Hello,

1. I red a lot about disconnect problem w/ nv2. I experience it also in RX SXT 5.5. If i want to keep nv2 what settings should i use?
2. How to reach best speed/pps between too RB SXT?

My distance is very tiny, 50m. UDP bandwidth test tell me send/receive about avg 80mbps, top 120mbps. TCP tell avg 35mbps, top 45mbps. That's all?
My SXT are not routing anything, they just make a wireless link between two networks.

Please see screenshots w/ my config.
Client bridge below.
Station bridge in next post.

Thanks,
NIO

Screenshots for second station bridge sxt.

please upgrade both boards to RouterOS v5.14 and report back.
Also how are you testing that speed?

>>Please upgrade both boards to RouterOS v5.14 and report back.
OK will do.
>>Speed.
a. I launch bandwidth test from winbox between SXTs. May be SXT CPUs limit it.
b. I launch bandwidth test from winbox between SXT and network. Same results as in 1.
c. I launch bandwidth test from network 1 to network 2, SXT just like a bridge. Result is a bit better then in a and b, about 10-20% faster.

5.14 upgrade done. Configurations is here http://forum.mikrotik.com/viewtopic.php ... 24#p307278 Still no more then tcp 45Mbps plus nv2 problems...

Please note that I have a limited experience with nv2, so what I'm going to say about it is mainly conceptual matter.

I see no actual reasons, apart of some marginal latency improvement, to use a TDMA protocol (as nv2) in a Point-to-Point configuration like yours, where a proper nstreme setup tipically performs very good.

In addition I'm inclined to prefer "station-wds" mode over "station-bridge" for PtP bridges (but may be I'm too devoted to my usual configuration habits).

As already suggested, a proper speed test should be run end-to-end (i.e. between two devices/PC on each of the bridged networks) in order to let the wireless device's CPUs devote all their computation power to the routing/bridging job.

Finally, it appears that some hardware does not like strong signals with some nv2 configurations, so reducing TX power might be of help. Generally speaking on fixed installations with stable and well predictable link conditions I never use more power than what is strictly required to achieve a fade-margin around 15dB over the receiver sensitivity (or the "background noise", whichever is higher): this almost always gives the best results, reduces the power consumption (a critical factor in solar-powered setups), makes the hardware working more safe (less heating) and is more kind to the surrounding RF environment.

These are just few general hints.

Regards,
rock

In actual testing we have found that station-wds with NV2 delivers higher bandwidth test speeds, but performs quite poorly when copying multiple small files. The best results overall seem to be using MPLS/VPLS pseudowire.

I have configured units on bridge mode as per the configs, the tx/rx rate shows between 200/270 Mbps both tx and rx , tx/rx signal about -50/-50 but when i copy a file from the remote site i will hardly get 10 Mbps, can somebody explain why this scenario, distance is 2 Km

Ajust antenna. Adjust antenna one more time, and last time. Adjust antenna agane. Dont look at tx/rx CCQ. Look at speed when adjusting. This has fixed issues like this in 80% of this problem. last 20% change channel.

Also make sue both chane are selected. and using only-n

not that some hardware do not like strong signal the thing is why shout when you can whisper and hear each other laymans language , it always hurts when one shouts into your ear

Sent from my GT-I9100 using Tapatalk

there sure is alignment issue with NIO cause those singals at client site in the screenshot r far too high for 50 mtrs. or may be NLOS
I get such signals at 4 km clear LOS. check for alignment.
Btest gives me above 120 mbps with ptp and Nv2.

I have configured units on bridge mode as per the configs, the tx/rx rate shows between 200/270 Mbps both tx and rx , tx/rx signal about -50/-50 but when i copy a file from the remote site i will hardly get 10 Mbps, can somebody explain why this scenario, distance is 2 Km

May be it's more simple than you think.

First: how do you mean by "copy a file" ? FTP/HTTP download or what else?

Second: try to copy the same file in the same way (i.e. same protocol) on your local wired network. Ideally you should repeat such test on the local networks on both side of the wireless link and between the same system (this is not always possible, I know). Take note of the results and compare them with the values obtained on the wireless link.

May be this is not your case, but I've seen countless people wasting tons of time in trying to "fix" a perfectly working network only to discover the Most Obvious Thing Of The World: the speed bottleneck was NOT in the network (but in the harddisk's transfer rate, or in a cheap network card, a faulty or overloaded switch/router/firewall/computer, etc.)

Regards,
rock

Guys,

Speed is the smallest of my problems. I realized that i used tcp bandwidth tests from router to router and this is not the real bandwidth. I used iperf at pc from one side and at the other side and did a test. Now i see TCP around 80 Mbps. That is very good for me. The bad is i still have connection lost/rebooting problems as described here http://forum.mikrotik.com/viewtopic.php?f=7&t=60381