David

You are going to have to be more specific as to your intended use.

Assuming you are talking about being omnidirectional horizontally from
the antenna, virtually any form of collinear construction would suit.
This is provided the gain doesnt get so high and the vertical half power
beamwidth doesnt get so narrow that your use of it drops out of the
pattern. (These gain and beamwidth figures being inversely related)

For ease of manufacture I'd probably go for a centrally fed collinear
array with a number of 1/4 wave phasing sections. I'd feed it at the
centre phasing section in a similar manner to a jpole but with a 4:1
coaxial balun. You could make this antenna from (say) a single 3/16"
copper pipe/wire length. This construction may not suit if you want to
run it indoors but will be well suited to side tower mounting.

A "super" jpole or one with a phasing section and extra half wave on top
might be another option. Part the way through this page;

http://www.cebik.com/vhf/jp4.html

Something I havent proven to myself so please take this with a grain of
salt, when feeding a vertical collinear antenna from its base very good
feedline decoupling is needed to reduce the effect of skying the pattern
from the antenna . What is more of a problem though (and what I havent
researched) is that without allowing for power loss (through phasing
delays etc) from one element to the next means you have the same
radiation skying effect, the lower elements radiating more than the
upper. One trick here is to mount then antenna inverted or complicate
the construction with a centered feedpoint.

I havent checked websites for 900MHz plans but I have heard that they
are very popular. It may even be cheaper for you to purchase two medium
gain commercial products (that required a groundplane) and arrange them
like a dipole, feeding them through a 1/4 wave Q section. If an antenna
is designed to be 50 ohms resistive over a ground it will be close to
100 ohms in dipole mode. (Making the Q section 75 ohms will do the trick)

Hope this is helpful

Cheers Bob VK2YQA

David wrote:
The desired response is omni-directional with gain greater than or equal
to a 1/2 wave dipole, something that does not require a ground plane
and is reasonably straight forward to manufacture in-house.