts up against the upright. You'd want them thick rather than thin in this case, but stacking 2 or 3 could be an option given how cheap they are (also available in different colours!).

Ordered the nylon sleeve washers already, should be here tomorrow. I think the insulator method will allow me to maintain the high level of strength and clean look of the current frame.

Again would only need 2 corners to be fully insulated, or 4 with one rail insulated.

Good point, I can save a bit of work this way.

Hydron's version of fully-insulating two opposing corners saves a bit of work over my suggested version.  Only 6 corner pieces need modifying for shoulder washers rather than 8.  It functions the same electrically - one end of each 80/20 extrusion is insulated and another end is grounded to a vertical strut.  Only reason to do my version is if you like the symmetry of all four corners being identical.

This is a good idea, I'll go ahead and do both rails on two opposing corners. Thanks for the suggestions guys!

[acobaugh]

On the rail heating topic - since you want to get the thing up off the ground, why not take this opportunity to do just that? Use some sort of insulating material to replace the aluminum posts that hold up the primary, and double the distance from the primary to the top of the base? Just a thought.  Good to see we're learning where the limitations are. I bet it's even more impressive in person with those super fat arcs.

[fh89]

On the rail heating topic - since you want to get the thing up off the ground, why not take this opportunity to do just that? Use some sort of insulating material to replace the aluminum posts that hold up the primary, and double the distance from the primary to the top of the base? Just a thought.

This is a possibility, but I would need to order more 80/20 rail, and extend the coolant lines. I wouldn't use anything other than the extruded aluminum rails for the primary supports as the secondary and toroid assembly weigh over 100 pounds requiring very strong supports. Even then, I'm not sure I could go double without making it too top-heavy and unstable. I built a 2' tall stand as a temporary measure that I will try out next time.

Good to see we're learning where the limitations are. I bet it's even more impressive in person with those super fat arcs.

Absolutely! Almost terrifying close-up, I definitely get an adrenaline rush from it. The scale and magnitude really don't translate well on the gopro footage I posted. Here's a video from my buddy's phone that I think gives a little better impression of the scale even though the autofocus is having a hard time with the streamers. I think 14' to the ladder looks more plausible here (it was measured with a tape measure). For some reason the gopro video looks more compressed horizontally. I'm going to try to get some higher quality video made in the future.

[fh89]

I finished the two corner insulation last night.

Holes in brackets drilled out to fit nylon sleeve washers.


Trimmed flush


Sleeve washers for one corner installed


Kapton tape insulation on underside


Cut some  shims out of HDPE to insulate the rails from the verticals



Added a layer of the kapton tape to the rails


Corner with Kapton tape and HDPE shims


Completed corner


Did the same thing with the opposing corner. Hopefully this will be enough to eliminate the shorted turn effect and stop melting my bolts!

[davekni]

That looks great for insulation!

You'll still need to make sure the other corners are uninsulated all the way to ground.  Otherwise any secondary strike to to the rails will force its way to ground through some insulated path.  That path might or might not be through your newly-insulated corners.  (The only other option that comes to mind is to add strike rings around the outside to block all arcs from the rails.  I think grounded rails are preferable.)

One option for making corners grounded is to grind away the thin anodization layers (on both extrusions and corner pieces if they are anodized).  Another is to use the cut ends that aren't anodized - connect them with foil or wire or ...  Or drill small holes for small metal (not blued) screws, with wires under the screw heads between extrusion pieces.  This could be on the inside to be less visible.  No matter what technique you pick, I suggest a final continuity check between each frame piece and ground.

[fh89]

I think what I'm going to do is use stainless steel self-tapping screws to bond each corner plate to each rail (aside from the insulated ends of course.) I think this will be the easiest way to get a solid electrical connection without having to take everything apart and grind off the anodization.

[davekni]

Yes, that sounds easier than my ideas.  Of course, the holes in the corner plates need to be small enough that threads tap into them as well as the 20/80 extrusions.  I presume you are doing this through the lower part of the frame too - all the way until there is some solid connection to ground.

[fh89]

Slight clarification- I'm using self-drilling screws. They drill their own pilot hole a bit smaller than the thread size, so it will automatically be threaded in to both pieces. I will make sure to get every piece a path to ground.

[davekni]

Wow, I didn't realize there were self-drilling screws for more than sheet-metal and wood.  Sounds like an easy fix if they work on your aluminum plates and extrusions.

[fh89]

Oh, they're intended for sheet metal, but I'm hoping they will work in the soft aluminum! Worst case I can drill pilot holes.

[fh89]

I added 10,000uF to the bus bringing the total to 16,800uF in preparation for running longer on-times. I have parts for a new interrupter capable of 15ms on-times arriving this weekend.

4 Epcos 10,000uF 350V caps


As close to the bridge as I could get and still have a solid mounting point



Caps bolted to 1/4" polycarbonate front panel.

[fh89]

Assembled a new controller from Phillip Slawinski customized for up to 15ms on-times. Duty cycle limit set to 15%.








1.46ms at 100hz max frequency


5ms at 30Hz


15ms at 10HZ


Not sure If I'll get a chance to run the coil this weekend, but I will post results from these on-times if I do!

