From: Paul
Date: Thu, 03 May 2001 19:09:42 +0100
Subject: Re: [TSSP] Time domain modeling
Hi All, Gosh, this time domain modeling stuff does open up a lot of avenues of investigation. I'm just sitting down to another evening with the code, but even when I'm at work, thoughts keep coming back to the problems and possibilities. For example, I've always assumed that breakdown of the sec is likely if k is taken too high, due to the induced EMF from the primary exceeding the turn-turn breakdown of the secondary. Havn't got a full picture yet, but other possibilities arise. The familiar pri-sec beat is the sum of two normal modes, more or less equally excited by the initial bang. For small k, the modes are close and very similar sec I distribution I(x) for each, but of opposite sign, relative to Ipri. So with equal amounts of each, the two sec I(x) cancel out to around zero, and the two Ipri add up to the amount required for the bang. As k increases the modes move apart and also change shape. Cancellation is now no longer perfect, and thus higher modes need to go into the pot in order to recover the match to the initial conditions. Energy in the higher modes has a roughly proportionally higher voltage gradient on the secondary, eg 3/4 wave has almost 3 times the volts per metre than 1/4 wave, for the same terminal voltage. Hopefully we'll be able to tell if this mechanism contributes to difficulties of using high k. We'll also soon be able to see if the conical primary makes life easier in this respect - we might expect its 'smoother' application of EMF to couple less to the higher modes. Sorry for the handwaving, just a few thoughts. The math and computations should spell this out more rigorously. I wrote: > > >The reason I ask is that, depending on just how > > long the arc > > >discharge path can be considered to be grounding > > the top of the > > >coil, it looks like quite a substantial amount of > > energy can be > > >transfered into one of the higher modes, to be > > trapped there. Boris wrote: > Correct.The freq. of power discharge depends on a > isotropic C, physical size of isotropic capacity(not > just on C but there is difference wether toroid > /sphere etc.used),density of coil turns and I suspect > to some extent on geometry of outter path to ground > coil discharges into. Yes, but I'm not too concerned with the high frequency components of the arc itself - I wouldn't know where to start to calculate these. Just that, while the arc is present, the solenoid is effectively grounded at both ends, with an appropriate set of modes for this boundary condition, and the longer this condition persists, I think the greater is the energy is transfered from the two main modes into higher order resonances. When the discharge ends and the top short is released, the original grounded base modes are restored and the energy remaining in the coil is once again redistributed, this time amongst the original mode set. Can't say for sure yet, but I don't think the energy winds up back in the original main two modes, but is more spread out than before the discharge. > Terry's former experimental coil has shown power arc > impedance about 2kohm for 3.5" arc. > System I tested in Croatia by estimation of heat > consumption of watter resistor,has shown an average > impedance of more than 8000 ohms for 17" arc. > Dale Hall's single pulse operated coil with 24"*8" top > toroid gave about 500-1000 ohms for 20" arc. Quite some variation there, going to take some effort to get a general picture. I think that the load impedance will end up being the dominant limiting factor in ultimate accuracy, more so than, say, Q factor, and therefore we are justified in reducing the accuracy of the modeling itself. So I've been using around 200 steps rather than around 500 steps as used up to now by tsim. Cheers All, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.