Marconi, Thank you for taking the time to explain. I now understand the point you were trying to make in identifying which end of the cone would be the source for the CMC energy. I think we agree that the source of the CMC will begin at the end of the coax or in this case, the top of the cone loop that serves as the end of the transmission line to the top 1/2 wave. I think you're suggesting that because the bottom of the cone may represent a low impedance to the CMC that the current could not flow along this 1/4 wavelength resonant path.
There are cases where CMC is handled exactly this way. If you have an antenna with lots of CMC on the transmission line, you can connect the shield back to the feedpoint exactly 1/4 wavelength down the cable to suppress the current from traveling further down the line. You'll notice the bottom end of this 1/4 wavelength cone is connected back to the shield at the main feedpoint the same way. I also do not suggest that there are changes in the phase of this CMC or the normal radial currents we would see on a typical groundplane. The 90 degree phase shift from the source takes place on the vertical inside the cone. That should be obvious since the top 1/2 wave is now an electrical 1/4 wave away from the source.
It's not the upward radials that have experienced some magical change in the phase of their currents. It's the simple idea that the top 1/2 wave has its signal effectively delayed by a 90 degree phase shift that must occur since it's that distance from the source. That is the only phase correction required in order to form the "non apparent collinear" since that's all that is required to bring the top 1/2 wave into a constructive phase with the cone. Whatever radial currents that are not canceled on this unbalanced design can now radiate constructively from the cone along with the CMC on the outside of the cone.
I thought you may have found the problem with EZNEC when you first pointed out the direction of the wires in the program. You confirmed the corrections did nothing to make the antenna look any different than a half wave using this software. I then provided several drawings with explanations on how to add four simple wires to your model in order to test the programs ability to accurately predict the phase after your correction with no response. That information is quickly revealed by the length of the added phase shift wires since EZNEC predicts they should be 100% too long. Not some minor amount attributed to velocity factor or some other small issue. Enough to make all models of the Sigma done in the program useless.
DB, Your 3-D work on the currents is a new approach to me and I appreciate this insight. I suppose it is better to take a deeper look than to just consider the cones elements as one. I agree the cone does not provide a perfect Faraday cadge effect. I also wish that CST provided a written theory of operation in its data. In the absence of that, I have to rely on both my field testing and the analysis of the CST data provided by the engineer who made the model. The engineer described the action taking place inside the cone as "confining" the radiation of the lower 1/4 wave within the cone which suggest more shielding or the Faraday explanation. Without wrapping the cone around the base of the radiator, we cannot achieve gain in the field either. The pattern also begins to look more distorted like the J-Pole.
I've built all these models in copper for field testing and found improvements going from 3 to 4 upward radials. Adding another did nothing and wrapping it in wire mesh barely made an improvement that could be measured on test equipment. That was somewhat inconclusive due to pattern distortion around the gamma and wire mesh. The Sigma cone does share some other properties with the broadband dipole example in that it increases the bandwidth of this antenna too. When an element flares out and is terminated with a curved conductor, bandwidth is further expanded. There is no doubt that there are individual currents on these radials however, they are equally sharing the currents as one bigger element around the outside of the vertical to produce a uniform effect.
To follow up here is an example of single conductors used to form "coax" carrying a megawatt at lower frequencies: http://en.wikipedia.org/wiki/File:Solec_Kujawski_longwave_antenna_feeder.jpg With relatively few wires this method forms a very efficient and well shielded transmission line. The Russian OTH radar also used similar coax well into the HF band. I totally understand at higher frequencies it would require closer shielding wires to provide the same degree of confinement but then again we don't need 100% shielding of the signal inside to achieve noticeable gain from this antenna. I point this out to show the "skeleton" design can be used as both a transmission line and a broadband radiator since both appear to be happening with the Sigma IV cone.
