Problem with SM TX power balancing (ATPC)

Recently I’ve been upgrading our network of FSK radios, PMP 100s on 900 MHz, to 10.5. I’ve tried the SM TX power control setting on a few different towers with extremely mixed results. On some APs there appear to be no issue and it works as advertised, where the receive level at the AP is brought down across the board to near-ish the target, and jitter figures and efficiencies improve.

However on some APs almost the opposite happens. On both P9 and P10 hardware, with as little as 4 SMs all running at 2X up to an AP with 11 customers in a mixture of 1X/2X, when the automatic TX power control is enabled there is a lot of problem maintaining 2X links. Most of the time I will see all of the links on a given AP will run at 1X only. Sometimes I will see radios attempt 2X, but fall back to 1X almost immediately. And I have also had a couple of SMs that say they are connected at 2X, but the aggregate speed on the Link Capacity Test shows 2.6 Mbps where it was previously a perfect 4.3 Mbps link. I can’t find a pattern to this yet. The workaround appears to be to disable this feature. In fact, taking the TX levels determined by the mechanism, disabling the automatic control, and manually entering the power levels in the SMs makes the 1X problem disappear while giving me the advantage of the power level balancing.

As well I have noticed an issue while looking at the Link Status Table on an AP. Sometimes some SMs will drop to an upstream signal level of -105 dBm and a jitter of 0, which looks like a completely loss of link, but the Reg and Re-Reg counters do not increment. I suspect that this is performance impacting.

I’ve sent that in an email to support but thought I’d check here as well. Has anyone else seem similar?

I have an 2.4GHz AP with 14 SMs on it and I can see the same thing, you mentioned in my Link Status table. Jitter goes to 0 for a second or two, as well the signal’s power level drops to -99 ,but the Reg and Re-Reg counters do not increment.
Would like to see the response from their support once you got it.
By the way have you noticed how often the SMs are starting new sessions with the AP?

Interesting to hear that we’re not the only ones! The response we got back from support was to gather a ton of information like spectrum analyses and do an RF map and CNUT dump and get back to them when we can. Unfortunately we haven’t yet had the time to get all this done what with our PMP320 evaluation, IPTV, and MPLS projects.

We saw some other problems too and downgraded to 9.5 again. On some of the affected APs we’ve solved issues that may have been related to some readings (poor links, local interference, etc) but I have not yet upgraded the network to 10.5 again to see if things are any different.