From: Bert Hickman
Date: Sun, 28 Apr 2002 15:15:02 -0500
Subject: Re: [TSSP] Secondary voltage stress factor
Hi Paul, This sounds like an excellent approach, since it provides a way to compare, in one table, a significant variety of coil geometries. Nice job on defining geometric parameters to cover all common secondary configurations! Your proposal also got me to thinking more about nonuniform voltage distributions across the coil. The great mystery still seems to be the root cause of "racing sparks" and the seeming link to overcoupling. Some mid afternoon musings follow... Modern magnet wire insulation has a relatively high dielectric strength. For example for 18 AWG magnet wire, look at the following data (provided by Phelps-Dodge) for double build (i.e., 2 layers of insulation): Insulation: Mils: DC Breakdown (kV): Thermaleze-T (TZT) 2.80 11 Armored Polythermaleze 3.05 11 Imideze (ML) 2.90 12 Formvar 3.00 10 Nyleze 2.90 8.5 With breakdown voltages like these, turn-turn voltage stress may not matter very much for coils built with virgin (unscratched/nicked) magnet wire. On otherwise soundly wound coils, turn-turn shorts seem to appear only after significant physical damage has occurred to the insulation system such as abrasion, a power arc from the toroid (I've seen this occur on magnifiers), or a primary-secondary flashover. Unlike power arcs, "racing sparks" seem to be of comparatively low energy, and seldom seem to do permanent damage. Perhaps as an offshoot of the proposed study, maybe we could then look at the worst-case transient voltage gradients occurring across the secondary. This might be done by looking at the transient analysis during ringup or by stimulation from an out of tune frequency source. We'd want to look for the which regions on a secondary saw maximum voltage stress versus linear distance. Excessive transient voltage stress may stimulate an initial breakdown of the air immediately above the wire/insulation surface (especially if there's an insulation defect or conductive asperity enhancing the local E-field). Once the initial leader starts it can then propagate along the outer surface of the dielectric (wire insulation and any overlying conformal coating) towards the region of lower potential. This mode of spark propagation, called a "creeping leader", can extend considerably longer (a factor of 4-5X!) over the surface of a dielectric than a spark propagating in air under the same voltage stress, making it appear as if we had a much higher voltage differential between the points. Its known that creeping leaders can be initiated at significantly lower voltages than in-air leaders. If the voltage differences between sections of the coil become sufficiently high through excessive coupling, too rapid a ringup(?), mistuning(?), creeping leader discharges may account for the racing sparks. Looking at maximum gradients during transient analysis might shed more light on the conditions that might foster coil breakdown. This type of analysis would really make Terry's Sun Blade100 earn its keep. :^) Best regards, and excellent work! -- Bert -- -- Bert Hickman Stoneridge Engineering Coins Shrunk Electromagnetically! http://www.teslamania.com Paul wrote: > > Hi All, > > Here's an idea for your consideration. > > If a uniform secondary voltage gradient is assumed, ie a constant > volts/turn all along the coil, then the volts/turn is simply Vtop/N > where N is the number of turns. > > In practice, the voltage distribution is non-uniform and therefore > the highest volts/turn on the coil must be something in excess of > the value Vtop/N. The naive value of Vtop/N is thus a lower limit > for voltage stress. We can characterise a particular coil by > indicating what its maximum voltage stress is, in units of Vtop/N, > in other words by specifying a 'voltage stress factor' which is the > ratio of the highest V/turn on the coil, to the value Vtop/N. > > For example, a particular coil with h/d=6 and 1200 turns has a peak > top voltage of 520 kV, and a max gradient of 0.63kV/turn (at 22% > height). Thus the uniform gradient is 520/1200 = 0.43 kV/turn and > we have a 'stress factor' of 0.63/0.43 = 1.46, so that for this > coil, naive estimates of the insulation requirements need to be > uprated by at least this factor. These figures were calculated for > CW steady state resonance, and this provides a lower limit for the > voltage stress, since in a primary-driven coil the primary induction > must be added, along with the contributions of higher modes. > > I'm wondering whether it would be helpful and possible to produce a > table of these voltage stress factors, as a function of the shape of > the secondary. > > Consider an arbitrary secondary coil, in cross-section in > > http://www.abelian.demon.co.uk/tmp/vcoil.gif > > and with no loss of generality we can say that the 'r1' end of the > coil is the grounded end. > > The shape of this coil can be described by two parameters, for > example, let d be the average diameter, ie > > d = r1 + r2 > > then we could choose the two parameters > > A = h/d > > and > > B = (r1 - r2)/d > > to describe the shape of the coil independently of its overall size. > > Then cylindrical secondaries will have B = 0 and A between say 1 and > say 8. Flat secondaries with center-ground will have A = 0 and B > between -1 and 0. Center-hot flat secondaries will have A = 0 and B > between +1 and 0. Cone shaped coils, as per the diagram will have > A between say 1 and 8, and B between 0 and 1. Inverted cones would > have the same range of A but with B between 0 and -1. > > The voltage stress factors could then be tabulated thus, > > B > > -1 -0.8 -0.5 -0.3 0 0.3 0.5 0.8 1 > A 8 > 6 1.46 > 4 > 3 > 2.5 > 2 > 1.5 > 0 1.75 1.0 2.14 > > in which cylindrical coils occupy the center column, flat coils > occupy the bottom row, and the rest of the table covers the various > cone shapes. > > Whether this would work or not depends on whether the voltage > stress factor is mostly independent of the number of turns on the > coil, and of the overall size and position of the coil. > If it turned out that this was the case, ie the stress factor was > mainly a function of the coil's shape, then such a table would be > meaningful, and could be potentially useful to coilers. > > I propose we run a series of coils through the model to test > whether the voltage stress factor can be characterised as a > function of shape only, and if so, to fill in the above table. > > This would require processing of a few hundred coils. > > Terry, would your Sun computer be available for this, by any chance? > The program could be set to run at low priority to avoid slowing > down the qvar results processing. About 600Mbyte of disk space would > be needed, and about 200-300 hours of CPU time. > > Cheers All, > -- > Paul Nicholson, > --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.