[fh89]

Whelp, destroyed the coil running at 4ms tonight, but not the way I was expecting. I thought the MMC would be the first to go, but the secondary coil flashed over and is totally destroyed.



Here's a successful run at 3ms:

/>

And here's the video of the failure:

/>

Not looking too good here folks


Heavy carbon tracking at the bottom of the secondary


Bottom of secondary:


PTFE wrap on the lower end of the secondary was designed to prevent/reduce primary-to-secondary flashover. Not sure that's what happened here, or if it was just secondary flashover due to overcoupling.



Secondary completely destroyed.



My guess is this was caused by the coupling being too high for the very long on-time and power level. JAVATC puts k at .18
Any suggestions for a better k when I make the new secondary? The current secondary winding starts at the height of the primary. If I start it 3 inches higher, the k drops to .119, does that sound better?

Here's the JAVATC dump:

Code: [Select]

J A V A T C version 13.6 - CONSOLIDATED OUTPUT
3/5/2021, 7:48:18 PM

Units = Inches
Ambient Temp = 68ºF

----------------------------------------------------
Secondary Coil Inputs:
----------------------------------------------------
Current Profile = G.PROFILE_LOADED
6.375 = Radius 1
6.375 = Radius 2
0 = Height 1
46 = Height 2
1703.7 = Turns
23 = Wire Awg

----------------------------------------------------
Primary Coil Inputs:
----------------------------------------------------
Round Primary Conductor
7.5 = Radius 1
13 = Radius 2
0 = Height 1
0 = Height 2
7.5 = Turns
0.375 = Wire Diameter
0 = Ribbon Width
0 = Ribbon Thickness
0.57 = Primary Cap (uF)
0 = Total Lead Length
0 = Lead Diameter

----------------------------------------------------
Secondary Coil Outputs:
----------------------------------------------------
50.69 [kHz] = Secondary Resonant Frequency
90 [deg °] = Angle of Secondary
46 [inch] = Length of Winding
37 [inch] = Turns Per Unit
0.00443 [inch] = Space Between Turns (edge to edge)
5686.9 [ft] = Length of Wire
3.61 [:1] = H/D Aspect Ratio
114.8175 [Ohms] = DC Resistance
72877 [Ohms] = Reactance at Resonance
8.77 [ lbs] = Weight of Wire
228.818 [mH] = Les-Effective Series Inductance
javascript:PopupCenter('out3.html','out3','900','300')237.254 [mH] = Lee-Equivalent Energy Inductance
230.116 [mH] = Ldc-Low Frequency Inductance
43.083 [pF] = Ces-Effective Shunt Capacitance
41.551 [pF] = Cee-Equivalent Energy Capacitance
77.946 [pF] = Cdc-Low Frequency Capacitance
13.43 [mils] = Skin Depth
32.374 [pF] = Topload Effective Capacitance
193.688 [Ohms] = Effective AC Resistance
376 [Q] = Quality Factor

----------------------------------------------------
Primary Coil Outputs:
----------------------------------------------------
33.05 [kHz] = Primary Resonant Frequency
34.81 [% high] = Percent Detuned
0 [deg °] = Angle of Primary
40.25 [ft] = Length of Wire
2.97 [mOhms] = DC Resistance
0.358 [inch] = Average spacing between turns (edge to edge)
0.926 [ inch] = Proximity between coils
0 [inch] = Recommended minimum proximity between coils
40.691 [µH] = Ldc-Low Frequency Inductance
0.24227 [µF] = Cap size needed with Primary L (reference)
0 [µH] = Lead Length Inductance
552.257 [µH] = Lm-Mutual Inductance
0.18 [k] = Coupling Coefficient
0.155 [k] = Recommended Coupling Coefficient
5.56 [half cycles] = Number of half cycles for energy transfer at K
82.34 [µs] = Time for total energy transfer

----------------------------------------------------
Top Load Inputs:
----------------------------------------------------
Toroid #1: minor=10, major=36, height=52, topload


[davekni]

Yes, I suspect it was local secondary breakdown where voltage/turn is highest due to coupling.  Unfortunately, reducing coupling is likely to reduce long-arc performance.  It will reduce how low the primary can be tuned while still generating enough voltage to start an arc.  Others here may have more specific experience.  For short on-times, I think there can be advantages to lower coupling.  I can't think of any advantages with long on-times.  I had to reduce my DRSSTC coupling from 0.157 to 0.14 and add more resin layers to fix secondary arcing.  In my case, arcing from ~1/3rd of the way up to ground guard ring.  Would love to have 0.18 coupling, but my 160mm secondary diameter limits getting there.

That PTFE wrap may have induced the failure.  I've found (the hard way) that thin trapped air in areas of high AC electric field makes breakdown more likely.  I only half-understand why.  With thicker insulation material, the mechanism is more clear.  With a thin layer like your PTFE, it isn't as obvious to me.  It may be a combination of trapping heat and trapping any tiny bits of corona discharge, allowing the resulting slight insulation degradation to accumulate into a failure.  I've seen that on small HV projects myself even with thin layers over a thin air gap.

