Re: Rotary Spark Gap Design
In-Reply-To: tesla-at-pupman-dot-com -- 05/15/96 12:20
Regarding the xmsn line vs. lumped element methods of modeling the secondary,
it seems that the most accurate model is a mixture of the two, but much closer
to the lumped element side. There are several reasons why a close-wound,
single layer coil cannot be treated as an xmsn line:
1. Turn-to-turn coupling - If a xmsn line is divided into segments, each
segment has almost no 'lumped' coupling with the next. Adjacent turns in a
coil have a coupling of about 0.15, which would cause massive dispersion for
any wave that tries to propagate along the secondary wire.
2. Impedance - The highest line impedance possible without ferrite cores is
that of free space, about 377 ohms. For two parallel wires, it's around a
few hundred ohms. But the effective output impedance of a tesla coil (Vs/Is)
is usually between 10,000 and 30,000 ohms, clear evidence of some lumped
parameters in action.
3. Resonance - Secondary coils are highly resonant, as everyone here has
witnessed. Xmsn lines are, by definition, wide-band and therefore non-resonant.
However, 'lumped' isn't a really good description of C sec either, since it's
more or less distributed over the surface of the secondary. And to make
matters worse, the surface is not all at the same potential, resulting in
a horrifying modelling challenge for even the most avid 'rainy day modeler'.
One thing for sure, It's a tough problem. I've tried modeling it in PSPICE,
treating each turn as an xfmr winding, with a stray cap, and k=0.15 to the
adjacent turns. The results are sort of believable; perhaps the most
interesting result was that each turn contributes the same energy, instead
of voltage, resulting in sort of a'square root of X' voltage curve along
the winding. (Looked a lot like a quarter-sinewave at first).
I don't have high confidence in my model at this point, and will gladly trade
info with other 'rainy-day modelers'.