We have a tower with 4 ePMP 1000 2.4Ghz GPS AP's orientated N, S, E, and W. We are using A/B freq reuse, and KP sectors with the metal case on the back. Last summer we started having connectivity issues with our south sector which gets the brunt of the sun most of the day. Any SM lower than a -70 would have random drop outs. Units would immediately reconnect and it did not seem anyone noticed but me ( I watch our network like some guys watch p0rn).
We had a SM that varied from -68 to -72 that would drop out, then we had 2 that varied from a -71 to -74 that also dropped out. I will not connect anyone lower than a -75 in the summer months (and very few will I allow to be connected at those levels even then), so I did not have anything lower to watch constantly. There had been a close by lightning strike earlier in the spring that I thought may have driven the front end of the receiver into the dirt. But, I noticed these drop outs only happened later in the day and only on days when temp was in the upper 90's. In the fall we replaced the radio anyway, and all was good.
This past weekend, we had 3 days in the mid 90's. Guess, what??? Yep, you got it, the random fall outs began again. Always started in the hottest part of the day, around 3pm and lasted until 6pm or so. The last 3 days we have had rain here, and no fall outs. I even moved a customer over from another AP, that registered at a -77 on the problematic AP for testing and they did not fall out for over a day during this rain period.
I am not sure if the PA (Power Amplifier) is getting hot and acting up (spurs) and getting into the receiver when transmitter should be off, causing the AP to not "hear" the SM's replying, or if the heat is actually messing with the receiver. Either way it seems the AP is not hearing the SM's and booting them off. And...it definitely seems as a heat related issue due to the enclosed metal case on the KP's.
This tower is in a rural area with a few houses that it sees close by. There were some SOHO routers picked up on edect when we first started, but now not many at all. Seems our AP ran most of them to an unused channel. Spec-an looks pretty clean too. If it is something interferring, it is very random, fast acting, and also must be heat related as well due to times when we have this problem (temp in mid 90's or higher).
I do see, "no pong for last ping" in the syslog, but it seems to be at different times from the disconnects. Maybe this is a prelude to the disconnects happening???
Anyone had this problem??