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

Re: Base current......Re: Tesla Coil RF Transmitter



Original poster: Paul Nicholson <paul@xxxxxxxxxxxxxxxxxxx>

Dan (Dullfig) wrote:

> 2 -- The GeoTC library clearly states that it calculates the
> inductance from Grover's tables. I have looked at the actual source
> code, and this is indeed the case.

No surprise there, then :)

> Inductance calculations are done by sepparating a coil into
> individual turns, and then looking up the inductance in grover's
> tables for that turn.

Yes.  That method is fast, and accurate enough for the job.
What the program needs in order to calculate distributed inductance
is the mutual inductance between every pair of turns, ie for N
turns in a coil, there are N * (N-1)/2 mutual inductance terms.

> As stated elsewhere, inductance of a resonant coil cannot be
> predicted on the basis of tables.

No, it was stated that the lumped element formulas for solenoids
don't give the correct effective inductance at high frequency
because the current is not uniform.   Because the current is not
uniform, you have to break the solenoid into individual turns,
compute (or lookup in a table) the effect of each turn, and
multiply that effect by the current in that turn.  The total
effective inductance is then added up from all individual turns.

'Lumped' formulas assume every turn in the coil carries the same
current, which is why they fail at high frequency.  This is
nothing to do with being 'lumped' (that's a fairly meaningless
term when applied to real coils).  Instead, it is the built-in
assumption of uniform current, which is only guaranteed to be
valid at low frequency, which can fail at high frequency.
Hence we prefer to call the 'lumped' inductance the 'low frequency'
or DC inductance.

See section 7.4 and 7.5 of

 http://www.abelian.demon.co.uk/tssp/pn2511.html

for definitions of the overall coil inductance in terms of the
mutual inductance between infinitesimal elements of the coil.

> 3 -- ... just because tssp may be accurate, does not mean
> GeoTC is accurate.

There is some loss of accuracy because GeoTC breaks the resonator
into some 10-50 elements for modelling purposes, whereas the
original tssp software uses a few hundred elements.  The
resulting tradeoff between program speed and accuracy is appropriate
for practical coil building purposes.

> As a matter of fact, since GeoTC calculates top load capacitance,
> and coil inductance, and derives resonant frequency from that,
> it seems it is indeed considering the TC as a lumped element
> circuit,

No, the program doesn't calculate resonant frequency by working
out the total inductance, total capacitance, and then applying
the lumped model.   Instead, it works out distributed inductance,
distributed capacitance, then computes Fres from a network model,
and finally computes the voltage and current distributions.

A separate, extra step takes all this distributed information and
determines the effective equivalent lumped L and C parameters, so
that the user can proceed to use a lumped model accurately.

> ...which was disproven by J. & K. Corum.

I don't think the Corums successfully proved or disproved anything
to do with Tesla coils, did they?
--
Paul Nicholson
--