RB433AH and R52Hn problems

We have been having point to point and point to multipoint link problems with the RB433AH and the R52Hn.

Recently we installed a network with a couple of repeaters built using RB433AH and 2 ea R52Hn. Generally the links have a point to multipoint backhaul with one radio as StationWDS looking at an upstream AP. They both use 5GHz, with one radio set at 5180 and the other at 5825. SSID are set differently so that the radios do not connect to each other and an ACL also prevents this. On the STATIONWDS radio we have a panel antenna and on the AP side a 90 deg sector. Both antennas are cross polarity dual feed with separation of at least 25dBi (according to the antenna spec).

When initially installed the links work great with good SNR and CCQ. Then after a couple of days the links fail. What we see is imbalanced chains with chain 0 having a reported signal of say 65 and chain 1 having a reported signal of 89. When the link was working they were both 65. If we disable chain 1, then the link comes up. But as soon as you enable chain 1 the system goes bad. We are using NV2 with dynamic WDS and AES encryption.

At first we thought this might be a power issue. But we tested with a single radio system and it does the same thing. Boards report 17.8V from a 18VDC 1A supply. Should be plenty of power. Cable is only a couple of feet. After a couple of days the chains become imbalanced and it does not appear to be related to power fluxuations. We have seen this a few times now at different installations.

So then we thought it might be driver related. We know that there were problems with the ath9 driver on the RB433AH boards under linux, but we thought ROS was ok. Looks like this might not be the case. So we tried using ROS 5.21, then ROS 6.0rc6 and are about to try 4.17 to see if we can get any of these to work.

We have a couple of similar links that use RB435G that do not appear to have this problem. We also have some links using RB411GL that also run OK.

Is there an issue with the PCI bridge or something like this that could break under load? Seems like both Linux and ROS have issues with RB433AH, but not with other boards. Any other ideas?

Hello,

I am not sure if you are experiencing the same issue as me but I will share my experience. We have 500+ Mikrotiks deployed and I have not had very good success with the R52HN in any area that has very much noise or is near other 5Ghz equipment. I have three deployments that were perfectly aimed and signals in the 60’s on both polarities and they would work great until another radio started pushing any traffic then they would blow chunks. and one wouldn’t work out of the gate as it cannot handle being within 10’ of an SR5.

Ultimately the DBII and UNBT cards are about the only cards that work reliably in our noisy environment. In a lab the R52HN cards work great but they don’t have enough shielding to reliably operate for us. We were using metal enclosures and shielded Cat5 etc. not our first rodeo, and we have gear on 50kw FM radio towers so we know how to shield properly to get things to work.

Hope you get it figured out,
Josh

Thanks Josh, but this does not appear to be a shielding or noise issue. The links were working fine for a period of time and then all of a sudden chain 1 failed. And it did it on several units at one time. No new interference sources. No high noise floor. Just weird behavior.

Hope you get it figured out. so now to figure out if its a bug or if its interference. if its the same chain on all affected units perhaps someone lit up some new links or AP’s on H or V and it messing with one of the polarities.


Regards,
Josh

The last time I had issues like this it was due to an R52Hn failing. Haven’t had any problems since, but shortly after they were released I had two fail within a week of testing on the ground.

All that passed a week in testing are in the air and have been for years now.