From: Bert Hickman
Date: Tue, 28 May 2002 07:44:20 -0500
Subject: Re: [TSSP] Topload breakout potentials
Hi Malcolm, I think this would be a much more achievable approach! I was thinking along the same lines over the weekend. Envision a graded column consisting of identical RC sections sandwiched between grading rings or toroids. Each section would consist of a parallel combination of series connected capacitors and resistors between grading toroids/rings whose ROC (at least at the upper sections) was large enough to inhibit corona in the space between it and the inside surface of the coilform. This should result in a stepwise linear grading of voltage along the entire length of the column. Placing the divider chain in the center of the secondary should also reduce the need for large ROC grading terminals, since we only need to prevent corona formation due to local e-field stresses between the toroid and nearby coil region which would also be at a (hopefully) similar elevated potential. By using identical parallel RC values in each section (combined with a larger C/smaller R at the base of the divider chain with an adjustable "tweaking" adjustment), we should get a compensated divider capable of measuring hundreds of kilovolts with reasonable DC and AC accuracy. By using relatively high valued resistors in the chain (>10's or hundred of megohms), prebreakout resistive loading can be minimized, and capacitive loading of the divider can be compensated via system retuning. Although the linear voltage profile along the column of the divider won't precisely match that of the coil, it may be satisfactory. Capacitors in the divider chain would need to have low a voltage coefficient, ruling out most HV ceramic caps. And a chains of vacuum caps might create undesirable X-radiation. Pressurized gas capacitors would work, but these are not commonly available... it might be possible to make them however. A series string of identical high voltage polypropylene caps would work nicely. The resistors should ideally be high voltage, low voltage coefficient type, but a series chain of 1 watt carbon resistors might also do the trick. Wirewound resistors, while having other desirable characteristics, will not provide the high resistance desired to minimize resistive loading of the topload. All of the components in a section should be contained well within the shielded portion between end toroids to prevent corona. The spacing between outermost portion of the toroid and the inner wall of the coilform must be sufficient to prevent corona under all loading conditions of the secondary. Further, the distance must also be sufficient to keep electrostatic coupling between the secondary and sections of the voltage divider to a low enough value so as not to unduly perturb the measurement. It would be nice if we could avoid the need to fill the coilform with oil. This implies that we may need a comparatively large secondary diameter. A small chain in a relatively short/stout coil may provide us with a proof of concept. Will bend some numbers this evening... Best regards, -- Bert -- -- Bert Hickman Stoneridge Engineering Coins Shrunk Electromagnetically! http://www.teslamania.com Malcolm Watts wrote: > > Hi all, > I have been thinking some more about the possible forms a > probe might take. My initial idea was to effectively pipe the HV out > through the bottom of the coil. That approach might still be on but > the issue of loading is probably going to cause more than desirable > deviations in coil characteristics. So I have another which might > eliminate the need for a central shield. > > Suppose the structure is graded such that it has a voltage > profile matching that of the coil. Will that do it? Probably only > only at Fr. If several probes having different gradings were made, > comparisons could be done. Equalization along the probe's length > might well be necessary but it would get us to the point of having to > deal only with low voltages/currents at the base where it exits the > coil form. And the shunt impedance could be made arbitrarily high. > > There is the nasty impication that we would measure what we expect to > measure but different gradings might get around that problem if all > shunt impedances are equal. > > Any thoughts from anyone? > > Malcolm
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.