On Thu, 26 Mar 2009 09:35:06 +0100, Jon K Hellan LA4RT
wrote:

I'm very naive in these matters. Could a coaxial stub be modeled as a
cage of wires around the center conductor? Would the orders of
magnitude difference between shield/center distance and wire lengths
cause problems?

Hi Jon,

I offered that model long ago in this thread - as it was ignored, I
was condemned to use it myself.

I had given some thought ahead of plunging ahead into the model (it
was originally a thick radiator for which the model was perfectly
suitable). The concept of coaxial tube shielding proceeds along the
premise of the shield supporting separate conduction paths, isolated
by skin effect of the tube conductor. That is, the currents of the
shield on the inside surface are separate and distinct from those on
the outside surface. I knew full well that NEC would not maintain
that distinction for any wire in a cage simply because it lacks the
ability to report separate currents along the same wire as would be
found in this inside/outside tube surface.

The model I published and provided the link to here in this thread was
not strictly faithful to the concept of the cage model for a coaxial
tube, however. I enhanced it into roughly 1000 wires emulating a cage
10.5M long, 2M in diameter, with hoops every 33cM along its length,
and closed at both ends. Think of it as a roll of mesh with a 1 foot
grid capped at both ends with radial wires. Within it is a length of
wire that is roughly 10M long and isolated from the cage at both ends.

With the wire loss set to perfect, the central wire was driven and it
was as though no shielding cage existed. Within tenths of a dB, the
radiation characteristic across HF was roughly the same as from a
simple wire dipole. Conceptually, it would appear that the Faraday
shield does not exist in the world of NEC.

When I introduced the copper setting for wire loss, this assemblage
exhibited the following "loss"
MHz
1 24.2
2 16.6
3 14.2
4 12.8
5 12.1
6 12.1
7 13.4
8 18.8
9 19.0
10 6.6
11 3.0
12 2.2
13 2.0
14 1.8
15 1.6
16 1.7
17 1.5
18 1.4

Following this, I connected the ends of the coaxial interior wire to
the caps at the tube ends (a complete short circuit). Losses in the
left column (where significant); lobe peak in the right column (where
significant):
MHz
1 56.1
2 39.4
3 29.3
4 21.9
5 15.8
6 10.5
7 6.0
8 2.4
9 0.2 1.8 dBi
10 2.7 dBi
11 2.8 dBi
12 2.7 dBi
13 2.5 dBi
14 2.3 dBi
15 0.2 2.1 dBi
16 0.5
17 0.7
18 1.1

So, to your question:
Could a coaxial stub be modeled as a
cage of wires around the center conductor?
No, not if my experience bears any relevance.

73's
Richard Clark, KB7QHC