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

Re: Measuring secondary resonant frequency



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

Hi Alan, All !

>Original poster: "Kelly & Phillipa Williams by way of Terry Fritz

<snip>

>Question - does this mean that when my coil is operating, and I walk
closer
>to it and further away from it, (while remaining safe from arcs) I am
>actually changing the resonant frequency slightly?


Yes!  But it's so small and the Q under load so small (as you noted)
that it really doesn't make any difference to performance, and the
bandwidth of the signal generated by the TC when operating is so great
you are going to be hard pressed to measure the shift.

>Also, if you touch the toroid during the test, the wave on the scope
cuts
>completely to a straight line. I suspect that this is similar to what
>happens when the secondary coil is arcing to a grounded rod, when the
Q of
>the coil and the output voltage drops.


Yes, severe loading can completely ruin the Q of a resonant circuit,
basically you are adding a resistance in parallel ("shunt") which
dampens things down.  The lower the shunt resistance, the more savage
the damping.  Sufficient (in the shunt case low) resistive load also
changes the frequency, and then the equation f = 1/ 2.pi.root(LC) is
no longer a good enough approximation.

FWIW, F. Langford-Smith in his 1482 page tome "Radio Designer's
Handbook" gives a number of ways of measuring self-resonance,
capacitance etc of coils at rf, some only 10% accurate and some
better, and ends up saying effectively that at rf you can try
measuring what you like, the stray capacitances and inductances will
get you in the end.  Only low frequency measurements can be made truly
accurate.  Grover essentially says the same in his book on inductance
regarding the calculation of rf inductance, the chapter on high
frequency inductance being remarkable for its brevity.  When
attempting accurate measurements at rf, you have to take into account
that capacitors are not pure capacitances and inductors not pure
inductances, nor resistors pure resistances (though that one is
usually the simplest to get close to ideal).

Your experiment demonstrates a stray capacitance between your coil and
you!  Simply attaching the signal generator will add capacitance to
the base of the coil and shift the frequency.  Likewise there exist
strays between the coil and its surroundings (hence the separation
information between coil and floor/walls/ceiling is required in e.g.
Terry's ETesla program) all manner of irreproducible loading due to
humidity in wooden floors and ceiling joists, and heavens knows what
else.

Although this is worlds removed from quantum mechanics (Alan Sharp's
old subject!) it does illustrate the difficulty of making a
measurement without changing the thing you are measuring.  I recall
Malcolm Watts commenting on this list some time ago that he has taken
to regarding the Tesla secondary as a complete system (maybe each
secondary system ought to be described in terms of a numerical Green's
function like the Numerical Electromagnetics Code uses to describe the
composite of an aerial system and its surroundings - very cumbersome)
and I have to say I feel this is the safest approach.

I _still_ think that the NEC2 code -
http://dutettq.et.tudelft.nl/~koen/Nec/welcome.html
- given enough memory (say 0,5GB+) and a fffffffaaaaaaasst cpu - might
form the basis of a very potent secondary simulator, and the source
fortran is public domain and downloadable via the web.  It allows
simulation of real ground using the Sommerfeld method (I think
Sommerfeld published his mathematical ground model around 90 years
ago!)

The problem with the method of moments approach which NEC2 uses is
that it requires the manipulation of gigantic matrices and is *very*
computationally-intensive, calculating the influence of each current
filament on each and every other current filament, with compensations
for skin effect etc.  (Paul - does the tssp project use method of
moments or something else?) I did do some crude modelling with it
months ago on completely unreasonable coils, physically huge with few
turns (100 turns, 5m high, 5m diameter, 10m diameter primary!) and
even so it seemed to pick up self resonance with no trouble, it will
even generate reports of electric field strength at close quarters.
Unfortunately, my computer has a mere 128 meg of memory, and this is
nowhere near enough to test NEC2 on a real coil as it is limited to
maybe 2500 current segments and in any case a run of this size on a
Celeron 300A takes two hours or so!  I'd think you need 10000 segments
at least for this application.  Then of course it becomes worthwhile
doing fancy stuff with LAPACK libraries etc, but I've strayed far
enough on the borders of the topic as it is.  Consult the link and
details therein for more info.

Dunckx