Background:
I have 3 900MHz APs all located on 1 tower. Each sector covers 120[degrees] and is running on 10MHz channels at single (horizontal) polarity. The APs are made up of RB600s with XR9 cards, and the clients are all RB411s with XR9 cards. I have heard some pretty amazing stories centered around the TDMA protocol and was astounded MK came out with their own version coupled in nV2. Therefore, I immediately jumped on the chance to implement this in my network.

So I updated all APs and clients to the same routerOS version and implemented nV2 on all APs and clients following the exact guidance of the MK Wiki, forums, reference manuals, etc. (who said jumping both feet in wasn't fun - right). The result - horrible to say the least!! However, when I disabled nV2 on just 2 of the APs and the clients associated to those APs, nV2 worked flawlessly on the 3rd, running nV2 all by itself.

I think it prudent to mention that each AP only had about 10-15 clients associated to each.

Question:
Can anyone explain why nV2 only worked as it should with just 1 sector AP enabled and the other 2 running normal CSMA protocol, and not with all 3 sector APs running nV2? Could it have anything to do with the timing or time slot overlap between the competing APs? I know in UBNT's TDMA version (airMax), they now have GPS syncing capability between the APs, could the lack of some sort of "syncing" capability have anything to do with the problem as well? If so, does MK have a roadmap to GPS syncing between RB systems.

Just curious if anyone could provide some insight - Thank you!

NV2 needs AP's with high signal rejection off axis, in other words in your case beyond 120 ° the AP attenuates incoming signals, what you really need is to reduce is the signal from the other sector AP’s,
My own experience of NV2 with co-location interference was solved by RF screening each AP,
With syncing of AP’s you will not have max throughput from an AP as you would running the same AP on it’s own.

http://forum.mikrotik.com/viewtopic.php?f=7&t=49835