Re: [femm] DXF Import Tricks & Tips (and questions) from a FEMM Newbie

[David Meeker <dmeeker@xxxxxxxxxxxxxxxxx>]

"Jay D. Greener" wrote:

Hey Everybody,

I just recently started with FEMM and E-mag analysis in general; please forgive any
ignorance.  I?ve encountered a few problems with importing a .dxf and would like to
share them for the new guys and ask for any insight the more experienced users
might be able to provide.

Problem Background:

I?m using SolidWorks2000 to generate an axisymmetric dxf model of a pot core
solenoid.

Findings:

   1.The .DXF seems to import in a random position w.r.t. the FEMM coordinate
     system ? even drawing the model from 0,0 in Solidworks seems to result in
     (random?) placement in FEMM. The model crashes upon analysis (perhaps
     because of the undefined region of space between the model axis of symmetry
     and FEMM?s r=0 axis?).
   2.The model scale may or may not import correctly, on occasion the import
     scale is 1:1, I?ve also had models import at 2x.
   3.Hidden dis-jointed lines in the dxf model will result in unexpected results:
     Occasionally the model won?t mesh, I?ve found that erasing all lines (in FEMM),
     then re-drawing between the points fixes the problem.
   4.Be careful about ?point imports?.  In one model two points were very, very close
     together, appearing as 1, it allowed a small connection between two regions
     that I thought were separated ? one defined as air, the other defined as iron ?
     the model solved as if the total region were iron!  It is (apparently) possible to
     define 1 region with two different material types.(?)

Questions:

   1.The FEMM manual specifically say?s ?AutoCAD? DXF? is SolidWorks DXF
     different enough that my problems may be solved by using AutoCAD?
   2.Because I can ?move?  the model in FEMM, the import location isn?t a big deal,
     but the only way I can find the location of a point (to define ?translation
     coordinates?) is to ?hover? the mouse pointer over the point.  I?m worried that I?m
     getting some erroneous results by having a ?sliver? of undefined space between
     my model axis of symmetry and FEMM?s r=0.  (I?ve seen flux densities of an
     erratic nature down the model centerline, wonder if it?s because it?s not
     perfectly aligned at FEMM z=0)
   3.I?ve attached 3 .fem files.  One contains an example pot-core solenoid model
     (without an air region outside of the magnetic circuit), an example problem from
     the Ansys guide (thanks Si Hang).  One of my pot core models without a
     region of air defined around the magnetic circuit and the same model with the
     region of air.  Why do I have to add the region of air to get the flux lines to
     close in my model, but not in the model provided by Si Hang?

1) Femm could still use a little work on the dxf import filter, but a lot of problems stem from the fact that dxf is really just a drawing format rather than a finite element format.  Problems like lines not quite meeting, points almost being coincident, etc, aren't problems for a drawing format, but they are a headache if you are trying to specify a consistent geometry.  There's a really good explanation of problems and possible solutions w.r.t. dxf import on the IES website at http://www.integratedsoft.com/demos/2DDXFImport.htm. Although this site is specifically about dxf import issues with IES Magneto, all of the wisdom applies to doing dxf imports in femm as well.

2) Do a right button double-click when the mouse pointer is close to the point you are interested in.  The program will then pop up a message box with the exact coordinates, applied property, and group membership of the nearest point.  The right button double-click works in other modes to enquire about the location and properties of the nearest object without actually opening it.  This is an "undocumented feature" that I really should get around to documenting...  Anyhow, if you are worried about trivial displacements from z=0 due to import inaccuracies, you can explicitly define an A=0 boundary condition and apply it to your centerline.  This ought to fix things.

3) Hang applied an A=0 boundary condition to all the outer edges of his model.  This makes it so that no flux can cross out of your problem domain.  In contrast, if you don't apply any boundary condition (like in no-air your model), that's as if you surrounded the actuator with an infinitely permeable iron.  The flux would rather travel in that lower reluctance path outside the solution domain rather than through your actuator.  When you add back the air, it puts a high reluctance barrier between your actuator and the "infinitely permeable iron," so the flux again prefers to stay inside your actuator.

Dave.
 

begin:vcard 
n:Meeker;David
tel;fax:781-890-3489
tel;work:781-684-4070
x-mozilla-html:TRUE
url:http://members.aol.com/dcm3c
org:Foster-Miller, Inc.;Electrical and Electronic Systems Group
version:2.1
email;internet:dmeeker@xxxxxxxxxxxxxxxxx
title:Senior Engineer
adr;quoted-printable:;;350 Second Avenue=0D=0A;Waltham;MA;02451-1196;USA
fn:David Meeker, Ph.D.
end:vcard