Roy Lewallen wrote in
:

Owen Duffy wrote:
....
I believe that NEC can do a fine job of modeling any of the variations
we've been discussing. But like all modeling systems, it has to be
used properly -- the transmission line object isn't an adequate model
for either a coaxial structure or an open wire stub, if either is
carrying any common mode current. And in these antennas it is, so you
can't insist on using nothing more than a transmission line object and
then bemoaning that the result isn't correct. The wire stub variation
can be correctly modeled as wires. The coaxial structure can be
correctly modeled as a combination of a wire and transmission line
object. In either case I have high confidence that carefully and
accurately measured results will agree closely with NEC predictions.

Taking the W5GI as an example, here is a deck that models the coaxial
stub section as a conductor of 5mm dia, whilst the wires for the other
sections are 2mm diameter. I have calculated the impedance looking into
16.5' of RG8X (W5GI's specified stub) as 14.5-j179 at 14.2MHz, and
inserted that load in both of the segments where the o/c end of the stub
is located. I have not used a TL element, rather I have separately
calculated the input Z of the stub using the technique used at
http://www.vk1od.net/calc/tl/tllc.php , that should be more accurate than
using a lossless TL element.

The model assumes an effective balun, ie that there is no common mode
feedline current since I have not provided such a path.

CM W5GI Mystery Antenna
CM Extended thin wire kernel used
CM
CE
GW 1 31 -5.033 0.000 10.563 5.033 0.000 10.563 0.001000
GW 2 15 -10.067 0.000 10.563 -5.033 0.000 10.563 0.002500
GW 3 15 -15.100 0.000 10.563 -10.067 0.000 10.563 0.001000
GW 4 15 5.033 0.000 10.563 10.067 0.000 10.563 0.002500
GW 5 15 10.067 0.000 10.563 15.100 0.000 10.563 0.001000
GE
EK
FR 0,1,0,0,14.200
EX 0 1 16 0 1 0
LD 5 0 0 0 5.7E7
LD 4 1 1 1 14.505 -191.739 0
LD 4 1 31 31 14.505 -191.739 0
GN 2 0 0 0 13 0.005
XQ
EN

This model indicates out of phase operation of the antenna, a multi lobed
pattern and feedpoint Z of 115-j179. (Although there is a half wave of
300 ohm line in between, this feedpoint Z would cause VSWR=8 on the 50
ohms line.

I think that I have dealt with the common mode path properly.

Try as I might changing stub lengths etc, I cannot get this configuration
to deliver in-phase operation of the radiator.

I suspect the model is not valid.

Owen