I reran my EZNEC coax model with a cage with an octagon loop every foot
in height... This is a pretty good shield at 40m, I think. The holes
are .002x.007 wavelength.

The difference is apparent.

Not much confidence that this is modeling the problem... but maybe it's
better:

Ran it with wire inside/outside the shield. Did #2 wire this time to
simulate thin wire down the center of a pipe. Before, the
inside/outside comparison yielded almost identical results. Now it
doesn't...

Wire 6 inches outside shield:

Wire No. 26:
Segment Conn Magnitude (A.) Phase (Deg.)
1 Open .00356 86.70
2 .00776 95.54
3 .01039 103.14
4 .01208 109.03
5 .01323 117.85
6 .01373 126.69
7 .01297 133.85
8 .01159 144.14
9 Open .00684 154.69

Wire centered in shield:

Wire No. 26:
Segment Conn Magnitude (A.) Phase (Deg.)
1 Open .00201 53.19
2 .00298 65.98
3 .00324 71.00
4 .00509 45.58
5 .00387 78.85
6 .00401 118.35
7 .00308 63.54
8 .00347 137.70
9 Open .00503 164.46


Anyway, the point that the original cage was a bad model is taken. An
additional approximation toward a full shield changes things a great
deal.

Should anyone want to take a look: http://www.n3ox.net/cage_coax.ez

Dan