TSSP: List Archives

From: "Malcolm Watts"
Date: Fri, 6 Oct 2000 09:51:59 +1300
Subject: Re: [TSSP] Progress report 28th Sep 2000

Paul,
       Thanks for your considered comments. I have a couple to 
add:

On 5 Oct 00, at 12:32, Paul wrote:

> Malcolm Watts wrote:
> 
> > The obvious possibility is msmt error.
> 
> I'm inclined to agree with you that the overall consistency of your
> set of measurements is encouraging. You have already highlighted the
> one doubtful reading. There is always the possibility of systematic
> error but I think that suggestion should be kept decidedly in
> reserve.
> 
> > Therefore, I would be interested to know what an approach other
> > than Medhurst suggests and how Medhurst arrives at his result.
> 
> I have doubts about the applicability of Medhurst's Q results to
> tesla resonators. Medhurst made his AC resistance measurements at
> frequencies well below the lowest self resonant frequency, by
> swamping Cself with large external capacitances. This forces the
> coil's current distribution to be almost uniform. Thus we might
> expect an additional factor to occur when considering coils operating
> near their bare self resonant frequency. This effect would be more
> pronounced in the absence of a topload, since then the current
> non-uniformity is highest. This discrepancy would not be noticed in
> the conventional applications for Medhurst AC resistance estimates,
> since most HF inductors work with circuit capacitances significantly
> higher than the self capacitance and therefore have uniform current.

I suspect that much previous research of this type was done 
with radio applications in mind since tuning small inductors 
with capacitors is the norm for those applications. Bare 
resonators are out for such apps since the influence of the 
confinements is much too great to be useful.
 
> As a result I'm looking to find a way to calculate the effective
> AC resistance independently of Medhurst.
> 
> On primary ringdown measurements,
> 
> > The *relative* times and levels give all the information needed
> > to calculate differences in proportional losses on a per-cycle
> > basis.
> 
> Your observed scale invariance of the ringdown (with respect
> presumably to bang energy and frequency) suggests to me that
> if the gap resistance is significant, then it must also be more
> or less constant with current and frequency. Alternatively, if the
> gap resistance is not constant (which I might reasonably expect to
> be the case) then it must be relatively small.

It is very low in general. It does appear to have a strong 
dependence on current and this is confirmed by equations which 
others have derived to descibe it. For such reasons, I am 
loathe to take figures like "200k Ohms" for output discharges 
to air too seriously. Capacitance/unit length I can pretty 
much live with.
 
> > General rules can be derived such as constant (or nearly so)
> > voltage drop per gap etc.
> 
> Presumably then the (nearly constant) grap volts drop is small.
> Would that indicate that the dominant factor in primary circuit
> loss is not the gap?

Intuitively, it seems reasonable that it is the major loss - 
where the heating occurs is the main indicator. 
Experimentally, varying the surge impedance of the circuit for 
a fixed firing voltage changes the decrementing slope of the 
circuit somewhat proportionally and way out of step with the 
increase in changes in copper resistance.  

> > I think the major thing to come out of my secondary
> > research was that for non-output discharge applications,
> > secondary characteristics really matter but if one is
> > designing for long sparks, those considerations, while
> > important to a certain extent are relegated to the stalls,
> > because the emphasis now shifts to dealing with the
> > characteristics of ionized air and how best to make it do what
> > is required.
> 
> Agreed. Terry has described satisfactory performance from his
> coil despite a low measured Q. Any predictions of voltage and
> current from our software model will only apply to small signal
> operation. We might hope that one day we will be able to predict
> Q, but will that help in optimising real world coils? Perhaps
> not.

I think it will matter, but it will be only a part of the 
picture. It can be shown that a poor secondary will not 
produce the goods at all well. Meanwhile, I am interested in 
focusing on the behaviour of the resonator alone since its 
modelling is of great interest to me. In particular, I am 
curious to know just how closely the Corum's model and its 
consequent predictions of enormous voltages approaches 
reality. I have seen some indications in low level tests that 
left me flabbergasted and suspecting they might not be too far 
wide of the mark. But is it realistically modelled by a 
uniform transmission line? And just how close can one get to 
actually measuring the real voltages produced by a non-
toploaded resonator? The sparks can live for another day.

Regards,
malcolm


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