This is a description of the evolution of my coil, detailing the chronology and the incremental improvements I made to it.

It all began when my wife got me a pair of oil burner ignition transformers (OBIT's) for Christmas, 1996, after seeing my eyes light up upon seeing a Jacob's Ladder made by a friend. Had she realized where these would lead, I'd probably have a new shirt and you'd be looking at somebody else's web site! A friend sent me a pointer to a Tesla web site, and I realized "I can build one of these!!" My ambitions quickly exceeded the Jacobs Ladder and I began seeking information about Tesla Coils, scouring every web site I could find.

As I began gathering materials, I sensed that using OBIT's were a second choice for power supply as compared with Neon Sign Transformers (NST's). Even just from a mechanical point of view, the OBIT's were awkward to mount, so I began looking for used NST's. The very first sign store I tried was a score! The guy said that every once in a while someone comes in looking for a used transformer, so he keeps 1-2 around for that reason, and he gave me a nice clean 15KV/30mA unit, for free!

My first capacitor was then sized for this NST, targeting .005 uF. I made a rolled poly/flashing/oil unit, using two layers of .04" poly between the plates. No vacuum pump-down. Achieved .0066 uF.

My first primary was a flat pancake constructed of 0.25" copper tubing. I decided to buck convention and locate the primary below a 3/8" Lexan support plate, rather than above it. My thinking was that it would be easier to change primary taps from below, and that the lexan plate above it would provide strike protection as well as keeping dust off the primary.

My first spark gap was a 6 x .03" RQ cylinder static gap with fan cooling.

Top terminal was a 4" x 18" toroid made from aluminum flex duct, stiffened and smoothed with wood filler and spackle, covered with aluminum tape.

The secondary (the one I still use) is 4.25"D x 23.25"L 22 AWG wound on 1/16" wall dried/sealed PVC SDR drain pipe. Many coats polyurethane.

For the NST protection network, I was frustrated in not being able to locate a clear consensus on what's best. I decided upon safety gaps, a pair of home made 600 pF bypass caps made from .09" copper-clad G10, 750 Ohm 50W series damping resistors, and several ferrite core chokes.

My ferrite core chokes had serious arc-over problems that could not be overcome with insulating goop. With the voltages involved, ferrite is NOT an insulator. I got rid of the ferrite's and wound some #30 AWG 2.25"D x 7.5"L air-core chokes on PVC pipe, measuring 7.9mH.

The air-core chokes had their own problems. Despite series damping resistors, some sort of choke resonance caused the entire tank circuit to want to arc to ground. This didn't happen with the ferrite chokes. I was getting arcs from the primary to the base of the secondary, and arcs through the PVC case of my rolled cap to the wood base. I tried changing the choke's orientation relative to the primary, and moving the chokes several feet distant from the primary. The only thing that helped was putting the damping resistors in parallel with the chokes, but then that diminishes their low pass filtering characteristics.

I finally gave up on the air core chokes and got some big 3.25" OD x 2.25" ID x .50"Thk ferrite toroids, heavily insulated them with .04" poly and lots of hot-melt glue. 51 turns of heavy poly insulated #22 AWG yielded 14mH. This solved both arcing problems.

Performance of version 1.0 coil: 10-12" sparks from nail on toroid, no breakout without nail.

I realized that my total gap setting of 0.18" might be somewhat small for a 15KV power supply, so I built a second set of 6 x .03" gaps in series with the first set for a total of 12 x .03" = 0.36". I believe this boosted performance up to 16" sparks.

At some point in this configuration, things came to a sudden, silent stop. A post-mortem on my capacitor revealed a dielectric puncture several inches inside from the edge of the flashing. The flashing at that point looked like it had a raised pimple, though I can't imagine not noticing it as I was assembling it.

I wanted MORE POWER, a bigger NST. This also meant a bigger capacitor.

First order of business - a new NST. It seems that while 30 mA NST's are used everywhere, 60mA jobs are almost unheard of. I exhausted all of the nearby sign shops and let my fingers do the walking. I finally found a place that had a used 15KV/60mA, and bought it for $73.

Next, a new capacitor. I saw that Fair Radio Sales was selling .01uF/100KVDC units for $99. The specs sounded right so I bought one, I had too many other things to build and didn't look forward to rolling another.

