From: Paul
Date: Sun, 19 May 2002 20:10:04 +0100
Subject: Re: [TSSP] Racing arcs
Hi All, I'm begining to wonder if this racing arc thing isn't simply a case of the secondary breaking out before or together with the topload. Effectively the ROC of the sec may be as small or even smaller than that of the topload. We see from the gradient animations that the surface fields could be getting up into the 20kV/cm range, and are not that far behind the surface fields of the topload. If you look at gradients of the k=0.2 in-tune model http://www.abelian.demon.co.uk/tssp/pn040502/tfsm1-h1.grad.gif in which the primary is placed quite a way up the secondary in order to achieve the k, the presence of the primary draws a whole bunch of charge to the surface of the coil there, with the result a very high surface field just level with the primary. Little wonder that such configurations are not used. With the primary low down and out of the way, the surface field still wants to go very high, for example in Marc's coil, http://www.abelian.demon.co.uk/tssp/cmod/mm3p.grad.gif the gradient would like to rise to over 20kV/cm. Remember though that the coil is breaking out, so it is not reaching in this case the 960kV predicted for it because that would give a topload surface field of 50kV/cm. If breakout limits that in reality to 26kV/cm, the topload will be limited to 960 kV * 26/50 = 500 kV. What of the effect of this topload breakout clamping on the coil surface field - is that clamped by the same proportion? If so then the 20kV/cm or so predicted surface field is held down to the 10kV/cm mark, hence no breakout from the secondary. I can feel another ratio coming along. How about defining a relative breakdown threshold for the secondary, valid for a given topload. If the topload is clamped by streamers to some voltage that gives 26kV/cm at its surface, then somewhere on the secondary will be the point that has the greatest radial (or perhaps vertical) surface field. Suppose we take the breakdown ratio of the secondary to be that highest surface field divided by 26kV/cm. This is the same as defining highest surface field on the secondary -------------------------------------- = secondary breakdown ratio highest surface field on the topload Thus if Malcolm's coil where driven up to 26kV/cm on the toroid, the secondary would be peaking at around 12.5kV/cm and the ratio would be 12.5/26 = 0.48 Similarly, for all the coils, we have very roughly ba0 ratio = 1.24 racing arcs every time mw1 ratio = 0.48 none jftt42b ratio = 0.58 none jftt42a ratio = 0.56 very close to mm3p ratio = 1.30 never Now the problem here is that Bart's and Marc's coils seem to exhibit quite a high field strength off the top winding of the secondary, much higher than the rest of the coil, so their breakout ratios come out based on this figure. (I might have guessed Marc's topload height incorrectly but Bart's is right). The top winding on Bart's coil should be breaking out ahead of the topload. However for both these two coils the more typical surface field on the rest of the coil is quite a bit less. To obtain a ratio more descriptive of the coil as a whole, we could use the average surface field, or perhaps the peak field ignoring the two end 10%'s of the coil. This latter approach gives ba0 ratio = 0.65 racing arcs every time mm3p ratio = 0.54 never The above figures are based on radial gradients. The same arithmetic could be done for the vertical gradients, or for some combination of the two that was felt to represent the total field gradient. The field around the secondary falls away quicker than the field around the topload, so that breakout from the secondary would tend to be confined close to the surface rather than forming long streamers. Note that altering the k factor alone would not affect the breakdown ratio because the peak voltages achieved would be around the same, only the timing would be different. But the act of raising the secondary to lower the k would have the beneficial side effect of reducing the secondary surface field and therefore reducing the secondary breakdown ratio. John wrote: > Magnifier drivers show racing sparks too I think, and others have said this, so clearly a close proximity primary is not essential to promote racing arcs. I've often seen magnifier tertiarys set quite high off the ground. I'd guess that would improve (reduce) the tertiary's breakdown ratio, compared to the same coil fixed at a lower height above ground. > In Richard Hull's magnifier, he used a 1 foot long resonator, which > gave an 11 foot streamer. To eliminate racing sparks on the > resonator, he used 6" x 20" toroids at each end of the resonator. Yes. I do hope that Richard's detailed notes, measurements, etc are available somewhere. I think this configuration would reduce the overall radial field of the coil quite a bit, so a very low breakdown ratio, probably limited by the *vertical* breakdown ratio rather than the radial limit that we see with the 5 cmod coils. I'm looking forward to modeling that one! > In general for two coil systems, the spark length tends to increase > as the coupling is increased. This suggests that more energy is > reaching the secondary. This makes sense since the spark gap has > less time to lose energy. This greater energy may directly be > causing the racing sparks, by over-powering the winding in some > fashion. Yes, all the voltages would be reduced by some factor depending on the overall Q. If the Q were not high for some reason then the peak voltages would be