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

Re: Variable Capacitance and Inductance

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

----- Original Message -----
From: "Tesla list" <tesla-at-pupman-dot-com>
To: <tesla-at-pupman-dot-com>
Sent: Saturday, May 25, 2002 9:48 AM
Subject: RE: Variable Capacitance and Inductance

> Original poster: "Paul Nicholson by way of Terry Fritz
<twftesla-at-qwest-dot-net>" <paul-at-abelian.demon.co.uk>
Well said..
> Fortunately, I've noticed that you don't need to know anything
> very advanced - you just need to be really solid on the basics and you
> can make great progress. Tssp hasn't turned up anything original or
> made any new laws.  It's just an example of carefully applying some
> very basic laws of electrics to the TC.  We're applying Kirchhoff's,
> Faraday's, Coulomb's, Ohm's, Biot-Savart laws - all really basic stuff
> that you learn at school, not university. We call it 'research' when
> really it's just an exercise that might help us make better coils.  We
> apply the methods of science, not because we're doing 'new' stuff, but
> simply to try to avoid making errors.  It's as well for list members
> to remind themselves of that from time to time, and not get carried
> away with the fanciful notion that any of us are doing 'New' science.
> --

In fact, modern computer performance has made the job of applying basic
theory easier. Rather than struggle for an analytical solution to a
particular integral, we can numerically integrate the basic underlying
equations (which are truly simple) at ever finer resolution. Maxwell's
equations aren't particularly complex or high order, particularly in their
differential form.  It's not like we are trying to find a closed form
solution for arbitrary helices, toroids, and coupled inductotrs... we don't
need to... we can just let the computer crunch for a minute, hour, day, what
have you.

Then, we can take that computer produced result and see how closely it
matches measured performance: as a test of the computer algorithm and our
measurement technique, more likely than not. Although I am not a craftsman
(by any means), I DO take pleasure in gradually beating the various
measurement errors out (at least in small doses... I'm not a metrologist..
those folks get positively ecstatic about it). There is satisfaction in
removing various biases and factors until you can detect phenomena in the
parts per trillion range: in my case, it was relativistic changes in a TCXO
oscillator frequency on an orbiting spacecraft where the relative velocity
between me and it varied during a pass (7km/sec, 800 km closest approach,
exercise for the reader).

There is, of course, a need for "parametric representations" of the
underlying phenomena.  These are useful as design tools (until computers get
faster).  It's a heck of a lot faster to use Wheeler and Medhurst
approximations in a lumped model to do the quick tradeoffs to get a ballpark
answer: how do I get the longest spark with this transformer I found in the
scrap heap (or whatever your particular figure of merit is).  How much wire
do I need to order? Should I use a 4" or 6" or an 8" secondary form,
etc.etc.etc.  This is what parametric representations are really handy for
(and who cares whether the parameterized form happens to represent some
underlying physics?).  The equations are simple enough that I can run the
calculations on a slide-rule, piece of paper, or hand calculator while
standing at said scrap heap, to decide if I want to drag that piece of pipe,
spool of wire, or transformer home (cost/benefit analysis: look on spouse's
face vs value in producing sparks).

Frankly, I think that as computers get faster, the approximation gets less
useful. Why not use my wireless internet enabled phone to hit a web site
with a beowulf cluster behind it to run the numerical analysis in seconds?
To me, this is much more handy thing than being able to download customized
ring tones, play a spiderman game, or find out what the score is on some
obscure minor league sports event in lower Angola. Who's going to put up the
first WAP enabled website with all these numerical models behind it?

(Some clever algorithms to go with it helps; although, CPU seconds are a lot
cheaper than engineer and software developer seconds, and the trade space is
moving towards brute force every day.)

In the RF/mixed signal world, this is a big problem... Sure, SPICE works
great, but there's not a lot available at higher levels of abstraction, and
designing/simulating a full up radar design in spice would be a real
challenge, much less iterating the design over temperature, parts
tolerances, etc.  Until computers get several orders of magnitude faster, we
design RF systems by using skilled engineers who have tried (and failed, and
fixed the problems) it before.

In Tesla coiling, we're probably all the way there on the mechanical parts
of the system (capacitor, inductor, etc.) and good models, parametric or
otherwise, for the spark gap and streamers are the remaining challenge, and
a very hard one, because the physics isn't well understood (unlike the
physics for the wires, windings, and capacitors)