Tesla Forums

Yes, your configuration and aspect ratio mean a lot, here.

 You might also try an inverted cone primary like mine.  This will make  the turns, as they approach center of secondary,  further and further apart. That is, as the inductance increases, and hence,  the primary approaches resonance and highest resonant voltage rise,  the distance increases also. My small coil has a  1 1/2" torus about 9" across (center of torus)  on a secondary about 16" high. The primary "cone" goes about 6" high. at most.  You can twist an experimental, freestanding one of 1/4" Cu tubing in a jiffy. (Don't pay $ for new tubing---swap meets or salvage yard)   If it fixes the problem, you cant then build neat Plexiglas spacers for it .

Try y experimenting with the distance from torus to top of secondary.  Just suspend it with a nylon or poly  string to the ceiling.  Use, of course, a flexible   jumper wire to the top. Try raising it while operating.

It could be that the added capacitance of the torus is changing the resonant frequency's wavelength to  to 1/2 the secondary length.  If nothing else helps, try an experimental secondary  of the same physical size but  1/2 the amount of turns.

Varathane or Formvar insulation on the secondary increases dielectric constant and lowers efficiency, also changing resonant frequency, Older designs used shellac, or single-spaced bare wire.

I'm noticing a similar discharge from the base of a secondary specifically where the magnet wire enters to connect with RF ground.  The dendrite leader points upward to the toroid parallel to the secondary and terminating very much like a point to point discharge about 5" above the breakout point.  The presence of this discharge does not always occur, but once it starts, it tends to remain and start again shortly after the coil is restarted after a cool down (10% duty cycle in part due to the fan also being out--though fans are NOT included in later designs from the same designer).  The breakout point seems to be pretty consistent.  I rotated the secondary a quarter turn to ensure that the issue was not related to anything directly below.

The observed breakout discharge is absent in point to point toroid discharges or when the safety ground is positioned (through the aid of a conveniently located vice) slightly closer to the toroid than the breakout point.

The coil itself is an Information Unlimited BTC30 250kV table top model approximately 7-9 years old with a new capacitor (due in part to Information Unlimited choosing to use whatever Condenser Products Corp had on hand and that was glass -- since changed when they added a Terry filter to the design).   The secondary has had two additional coats of shellac to fix some abrasions recently.   With the current tuning, open air sparks of at least 8-10" inches were observed (the plans indicate the coil should easily produce 10-12" long open air sparks) and a compact fluorescent bulb illuminated about 5 feet away.

For now, my (not the OP) breakout problem seems resolved.  What I did was:

1. Narrow the spark gap slightly
2. Tidy up the inner bundle of wires in the primary.
3. Detuned by moving the tap clip outward.
4. Arranged the tap lead to come in perpendicularly to the primary.

Once all four of these steps were taken, the  breakout was not observed at all and the coil did not once cut out during 15 minutes of on and off testing/experimenting.