[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Fasthenry program does much more than I originally thought

Original poster: "Jim Lux by way of Terry Fritz <twftesla-at-qwest-dot-net>" <jimlux-at-earthlink-dot-net>

> Original poster: "rob by way of Terry Fritz <twftesla-at-qwest-dot-net>"
> I had some free time at work today and already have a working core of a
> Python language preprocessor for fasthenry.  I'm calling it "fastTesla"
> :)
> I was amused at some  assertions I saw that it wasn't possible to
> accurately model inductors and capacitors.  The programs that do this
> are full-wave EM type simulators, and unfortunately they cost about
> $10000 per node license (e.g. Zeland Software and Ansofts HFSS).  If it
> wasn't possible to accurately model inductors and capacitors, my company
> would have been in Chapter 11 long ago.  We manufacture MMIC's, and our
> biggest problem is modeling the nonlinear capacitances of the active
> devices that are attached to the inductors.  Passive devices are no
> problem.

Just like when modeling antennas, I suspect the real problem for TCs will be
describing the model sufficiently well, not the fidelity of the simulation.
It's all those construction tolerance things... nonuniform windings, etc.

That said, some basic approximations and shortcuts (assuming constant turns
per inch, etc.) would get us started, and then, you could do some Monte
Carlo'ish sensitivity analysis. Practically speaking, it's probably not
worth it... Construction tolerances are probably in the few percent area,
and the existing fairly straightforward models are more accurate than that.
Heck, a simplistic Wheeler and Medcalf approximation is almost that good,
except for the topload/secondary Cself problem...

I think the real value of high performance simulation will come in
accurately modeling the spark growth, top load, secondary coil interaction.
The primary/secondary interaction is pretty straightforward.. The other area
of some interest would be integrating a suitable model of the primary spark
gap in a timedomain simulation..

Correct me if I'm wrong, but the great work by Paul, etc., is basically
quasistatic steadystate for the fields and fluxes.