Jitter difference from AP-SM vs SM-AP

OK, I’m making BIG progress here on my private 900MHz point to point link. I had a eureka moment on this forum when there was a suggestion to put AP into SM mode. That made me think of using the spectrum analyzer on the AP side (in SM mode) instead of from the SM side as I did in the past.

On my previous analysis, the AP was active when I measured stuff from the SM side and I used the freq with the strongest signal. This time, with the AP in SM mode I assume what I’d see is the noise level. So I changed to the frequency with the lowest signal and – voila! – now I have by far my best connections yet. Night and day difference, actually.

In 7.5 hrs, I’ve had the session drop just 9 times. While 9 more than ideal, it is a far cry from what happened in the past. I’m suspecting the drop or re-regs are due to jitter, but what I don’t understand is the values I am seeing.

On the AP session page, the RSSI is about 1400, the jitter always 3-4 (with an occasional 2), and the signal is typically -69 to -71. I take this to be the values for SM transmissions to the AP.

On the SM status page, I see RSSI of about 1600 and signal levels of -65 or -66 pretty consistently. But while jitter is sometimes 3-4, it is often 10-12 which I’m assuming is why I sometimes drop the session or need to re-register

So the question becomes why the big disparity in jitter depending on the direction? It isn’t as if the AP is on a tall tower – both ends are about 50’ up. If it was the trees, wouldn’t it have similar effect on each direction?

Does this imply more interference in the 900MHz spectrum on one end? Since the high jitter is on the SM status page, would that suggest the increased interference is on the SM side (ie cordless phones, etc)?

I am running with 2X enabled (version 7.3.6 on each end) and the AP session shows the rate as VC 18 Rate 2X/2X. Does that mean a jitter of 10-12 is really comparable to 5-6 if running 1X?

Can mast sway of a few inches increase jitter on a 7000 ft link? Or is jitter more a matter of interference? If the latter, what is the proper way to use the spectrum analysis to optimize the frequency selected?

I’ve now got a very usable link – over 3500k down and 900k up and I’ll likely push the ratio above 75% to increase it even more. But if I could minimize the session drops or rereg count, it would obviously be preferrable.

Time will tell if my connection stability varies with time of day (phone usage?), rain, etc. But what should I be looking at to reduce jitter?