| T O P I C R E V I E W |
| aryajur |
Posted - Dec 08 2016 : 18:48:53
Hi,
In the uniformly discretized ground plane there is a lot of overlap in the segments. Would that cause inaccurate results? I am just thinking about what I should consider if I want to discretize a shape myself. Does the overlap cause inaccuracy? In the plane the area of the segments would be almost twice the area of the plane as I understood.
Thanks |
| 1 L A T E S T R E P L I E S (Newest First) |
| Enrico |
Posted - Dec 09 2016 : 11:42:53
Overlaps of perpendicular segments are not only allowed, but also necessary for a correct modeling of currents in the different directions. In fact, there is no magnetic field coupling for filaments with currents flowing in perpendicular directions. So you may correctly have up to three perpendicular segments sharing a common node in the 3D space. In the ground plane, the current flow is assumed to be 2D only, so you have perpendicular segments only lying on the plane.
On the other hand, you should not overlap at all parallel segments. Note that in FastHenry's ground planes the parallel segments never overlap, but just completely 'seal' the plane; please refer to the User's guide for details. If you overlap parallel segments, you will create numerical issues; most of the times this causes a warning from FastHenry, but you may also end up in corner cases, anyway affecting the precision of the result.
As a final note, please whenever possible prefer the internal ground plane definitions instead of your own, as FastHenry needs to discover all the meshes (current loops) in your input structure. For the ground planes this is easy as FastHenry knows exactly how they are formed, so the meshes in the plane are already known; while if you use arbitrary constructions, FastHenry must use a more complex algorithm to find them all.
Best Regards,
Enrico
|
|
|
