Re: Measure that frequency shift. (or, at least, change the subject) :-) (fwd)

---------- Forwarded message ----------
Date: Sun, 19 Jul 1998 08:53:52 -0700
From: Ed Phillips <evp-at-pacbell-dot-net>
Reply-To: ed-at-alumni.caltech.edu
To: Tesla List <tesla-at-pupman-dot-com>
Subject: Re: Measure that frequency shift.  (or, at least, change the  subject) :-) (fwd)

"Apparently, with big streamers, we need to tune to a slightly lower
frequency for best transfer.  The arcs seem to consist of a resistance
the megaohm region and a roughly 5 pF of parallel streamer capacitance
is switched in after the output voltage reaches its peak (at least
what this afternoon's model suggests.)  Richard points out that it is
complicated, but on average, the models do predict real behavior fairly
well.  Antonio suggests we should tune the coupling to certain sweet
to help quenching also.  I think this is somewhat important but in
very difficult to control.  And there is the thought that the ring up
occur before breakout and the original frequencies may be correct.  Then
there is the transformer effect.  We are getting to the point of
hairs.... Which is good!"

	I've been watching this thread for a while and thought I'd throw in
some comments.  First of all, I'm sure there is reactive loading due to
the sparks.  Secondly, the shunt load resistance on the secondary is
lowered, lowering the Q.  Working out the equations for a double-tuned
circuit you will find that the effect of more secondary loading is to
increase the required primary capacitance (or inductance, if you varied
it as we do with tapped primaries) is increased with the increased
loading.  I find this effect very noticeable with transmitters coupled
to antennas.  As the coupling is increased more capacitance (in this
particular case) must be added for maximum secondary current.  With
really low Q circuits the effect is quite pronounced.  So, the secondary
resonant frequency IS lowered as the sparks form, due to multiple
effects, and the primary must be tuned to a lower frequency.

	As for the coupling factors creating the "sweet spots", to a first
degree those depend only on the circuit parameters and not on the
secondary loading and should not be hard to achieve.  I've run a number
of simulations using Electronics Workbench (a spice-based simulation
program) and find that proper tuning to achieve one of the "sweet k
factors" can make a significant difference in peak secondary voltage,
provided the gap quenching is good.

For what it's worth,