Hello, i post below my config and my results with two SXT AC, distance between is about 60mt and the antenna can seen each other. I’m not an expert in Mikrotik OS and HW, this is my first job with this devices.
Do you think ai can optimize something for get more throughtoutput? Currently i can get arount 500mbps UDP in receive mode, and it seem to works better in 802.11 than NV2.
Thanks in advance for your help!
The first major issue is not to use the devices themselves for b/w test because it will max out the CPU before you reach the possible link saturation.
Better use two computers with either MikroTiks btest tool or use iperf to test the actual link bandwidth.
The next issue which I can only assume because you haven’t posted screenshots of their wireless interfaces status tab is TX power:
With only 60 meters between them, I’d advise to generously cut down TX power. Anything between -70 and -60db RX level on the remote site is good - I suspect your setup to be somwhere around -35…-30 which is too high - this oversaturizes the RX chains.
First thanks for your much appreciated help Chris!
I lowered the TX power of both SXT to 0db (all rate fixed), i started from 5 and after 4-3-2-1 and the link seem to be the same also with 0 db in the tx tab and 0 in the atenna gain field.
I did an iperf from my hardwired laptop to an antenna and my debian server on the other, resulting in 251-256mbps (i can obtain 930 mbps with the cable).
Do you think i can obtaion more from this wireless ptp link? I attach a screen of the link status, sorry to didn’t post before and thanks again for your help.
Andrea.
As a gut feeling there should be more headroom.
How many streams did you use for the test ( -P option on iperf client).
Did you also run a UDP test?
Is there a reason you use 802.11 as wireless protocol?
I’d strongly advise using nv2 - using 802.11 on a simple PtP link brings some drawbacks:
Any wireless client passing by will poll the AP for known networks - and this blocks precious air time for your link.
802.11 will always listen for new clients. Air timing is sort of unclear and latencies may vary widely.
CCQ looks really good but to my perception RX levels are still too high.
Try increasing antenna gain settings.
And finally: did you scan for channel usage on both ends and select the channel based on the least utilized frequencies on both ends?
Do you really need 80MHz channels?
Ok, thanks both. I did some test and 802.11 is performing a bit better compared to NV2, about 80mbps more in advantage for 802.11.
I’m trying to decrease the power of the signal but without success, i tried all: tx power 0, 1 and fill the “antenna gain” but nothing i did move the signal streight below -45
I tried also to change to 20/40 mhz but with this the bandwith stops at about 350mbs (compared to 550). Any idea for lowering the signal? Thanks again!
Since I remember, you will never reach 500mbps in a real situation. Under 300mbps for me in excelent conditions.
What speed are you trying to achieve anyway? 866mbps?
The different RX levels on Chain0/1 sometimes indicate poor alignment and/or partially obstructed path. For instance the radios are too close to the ground.
Thanks all. Yesterday i tried to move a bit the antenna but this is the best result… 551mbps UDP. I switch to superchannel and move the bridge on a local blocked frequency for don’t have noise in the near future. I hope to see and imprevement on the NV2 link in next firmware, i would like to move on a different protocol and hide the link as much as possibile…
