Jeff wrote:

This is a problem that is particular to vertically polarized yagis that are
mounted with a pole going between the elements. The pole has to come up
parallel to the elements to the boom. Some people overcome this by using
non-conductive pole, BUT what do you do with the feeder, if you tape it to
the pole you might just as well have used a metal pole!!!

However, as I said in my previous post, my simulations seem to show that
with careful positioning the effects can to quite small

This problem has been investigated very carefully by moonbouncers who
want to use selectable horizontal/vertical polarization.

The best solution from the electrical viewpoint is to mount the yagi on
a fibreglass mast and route the feedlines out the rear, but this is not
mechanically practical for very large steerable arrays. Simulations have
found that metal masts in the plane of the elements can be used, and
feedlines can follow the same route, if - BUT ONLY IF - the yagi is
suitable and it's done at the right place along the boom. Under these
conditions, the penalty (in terms of forward gain) can be less than 1dB.

However, only certain yagi designs will be suitable, because the support
point must also be close to the mechanical centre of gravity. These
electrical and mechanical requirements may conflict, so I strongly agree
with Jeff that the effect of the mast and feedline will always need to
be simulated and computer-optimized. There is no way to guess at this
one.


Coming back to the original question about a J-pole feed for a
vertically polarized yagi, this technique was used in the 1950s-60s by
the British company J-Beam for TV antennas around 50MHz. The driven
elements that I remember were not actually J-shaped, but used a coaxial
decoupling sleeve that also served as a mast. The bottom end of the
sleeve was typically clamped to a chimney.

It worked... but in those days, nobody really understood how well or
poorly it worked.


--
73 from Ian GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek