NEC2: Should I model a "tube" or just solid wire (GW) for an aluminum-tube-based antenna design?

I'm considering using this 1/4" aluminum tube to build a cross-yagi antenna in the 2m band. The tube dimensions are 1/4" OD x 1/8" ID x 1/16" Wall (.250" x .120" x .065").

It's certainly easier to model this as a ~6mm wire (GW) in NEC2. I could model real tubes with circles (GA) and wires, but its going to be slow and I am concerned about simulation accuracy since $$\lambda$$ is so much bigger than tube's cross-sectional dimensions.

Questions:

• Should I model this as a thick wire, or really model the tube?
• Is NEC2 accurate enough to model the tube with wires and arc's, or will that just introduce error?

Definitely model the yagi elements as wires.

From the structure modelling guidelines: NEC2 segments should be no longer than about $$0.1\lambda$$ long (a bit longer is OK for straight wires with no junctions or corners). So a halfwave dipole or yagi element should be made of 5 segments, or 6 if you need a centre junction for some reason.

For the wire radius: they recommend segment Length/Radius to be greater than 8, so at 145 MHz the segment diameter can be up to $$\lambda/40$$ or 50 mm or 2". The extended thin-wire kernel (EK) allows thicker wires, with a segment length/diameter as much as 1:1, fully cube-shaped! BUT remember always that segments are an approximation of a wire and a thick segment is not a tube. Current can only flow down the length of the wire, not around it. Current is constant on the segment. So you can use thick segments to accurately model a 2m yagi made of 2" pipe, it will correctly calculate the self and mutual impedances, but you can't use thick segments to model a gamma match.

The segments getting too short and fat usually limits the number of segments. Also more segments are slower to simulate.

It wouldn't be possible or safe to model such a thin tube with a mesh. I would use solid segments up to about a 3" tube, and after that I'd consider approximating it with four wires, aiming for $$\lambda/10$$ or maybe $$\lambda/20$$ grid. Here is a cylinder 200 mm diameter, 1 m long, made with $$\lambda/20$$ segments:

Even if you have a metal boom, it's probably best to model the yagi with no boom. Then apply the boom correction to the elements before you build it. If the elements are mounted through the metal boom, then you need to make them longer by a small amount, the graph is as follows:

You could also experiment with adding a boom in NEC. Junctions with different segment diameters are a bit scary, so treat this with caution. If you add a metal boom and use an even number of segments for the wires, then you need to feed it with a pair of AFVS sources, one each side of the boom (with the voltage chosen to add not cancel of course).

The graph above is from the excellent book "Antennas in Practice" made freely available online by the authors.

• Thank you for that write up! I read the docs that you referred to, I appreciate that. You said "segments should be about 0.1λ long" and it might be a good idea to adjust that to say "segments should be no more than about 0.1λ long, but absolutely no smaller than $10^{-3}\lambda$", just in case somebody gets here thinking that 0.1λ is a lower limit. it looks like smaller segments are acceptable for higher accuracy. Does that sound right to you? Commented Jul 10 at 4:29
• Yes they can be shorter, but - there is a lower limit, when the segments get too fat, in your case about 30 segments in each wire; also for uncomplicated dipole type wires there is no benefit in smaller segments, because the sinusoidal currents are perfectly modelled by the sine,cosine,constant basis function. Commented Jul 10 at 5:23