[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Duplicated points
David,
I just became aware of a problem with FEMM.
If points are used for "wires" as point currents
and the copy or mirror functions are used
duplicate points can be created and give
erroneous results.
Placing one point on top of another is not
allowed when doing it manually. You can't
have one on top of the other that way.
But if you do a mirror or copy function and
there are already some points at the copy
region then you can get duplicate ones that
are invisible. If these are set to have
current properties then you can get unexpected
extra amp-turns and get misled by the results.
Please check into this. Here is the report I
got about this problem.
Dave Squires
--------
Hi Dave,
There's a bug in the FEM file you sent. It's got overlaid nodes... guess
Dave M. still hasn't got rid of this flaw. It's easy to make the mistake.
Try highlighting (right clicking) some of the nodes around the exciter
coil at the center of the PFT in your file. You will notice, that some
of the nodes will turn red (selecting), but others WON'T on a first
click. When you try to right click some of the nodes, it's as if the
computer "isn't listening" or doesn't notice.
Sometimes, if you look carefully, the "selected" node will flash red for
a split-second; then turn black. Most of the time you won't see that
though, when there is this problem.
As an experiment, when that happens (and you right click, but no node
turns red) press Delete. (You have actually deleted a node now!) Then
right click again without moving the mouse.
Now "the" node will probably turn red!
This is because there were TWO nodes in the same location, overlaid;
FEMM selects the "bottom" one first, so even though it's selected by the
right click you can't see it, and the "top" node overlays it, unselected
and black.
When nodes are overlaid and both carrying current, you get 2x the mmf
(or more, say if you have 3 or 4 nodes at the same point) - without
usually realizing it. It skews the math (and may also be the reason why
Stan is wary of using these; he may have unwittingly ran into the same
FEMM bug!)
This happens when you use copy, mirror and rotate functions to multiply
nodes, but "copy" more nodes onto places where there are already nodes.
FEMM won't let you manually overlay two nodes in the same place with the
mouse, but the Mirror function, for example will do it. Sometimes. On
Thursday.
My advice is to treat current carrying nodes differently than structural
nodes. It pays before simulation (if you are in doubt) to go around
right clicking current nodes that could be overlaid - watching for nodes
that won't turn red when you click them.
The best approach is to envision where your Copy or Mirror instructions
are going to create new nodes - and make sure there are NO pre-existing
nodes in any of those places.
I have had former analyses totally spoiled by only ONE of these overlaid
nodes, hidden in the file; I had to keep deleting various nodes (and
undoing the action right afterward) to see which one was doubled up. Can
be irritating.
-------