Hello,

I have just begun interfacing with
Ansoft's Designer (SV). I am trying to model a magnetic
dipole in free space, operating at 10GHz and extract
the radiation resistance.

This is well known
theory:

Rrad=197*((a*omega)/c)^4 [Ohms]

where:
a=the radius of the loop

omega = 2*pi*f

c=the
speed of light

For a<<lambda

Here
is what I have thus far:

Two layers of air
dielectric both of thickness = lambda/2 =15mm

A lossless conductor for my signal layer. The signal
layer is sandwiched between the two dielectrics.

Defined "Drawing extent" (Layout>>Settings)
such that each edge (Left, right, top, bottom) are
2*lambda from the center where the dipole is located.
I have done this too provide adequate space for
the far-field approximation to be accurate, but
I would like to know if this is necessary. Any thoughts?

The geometry:

I have constructed a square
loop (at this level of approximation (a<<lambda)
it shouldn't matter if the geometry is square or
circular) by making a square cell 60x60 microns^2.
I copied and pasted this cell four times then assembled
to form a square horseshoe. I closed the loop with
two rectangles 60x90microns^2. This leaves a void
of 60x60 microns^2 in the middle.

I would
like to have one port located between the two rectangles
(centrally located rather than on a corner) to provide
a voltage potential and setup circulating currents.
Is there a better design choice? Any suggestions?

Results (so far):

I tried the single
edge port (which then defaulted to a gap source)
and Ansoft gave me the following error:

“Edge
port is not represented in the mesh: port 1”

Then I tried separating the loop just slightly
and defined two ports one on each edge (now facing
each other) so the geometry looks like square “C”.
No errors. However, the currents are virtually non-existent.

I would appreciate any suggestions:

Meshing: Fixed (with or with out edge mesh)
vs. Adaptive

Port placement ideas

etc…