TSSP: List Archives

From: Paul
Date: Thu, 28 Mar 2002 08:25:35 +0000
Subject: Re: Ready :-)) - Re: [TSSP] short H/D and stuff

Hi Terry,

> I wonder what the pinger would reveal on the full coil setups
> and such?

Well I'm hoping that tcma will print out the beat frequencies and
their Q factors.  From the frequencies the k factor can be
calculated, and knowing the Q factors, we should be able to give
an efficiency figure for the resonator.

The challenge with the coupled resonators of the full system is to
reliably extract the beat frequency parameters.  At low k when they
are close together, there are two options: 1) Use a very narrow
filter to separate them and process each separately, or 2) match
a dual resonance waveform to the beat signal, ie simultaneously
optimise the 8 parameters of the beat envelope, rather than doing
independent runs for the two frequencies.  Option 1 is much quicker
but fails when both k and Q are very low.  2nd option has wider
range of application but is much slower.  My approach is to use
option 1 first, to get all the modes, then refine the whole set
together.

> Let's see:
> 22nF and 32.65mH resonates at 5936Hz so that's not it.

I was thinking of the current loop involving the 22nF and the
Victoreen, but looking at it, there's not much room to get the
necessary inductance.  Then again, it could just be an artifact
of the software!  Time will tell - it will either go away or
become more prominent.

> I am no expert on the Sun UNIX stuff but I think I can
> figure out the compiling and stuff.

Do you know whether the GNU C compiler is installed, or the native
Sun?  If in doubt, type gcc to the dollar prompt and see what
happens.  We really should be doing something with that machine.
Is it possible to reach it from the Internet with telnet and ftp?

> http://www.metratek.com/
>  I always like "little guy" programmers like that

Yes, I agree.  In the commercial programming world I come across
a great deal of (perhaps the majority is) utterly crap software,

>  designed by some kind of Dilbert committee...

is a good description of the category!  Specified by marketing
'experts', designed by visionless committees, coded by bored and   
disinterested programmers...    
                                    
> I'll concentrate on getting the Q vs. weather thing going

Yes, the Q variations.  We should focus our attention now on       
considering just what experiments should be done.

First off, to prove that things are working, we should be able to
observe a Q variation of the order of 0.4% per deg C resulting from
the increase of copper's R with T.  Whether we see greater or lesser
change in the Q with temperature will depend on what fraction of the
total loss the copper loss accounts for, and on the temp coeff of
other the lossy materials in the system.

For this test, we should try to exclude as many other sources of
loss as possible, so a foil groundplane and faraday cage to
contain the E field would be nice.  Completely enclosing the coil
in a foil cylinder would do the trick, and the resulting eddy
currents will have the added bonus of containing most of the B field
too.  By this means we ought to be able to obtain a baseline Q
variation due to the coil materials alone.  Then work forwards by
introducing each bit of the environment separately.

The other approach would be to just put the coil out there and get
a look at the full variation, then work backwards to isolate or
eliminate each possible contribution.

Either way, no harm in getting started with some measurements.
tcma will give some reasonable answers as it stands, and we can
always re-run it over old data as and when it is improved.

I'm just wondering whether it would be possible to ping several
coils all at once, ie bases all converge to a junction at the
coil side of the Pearson.  The trace records several overlapping
ringdown waveforms, and tcma untangles it all.  If you want to
try this - ping two coils separately, then ping them both together
and we'll see how the three CSV files compare.  If the coils are
far enough apart, the virtual ground through the spark gap should
ensure the coils are decoupled from each other.   Would be a good
way to compare sonotube and plastic formers.  tcma uses +/- 3%
bandwidth to separate the frequency components, so the coil's
Fres need to be more than 6% apart if high Q, and further apart
if low Q (because the faster ringdown has a higher bandwidth, so
the ringdown from one component spills over into the bandpass of
the other, hindering convergence).
--
Paul Nicholson,
--


Maintainer Paul Nicholson, paul@abelian.demon.co.uk.