Original poster: Paul Nicholson <paul@xxxxxxxxxxxxxxxxxxx>
Gerry wrote:
> I'm thinking that the fundamental frequencies (the two that
> split further apart when the coupling is increased) also can
> beat with each other ...
Yes, they form the 'beat envelope', for example in
http://www.abelian.demon.co.uk/tssp/tmod.html
http://www.abelian.demon.co.uk/tssp/tfss270501/
http://www.abelian.demon.co.uk/tssp/md110701/
which is essential to the complete transfer of energy from
primary to secondary. This occurs half way through the first
beat, when the two frequency components momentarily cancel out
the voltage and current of the primary, leaving all the stored
energy in the secondary.
Incidentally, in the last of the above links you can see some
evidence of primary overtones appearing just after the bang.
These are inducing secondary base currents peaking at around
10 amps, which leads to very high RF voltages along the secondary.
They've pretty well died out by 5uS after the bang, but reappear
in a short burst with each zero crossing of the primary current.
(Another benefit of low-k here: the beat cycle is long enough
for HF modes to decay before the underlying 1/4 wave voltage
has developed to the full amount).
Cases of racing arcs involving short faint sparks over the whole
surface of the coil would suggest very high frequencies are
involved, perhaps 20 or more times the operating frequency. If this
was the case, a resistor connected between coil top and topload
might act as a good damper. A suitably chosen resistor would take
advantage of the very low impedance of these HF modes as they pass
from coil to topload, without losing much energy at the high
impedance of the operating frequency. The same effect might also
be achieved with the resistor between coil base and ground.
Gerry wrote:
> My understanding is frequency splitting occurs at all levels
> of couplings greater than zero.
Two normal modes appear, but are indistinguishable on an impedance
sweep when the coupling is below about 1/Q. In the frequency
domain the two peaks are merged into a single peak which has a
flatter top than a single resonant peak would have. Above
the transitional coupling, the resonant modes have separated
enough for the flattened top to start to form a dip in the middle.
This is the state of affairs desirable in a bandpass coupling
circuit, eg a dual resonant I.F. transformer.
I have my doubts about D.C.'s response measurements because a
system which failed to display a distinct double humped impedance
response due to very low coupling, would not support a beat
envelope - lack of distinct peaks in the impedance spectrum
indicates that the beat cycle (ie energy transfer time) is longer
than the decay time of the stored energy. We would have to look
at the measurement procedure to work out what is wrong - obviously
the coils work so maybe the measurement setup is introducing some
major Q spoilage. In other words, with coils of reasonable Q,
even with the primary quite a few inches below the secondary,
there should be enough coupling for a distinct double humped
response to be seen on an impedance or transfer sweep.
--
Paul Nicholson
--