noise floor is recalculated each time when the periodic calibration takes place.
To be honnest I don't remember If I enabled the periodic cal on such trial. But I remember that this changes on NF readings was noticed every couple of seconds, back and forth. May be turnin' it on make it to settle down to the real value instead of keeping such initial values (eg digital AGC, if any, with 50% range, driving RSL ckt/routines and this initial value will only be changed when periodic cal takes place).
I even tried to decrease R52Hn power to 5dBm and for a minute (as AP Bridge) and switched to 5GHz band, cos I want to see if there was some out of band signal overloading the LNA. Other noise sources may be 433UAH clocks and data lines, the XR9 and R52n sourrouding it & etc but they can't affect its IF because I guess it is a zero-IF transceiver.
But: I don't believe R52Hn could be more sucetible to such out of band noise sources than R52n.
Please do not power the unit if you don't have antenna connector to the antenna connector as you might damage the power amp of the card. Check if you card doesn't heat up too much. Here is the instructions how you can check the card:
http://wiki.mikrotik.com/wiki/Manual:Wi ... iagnostics
All this tests was using or sector antenna or terminating the MMCX to N pigtail with Hewlett Packard 50 ohm loads (good to 26GHz).
I guess the same terminator may be left at chain 1 if not in use because those RF switches have 20 to 30dB isolation only when properly terminated.
The bench tests later show a healthy R52Hn when connected to a probe antenna aimed to an AP abt 300m, the same tx/rx signals than a 411AR client.
Please tell us more about hose interface hangs with v5.x - we are not aware of that issue.
Well, I'll try this again next days despite this 433UAH is a production unit. I say abt v5.x on it. There is 60 possible clients with NS2 Loco except one (that uses a temporary CPE with Engenius ESR1221). May be this ESR is messing things up.
There was 4 freezes, spaced an hour or so, not the router itself, all clients still connected but the traffic at this WLAN interface dropped to zero for 10 ... 15 min and went up again alone. Nothing hapened on XR9 or ETHs, all up and runnin'.
The regular log was full of messages related to this ESR CPE. I did not enabled the debug because this ESR is temporary.
I saw this once when I decided to allow a single client notebook to have access to a remote AP (WRAP, X86, R52n & R52Hn fitted on it). After 30 minutes the AP freezes. I repeated this 4 or 5 times to be shure and to avoid this I needed to hide SSID and delete the correspondig network profile on such mobile device. Up to now 120 days without problems.
Neither NF issue nor hangs was noticed on a X86 with v5.10.