So you're actually the first site with all 3 sectors installed, and this has exposed a problem I hoped would be better mitigated with the open air environment your site provides. I knew there was antenna crosstalk, I measured it @ 60dB of isolation in my garage. I was hoping it was mostly reflection-based, and it was. But the crosstalk up on the tower is as bad as 72dB isolation, which is still more than I'd like to see. If we can chop another 40-50dB down from that, I'd be happy. The fact that there are other antennas in front of the sectors isn't helping either, as the signal scatters and bounces back to neighboring sectors. I'm gonna fab some retrofit isolation shields to put on the sectors. Aluminum alloy 1100-H14 seems to be good for microwave work. First test, we'll try to eliminate the backwards radiation from the dish making it into the north sector. It's hard to verify this kinda stuff on the ground cuz the isolation levels required are so high, and ground reflections are a plenty. It's really gotta be done without echoes, so the tower provides a good environment for that. Here's what sector 1 (north facing) is seeing. The part shaded "WiFi" is the dish pointing back @ Capitol Hill. That -55dBm signal should not be entering the north sector. It's also seeing a lot more signal bleeding from sector 2 (120 deg facing) which peaks around -40dBm. And finally you can see sector 3 (bearing 240 deg) leaking in at about -55dBm. The "WiFi" marked leakage is only 5MHz bandwidth, whereas the other 2 channels are 20MHz wide. Hopefully with the shield in place we'll bring that leak down. If not, we might also have to augment the dish itself, as it's probably scattering the signal around the main reflector edges. Fabricating a shield for that is a little more tricky, so I'd like to augment the sectors first. BTW, a pretty nice little spectrum analyzer built into the modems, huh? :) --Bart