You are left with the same quandary that I had:  How much to reduce coupling and how much to rely on other improvements.  If you gambled on just leaving off the PTFE wrap next time (and perhaps thicker resin at the bottom), there is a lot of work making a new secondary that could still end up with the same failure.  Perhaps that combined with a slight reduction in coupling.

If you do reduce coupling, I suggest doing so by increasing primary inside diameter if that isn't too hard mechanically.  Do you have enough margin to remove the inside turn without running out of tuning range?  The reason to lower coupling this way is that it spreads out the field along the secondary.  Volts/turn will reduce more for a given reduction in coupling.

Such failures sure are frustrating.  But they are always part of pushing limits.  Good luck with your secondary rebuild!

[alexhanyuan]

Nice work! As nasty as that looks, you should be able to clean that up no problem. I've had no problems doing solder splices in the middle of my secondary. Just coat it back up with several thick layers of epoxy.

I didn't realize that was PTFE wrap on the bottom of your secondary. I half agree with davekni; it looks like your epoxy layer is relatively thin and the PTFE delayed the failure.

Keep the updates coming!

[fh89]

That PTFE wrap may have induced the failure.  I've found (the hard way) that thin trapped air in areas of high AC electric field makes breakdown more likely.  I only half-understand why.  With thicker insulation material, the mechanism is more clear.  With a thin layer like your PTFE, it isn't as obvious to me.  It may be a combination of trapping heat and trapping any tiny bits of corona discharge, allowing the resulting slight insulation degradation to accumulate into a failure.  I've seen that on small HV projects myself even with thin layers over a thin air gap.

This explanation makes sense to me. The PTFE layer is actually relatively thick at .093" I could definitely see how it would trap heat and concentrate corona in small gaps. That stupid piece of plastic cost $275 just to end up destroying what it was intended to protect! Oh well, lesson learned!

Unfortunately, reducing coupling is likely to reduce long-arc performance.  It will reduce how low the primary can be tuned while still generating enough voltage to start an arc.  Others here may have more specific experience.  For short on-times, I think there can be advantages to lower coupling.  I can't think of any advantages with long on-times.  I had to reduce my DRSSTC coupling from 0.157 to 0.14 and add more resin layers to fix secondary arcing.  In my case, arcing from ~1/3rd of the way up to ground guard ring.  Would love to have 0.18 coupling, but my 160mm secondary diameter limits getting there.

You are left with the same quandary that I had:  How much to reduce coupling and how much to rely on other improvements.  If you gambled on just leaving off the PTFE wrap next time (and perhaps thicker resin at the bottom), there is a lot of work making a new secondary that could still end up with the same failure.  Perhaps that combined with a slight reduction in coupling.

Well assuming the PTFE layer contributed to or outright caused the failure, and considering that reducing the coupling will reduce performance and may not be necessary, maybe I will just rewind it the same height and like you said, leave out the teflon and add a few extra coats of epoxy on the bottom.

If you do reduce coupling, I suggest doing so by increasing primary inside diameter if that isn't too hard mechanically.  Do you have enough margin to remove the inside turn without running out of tuning range?  The reason to lower coupling this way is that it spreads out the field along the secondary.  Volts/turn will reduce more for a given reduction in coupling.

I don't have that margin the way it is now, I'm tapped all the way out. I would have to move the primary supports out, and rewind the primary from scratch. PITA factor is also increased by the liquid cooling setup- having to drain the coolant, make new coolant lines etc. Hopefully this won't be necessary and the above idea will work. However, I do think I might benefit from being able to tune the primary lower and I could add half a turn if I remade it. That or build a bigger MMC to drop the Fres.

Such failures sure are frustrating.  But they are always part of pushing limits.  Good luck with your secondary rebuild!

Thank you sir! I will get it repaired and working again soon. I was expecting failure at some point for sure, but was expecting it to be the MMC or IGBTs!

Nice work! As nasty as that looks, you should be able to clean that up no problem. I've had no problems doing solder splices in the middle of my secondary. Just coat it back up with several thick layers of epoxy.

Thanks! I haven't decided if I am going to attempt to repair it or just rewind the whole thing. You guys think a splice would hold up on a coil of this power level? Seems to me like the slight imperfection would lead to heavy corona at the splice point and quick re-failure? It would certainly be a ton easier to splice the wire and only have to re-wind the last 1/4 or so of the coil. I'd give this a shot if the consensus is that it will hold up.

I didn't realize that was PTFE wrap on the bottom of your secondary. I half agree with davekni; it looks like your epoxy layer is relatively thin and the PTFE delayed the failure.

The epoxy was 3 coats of famowood glaze coat- used nearly a gallon on the coil! Apparently even that was not enough though!

I think Dave was saying that possibly the PTFE actually caused the failure by trapping heat and causing uneven corona. I think his explanation makes sense. A classic example of poor engineering, fixing one problem and causing another with the solution! Lesson learned the hard way!

[davekni]

How symmetric is your secondary coil?  Repair may be possible if you can install it upside down.  Voltage stress is lower at the top, so may be able to handle the patches.