Hello,

Have at the moment next problem:

Link operates at dual nstream, ~54/54 x2 @5ghz-a turbo
Signal strength ~ -55dbm
signal to noice ~ 40 db
lowering db (manual to 10,13,5 does nothing)


real traffic push ~ 4,6-6,1MB/s


both sides 2xR5h +F50 pro(dbii) similar card
==========================
If cards are changed to DBii F50 N-card (atheros N11)
Then, link connects RX CCQ almost 100% TX CCQ goes suddenly to 25%;
speed 130ht / 6Mbps (jumps for second to 130mbps, then back). very unstable
signal strength -18 ~ -30 dbm
signal to noice ~ 80! db


RX CCQ could change with TX CCQ it's baddnes, -> TX goes good, RX bad ; and backwords

Both sides dbii Atheros 11N cards; I have tried to put then work on 5800mhz, then 5200mhz same thing. Scanning/frequency usage -110 almost all..

Where could be "mistake"?

Oh, ROS4.14

Without actually viewing the link itself a few things come to mind in order of likelyhood.

1. Antenna alignment
2. Freznel zone obstruction.
3. Noise in channel selected
4. Spacing between antennas.
5. Distance of link and antennas used.

Perhaps separate the links, and measure performance of each to isolate the problem.

Performance is also affected by device and rules etc.

ie I have a rb411 push 24Mbps tcp on 5.8 A turbo, (cpu limitation) but rb pc push 70Mbps tcp. (Single feed antenna, not duel nstreme)

Also upgrade to 4.17 Ros.

Hope it helps.

Without actually viewing the link itself a few things come to mind in order of likelyhood.

1. Antenna alignment
2. Freznel zone obstruction.
3. Noise in channel selected
4. Spacing between antennas.
5. Distance of link and antennas used.

Perhaps separate the links, and measure performance of each to isolate the problem.

Performance is also affected by device and rules etc.

ie I have a rb411 push 24Mbps tcp on 5.8 A turbo, (cpu limitation) but rb pc push 70Mbps tcp. (Single feed antenna, not duel nstreme)

Also upgrade to 4.17 Ros.

Hope it helps.

Between antennas space about 2m; cable CDF400 low noise ~1/1.5m; noise on frequency table -100 ~ -118; no near other networks etc; distance between point a and point b 1km, LOS; antennas svenska antennispetsialisten 28dbi dishes; (1xV and 1xH on each side);

In dual nstream same link works perfectly 4.1-6MB/s; trouble comes when change cards to atheros 11N and put 5ghz-only-n, installation is the same. RB433AH are in use..

It appears I misunderstood the question.

Its possible that nstreme-dual doesnt support N links, as N is supposed to be its replacement.

Have you tried to bond the 2 links ?

http://wiki.mikrotik.com/wiki/Manual:Interface/Bonding

It appears I misunderstood the question.

Its possible that nstreme-dual doesnt support N links, as N is supposed to be its replacement.

Have you tried to bond the 2 links ?

http://wiki.mikrotik.com/wiki/Manual:Interface/Bonding

eee, point is not in that. I am trying to set up existing Dual nstreme link with 802.11a-turbo(2x cards at each side) to 802.11n with 1 card at each side, and both channels (antennas on ch0 and ch1)...i suspect that going to 802.11n will increase throughput. each side 2x antennas

problem is in that, that with 802.11n it shows bad CCQ - for example TX is almost 100%; RX jumping from 25-85%

i have one mistake at my point of view - dishes are not under angle, but should to be;
point A is about 10m higher than poing B, but with nstreame dual problem doesnot persists, so i think it's not the problem

another thought could it be the problem with cards, perhaps someone has experience in such case

distance between point a and point b 1km, LOS; antennas svenska antennispetsialisten 28dbi dishes; (1xV and 1xH on each side);

28 dbi antennas for 1km is overkill beyond words. That matched with dbii 28 db N cards and theres just to much power. as a comparison, I use 28 dbi antennas for 20km links and get -57.

reduce your power on the dbii cards on both sides until your signals about -65, your ccq should rise. use card rates on tx power tab, and lower the power. with dbii n card calc power as card rate + 5.

post your /int wi pr advanced and /int wi reg stats with the n cards in.