Version 2.0 - Flat pancake primary, 15KV/60mA NST, .01 uF Fair Radio Sales capacitor, 14mH ferrite chokes, 500 Ohm/75W damping R's, 600 pF bypass caps, 12 x .03" RQ static fan cooled gap, 4.25"D x 23.25"L 22 AWG secondary, 4.5" x 15.5" toroid. Performance was about 28".

After not too much run time, the performance suddenly went from 28" to zero. Diagnosis: One side of the NST was shorted. This, despite my obsessing over chokes and protection networks! Well, I had read up on the NST unpotting and repair procedures. I wasn't looking forward to this smelly, messy job but figured if successful, may result in a better, more robust, and possibly more powerful transformer if I can remove some shunts.

The top cover of my NST unscrewed, revealing the cracked asphalt surface covering the transformer guts. I placed the transformer in a disposable aluminum roasting pan and put this in a covered charcoal barbecue. Several hours later the asphalt was liquid and I poured off what I could, and dumped the hot core and coils into paint thinner. After cleaning everything off with many additional baths and reconnecting the primary and secondary connections, I repotted the works in the original enclosure and bushings, this time using melted Vaseline petroleum jelly. Due to the construction of the core, I was unable to remove any current shunts.

With my primary located below the lexan support plate, the closest I could get my secondary was about 2" or so above the primary, and I believed that I was undercoupled. I decided to rebuild my primary as a 15-degree conical shape, above the support plate, reusing my .25" copper tubing. Now with the secondary in its lowest setting, the first turn of each were at the same level.

I discovered that with the larger power supply and tighter coupling, I could no longer get just a single streamer breakout from a nail on my toroid, there were always several streamers. This indicated that I needed a larger top load. I constructed a second toroid, 6.25" x 24", and placed this on top of the original 4.5" x 15.5" toroid.

The new conical primary provided the single greatest improvement in performance to date. Maximum spark now - 42" from a nail on the 6.25" x 24" toroid! As far as nice looking sparks however, I prefer the multiple (shorter) streamers from just the smaller toroid.

Prompted by discussions on the Tesla List and the continued lack of agreement on what constitutes an adequate NST protection network, I began studying different networks, using PSpice to simulate the consequences. What I found was that LCR networks, which is what I was using, has the potential to cause as much high voltage oscillations as it filtered out. I pulled out the ferrite chokes and replaced the 500 Ohm/75W wire wound resistors with 1.6K/113W non-inductive resistors.

Terry Fritz published a paper on his web site, comparing the performance of a single gap with a multiple gap. At higher couplings, the multiple gaps showed superior quenching, but the single gap showed higher secondary currents. No forced air was used. I reasoned that if forced air were used, the quenching of a single gap might be improved, and if so, the higher secondary currents would result in better performance (as measured by spark length), even if quenching were not as good. I built a single gap with vacuum cleaner motor forced air to replace my 12-segment RQ gap. Total gap spacing for both was 0.36 inches.

With the new single gap, spark performance increased from 42" to 51".

My Fair Radio Sales capacitor would always get slightly warm, but with the single gap, it got warm fast, possibly due to inferior quenching. I decided to try my hand at building an extended foil rolled capacitor. Initial results see comparable spark performance, but there's no sign of heating.

Ummm, an update to the last paragraph. After being asked if my new extended-foil rolled-poly capacitor is really robust and has seen enough use to declare it a good design, I attempted to run it for about 7 continuous minutes. I noted that the ends of my new rolled caps did get slightly warm in the area corresponding to the place where the foil ends (and where corona and dielectric stress would be greatest). Then at about the 7 minute mark, there was a loud POOM and a spray of mineral oil all over my garage floor. A dielectric puncture occurred at the edge of the foil and I failed to provide any pressure relief mechanism.

Next project is to build a dry capacitor, using 20 elements in series to achieve a voltage per element low enough that corona doesn't occur and oil immersion isn't necessary. It will be a flat plate design, using aluminum foil and 13"x10" 2 mil poly bags. Each element will be 0.2 uF, and will have 6 mils dielectric.

Well, you can't imagine how tedious it is to cut down something like 800 plastic bags and foil sheets to just the right sizes, and to get them all neatly stacked. I built three of the 0.2uF sections and started thinking about doing other things. Discussions on the Tesla List described using polypropylene caps. That sounds a lot easier than cutting all of those bags.

