[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Back on the list after a few years



Original poster: "Paul Nicholson by way of Terry Fritz <twftesla-at-qwest-dot-net>" <paul-at-abelian.demon.co.uk>

Rob wrote:

> And you can get pretty close to distributed models by modelling
> them as a large number of lumped elements.

Funny, I've never seen a distributed model that doesn't involve
a large number of lumped elements.  Except by analogue methods we
don't have the option to produce a true distributed model.  In the
digital computer all we can ever do is devise ever more effective
methods of discretisation in our attempts to extract numerical
solutions from the continuum equations of EM.

> Still lumped models work pretty well.

Yup.  Consider the lumped LC secondary model as a distributed model
with just a single element.  You can make it exact at a single
frequency for any two of the three properties { base current, top
voltage, stored energy }.  Add more lumps for more properties at
more frequencies.

> I am very interested in modeling secondaries and/or magnifiers. 

Well you've come to the right place then.  The frontier involves
tackling the problem of modeling the losses of the coil, which
current models such as tssp fail to do.  I see you're working in
Python, so you obviously have the necessary patience.  If you drop
down to the quasi-static domain, tssp can provide C code to construct
matrices for all the operators defined in 

 http://www.abelian.demon.co.uk/tssp/pn1401.html

and these you could convert to python.  But the trouble is, by doing
so, you loose the physics which determines many of the losses. 
In order to proceed, we need to find ways to model at least portions
of the conductor as something more than a filament.  This can be
done fairly easily for the case where the wire radius is small
compared to the skin depth, but this regime does not apply too well
to Tesla secondaries.  Another frontier, if you're adventurous enough
to move a little away from the secondary, is to try to model the
interaction of the coil's fields with the environment, particularly
the ground beneath the coil.  At present we are unable to design a
coil to a given base impedance, or predict the base impedance of a
given coil.  Some fairly tough work on the subject of coil losses
and environment losses would be necessary to obtain quantitative 
models.

> I've read some of the Corum and Corum stuff, and I think
> that they may be right. 

Couldn't have been any of the Corum stuff that I've read then. Not
that awful 'Class Notes' thing, surely?
--
Paul Nicholson
--