[Home][2016 Index] [TCML] no NST, no worries [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[TCML] no NST, no worries

Hi Group,

I'm building Tesla magnetics on a semi-industrial basis here with the help of a transformer winding company in Poland and a Swiss plastics supplier who provides phenolic paper sheet and
epoxy glass tube stock.

We managed commercial sourcing of some very pretty wound and coated secondaries on 3.5 inch (or 89mm) epoxy-glass tubes n batches of 12. There's also a multi-chamber phenolic transformer bobbin set for custom mains power transformers. As people are lamenting, NST are a bit rare now
and it's time to tackle this problem.

I'm presenting the demo coil in Zürich at the Maker faire next week. This one is billed as the Tesla Time Machine model due to an attractive and impractical Nixie tube alarm clock feature. Ultimate wake up call!

Avoiding the horrors of plastic drain pipe has reduced weight and stabilized the secondary coil against thermal expansion damage (the copper and laminate are matched as in a printed circuit board). FR4 secondary coils are stable enough to be posted almost naked or can be left in hot vehicles without
turns slipping. Anyone else had this accident?

Epoxy glass does not seem prone to holding a static charge after operation and does not absorb water.

As the mains transformer had to be home-brew with a free choice of design parameters, it seems that 6 kVAC @ 200mA is a better platform than any NST voltage; a static spark gap needs only two gaps in series and the primary coil is trivial to engineer compared to classic NST coils. There is no longer any corona.

The mains transformer employs air insulation (no oil or tar). It uses a snap-together bobbin milled from phenolic paper sheet to create eight wire chambers, each for 1000 turns. This transformer uses laminated iron C-cores which make it impractical to replicate the customary magnetic shunt.

The empirical "right" ballast solution deviates from anything I've seen on the list before. There are a couple of interesting universal results that drop out if you are holding a chunk of iron and want to know what it will deliver
for a Tesla coil.

For best results, the main transformer should be air-gapped (NSTs aren't) so it stores enough energy to fully charge the capacitor bank from stored magnetic energy alone. The coil will fire (in your face) readily when the primary is tickled with a car battery. Ultimately, gapping the core wide allows more energy to be stored per kilo but excessive gaps make
excessive copper loss.

As the rule of thumb, 1.5-2% of the magnetic path length should be devoted to air. Each kilo of transformer provides one Joule of energy in the capacitor bank for conservative designs. Pushing it further doesn't seem to help things either.

The primary side is normally driven from the AC line through an much smaller and ungapped saturating inductor that can hold off only 1/3 of the applied AC line voltage. (Another magic number.) The onset of saturation is sudden and causes a resonant LC charging event using leakage inductance; not the same charging mechanism described above.

The waveforms are a bit complicated but the overall effect is to allow easy gap commutation for the static gap with hits on both leading and trailing edges of the mains cycle. This gives 200 Hz operation from a 50Hz line
with a static gap.

A suitable spark gap is built from 5mm fuse holders holding tungsten slugs with TO18 transistor heatsinks
on them all aligned across a 40mm fan diameter. Nothing fancy.

The leading edge hit uses resonant LC charging exploiting the leakage inductance of both transformers while the trailing edge charge is recovering the energy stored in the air gap. It's odd but it works well.

Having full control of the transformer design allows a better power throughput per kilo of construction material. With a little regard for known torroid characteristics (breakdown voltage and capacitance), one can right-size the top load for the c-cores on hand and find out how many of the lovely 942C capacitors we need.

For a good first guess using long shape cores from the catalogue and torroids that are not bulbous, the external diameter of the torroid has to be equal to the magnetic path length of the magnetic core set.

Need about 3 pcs of the 942C 0.15u/2kV for every Kilo of iron. The system always charges resonantly so no safety margin if needed. People who ignore the AC voltage rating of their capacitors may not achieve capacitor lifetimes measured in
thousands of hours (or whole hours?) but these things are consumables.

The power supply can also be simulated in LTspice using the "chan" non-linear core models.

If anyone is interested in CNC milling their own transformers then email me for more info.

Tesla mailing list