So in March 1999 I shelled out well over $100 for a big bag of polypropylene caps and in about 1 weeks time, completed a new tank cap using the so-called "MMC", or Multi-Mini Capacitor approach. I'll probably abandon completely the dry poly bag cap now.

What next? I may try to increase my pri/sec coupling. Right now with my 15 degree primary even with my secondary base, the coupling constant k = 0.139, and if I raise the lowest pri turn 0.25" above the sec base, I can get k = 0.144. I'd like to shoot for one of the so-called magic values of k at 0.153 or 0.18, but to do this, I'll probably have to construct a ribbon or two-layer primary.

For kicks, I tried paralleling my .01uF Fair Radio Sales cap with my .012uF MMC, for a total tank capacitance of .022uF. Of course I had to tap inward about two turns on the primary, but now with this larger cap, I see 58" (max) streamers. I also see my 120VAC input current pegging a 20 Amp meter and tripping the 20 Amp breaker in about a minute's time, but I like it!!! So much for sizing the cap for mains resonance!

In June '99 I made a fabulous buy of some .02uF/45KV Maxwell Pulse caps at the MIT flea Market, so I'll be trying to use them wherever possible now. Going from my MMC and FRS cap to a single Maxwell cap pushed the maximum arc length to a nice even 60".

It's September 1999 and work on my synchronous rotary spark gap is done. I used a small 3600 RPM sync motor pulled from an old Teletype machine. The disk is 6.25" diameter 3/8" thick Lexan, with two 2.75" long x 5/8" diameter copper pipe electrodes. I had to build a custom starting circuit to enable the motor to spin up such a heavy rotor assembly. I also increased the tank capacitance to .03uF so as not to over-volt the cap, and I tapped further in on the primary. So far I've had nothing but trouble with racing sparks along the length of the secondary. I've tried raising the secondary up by 3/4" but still got the racing sparks, and burnt an area near the base of the secondary. I'm now wondering if it's not time to move on to a 6" diameter and longer secondary…

It's December 1999. I've built a new 6.25" by 28" #22AWG secondary, and I've modified my conical primary, removing the two innermost turns to make clearance for the larger secondary, and splicing them onto the outermost turn. System resonance is now at around 138KHz. With the .02uF cap and my single static gap, the performance is, well, different. Due to the taller secondary, my top load is now too close to the garage ceiling to allow me to make a comparative maximum arc length measurement to the wall. By "different", it used to be that the large toroid would only intermittently break out without a breakout point. Now it breaks out into several arcs, with or without a point.

I tried hooking my synchronous RSG up in place of the static gap, and I added my .01uF MMC in parallel with the .02uF Maxwell cap. For about 10 wonderful seconds, the toroid screamed with a long, sinuous, 120 Hz-sounding arc. Then the smell of burnt plastic became apparent, and the show was over. The aluminum bars that held the stationary electrodes had sharp corners that were aimed towards each other, and there was a piece of Lexan plastic bridging this relatively narrow span of about 1.25". I suspect that the sharp edges generated corona which promoted surface arcing across the plastic electrode supports, shorting out the two electrodes.

I've since rounded all edges of the electrode supports and replaced the burnt plastic piece, but I'm now having problems with the sync motor starting circuit. Now, more often than not, the motor blows its 10 Amp fuse during startup. I'm hoping it's something that can be fixed just by altering the timing of the starting relay…

Fast forward to 2004.  I've been lax in updating this page, there's probably been several intermediate incarnations between what was last reported and how my coil is now.  Many things have changed - ya can't change just one thing without affecting many others.  I built a 2-layer primary, the theory being that a high-inductance primary has a higher impedance, which means lower current through the spark gap, which means (I think) lower gap losses.  A high inductance primary also means a lower operating frequency, which means lower skin-effect losses throughout the coil.  I changed from a 6.25"x28" #22AWG secondary to a 6.25"x28" #26AWG secondary , and I built the 7"x30" foam-core toroid top load to match the lower primary frequency.  Experiments to determine the best cap size revealed that I should be using a .04uF cap with my 15/60 NST.  I rebuilt my sync RSG to a propeller configuration.  I was now getting 66" arcs, and I can no longer run my coil in the garage without hitting just the ceiling.  Racing sparks are a problem as I advance the SRSG phase and the bangs get bigger.  I found that by using two toroids, raising the main one higher, the tendency to generate racing sparks is reduced.  This allowed me to hit a personal best of an 80" streamer.

Back to Gary Lau's main Tesla Coil Page