I have a ~1.7 mile NLOS 907MHz, 5MHz wide PTP link. I’m using a rb532a at one site and a rb411 at the other, XR9 cards and a 13db pacwireless yagi. The signal is all over the place and at times the noise floor will hit -80dbm or higher. I’m looking at getting a rflinx bandpass filter. Has anyone had experience with these? I can’t find the ubnt cavity filters in stock… Are there any wireless settings that could help? I’m using adaptive noise immunity on both the client and the ap.
signal strength
A few of us have had better results in B mode than 5 or 10MHz.
My current config is 10MHz channel, just isn’t stable at 5MHz and B mode catches interference from above and below.
what I’ve found that helps push through noise is 10MHz Nstreme disable CSMA.
Best bet, get a filter. Streakwave has some for around $50.
Thanks for the reply. I’ll try the suggestions above with B-mode and the 10MHz wide channel. I’ve tried using Nstreme with the 5MHz width and it seriously degrades link quality. This is RouterOS 4.5 and I don’t think Nstreme is stable, I’ve tried upping the hw-retries to 15 and still no luck.
Any link I have that has any interference or weak signal, I’ve had to go back to 3.30.
Several customers are 100% stable, better than 70% ccq, 3ms pings, 36mb connection, upgrade to 4.x and I can barely even get into the radio to check signal strength. A couple have signal levels in the mid 70’s.
Reset the wireless config to defaults, reconfigure, no change. Downgrade to 3.30, works fine. Installed 3.30 test and run CTS/RTS without issue. Upgrade to 4.x again, same thing.
Have you used any of the $50 filters? The commercial grade filters usually run around $250-300 and have 8 poles.
Ugh that’s definitely not what I wanted to hear as I’m using MPLS which has several patches since 3.30… I’ll have to change the link over to WDS and do some testing with 3.30.
Bud of mine, he’s on here, can’t remember his user name, says the $50 filters are pretty good. He’s used the UBNT, home made and the $50, says the cheap ones are good enough.
I’ve tried testing with the 10MHz channel widths on 907, 912, 917, and 922MHz. All perform worse than the 5MHz width… Signal strength and noise floor is about the same at ~-75 and -90 respectively, but CCQ is way down.
I am not saying it is the answer, I’m saying it works better for a lot of us.
I can’t use 5Mhz or B mode stable for 2 of my 900MHz customers, but 10MHz has been for months now. I can’t get either stable with anything past 3.30.
I too was having lots of trouble getting XR9s to work in anything but 10mhz at 2427. Any other configuration would not allow all links/clients to register and would give poor results and frequent disconnects.
Then I came across an alternative to the XR9, it is a new card caled the Xagyl FLR9G30 and it has more power (1000mw) than the Ubiquiti card and better filtering for out of band interference. The FLR9 is compatible with the XR9 so I installed one of these cards on an AP with 5 clients and I immediately saw my noise floor go from -85 to -105 and all of a sudden I could actually change the channel and the channel widths without loosing some/all of my clients. The disconnect rates also improved dramatically. Bandwidth tests also showed an improvement of close to 20%.
Creating a link with 2 of these cards has proven to be much more stable and flexible than the XR9 for me.
The real bonus is that the FLR9 cost less that the inferior XR9.
I suggest anyone having issues with Ubiquiti XR9s should try these new cards. Just google for “FLR9G30”