"rough draft" of updated version

["David Meeker" <dcm3c@xxxxxxx>]

All-

I've posted sort of a "rough draft" of a new femm release in the
Files section of the femm yahoo groups site at:
http://groups.yahoo.com/group/femm/files/
If you have trouble getting into the files section, the same files
are also mirrored at:
http://members.aol.com/_ht_a/gmagnetics/roughdraft
Both the executable and source have been posted.

If you download this release and you see it do anything "weird", let 
me know.

This version isn't monumentally different from the version currently 
available on the femm website, but it does have a number of minor 
fixes and additions:

Robin Cornelius' lua extensions have been included into this latest 
version. I've also included a file called "coilgun.zip" that shows a 
relatively simple example of the kind of thing that you can do with 
the lua scripting. This particular case considers the force on a 
small steel ball, stepping the position of the ball in small 
increments, evaluating the force, and writing the force to a file. 
Since the "gun" is just an air-cored coil, I evaluate force in this 
example by computing the Lorentz force on the coil and infer that the 
force on the ball should be of equal magnitude and opposite direction.

Both femme and femmview now have a "preferences" selection off of the 
"Edit" selection of the main menu. These preferences allow the user 
to set a number of default behaviors, colors, etc. This was a 
suprisingly difficult addition to make, because I kept running into 
instances in which the preferences were not getting applied in a 
consistent way. I think I've rooted most of these instances out, but 
if you find any, let me know.

The font sizes have been increased, both in the status bar and in the 
point properties window of femmview.

The documentation has been (more or less) updated to reflect the 
features in the current version. Probably some more editing is 
required, however....

I've included a pdf version of Ian Stokes-Rees' "Introduction to FEA 
with FEMM" tutorial in the standard distribution. Since the original 
version of this tutorial had some stuff that was specific to the 2.1a 
version of the program, I've updated it a bit to conform to the 
present release.

I made a mild change in the way that the program interpolates B-H 
curves. Now, the program fits a cubic spline to the data points and 
tests to see if the resulting fit is single-valued. If not, the B-H 
curve is smoothed using a 3-point moving average filter, and the 
process is repeated until a single-valued curve is obtained. In 
contrast, the previous version smoothed every B-H curve initially and 
then fit cubic spline to the smoothed data. If the B-H points
weren't relatively carefully chosen, the result could be a B-H curve 
that is not single valued. A B-H curve that isn't single-valued can 
cause convergence problems.

Bug fixes were made to both the fast point location algorithm as well 
as the B-field smoothing algorithm to accommodate some sort of
obscure but pathological geometries.

Dave.