reduced significantly as k was lowered. But I think you'd have to have quite a low Q and/or fairly low k to begin with. Could it also be possible that altering k alters way the streamers develop and therefore alters the effectiveness of the streamer's clamping of the topvolts? As k increases, the beat envelope contracts so that the middle cycle topvolts peaks much higher than the others. Thus, per beat, the topload only really gets one or two half cycles in which to form streamers, and most of the energy and charge is delivered to the streamers in this middle cycle. Perhaps the short duration of this peak allows the secondary volts to overshoot the value commensurate with the topvolts clamping. We shan't be able to model that without knowing a little more about how the streamers actually load the resonator. > Lowering a toroid tends to reduce the spark length, and also tends > to stop racing sparks. > ...But could some shielding effect be occuring? Yes, we might say that the lowered topload is shielding the coil better and reducing that breakdown ratio. Certainly the shielding occurs - we can calculate that very accurately. > Another question that occurs to me is; do racing sparks begin > with a turn-to-turn breakdown, or does the spark jump over a > number of windings? Well I think the breakdown I've been talking about here is described by your comment about magnifiers: > often a great amount of corona and streamers that creap along the > insulating sleeves because I've been concentrating on the radial gradients. But then > Often, they appear to jump over and skip about > an inch of winding or so, but maybe it's an optical illusion. If the > sparks are jumping over windings, this would suggest an uneven > voltage gradient on the secondary, due to higher frequencies in > the winding. This may be a different flavour of racing arc, a vertical gradient thing, and high frequency ringing and transients is my favourite candidate. There would I think be a characteristic 'skip length' to HF-induced racing arcs, associated with 'favourable' range of frequencies. Too high a frequency and there is too little energy available. Too low and the wave peaks and troughs are separated too far apart along the coil to break down. These HF components can have three possible sources: a) Present from the start of the bang, part of the characteristic 'tone' of the coil, these are derived by requiring that the spectrum of normal modes of the coil reproduce the initial conditions of the resonator at the firing point. These definately exist and we can model them quite accurately, eg as in http://www.abelian.demon.co.uk/tssp/md110701 They increase with k, but are not particularly large in the examples modeled so far. They make up the fine detailed rippling that you see on both the base current traces and the gradient animations. b) Transients injected into the secondary by suddenly changing boundary conditions aroung the topload. Eg sudden streamer formation, discharge to earth, or sudden end to streamer loading, etc. This forces the energy of the coil, which to begin with is fairly well concentrated into the two 'beat' modes, to redistribute itself to match the revised boundary conditions. This 'scattering' process is kind of like a thermodynamic process, and would eventually spread all the coil's energy evenly across the full spectrum of modes - it would look like a bathtub of water after you've had a good slosh around in it. Whether this process gets very far before the envelope decays is something I'll be looking for in the scope traces. Lots of small gradual changes would have a less drastic effect than sudden changes - in the latter case the sudden scattering of energy into HF modes appears as an alternative and equivalent way to describe the transient brought on by a step loss of charge from the topload. > Another thing I was thinking is that maybe the racing sparks begin > with a arc-short between two adjacent windings, and this short > causes a pulse or ringing along the coil, which then propagates > along the windings, causing more racing sparks, or extending the > racing sparks. Does such an idea seem plausible? Indeed, and that's a variation on the theme of (b). And finally, c) Transients induced at the firing point by unwanted resonances of the primary and its circuitry, for example the bit of fluff at the start of Marco's base current in http://www.abelian.demon.co.uk/tssp/md110701 I estimated was putting some 80kV across the bottom 6 cm of the secondary. These kind of things, especially if they involve the pri-sec mutual capacitance, would be very sensitive to relative position of primary and secondary, and could well be localised to that region. If they were of such a frequency that it could excite one of the secondary modes, then high gradients could appear all along the coil (with racing arcs showing the appropriate skip distance). Just going back to the type (b) transients, those wouldn't tend to show a well defined 'skip' because they contain a whole range of modes forming themselves into a single 'step' transient moving up and down the coil. The type (c), if they affect the whole coil, would tend to emphasise the pattern of nodes appropriate to the particular excited mode. One way or another, we're accumulating quite a few reasonable candidate explanations here - for lots of different types of racing arcs. With all these species of racing arcs, it's getting more like botany than coiling... -- Paul Nicholson, --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.