On 8/14/2011 11:53 PM, Helmut Wabnig wrote:


When optimizing for F/B also 4nec2 results in a too short driven
element, same as YAGICAD, they obviously use the same formula,
(a long two-liner) from literature.

Unlikely...
4nec2 is just driving the finite element NEC engine underneath. YagiCad
uses analytical approximations.

NEC is *very* sensitive to segmentation, spacing, and wire diameters,
etc. particularly when close together or angled. The NEC4 engine is
quite a bit better, but you're probably using the default free NEC2 engine.

To effectively use this kind of thing, you need to be pretty aware of
the limitations and peculiarities of the underlying FEM codes. While I
wouldn't expect someone to have read the theory manual for NEC2, you
might want to check out some of L.B.Cebik's writeups on using MoM codes,
or the stuff in the ARRL Antenna Compendium from time to time.



I actually built antennas following that prescription,
and they do not work without additional corrections,
e.g. a very long hair-pin match.



Imagine a 3 element 2 meter yagi,

|
| |
| | |
| | |
| | |
| | |
| | |
| |
|

Something is wrong here with the capacitive driven element.

Depends on what you were optimizing for.
If all you said was "drive F/B to maximum", then it won't try to get you
a decent match, all it will do is drive to having element currents that
optimize for F/B, which will almost certainly have an evil feedpoint
impedance.



Now I am having a hard time, because I want to find out in
what differs a gain optimized vs a F/B optimized antenna
and it is questionable whether that can be done with software.
Have not tried other programs yet.

What I do is not try to use the optimizer to design the whole antenna
including matching network. I optimize the basic element design first
(without matching network). Then I add the matching network, hold the
element spacing and length constant, and let that optimize. Then I go
back and allow element length to change, but hold matching network
dimensions/component values (I tend to use lumped element matching
networks) fixed.

It's an iterative thing because it's tough to specify an appropriate
optimization cost function with a simple interface. What you really
want is something like "optimize F/B, but keep the feedpoint impedance
20 ohms and 100 reactive ohms"





A Yagi with fixed element positions can only be used for 1 frequency
(within a very small range) and therefore may be impractical
in many situations. The CEBIK tape measure antenna with their
water-pipe construction could be easily modified with a moveable
center element.

this is just not true (otherwise the SteppIR wouldn't work.. it has
fixed element spacing and works over a huge (1 octave) frequency range)
with fairly good performance.

One can also do similar with fixed spacing, and fixed length, and
reactive loading of the elements at the centers.