Re: EMI and Tesla Coils, calculation of

["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 Jim, All!

Thanks for the comments.

>As an interesting exercise, you might put a small loop in the "near
field"
>( say, 4 50 cm wires in square about 10 meters away), and see what
sort of
>currents are induced.  In the near field, the energy is in the
magnetic
>field rather than the E field.  Get very far away and the E field
>contributions from the various parts of the system all cancel (why
small
>radiators are inefficient).

OK, that's on my "to do" list.  By this evening, it might even be done
:-)

>> Perhaps it's saying that
>> with a low resistance loading, most of the power simply goes to
earth
>> and cooks the soil/concrete.
>
>Exactly... (that, and heating up the secondary and primary..)
Intuitively,
>you know that the spark isn't a great radiator (it's too small) If
you do a
>thermal balance on the spark (i.e. figure out how many joules it
takes to
>heat the air up to 10,000 degrees)(1m * 0.1 cm diameter = 0.7 cc  = 1
mg of
>air = .24E-3 cal/deg = 10 J) you can see that a fair amount of energy
is
>going into heat.

Well OK, that makes good sense then.  This is evidently the energy
which was used to separate the charges in the first place being
released when they come back together again.

>Why don't you post your NEC model file...?


THAT is an excellent suggestion, and I have done exactly that.  Now
anyone who feels mad enough can play with the thing.  You should find
shortly upon:

http://hot-streamer-dot-com/temp/skipnec.zip

a non-self-extracting zip file skipnec.zip a whopping 1k+ in size
containing two NEC2 input "card decks" i.e. text files.

One is called basefeed.nec and is the model of Skip's secondary coil
plus topload over perfectly-conducting ground and is base fed.  The
model as it stands will not generate the printing of coil currents
when you run it, but that can be changed simply by deleting the line
which reads "PT -1".  Be prepared for megabytes of output if you do
this.  As it stands, when run it will model 45 spot frequencies from
500kc/s to 11,5Mc/s at 250kc/s spacing.  This takes 30 hours on a
celeron 300A with 128MB ram and a suitably-compiled version of NEC2
(see below) and the output file will be nearly 900k.

The other is called "inductiv.nec" and is the model of the same coil
plus primary, which on the same system runs in around 40 minutes.  The
resonant frequency of the modelled secondary system is 524,8kc/s (the
real thing is 240kc/s !) and to get the model of the primary resonant
at the same frequency requires a primary series loading of 6,43nF, all
of which is built into the model.  Thus you can see that whatever
comes out of this will not apply to a real-world model directly, and
the reason for doing it is to get overall trends (in current,
impedance, voltage gradients etc) rather than direct predictions.
This model is geared for only the resonant frequency, but produces
coil currents and a bit more near electric field data.

To run these models you will need a version of NEC2 compiled for a
minimum of 3100 segments.  On a machine with 128MB ram this means
unfortunately that swapfiles must be enabled on compilation; 2048
segments in memory and 3100 using swapfiles are appropriate
specifications at compile time and you will need 700MB of free disk
space for the swapfiles (it uses 4 x ~170MB) as each swapfile is much
larger than the amount of memory it would use if it could run the
whole thing in ram.  To run without swapfiles you will need at least
155MB free ram, on top of whatever the system is using, i.e. a machine
with 256MB will do this OK.  Running with swapfiles takes an enormous
performance hit, so if you have the memory, it will be a lot less
painful !

You can find the NEC list archives (mainly pro engineers' forum, list
members include some of the authors of the code) at:

http://dutettq.et.tudelft.nl/~koen/Nec/welcome.html

and the unofficial NEC2 archives (source code, compiled executables,
utilities, manuals) are at:

http://www.qsl-dot-net/wb6tpu/swindex.html

and someone who _really_ knows how to use this thing for modelling
aerials is at

http://www.cebik-dot-com

These ought to be enough to get you under way (well, that's all I had
when I started.)

For those with deep pockets, I reckon you could do a half decent job
on a 1600 turn coil with 6GB ram . . . of course, if your motherboard
won't take 6GB, you should just about be able to get by with a mere
2GB ram and four 18GB SCSI 160 hard drives, one for each swapfile plus
code compiled accordingly.  A fast processor would be advisable. ;-)

Good luck!

Dunckx