From: Paul
Date: Wed, 06 Jun 2001 15:09:00 +0100
Subject: Re: [TSSP] Interesting article on Medhurst, Wheeler, modeling, etc.
Hi All,
Terry wrote:
> It seems like the first step is to actually have a fairly finished
> paper to send around.
Yes, that could be some way off.
Bart wrote:
> But, are the **conclusions** of TSSP ready for outside review?
The more I reflect on our project, the more holes I find. Just focusing
on the Cint issue, there are many fair criticisms that can be made, and
which we must address before something can be put together.
Some criticisms might be:
Where are your experimental results for h/d=1 and below? You say this
is where Cint shows up strongest, but you've no results for that
region.
Your predictions for high elevation don't match the measurements. 10%
to 25% error. This is another region where you say Cint is more
influential. Doesn't that suggest that Cint is either calculated wrong
or a nonsense altogether?
Prove to me that you are calculating Cint correctly. What are your
error estimates? Demonstrate that the predictions converge uniformly
to the measured values as Cint spatial resolution is increased.
Plenty more where those come from! Those are the kind tough questions
we have to ask ourselves - because a review to a professional research
standard would be a lot tougher.
Maybe Cint is just a blunder, since it's rarely mentioned in other
modern papers which investigate solenoids (eg Bodlovic's, and another
paper which is very closely related to our stuff -
Konig and Kreb,
'Study of Power Coil Resonance Phenomena Base on Travelling Wave
Theory', IEEE Transactions on Electrical Insulation,
Vol 26, No. 3, June 1991.
It's scanned on the web somewhere as a PDF called I think, konig.pdf.
Can't recall who's hosting it - probably Terry :)
They only consider C between immediate neighbour turns, not the longer
range C, but their paper is not actually modeling anything, they are
just reporting qualitatively an empirical spectrum, rather than
predicting the mode frequencies, so it's quite basic in that respect.
Meanwhile, the best we can do at the moment is to try to get some
interest in the project from those whose job it is to do this stuff
professionally. Someone who can advise us whether our fun project has
any relevance to the real world of electrical engineering.
Personally, I *believe* at the moment in Cint and its effects, but that
belief effectively disqualifies me from being a capable judge on whether
it is real. I'm open to lots of bias, eg when someone sends me a set
of measurements that agree nicely with predictions, I tend not to query
them. The results that don't match, we look closer at, and find ways to
get a better result. Do you see that the effect of this is to narrow
down the standard deviation of results, bringing the outliers in more
towards the center?
Folk often complain that scientists are too skeptical about things,
implying that they are narrow minded. But strong skepticism is vital in
order to separate out what's real from what's just a nice idea. It's a
way to counteract the natural (survival advantage) tendency of people to
believe things for the most flimsiest of reasons. A high intellect
doesn't help at all - just makes it easier for the perpetrator to make a
more convincing justification, but doesn't reflect on the reality of a
phenomena.
Our investigations seem to have forked into a number of branches.
I'll list them, because I like lists of things - they help me get a
clear picture,
a) Figure out how to predict the unloaded Q of a secondary coil. The
continuing mystery of the missing loss - the energy budget never
seems to balance - what is the I^2R loss, and what are the other
losses? Can we please go beyond Medhurst?
b) Firm up our picture of the secondary's physical reactance. Make the
computations more accurate, and more defensibly correct. Need to put
together a solid body of evidence to confirm these, especially the
Cint. This might include finding a way to measure a current
profile, since the elevated maximum is a direct effect of Cint.
c) Explore the dynamics of streamer loading from toroids and spheres,
find justifiable relationships between topvolts and streamer length,
confirm that streamer formation begins at a topvolts commensurate
with 26kV/cm at surface.
d) Similar to above, but for CW brush discharges from needle-like
electrodes. Any practical relationships between power and plasma
volume? Topvolts and volume, or length? Can we predict the load
impedance?
e) Primary gap behaviour. Can we do anything to predict voltage drop
or instantaneous resistance? Can we predict when a gap will quench?
f) Incorporate the non-linear stuff from c,d,e into the time domain
model.
Plus a few more, purely software issues, that I can think of. Plenty
there to go at!
As regards measurements, we don't need vast quantities, but what we do
need are unequivocal results - of the kind that Terry has demonstrated
on a couple of occasions.
Terry wrote:
> I have some short coils too when you need some data on them.
> I normally don't think of them as Tesla coils but they sound like
> just what you are looking for.
We need h/d=1 down to h/d=0.5, quite large coils, say at least 40cm
diam, on thin formers. Not too many turns, 500 is nice. No top
terminal - wire just ends abruptly, no tail. Base terminal small.
Coax feed to beneath center of ground plane. Very thin feedwire
rising up through groundplane to base. Need to be able to measure
resonances at a range of heights. Best way for you might be to hit
the coil with a transient, say a low frequency square wave, and
capture the resulting base current trace. We'll FT that to get all the
modes at once. Throw in a wiff of counter'd CW for a scope timebase
calibration. Test outside or well away from walls and things.
Now if such a coil was available with series tapping points, and if
you could figure out how to measure Q really precisely, you could use
the resistance insertion method to deduce the all-important current
profile. Alternatively, some sort of tiny in-line current meter
that could be moved along the tapping points..
One point, demonstrating an elevated current maximum, would I think,
prove beyond reasonable doubt that long range Cint is behaving the way
pn2511 says it should. Doesn't matter if the maximum isn't quite
at the right magnitude, or the right height - a qualitative result would
be strong in itself - there's just no other mechanism capable of
elevating the current maximum. Such a measurement would decouple the
issue of whether Cint exists and is effective, from the issue of whether
the software is actually calculating the right value.
In summary, we're a long way off having anything worth reporting, but
some kind of up-front review of our efforts would help put things into
a proper perspective, after all, we're just a bunch of amateurs dabbling
in all this.
Cheers All,
--
Paul Nicholson,
Manchester, UK.
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.