[Antennaware] EZNEC how to choose number of segments

[Terry Conboy]

Dave,

In general, using real/Mininec ground will give you slightly optimistic gain numbers, especially with horizontal conductors under 1/4 wavelength above ground.  This is because for impedance calculations, the ground is assumed to be be perfectly conducting.  However, for far field patterns, the specified ground conductivity and permittivity are used to determine the pattern shape.  (A pet peeve is seeing published articles that use models with Mininec ground specified, and getting inflated gains.)

I find the only time I need Mininec ground is when some part of the antenna is connected to ground when using the NEC2 engine, and then you will need to add resistive loads to simulate what you think the ground loss resistance is, which is never easy to do with any accuracy.  Of course, if you have a NEC4 engine, you can model the ground system with much better accuracy, although there is certainly an art to that, as well as many who will say NEC4 is far from perfect.

It’s likely that your metal post (below the fiberglass section) won’t have much impact on your pattern or impedance, but it is possible that if you don’t have a feed line choke with high enough impedance, there may be enough radiation from the outside of the coax shield to impact both the pattern and impedance.  Place a “load" with suitable resistance and/or reactance on a 1/4 wl wire (representing the coax shield) near the radial junction to see if it’s doing the job.  You may find that the old, very-approximate rule of thumb of a choke with 10x the feed Z isn’t nearly enough in many cases.  You also may be surprised how much power is burned up in the choke!  Of course, good luck finding out the complex impedance vs frequency of commercial common mode chokes!

None of this relates to your original question.  My approach is to use segments as small as possible without EZNEC complaining (>0.001 wavelength long), but as Dave Cuthbert suggests, it’s a good to make the segment lengths 4X the diameter.  If the calculation time is too great, or programs limits are exceeded, double the segment length and test for convergence, as well as testing for nearly 0 dB average gain (3D) with all losses zeroed out.

BTW, moving the source a segment away from the radial junction probably will have little impact on the apparent tuning of the element or radials, but it may show a slightly higher Z than it would actually be at the bottom of the vertical element.

73, Terry N6RY