Re: VTTC output / NEC2 (long posting)

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

Original poster: "Dr. Duncan Cadd by way of Terry Fritz <twftesla-at-uswest-dot-net>" <dunckx-at-freeuk-dot-com>

Hi Ed, All!


My apologies in advance for the bandwidth!

>Original poster: "Ed Phillips by way of Terry Fritz
<twftesla-at-uswest-dot-net>" <evp-at-pacbell-dot-net>
>
> I'll try to find it amidst all the mess here, or go back to the
source
>and try to get another copy in "electronic form" so I can sent it.
>Original I have is printed text with some hand-written notes in it.


Thanks Ed, much appreciated.

> Are you modelling a TC with the same kind of NEC with which I am
>familiar?


Er . . . you mean the one at:

http://www.qsl-dot-net/wb6tpu/swindex.html
http://dutettq.et.tudelft.nl/~koen/Nec/welcome.html

(Wibble mode set to 'on'.)  Um . . . well, yes.  I know it isn't
exactly ideal (translate: "it was never meant to do this in the first
place and if the numbers mean anything at all it will be not far short
of miraculous") and 2887 segments plus 29 surface patches (I'm
modelling Skip's toroid as a doughnut without the hole - chickened out
of the patch geometry for a true toroid ;-) uses 4x 168MB swapfiles on
the hd as 128MB ram isn't enough . . . it's **SLOW**.

It's also a bit soon to draw any conclusions worth having from this
fit of near-insanity (that's assuming there will be any conclusions
worth drawing) but I'm afraid it was one of those things which just
_had_ to be tried for no better reason than the fact that it could be,
just about, with a large shoe-horn and a lot of patience.  Plus, like
most people's pcs I expect, aside from surfing the web, emailing and a
bit of word processing, my computer generally sits around power off
doing nothing, and if it wasn't doing overnight runs, what else would
it do for its keep?  Not a lot.

For those who have not yet made acquaintance with this program . . .

NEC2 was created for Uncle Sam in the 70s via the Lawrence Livermore
Labs as part of an ongoing project to study radio aerials and EMP
phenomena etc.  When he went for the upgrade, around 1982, Uncle Sam
deposited his old fortran77 code into the public domain so that the
likes of me could do silly things with it.  You can download various
versions of it and the manuals as .pdf files (there are three and it
will take around 650 pages to print them all) from the above URLs and
see the archives of the NEC2 list to get a flavour of what you're
_really_ meant to do with this software.  I'd also recommend a look at
http://www.cebik-dot-com for a good example of its proper, intelligent
application.

What is interesting about it, is that it models metallic structures
(aerials, ships, etc) as short, straight segments of wire or surface
patches for each of which it will calculate the current which is
induced given a suitable voltage excitation.  It does this using some
pretty rigorous physics.  From this may emerge interesting stuff, like
electric field gradients (from which you can calculate Q if you step
the frequency) feedpoint impedances etc. Unfortunately, its relatively
few assumptions begin to break down when:

1) The segments are very short with respect to a wavelength (<0,001
lambda)
2) The segments are used to model bent wires with a small bend radius
3) There is a tendency for there to be a circumferential component of
the current in a segment.
4) The wire segments are closer together than a few radii.

Guess how many of these are compromised in a TC?

Despite this, it is interesting to observe that the phenomenon of
self-resonance _is_ reproduced by modelling a coil, albeit not at the
frequency you would measure it.  Just how much more will come out of
this is something I hope to discover in the weeks to come.  I know
that NEC2 has been used with some success to model mobile whip aerials
which use loading coils, but think in this case that the loading coil
was a "lumped" element and was not allowed to radiate, i.e. the NEC2
model was not in fact applied to the loading coil.

It's a dead cert that this attempt will *not* yield accurate results
on a par with e.g. Paul's TSSP efforts, leastways not with NEC2 in its
present form.  However, it might be possible (in years to come ;-) to
modify the NEC2 program (I have hacked it about in order to design
Landstorfer-Sacher arrays - the original legacy fortran code isn't
pretty, but with the help of the manuals it can be done) in order to
get access to e.g. TC feedpoint impedance predictions over real ground
models which might otherwise be impossible to obtain.  Alternatively,
the NEC2 methodology might be useful in extending Paul's and Richie's
predictions of feedpoint impedance to include realistic ground
simulations.

I do think there is a genuine chance of pinching some of the
techniques used here for other purposes, especially the ground models
might be useful elsewhere e.g. TSSP.  To those with a _strong_
mathematical bent, I would recommend the theory manual, volume 1,
which amounts to a mere 80 pages or so.  Modifying the code to take
account of short segments and close proximity of other wires is going
to be a major task, but I bet there are others out there in the radio
business who are doing exactly that.  And if it comes to modifying the
basis functions I am _well_ out of my depth (hey - as a chemist rather
than a physicist or mathematician I'm out of my depth anyway, but if
experimentation doesn't involve the unknown you're not likely to learn
much. ;-)

(Wibble mode set to 'off'.)

Dunckx