Try N with nstreme if your not doing so already. Please state how your testing the capacity of your link when using nstreme duel and N, also remember ccq drops when the link is not in use. Please give the model of the cards you were using for Ntreme dual link for reference purposes.

Lastly , you should be getting about 100 mbps tcp (roughly) with dual N, your data rate should be in the 270 Mbps range when its working properly.

note (informational):
Ive noticed 802.11 with nstreme, 802.11 with nv2 , N with nstreme and N with nv2 handle slight alignment problems differently and give different performances. Some just tolerate it better.

poor ccq is usually down to
1.poor alignment of antennas even with good signal

TX is almost 100%; RX jumping from 25-85%

indicates the recieve antenna alignment is out.
2.noise on channel

Scanning/frequency usage -110 almost all..

noise floor is just an indicator. use : /interface wireless spectral-history <wireless interface name> or /interface wireless spectral-scan <wireless interface name> for more accurate indication.
3.blown radios : running dbii high power radios without antennas (or dummy load) for any length of time could blow them. (True for any high power radio)

translate.google.com could help overcome language barriers.

distance between point a and point b 1km, LOS; antennas svenska antennispetsialisten 28dbi dishes; (1xV and 1xH on each side);

28 dbi antennas for 1km is overkill beyond words. That matched with dbii 28 db N cards and theres just to much power. as a comparison, I use 28 dbi antennas for 20km links and get -57.

reduce your power on the dbii cards on both sides until your signals about -65, your ccq should rise. use card rates on tx power tab, and lower the power. with dbii n card calc power as card rate + 5.

post your /int wi pr advanced and /int wi reg stats with the n cards in.

Try N with nstreme if your not doing so already. Please state how your testing the capacity of your link when using nstreme duel and N, also remember ccq drops when the link is not in use. Please give the model of the cards you were using for Ntreme dual link for reference purposes.

Lastly , you should be getting about 100 mbps tcp (roughly) with dual N, your data rate should be in the 270 Mbps range when its working properly.

note (informational):
Ive noticed 802.11 with nstreme, 802.11 with nv2 , N with nstreme and N with nv2 handle slight alignment problems differently and give different performances. Some just tolerate it better.

poor ccq is usually down to
1.poor alignment of antennas even with good signal

TX is almost 100%; RX jumping from 25-85%

indicates the recieve antenna alignment is out.
2.noise on channel

Scanning/frequency usage -110 almost all..

noise floor is just an indicator. use : /interface wireless spectral-history <wireless interface name> or /interface wireless spectral-scan <wireless interface name> for more accurate indication.
3.blown radios : running dbii high power radios without antennas (or dummy load) for any length of time could blow them. (True for any high power radio)

translate.google.com could help overcome language barriers.

I know about overkill on 1km with antennas
3. is out, couse board had power after we have all connected
2. will try, and tell you later
1. could be.. they are not angeled A to B just on 90degree in side of each other.. point A is about 10m higher than point B

I was using 28dbi becouse, it's very important link, it should stay even if someone will try to make interference, to down it. At the same, side A is on power plant 110kV, side B is on civilian living house.. At the same somewhere at half way, about 15degrees at right there are mobile antennas of 3 operators.. 3x3+3x3+3x3 sectors + some links i suppose on 20-40ghz;
ah, and there also a prison near, there are also generators that should down signals from mobile operators antennas... they are about 500-600m from antenna place

Anyway, better to have bigger antennas, than troubles with power, when it's needed

I'll upgrade to ROS5.x and continue at night works on that link..

ROS5.4 made my life easier

totally ~270mbps chain0+chain1;
fully loads 100mbps with tcp traffic rb i suppose sometime later i should try with atom minipc and gigabit lan try what could 802.11n drift out

ping is good.. if downloading 7mb/s ~5ms, no losses.
Protocol used nv2 nstreme; 20/40mhz above

used DBii pro 802.11N cards