How should I represent the curved ends of a folded dipole with MoM?

I'm trying to simulate a folded dipole using MMANA-GAL, which is a method of moments package based on NEC2. The diameter of the dipole is 12 mm, the radius of curvature 35 mm, so the two arms of the dipole are spaced by 70 mm.

How should I represent the ends? Clearly the method can't use a curve, so I wrote a C program which worked out the coordinates based on a number of wires I chose. It split the 180 degrees into an even number of angles, then used:


x=r*cos(theta)
y=r*sin(theta)

This means the ends of the wires lie on the circumference, but the total length of the wires is smaller than the length of the true arc.

How many wires should I use to simulate such a structure? I've used 6, but I don't know if that's too few or too many. Note the length of the arc is going to be pi*r, or 3.14*35=110 mm. The diameter of of the element is 12 mm. So if I break this into 6 wires, each is going to be about 110/6=18.3 mm, which is not much more than the diameter (12 mm).

I also wondered if it was more sensible to ensure the arc length is the same as the length of the wires.


I'm interested in knowing this for a range of frequencies in the range 50-500 MHz.