Indeed I do! First note this: Antennas are voodoo. Pure and simple. If you get better amplification by swinging a dead cat three times under a full moon, then by god, ship every unit with a dead cat. That said, there are a few general principles:
- 1/(2^n) wave antennas are better than other oddball ratios. Full wave are generally best.
- Vertical antennas are omni-directional, but that just means that they suck equally in every direction.
- Ground planes are great and bigger is generally better. A square mile sheet of copper should do it.
And something that applies to your case:
- The antenna is not the antenna. The antenna is the antenna plus everything leading up to it that isn't completely shielded, minus the parts that are inadvertently shielded, plus an unknown (and unknowable) factor based on god's will and the signs of the zodiac.
Seriously, antennas are voodoo. If you want repeatability, then L-com makes some fine antennas that almost always conform to spec. Except when they don't. But then you can return 'em. For tiny, do-it-yourself nodes, I'm (so far) having okay results with quarter-wave-ish, 18 awg wire pointing more-or-less straight up.
Now for the "optimum" case. Note that this is theoretical (doable) optimum and actual optimum may involve the aforementioned voodoo rituals and dead cats. Circular hunk of solid copper as a ground plane, one-wavelength in diameter, with everything below that shielded and sunk to earth ground. From that, a full-wave vertical, as straight as possible, and preferably superconducting, or at least solid copper.