A strange issue with rb 921 access point

RouterOS Wireless Networking
1

HI every on
just focus on the image down
i have rb921+ubnt sector anenna +15 clients useres using nanostation m5
this the ping test between my access point and one of the 15 clients connected with my sector antenna
perfect signal strength+perfect tx/rx ccq+perfect signal to noise = time out ping test !!!
what is the wrong . is that mean something wrong with my wireless chipset ( hardware failure) , noise proplem ,etc.
important note /
the stranger issue that if i use channels between 5100-5800 = the proplem above is still found
if i use the channels between 5810-5900 the proplem is gone and ping test become normally =1-10ms !!!
need an explanation please.. with thanks..




 
/interface wireless
set [ find default-name=wlan1 ] basic-rates-a/g=\
    6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps disabled=no frequency=5660 \
    frequency-mode=superchannel l2mtu=1600 mode=ap-bridge \
    proprietary-extensions=pre-2.9.25 rate-set=configured rx-chains=0,1 ssid=\
    "ANDERSON" supported-rates-a/g=\
    6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps tx-chains=0,1 \
    wds-default-bridge=bridge1 wds-mode=dynamic
2

How strange. When I read your symptom, my first thought was that you just had an incorrect routing table in the CPE that was causing ping responses to get lost. But the information about frequency ranges complicates this diagnosis.

I see that you have WDS active. Is there any possibility that changing the frequency actually causes the CPE to abandon that AP and lock onto another one with more compatible routing, at which point your pings are going through a third party? (In fact, if the CPE is not also configured for superchannel, that may very well be what is happening.)


Sent from my iPhone using Tapatalk

3

thank you .
i think you misunderstand the issue.
im talking about cpe client user with very very perfect connection with my mikrotik access point but actually , that user have no connection!!!.. unless changing the frequency of the access point :confused: :confused: :confused:

4

Yes, I understood all that. You have a lovely layer-2 connection with no layer-3 connection at all. That’s why you should look at your routing. I would treat the frequency thing as a red herring unless I made double sure that the CPE was connecting with the same AP in both cases.


Sent from my iPhone using Tapatalk

5

with decreasing the TX POWER of the access point
the issue is became more accebtable ..the ping test is became more better :smiley: :smiley:

6

The registration strength information you published is for the AP side I believe, from the board ID in the window header. What is the same information from the CPE?

If you have 15 users on Nanostations and only one exhibiting this misbehavior, change out the CPE. It’s not worth beating yourself up over and you will get instant validation of whether or not the problem is solved.

7

I’m sure that you do realize you’re driving Tx power well beyond wireless specifications of your NetMetal and it’s a small wonder you’re getting something out of it … actually it’s almost fine for client stations with worse signal strength that get served with lower MCS, but it’s a catastrophy for clients with excellent signal which deserve highrr MCS but those transmissions get garbled due to insufficient HW TX power.

8

the answer of your first question/ I dont need to show the information of the another side (CPE) , when you get 1-10ms ping test with high tx/rx ccq ,high signal to noise you will know there are no proplem with that CPE ( the internet of that client cpe is very good) but when you have a ping test with time out( unreachable) ,how can i show you the information of unreachable device??!!!!

the answer of your second question is that i make ping test of the other clients with 90% improvement in results
with my respect to you

9

theoritically you are wright .but i think putting 32dbm value in 2000mw ( the original tx power of rb921 ) = 1600mw tx power so that what is the wrong ??!!
sometimes you find yourself trying to do something out of the controls but finally, you will see a better results

10

Setting 32 dBm (1584mW) is not the main problem, the main problem is that you set-up wifi to use 32 dBm for all MCS while HW capability at MCS9 is merely 25 dBm (316mW). Sometimes this is called “higher modulation order power backoff” … as you don’t allow it, higher modulation transmissions are quite distorted and receiver is not able to decode transmited symbols.

I don’t know from where you got the information about max tx power of 2000mW , the specifications I linked in previous post mention 31 dBm (1259mW) as max Tx power at 6Mbps rate.

11
12

Ok, so it is 2W …but that’s sum of Tx power for both chains (hence 1W or 30dBm per chain) and AFAIK tx-power setting is per chain. In case you wonder how can it support 31dBm (1.26W) at lowest rate: for 802.11n, 6Mbps (MCS0) is delivered using single stream … seems that wifi output driver is capable of a kind-of power boost when transmitting single stream.
Anyway, also on the linked page there’s information about max power for a few select MCS’ and it’s clear that setting 31dBm for all rates calls for trouble.

13

Im so sorry , you all right
i think that link is talking about rb921 triple chain but not double chain
anyway,thank you for all

14

Mikrotik wiki says different:
tx-power (integer [-30..30]; Default: )
For 802.11ac wireless interface it’s total power but for 802.11a/b/g/n it’s power per chain.

So, according to wiki, the Tx Power setting in Netmetal/ Netbox 5 & other Mikrotik .ac devices means total power, NOT power per chain.
Can someone confirm this?