From: Paul
Date: Mon, 13 Nov 2000 20:13:20 +0000
Subject: [TSSP] Software progress report
Hi All, You may recall that I've been getting on with a rewrite of the laplace solver tlap which is responsible for producing the capacitance matrices required by the secondary simulator. Well I completed this - a really neat multigrid solver, much nicer than the existing program. Unfortunately it still suffers from the same problems as tlap - that is, the long wave components of the field solutions are of poor accuracy and the cause in both cases is the same - the difficulty of finding coarse grid representations of the electrodes and boundaries, combined with errors introduced by the boundary contraction schemes. These seem to be pretty fundamental to the method and are not just coding issues. As a result, I've abandoned the relaxation solution altogether. Since we don't need the field, just the capacitance distributions, I can also abandon attempts to solve the laplace equation. Instead, I've opted for a boundary element method which gives the charge distributions directly without having to compute the field first. This week has been spent coding it up and now it's under test. Despite being harder to code it has some major advantages: It's a lot faster - the cap matrix is produced to 5% accuracy in around 40 mins and a couple of hours gives enough accuracy for the simulator (these times are for a single P450, not the cluster machine). Also there are no accuracy problems at high coil elevations, boundaries at infinity are dealt with easily in this method, and the method can be tuned to concentrate extra effort on those all important edge effects. The program no longer uses the X display. Right now it's approaching the accuracy of tlap for the reference coils without toroids and I'll code for the toroid cases just as soon as I figure out some horrid surface integrals. A preview of the code is in http://www.abelian.demon.co.uk/tssp/tcap.c.html Hopefully tcap will be able to replace tlap in a couple of weeks. Once that's done there's nothing stopping us from running the necessary simulations required to determine the secondary coil dimensions for optimum CW performance. Hope my big secondary survives! Won't be much progress this week - its that time of the year for setting up antennas and computers ready for the leonids meteor shower. (last years readings, http://www.abelian.demon.co.uk/leo99/) Just hope I can stand the antennas up in the mud out there! Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.