I have/had a CPE unit very regurlary (several times a day, up to several per hour) disconnecting to associate back to AP in same second.
Both AP & CPE running routerboards with v5.4 + latest firmware in 5Ghz-10Mhz band with NV2 protocol enabled on AP (and cpe follows)
This CPE has only 6 meters on each side brother or sisten in same AP network, same hardware, same confg, but they don’t have a single dissconnect for days!
Running a scan the AP shows a distant (16km!) AP working in same frequency, with same NV2. This AP never before gave any problem. Signal on the AP while running the scan for this remote AP was -95 or worse, pretty weak.
Disconnecting client has signals in the -50 range both rx/tx of that AP. And it was in a urbanized estate where ´other´ AP on same freq. only could be seen with binocular and through some concrete and metal constructions on houses in the LOS to that remote AP. So basically I never suspected that remote AP to be the problem on this particular CPE.
After studying the wireless debug log I noticed that this CPE does tries to connect to this remote AP but to be rejected inmediately with message “MT:no” whereafter this CPE connects back to his assigned (connect_to list) AP.
Since this remote AP is also mine, I decided to shift his frequency 15Mhz down. And see, disconnect problem is dissapeared!
Before using NV2 this remote plus the CPE’s ´mother`-AP have both been running on the same frequency for a year without same disconnect problem. Also because this distant AP is a sector actually pointing 90 degrees away from the troubled CPE. Hence I never saw the problem of using the same frequency on AP’s 15km away fm eachother where one of the AP’s is actually not even radiating its signal in the direction of the other!
So it must be traces of the signal from that remote AP now still finding this troubled CPE and the TDMA is now one way or another aggressive enough to give problems where they never have been without NV2.
My conclusion is that either NV2 or the software package that introduced it (4.1? and higher) are more vulnerable to cause, or endure, interferences from other AP’s.
While NV2 had the promise it should actually do better than 802.11a/g (b/n can be different story) it certainly also has its backdrafts…
I wouldn’t really say your issue is resolved, you just found out how bad NV2 deals with noise. With a 45 SNR you shouldn’t have any issues with the remote AP causing you problems. If this is the case NV2 is totally worthless and broken beyond belief.
Well, I used this subject title to get the extra attention to keep the ´heat´ towards MT guys in keeping the discussion as much as alive as possible, like you did in your other new post!
Anyway, although it is indeed not an issue solved, it does give a nice example in that NV2 is either more vulnerable (from CPE point) or more agressive (from AP point) towards other 802.11 usage.
I already pleaded many times before that imho it is the tdma protocol in combination with the poor selectivity of our radio cards (they always pick up other signals not mend for them) and the increased sensitivity of the latest radio’s creating the problem. And it should be the OS dealing with it. I don’t know about ubnt but I do have the feeling they have lesser problems with the same chipsets than MT has?
Like I said before, a horse and wagon never needed aerodynamics. But now a car does need it. Suddenly the wind becomes a factor that might interfere with our course or speed. But that doesn’t make the car a poorer product. But you do need a good design to work with it to the best…
Ok, run a scan, on both 5, 10 and 20Mhz band, on that SXT. See what other frequencies are in use at what signals on the clients location.
Do the same on the AP.
If there are others, see what the channel separation is and how strong are these other signals.
Than, find a freq. not in use and also having some free spectrum around its working freq. and try that one. See if the disconnect disappears (and maybe comes back on one of your other stations)
By looking at the mac addresses of the ´other´ spectrum users, you can probably guess what system they use. In case of ubnt, is this Airmax or not. Or Motorola’s canopy I believe can be an issue as well. Or other MT user with NV2.
For test reason temporarily switched that remote AP back to same freq as client and see, inmediately the disconnects start again…
As can be seen in the scan, that remote AP is coming in with a signal around -88 where own ´mother´ AP comes in with -71 (is much better at times).
This situation without NV2 was never giving any problems. Now it is.[
Ehh, by the way, you have to zoom in to read the picture. I work with high resolution which don’t print nice in webforum…
Hi Rudy, good to read some issues have been resolved, as regards AP radiation patterns just because in the data sheet it says 90 degree’s horizontal but there is also side lobes beyond the 90 degree’s, also if the antenna is mounted onto a metal support this can at these frequencies change the radiation pattern, it is well documented that FM band II antenna manfacturers will setup the antennna’s on a test site using a scaled down copy of the customers tower from which they can modify the antenna’s to achieve or confirm a desired pattern for the customer and the effect the tower has on the antenna’s.
From reading your post it appears to me that clients which have issues like this is to use a cpe with high gain say 23dbi narrow tx/rx and also if not commerically available use alumimum folded in a horiziontal U box shape around the cpe to reduce/eliminate signal pick-up from outside the 23dbi 10degree rx/tx, or relocate the cpe to a location which had better shielding, say if possible from chimney to side gable of house,
I think the days of using plastic antenna boxes are limited for WISP’s and much more careful planning in the selection of each antenna and it’s location.
I am aware of side lobs and that metal parts might interfere with the signals leaving/receiving the antenna. I use metal back-plated sectors so I would not expect too much issue from these. But yes, it could well be a side lob is in this case responsible for the unexpected high gain at this remote CPE.
From reading your post it appears to me that clients which have issues like this is to use a cpe with high gain say 23dbi narrow tx/rx and also if not commercially available use aluminium folded in a horizontal U box shape around the cpe to reduce/eliminate signal pick-up from outside the 23dbi 10degree rx/tx, or relocate the cpe to a location which had better shielding, say if possible from chimney to side gable of house,
True. I’m indeed in the process of changing clients towards better shielded antenna’s. I use the MT RIC units the first years and decided to face them out since they where a bit big and where all equipped with slow boards and I could use these relative high gain units better in region where client have to bridge bigger distances. But actually I’m coming back on that decision. Actually I find now just replacing the boards is a better deal. These MT RIC`s still in use are actually never giving the problems in my networks. It is indeed the plastic boxes that do!
As you could read in some other topics I made/joined I started to shield these with metal paint and/or alu foil to shield back and side of the units better against interferences. And yes, the ´higher gain=better directional=better interference protection from side penetrating signals´ are the best solution. It also reduces the overall spectrum noise in an crowded area.
think the days of using plastic antenna boxes are limited for WISP’s and much more careful planning in the selection of each antenna and it’s location.
I can only fully agree on that!
But heho, special for small starting WISP economy comes as important too and the difference between good and bottom prices can be 300% or more!
perhaps once a week or so.