From: Paul
Date: Tue, 17 Oct 2000 11:12:24 +0100
Subject: Re: [TSSP] Surprising secondary voltage profiles
Terry wrote: > Computerized digital scope... I pushed the little button :-)) OK, so no accidental square law response. Any DC offset from the probe adding to the RMS value? I guess such would have to be pretty big to give the observed profile. > don't remember tuning much but a few percent is pushing it. Hmm, I think the pull would have been quite big if the probe cap was enough to truncate the line. I wrote: > You could check for this with a simultaneous V probe on the toroid. > The toroid volts should remain constant-ish Terry replied: > I now have one of those 40kV Tektronix freon filled probes. I remembered you also have the top current probe, which would do just as well. All it takes is some means to confirm that the resonance is not disturbed as the travelling probe moves up. The top current would do just as well as the volts. Terry wrote: > I have probed around coils a bit and the concave curve seems > closer to me than the convex. Yes. All I have to go on are some unproven 'extensions' to standard transmission line theory. tsim reports the same profiles as the theory predicts, but it's bound to since its coded from the theory, so it adds no extra proof in itself. My feeling is that the theory extensions are 'reasonable'. (euphemism: I can't think of a better theory :)). Has anyone else looked at voltage profiles? Am I out on a limb with this theory? The trouble is all my understanding of transmission lines is based on lines without longitudinal coupling. I've been tripped up a few times already by making assumptions that are no longer valid in the longitudinal case. I did some math last night to figure out the effect on the profiles of mutual inductance. I was surprised by the results - a symmetric 'weighting' of mutual inductance on a long coil has no effect on the shape except near the ends, simply because a smoothed cosine is still a cosine. It only changes the shape near the ends because the weighting becomes asymmetric. Tentatively: the inductance profile would only flatten the V profile towards linear in the middle of the coil if its weighting was wide enough to 'see' the ends from about 1/3rd the way in from each end. I'll investigate further. A couple more questions Terry, were you driving the coil with CW from the base or was a primary winding involved? Was there a ground plane around the base of the coil? Concerning E-Tesla5, Terry wrote: > but the profile really should be a plain "V" boundary condition. Yes, for the E-field to make sense you have to use the plain V profile on the coil to get the correct field. I think you'd then have to sum the flux as it arrives at the surface of the coil, separately for each grid cell on that surface. Then weight the resulting 'external C per cell' by the V^2 at that cell and sum for the total stored energy in the cap, from which you'd get a properly weighted equivalent cap for Fo calculations. How does E-Tesla5 estimate the effect of internal capacitance? I think you use a gaussian surface which includes the whole coil, so only external flux will be detected. If I 'switch off' Cint in tsim, the Fo goes from around 150kHz to around 162kHz. > Don't get hung up on the convex/concave thing. If > you think it is convex, just "go for it" and continue onward! Yes, unfortunately this is the next 'onward' step, ie validating tsim predictions of amplitude profiles. If these are wrong then the Q and Zin predictions will be messed up too. What's at stake here is the theoretical model of longitudinal coupling, as implemented in tsim and assumed in pn1310 sections 5 onwards. Meanwhile, today the machine's busy running lots of simulations using some 'odd' distortions of M, Cint and Cext, in a desperate attempt to find any way to produce a V concavity in the top half of the coil. Results on this to follow later. Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.