From: Paul
Date: Mon, 27 May 2002 10:17:14 +0100
Subject: Re: [TSSP] Topload breakout potentials
Malcolm wrote: > In effect, a thin coaxial probe with good insulation between the > centre wire and shield... Yes, > and between the shield and the windings. Yes, could make it impossible to use with narrow coils. The outer ROC of the shield must not induce breakdown to the secondary. > There is of course the capacitance between the central > wire and shield, Not a problem...it's constant and lumps in with the bottom half of the divider. > not to mention distributed inductance. which ought to have negligible effect if we're drawing only a small current into the divider. I'm assuming that as a first off we're not going for HF response. > The disturbances would be reduced as the coil diamater got larger > relative to the probe's diameter. Yup.. we're entitled to use a non-standard h/d ratio because we're studying the topload/streamer interface. > the insulating material must have a low dielectric > constant and be as low a loss as possible. I don't think that matters for this. > The probe can of course be both tested and modelled at operating > frequency independently of the coil being measured to quantify its > characteristics so these may be factored in to the actual coil > measurements. Yes, I'm assuming we'll calibrate the whole system at low power so that we can later compensate both for the RLC characteristics of the probe, and for the loading of the probe on the topload. > Sorry for thinking out loud - I Keep going! I'm picturing this coaxial probe column as the lower half of a C divider with the top half obtained from a capacitive coupling of the coaxial inner to the topload. I don't think it matters a great deal what the various C values are, the important thing is that no part of the probe breaks down before the topload breaks out. Maybe that will take an oil-filled secondary, I don't know. We can afford to be quite unconventional with the coil if needs be, in order to accomodate the probe. If a probe can be constructed inside the coil so that it doesn't break down, then it will measure the desired topvolts and it will be independent of the space charge and streamer effects. The probe shield and the coil form a coaxial space and the highest voltage gradient is 2 * V/(100 d * Ln(D/d)) ...volts/cm where V is the coil voltage (say = topvolts), d is the central core diameter, and D is the coil diameter. For a 600kV coil, 8.5" diam with a probe shield 1/10th the coil diameter, I make it 240kV/cm! The highest voltage can be withstood when D/d = 2.718, so for the 8.5" coil, the gradient would be 150kV/cm. I'll have to model the effect of such a large core on the coil but it sure looks as if air will not do as the dielectric between the probe outer shield and the inside surface of the coil. Even for my big coil, at 60cm diameter, the gradient is around 55kV/cm on the surface of the core, so going to a very wide coil doesn't solve the problem. This doesn't necessarily mean that the entire coil needs to be filled with oil, perhaps just an oil-filled sleeve over the outer shield of the probe tube. Terry wrote: > Would it be of any help to take voltage measurements at a point > say 20% up the length of the coil? After some thought I've come to the tentative conclusion that this would work in principle, but would be very hard to do in practice. Given a base current waveform, and an initial guess of the topload C, the waveform from the coil tap could be calculated and compared with the measured signal..the top C guess would then be revised and the computation repeated, and so on, iteratively until a satisfactory match is obtained. That's just for one instant in time, so repeat for 0.1uS steps over a 50uS event... -- Paul Nicholson, --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.