From: Paul
Date: Thu, 07 Mar 2002 19:34:53 +0000
Subject: [TSSP] Progress on capacitance program
Hi All, At last I can report a bit of progress on the capacitance program. This is the program which computes the physical cap of the resonator - the first step in modeling the system. The program tcap's main defect is that it doesn't take account of varying material dielectrics - it just assumes that all the conductors are immersed in a uniform dielectric. As a result, we tend to under- estimate the capacitance of the system, most notably on small coils and small h/d coils, where the coil former thickness is significant. This is unfortunate, as in order to demonstrate some of the more interesting effects of capacitance, we need to be able to explore the small coil and small h/d domains more accurately. Thus the goal is to alter tcap to take proper account of dielectric materials, eg coil formers, bases and pedestals, insulating walls, coil supports, etc. The whole procedure is described beautifully in a paper http://faculty.smu.edu/tausch/Papers/mtt1.ps.gz and basically I just need to add in the stuff from equ (5). In order to make room in the program, I've speeded up the whole thing quite a bit, and at the same time made it more accurate! For example, a test case involves calculating the self capacitance of a unit sphere, which is known to be 111.2650pF. Tcap takes 9 seconds to deliver an answer of 111.2672pF. The accuracy falls to around 1% on unfavourable geometries, but the program will now compute the self and mutual capacitances for any shape of object that has cylindrical symmetry. In other words, any oddball shape that can be described to acmi can be also be fed into tcap. The calculation of the self-C of isolated objects like toroids, cones, discs, whatever, is now the most accurate that we've had, and also the quickest, thanks to a few computational tricks. It takes around 6 minutes to compute the capacitance matrix for the whole resonator. Total Cdc of objects like secondaries and toroids takes just a few seconds to 1%. The upshot is that there is now plenty of spare capacity in the program to allow the ECF code to go in. I just need to find the time and determination to do it. There are quite a few other tricks that we can draw on, some of which are described in http://faculty.smu.edu/tausch/Papers/advCM.ps.gz These two papers have been a considerable source of inspiration lately. I'm very grateful to their authors for choosing to use the same kind of integral operators that I've found useful for describing the resonator in pn1401. -- Paul Nicholson, --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.