NEC-LIST: Wire proximity in NEC

With regard to the issue of crossing wires, the problem runs even deeper. The
NEC approximation method uses the free-space Green's function (exp(ikR)/R)
for the coupling beyween ANY pair of segments. This function is highly divergent
as R->0 so there are going to be gross errors (i.e. singularities) whenever
this happens, regardless of the angle between the current vector of the two
segments. There is simply no a priori way of estimating the amount of error.
Conceivably the one element (or two) of the Z matrix in such a case could easily
overwhelm the entire calculation (the matrix is inverted by use of row/column
pivoting which will go crazy when the divergent matrix element is reached.)

The only foolproof conditions for successful solutions with NEC are cases
using thin wire segments, relatively far spaced (compared to lambda), in free
space or half-space with perfect ground. Anything else is a crap-shoot.

Eric von Valtier K8LV

-- 
The NEC-List mailing list <nec-list_at_gweep.ca>
http://www.gweep.ca/mailman/listinfo.cgi/nec-list