I've got a site with six 1310's installed. Prior to Hurricane Katrina we've had no problems with the setup.
Since the hurricane, we've replaced all but two of the units (those two didn't get flooded). On the units that were replaced, we also replaced the antenna's (12dBi omni's) and the cable's (100' ULL).
We are getting really poor signal strength, -85 to -90 dBm. The maximum distance between root and non-root's is about 425 feet with clear line of sight. I don't have a good calculation of Fresnel zone, but since it is 28ft at 2.8 miles, I think I'm clear on this also.
I've two non-roots that are about 200 feet away from root, on start-up they have a signal strength of about -50 dBm, but over 10-15 minutes, they degrade to -85 dBm.
The Cisco engineer I was working with was thrown in the towel, saying he doesn't know anything about antenna's.
I did a site survey and didn't come up with any interference.
You don't mention whether the 12db omnis are more or less gain, I'm assuming more. You also don't mention whether the 100' ultra-low-loss is longer or shorter (or the same) as what you used to have (I'm assuming the same length, but maybe better quality).
Anyway, here goes:
Higher gain omnis have to get their gain from somewhere, you can't just get a bigger signal passively without changing the shape of the radiating pattern. In many cases, a hgher gain omni has a pattern that resembles a champaign glass, where lower gain omnis look more like a dougnut or a frisbee, centered around the long axis of the stick.
So, it's possible that all of your "extra" power is being sent over the top of the receiving side.
The way to check this is to look at the radiation pattern diagrams that accompany most, if not all, antennas. There are usally two; one that shows the pattern from the side (important for an omni) and one that shows the pattern from the top (important for directionals, yagis, dishes, and panels).
Check your diagrams ... there's a good chance your transmitted signal is over the head of your receiving antenna.
Regarding the cable; look over the entire cabling system, including the (so-called, inappropriately labeled) "lightning arrestors," the grounding connections, and the connectors at each end.
Because your are working in a high-humidity environment, making sure the connectors are dry and well sealed is very important.
If the signnal degrades over time as you mention, it could be a breakdown somewhere (like the "lightning arrestor" begins to "leak," or it could be the transmitter scaling back because of some mismatch or other problem in the "intentional radiator" (which is everything from the transmitter's interface to the input of the antenna).
It may be worthwhile to try another segment of cable, or substituting the various elements between the transmitter and the antenna, to rule out components that are bad or causing the transmitter to cut back.
If you went with a longer length of lower loss cabling ... there could be an issue there.
Provide some additional details if possible and we can work through it.
The satellite pictures are from Google, so they are pre-storm. The first picture, is the entire campus. The next two are zoomed in shots, splitting the campus in half where the root bridge is located. I used line art to indicate where the antenna's are mounted.
How important is the grounding on the antenna's? We did some testing by removing the ground, but didn't see any change, so we put the ground back on and kept going.
Also, I've just noticed there is a new IOS, 12.3(8). Would it help to upgrade to that?
Thanks for the pictures, it helps a lot to see how things are layed out.
The move to replace 200' of low loss to 100 feet of Ultra low loss was probably a Very Good Thing and should have increased the signal strength considerably.
Grounding is very important, if only for the safety aspects. "Lightning Arrestors" don't stop lightning; they bleed off static electricity and (tend to) reduce the space charge that can accumulate on exposed elements.
It's always a good idea to put in a segment of fiber between the APs and the wired network to prevent the zap from killing the rest of the network (and maybe the phones too ....).
Check the coax for kinks (or places that were kinked), sharp bends, broken jacket, etc. At Gig+ frequencies, any malformed section of cabling an have a significant impact on the propagation of the signal.
Check each connector to make sure there isn't any "stuff" that is causing a casual short between the center conductor and the shield. Look closely, wire fragments from the braid can be pretty hard to see.
Also make sure that the center pin is ~level with the surrounding sleeve. Sometime the center pin is driven back into the connector (i.e., bad termination) and is not making sufficient contact.
The "starts strong and tapers off" symptom can only be something related to power, a failing component, heat (causing a component to break down), or some safety mechanism (i.e., high SWR causing the transmitter to throttle back)....
Try another known good power brick,
Try another antenna (any antenna)
and see if the symptoms change.
Probably the most suspicious places would be the connections around the arrestor, and the connection from the coax to the antenna pigtail.