Specifications primary coil: 11 turns of 1/4 inch copper pipe supported by eight 1/8 inch lexan supports epoxied to the acrylic base. javatc gives 35uH for inductance. secondary coil: 1150 turns of 28 gauge red magnet wire on 40 cm of 3.5 inch (thickwall) pvc. javatc gives 23mH for inductance. power supply: 9000v, 30ma neon sign transformer driven with a 0-140v 5A variac low pass filter: standard 'terry filter', but without MOV's, which i judged to be too expensive and not absolutely necessary. i also have a corcom 5A line filter on the low voltage side of the nst. spark gap: 3 short sections of copper pipe fixed to acrylic plates; total gap distance is somewhere around 2.5 mm capacitor: 12 cornell dubilier 942c .15uf in series for .0125uf total topload: 3 inch dryer duct bent into 34 cm diameter toriod and covered with aluminum tape. it's held in place on 1/8 inch threaded rod that's screwed into the endcap of the secondary coil. javatc gives 20pF for capacitance. ground: 5 foot copper pipe completely driven into the ground base: two 3/16 inch thick, 40cm square pieces of acrylic bolted together with 1/4 inch threaded rod wiring: i used 20kv high voltage wire for all connections; in retrospect, this was probably overkill - i think any fairly thick, well insulated wire would have been sufficient. ground wire is a length of regular residential electrical wire (awg14ish?) with all three strands twisted together - not ideal, but given the low power of the coil, i think it's sufficient.
performance: so far, the longest streamers i've gotten are around 15 inches, at 140v input. i think this is pretty good, given the low power of my transformer (300ish VA). although i experimented with larger/smaller toroids, and with moving the tap point on the primary in a full turn and out a full turn (i temporarily added another turn of scrap 1/4 copper pipe), the coil seems to work best tapped at the full 11 turns. javatc agrees that the coil should be in tune here; it gives secondary and primary and secondary resonant frequencies of 237khz and 239khz respectively with the primary tapped at 11 turns.
overview of the setup i used the first day i got it all working; the tripod on the left is supporting a grounded rod
terry filter without MOV's
safety gaps - just screws with their ends filed smooth
mmc capacitor - 10M bleed resistors across each underneath perfboard
spark gap - i have a small computer fan to cool the pipes, since i was worried they might get hot enough to start melting the acrylic. the fan also appears to help with quenching, although i don't notice any difference in performance without it
my way of connecting the secondary coil to ground - i soldered the magnet wire from the end of the coil to strands of bare copper wire that i clamped into one end of a small plug. the piece of plexiglass is there to prevent possible arcing between the primary and ground connector - i don't think its really needed, but for the sake of 30 seconds with the band saw i figured it was worth it
my method for connecting the spark gap to the primary. the piece of threaded rod is to hold the copper pipe in place, since i cut it just too short to make it to the last groove
one of the eight alrylic supports for the primary - epoxying them didn't really work too well, since one or more supports would break off every time the 1/8 inch acyrlic base flexed when i picked it up. i wound up using small blocks on either side of the supports to give added strength, and was really careful not to bend the base at all.
12 inch-ish arcs to a grounded rod at 140v input; the flourscent light in the background was off but was faintly glowing because of its proximity to the tesla coil; the long exposure time made it seem bright.
arcs from breakout point at 140v input
corona/arcs from grounded rod to breakout point with a piece of 1/8 inch acrylic suspended in between - makes for a cool effect as the arcs snake along the surface of the acrylic trying to get around it