Herrick's Transformerless Tesla Coil

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "K. C. Herrick" <kchdlh-at-juno-dot-com> 

Well, OK...not quite: there's always the one BIG transformer, isn't 
there?  But this scheme requires no >other< transformer, and also it 
has >>>noooo<<< transistors!

So once more I prevail on Terry: see 
<http://hot-streamer-dot-com/temp/tspk17s1.pdf>http://hot-streamer-dot-com/temp/tspk17s1.pdf. 
Given enough time away from the easy-chair, I'll begin building something 
like this.  But I'll surely welcome comments in the meantime.

Basically, my design incorporates one equivalent turn of primary 
"coil"--with only the primary capacitors and their interconnections making 
up that "coil", for the lowest realizable impedance.  Groups of capacitors 
are arrayed closely around the periphery of the secondary (in my case, on a 
nominal 12" diameter or a little larger).  They are to be interconnected 
with something like 3/4"-diameter copper tubing.  I would arrange the 
tubing segments so that the capacitors' leads would just be tack-soldered 
directly to them.

I configure 2 uF capacitors, as the drawing shows, in 6 groups of 
15-paralleled, yielding 5 uF (-at- 5100 V withstanding) for the resonating 
capacitance.  Perhaps "resonating" is not quite the right term: with only 1 
"turn", the Q of the assembly will likely be relatively low.  I notice 
that, in such a case in simulation, the secondary's "notch" is absent 
(using 50 m-ohms and ~1 uH in the simulation circuit, that 
is).  So...anyone have a comment on this?  Would one still need to quench 
the gap?  If so, when--given no discernable notch on the secondary's waveform?

I specify the CDE 942C8W2K capacitor. The 942 series has a high 
current-capability, per CDE's catalog data--1148 A peak per capacitor in 
the case of the -C8W2K.

I asked myself:  "With a rotary gap, why not utilize that mechanism to 
additionally distribute charge to the primary capacitors?"  That's what I 
propose to do, as the drawing shows.

The capacitor sections are to be sequentially charged via added rotating 
brushes, bearing against stator contacts, and incorporated into the same 
rotating assembly as the gap elements.  The gap is fired by a rotating 
element that closely interposes between the two stationary elements, twice 
per revolution of the driving shaft and phased with respect to the 
commutator so that that event occurs between a pair of 
commutation-positions.  Thus the discharge events always occur while the 
primary circuit is galvanically isolated from the mains--but the primary 
becomes connected to earth ground during that time through the grounding of 
the rotating element.  By virtue of the two pairs of input slip-ring 
segments on the stator assembly, the input circuit for charging is 
reversed, every 180 degrees of rotation.  The result is that the net dc 
current throught the gap is made to be zero over time, minimizing 
material-transfer across the gap.

Simulation shows that a  450 V charge (a practical level: with no 
resistance, about 600) on each 30 uF of capacitance is realized in about 
700 us.  Thus, allowing for suitable inter-contact spacing, the commutator 
may not rotate more rapidly than 1 turn in ~20 ms, for a 100/s maximum 
spark rate (of 2 sparks per turn).

This seems to me to be a rather viable design.  For a conventional 
RSG-style Tesla coil, one has to build a primary coil, a capacitor assembly 
and a rotary-gap assembly anyway--so those tasks are pretty much a "wash" 
between the two concepts.  Here, one adds the rectifier/resonant-charging 
circuitry and the commutating components, but subtracts the H.V. 
mains-transformer, eliminates the need to protect such a transformer, 
simplifies the "coil" and implements zero net dc current in the 
gap.  Plus...there's no lurking mains-frequency high voltage to give you a 
big surprise when least expected.  A reasonable trade-off, do you suppose?

Ken Herrick