jmorash wrote:

I agree that the sleeve dipole, coax-based J, and whip with radials
should all have the same radiation pattern, but I get the impression
that performance might vary quite a bit (specifically the impedance
matching?).

Impedance matching isn't much of a problem with any of those antennas.
The main issue is feedline decoupling. Current will end up on the
outside of the feedline unless, of course, your antenna is mounted
directly on a metal structure -- in which case the current will end up
on the outside of the metal structure. This current radiates just like
the current in the antenna, and this added radiation can cause a number
of problems. One is that it can modify the pattern and reduce the
radiation toward the horizon. Another is that the current can get back
into the transmitter and other circuitry where it doesn't belong.
Finally, it can effect a change in feedpoint impedance, since the coax
is part of the antenna you probably didn't account for.

The current can originate by two mechanisms, conducted and coupled.
http://eznec.com/Amateur/Articles/Baluns.pdf explains the conducted
mechanism. Current can be coupled from the antenna to the feedline even
if you're using a solid ground plane of moderate diameter, and all the
popular implementations have coupled current to some extent. If you
search for a while, you'll find people both raving about and raving at
J-Poles. I suspect this is at least partially due to the amount and
phase of coupled current they ended up with due to their particular
installation. The amount of coupled current depends on the length and
path of the feedline, as well as the path to ground or some large body.
If I were designing the antenna you describe, I'd use a "current balun"
(common mode choke -- see the balun article) at the feedpoint and about
a quarter wavelength down the line. At that frequency, a good size
ferrite core or two of the right type might provide adequate impedance.

Half wave antennas have relatively little conducted current because the
feedpoint impedance is so high. But feedline current can still exist due
to coupling.

Roy Lewallen, W7EL

Thanks Roy! I think I understand baluns quite a bit better now. Seems
like, indeed, a couple ferrite beads could make quite a difference to
my radiation pattern.

This question is for Roy or anyone else who might know the answer: what
tool(s) do I want to try and gauge how good of an antenna I've built?
An SWR meter? Do they make those for 900 MHz? What else?

jmorash wrote:
Thanks Roy! I think I understand baluns quite a bit better now. Seems
like, indeed, a couple ferrite beads could make quite a difference to
my radiation pattern.

This question is for Roy or anyone else who might know the answer: what
tool(s) do I want to try and gauge how good of an antenna I've built?
An SWR meter? Do they make those for 900 MHz? What else?

An SWR meter doesn't measure antenna goodness. The only thing it tells
you is how close the antenna's impedance is to 50 ohms, which has
nothing to do with the important measures of its performance such as
gain and pattern. The best test instrument is a low power transmitter
and field strength meter, one at each end of a simulated communication
link. For the other end of the link, it would be best to use a setup
typical of what you'll actually be using in the field. You'll be able to
get some idea of the antenna's effectiveness with over-ground tests, but
the ultimate test will be the strength of the signal received over water
with the antenna mounted as it will be for the real application. Precise
quantitative measurement isn't trivial at all, but qualitative relative
measurements are fairly easily made.

Roy Lewallen, W7EL