We hung a new CBRS AP at one of our sites last week and attempted to move customers over to it yesterday. Unfortunately, all SMs end up sticking at 1x on TX and RX most of the time. Randomly, they will jump to 8x as expected but then plummet to 1x and stick there. Eventually they de-register and will stay disconnected for hours at a time. I was only able to leave one SM attached to play with, but this is feeling like a bad AP to me unless there is some magic I’m missing?
Spectrum is squeaky clean at the AP and at station locations. This SA capture is from the AP when SMs are stuck at 1x. SA captures from the SMs are even cleaner.
Stations report good SNR even with the bad rates:
Every now and then rates return to what I’d expect, but only briefly.
I have tested different frequencies and widths (20 and 40Mhz) across the band and the behavior is identical. Half the time SMs won’t even register and have a high re-reg count.
There is an old UBNT M2 sector about 3’ below this AP and an ePMP 5Ghz sector below that. Otherwise, there is nothing else on this tower and no CBRS gear for miles.
The behavior is the same on 22.1.1 and 22.2 firmware.
Has anyone else seen similar behavior? I have a number of sites with the same build without any similar issues.
Do you have another AP you could try up there? At this point I’d probably start a ticket with Cambium.
I will open a ticket with Cambium and do have more NIB APs I can hang - But, that also requires getting the tower crew back up there. Scheduling can be a challenge and the costs are not insignificant, unfortunately.
I’m working with Cambium support on this, who also doesn’t see any obvious cause for the issue. One thing that has definitely become clear though is that there may be a temperature and humidity component to it, as the problem doesn’t exist when it is warm and sunny out. Perhaps some sort of strange and extreme thermal inversion issue that only affects 3Ghz 450i in this location, but no other equipment.
Just spitballing, but if this is a 450i 3G Connectorized AP, there may be an issue with the connection between the AP and the antenna, such as moisture in the LMR connections.
It’s the integrated unit and has exhibited this problem since installed, on a sunny day.
Did you get a resolution on this?
We are seeing something similar - poor performance but then for unexplained reasons it comes right for a period of time.
No resolution, unfortunately. Support wasn’t ultimately of any help and I moved on to other things. The AP remains hanging on the tower because I don’t want to pay the tower crew to come up a third time to remove the replacement unit…
In the case of your graph - I usually see the same for two reasons:
- Interference at the subscriber’s side. I have seen this when competitor’s APs are on the same or overlapping channel with mine and in a similar azimuth from the subscriber. It seems to happen when the competitor’s AP is busy.
- Reflections or multipath. One of my installers for a while was installing the 450b radios a bit too close to reflective surfaces, like the wall of the house or the edge of the roof in the shot. Randomly, RX rates will crash and stick for minutes to days for no rhyme or reason. Moving the radio away from the reflective surface seems to usually resolve that problem permanently. The 450 radios seem very sensitive to this.
There is also another thread about v24.0 and RX rate crashes. That may be relevant if you have upgraded to 24.
Similar to your one - when it’s showing the modulation problems, the RSSI chain imbalance shoots up big time.
Spectrum analyser reveals no interference either side of the link.
Haven’t upgraded to 24 yet.
Also - I’m in New Zealand. Our spectrum works a bit different, for our region my WISP has the exclusive rights to use the spectrum we are on.
Our problem was very similar to @Jacob_Turner, in addition to major chain imbalance, we also saw different metrics affected by the outside temperature.
Both our faulty units are from a batch with serial numbers starting M9ZD.
We have received a new unit and replaced it on the tower today and immediately the problem has been resolved.
I know of three other WISPs in New Zealand experiencing the same problem, one who has done a RMA / replacement and it’s solved the problem.
Hopefully, others who’ve ended up with faulty equipment will find this thread and avoid a few days of wasted time and money troubleshooting this issue.
I’m not sure how the serial numbers are sequenced, but both units I have that exhibit the problem have MSN starting with M9ZB04
