John E. Davis wrote:
. . .
Perhaps you can advise me regarding the numerical stability of the
following omni-directional, which NEC indicates has a gain of 4 and a
VSWR1.2 when fed with a 50ohm feedline at 151.75 Mhz.

The geometry consists of a 54.5 inch vertical with 4 35.625 inch
radials that are bent upward by about 11 degrees. The vertical is
made from 14 AWG wire, while the radials are 1/8 inch brazing rod.
. . .

With segment sizes of 0.05 and 0.025 lambda, the average power gain is
very close to 1. Changing the segment size to 0.0125, drops the
average power gain to 0.93, which indicates numerical instability.

Should I believe this model?

Actually, that's pretty good. It's not uncommon to see quite good models
hit around a half dB from perfect, i.e., average gain around +/- 12%
from unity. Within that range, and even when the average gain is
substantially worse, you can correct both the input resistance and gain
with the average gain figure to arrive at an accurate result. (Refer to
the NEC-2 or EZNEC manual for more information.) Poor average gain
doesn't usually indicate instability, but rather problems with the way
the assumed currents from the source overlap onto adjacent segments --
it's an excitation problem. Particularly problematic are wires of
differing diameters, different segment lengths, and with bends in the
immediate vicinity of the source. Your model has all three, so the
average gain you're seeing is really good for that model.

Roy Lewallen, W7EL