Re: [femm] current density

[Dave Squires <djsquires@xxxxxxxx>]

Dave,
You are right of course.

But, the problem I have run into is where you have odd shapes
that perhaps are not symmetrical and the areas don't match.

In particular with a toroid type winding, it is nearly impossible
to get the inner winding area to match the outer winding area.
If you try to use area current densities it is a nightmare and
you can't get the two to sum to zero without a lot of pain
and frustration.

So it is much easier to just use point properties and the
same number of points to represent real coil winding turns.
Then the sum is always zero with no headaches.
At least this is what I have found from experience.

Regards,
Dave Squires

David Meeker wrote:

> Dave Squires wrote:
>
> > Trying to use areas with current densities with cross sectional
> > areas is problematic if you don't get the areas for plus and minus
> > to match. The simulator seems to hang if you don't get exact
> > matches for return currents. Point currents are much easier to
> > deal with.
> >
> If you don't define any boundary conditions (in which case, things
> default to dA/dn = 0), the problem can only be solved if the currents in
> the domain sum to exactly zero. Subsequently, if the currents don't add
> up to exactly zero, the solver will hang when boundary conditions are
> defined. The "cure" is to define A=0 on the outer boundary of the
> solution domain (or use one of the "open" boundary condition strategies
> described in Appendix C of the manual). If some boundary conditions are
> defined, the solution will be uniquely defined and the solve won't hang.
>
> Dave Meeker
>
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/