why this link Can not transport More than 35 mbps real traffic ?

CCQ always more than 95%
RX / TX signal level ( 55-60 ) / dual chain
output power 17
Free LOS
distance = 1200 m
superchannel

Are u routing something or it is simple p2p link
Signal to noise is a bit high
have u try nstreme

are you doing a bandwidth test from AP to Station?

@ Lakis :
it's simple p2p link ( wireless Mode :Bridge - station Bridge ) /// bridgh port > wlan1-ether1

nstreme !!!? Nooo ...
plz post nstreme configuration (wireless mode - security and etc )

@ semakka

speed test result From AP to CPE
UDP ( transmit ) : 139 mbps
TCP (Transmit ) : 37 mbps
TCP ( Both ) : 17 /18

Signal to noise is a bit high

65dB STN is too high?.... It's almost "perfect" you mean.

@ hipro5
i think 65 dB Signal to noise is the best result for 1.2 km link with 17db tx power !

@ semakka

speed test result From AP to CPE
UDP ( transmit ) : 139 mbps
TCP (Transmit ) : 37 mbps
TCP ( Both ) : 17 /18

Never do a bandwidth from AP to CPE if you want to see how much your throughput is. You will overload the CPUs of your AP and CPE.
Place 2 big resourceful daddy routers at both sites and b-test between them through the PtP Link, in the end that's what you want to know.
You need to know how much traffic is passed through, not how much data your AP can send to the Station.

I'm sure this was explained before and you'll find reference on wiki@mikrotik.

cheers,
Steven

@seanet: with it's internal CPU running at 400MHz, while you run the bandwidth test between them, the CPU of the AP goes up to 100% load and it CAN'T produce more data to send to the CPE.
Try EXTERNAL traffic generator (from a desktop pc through a 1G LAN connected to AP) and re-test...


@ semakka : you beat me to that post by ~1min...

@seanet: with it's internal CPU running at 400MHz, while you run the bandwidth test between them, the CPU of the AP goes up to 100% load and it CAN'T produce more data to send to the CPE.
Try EXTERNAL traffic generator (from a desktop pc through a 1G LAN connected to AP) and re-test...


@ semakka : you beat me to that post by ~1min...

Bugger....

I'll buy you a beer )))))

Steven

Bugger....

I'll buy you a beer )))))

Steven

rb1100ahx2 ___eth(cat6)__Sextant(bridge)<---->Sextant(station bridge) ____eth(Cat5e)____>Rb450G

test result rb1100 to 450g >>> just 8mbps !!!!

rb1100ahx2 ___eth(cat6)__Sextant(bridge)<---->Sextant(station bridge) ____eth(Cat5e)____>Rb450G

test result rb1100 to 450g >>> just 8mbps !!!!

RB450G? that has the same CPU as your link RB's.
I'm sure you can do better than that
You need more than a RB450, maybe another RB 1100AHx2

Steven

450G with 680mhz cpu !!!
but throughput lower than sextant AP to CPE ...

Are you sure all eth ports of all RBs are negotiated at 100M Full Duplex?

yes of course

do UDP b-test between the rb1100AHx2 and the bridge and another between BR450G to the station bridge. Send and receive, but not both.
if they both turn to be close to 100mbps and between RB1100AHx2 and RB450G is still 8... you have a wireless problem

maybe use WDS?...

steven

rb1100 ---- sextant 98 mbps
rb450 ---- sextant 96 mbps

i don't use wds

We have seen this before, it will work just got to keep tweaking.

The bandwidth test that you are doing, are you using tcp or udp, and if using tcp are you using the default 20 connections?

To avoid the whole cpu peaking during btest's, use udp just to get a feeling of how the link is performing.

Are you colocating these at sites with other 5Ghz equipment? Sextants have a very poor front to back ratio, and are very susceptible to noise from other devices.

In NV2 make sure you have set the cell radius close to your actual distance. In the data rates section define it to use a rate selection of advanced.

What are the speed differences when changing the link from nv2 to 802.11. Generally speaking, unless its a very long link nv2 really only shows its benefits in a PtMP setting.

Signal to noise is a bit high

65dB STN is too high?.... It's almost "perfect" you mean.

whatever is best for u
in best practice SNR should be around 30-40db - never the less this does not mean in all conditions should be correct

@ Lakis :
it's simple p2p link ( wireless Mode :Bridge - station Bridge ) /// bridgh port > wlan1-ether1

nstreme !!!? Nooo ...
plz post nstreme configuration (wireless mode - security and etc )

Say Noo to racism but never say no to nstreme

There is noting to configure just select on AP side wireless protocol=nstreme

Signal to noise is a bit high

65dB STN is too high?.... It's almost "perfect" you mean.

whatever is best for u
in best practice SNR should be around 30-40db - never the less this does not mean in all conditions should be correct

If you get a SNR of 30dB as you say in 64-QAM MiMo OR single transmit, you're f(l)ucked....
Your MER, VBER and CBER will be bad, your quality won't be good and you're never gonna "lock" at 300Mbps. You're gonna have ups and downs in locking and your throughput will be worst than locking EVEN at 90Mbps.
40dB SNR and BELOW(means this number be higher/bigger), is OK for 64-QAM.
IF someone sees a lower number than 35dB SNR, he'd better MANUAL lock his link in 16-QAM 3/4 at 180Mbps(MiMo) and 90Mbps single or 16-QAM 1/2 at 120Mbps (MiMo) and 60Mbps single and below connections (QPSK modulation).

Just an idea:
Your wireless device got the traffic fo or to gigabit devices. Please confirm that!
(edit: i',e just so the structure, so yes)

In my experience, if the traffic is originated from gigabit - to a wireless device with 100m port AND using nv2, thruput will be much lass than expected. If you start the btest from and to 100m port device, it will show much better thruput.

Please try to use gigabit'ed sxt or 433ah instead of 450g, and let me know if that helped.

Try to use just one chain.

I found that dual chain using 40MHz channel modulating 300/300Mbit Btested over 200Mbit UPD is pretty slower than single chain on 40MHz channel.

I can do always over 65 up to 100Mbit real TSP data on single chain 40Mhz channel Nv2. But was never able to do at least 60Mbit on dual chain.

4.5km p2p 2x5HnD
u have Internet traffic + btest

Here is a screenshot on a RB600 what is the AP side of a 15km link. The client side is a RB433AH. They are linked with 216/216mbps (40mhz, dual polarity - HQ antenna)

[img]mikrotik_nv2_nk_screenshot.jpg[/img]

It's also possible to push ~200mbps with NV2 in good environment.
(the above was ~30mbps+5mbps real traffic, plus 2 btest in different direction, with TCP - initiated from different boards)

Please check also the following link: http://forum.mikrotik.com/viewtopic.php?f=7&t=70129
probably it better explains where the problem is. ( i bet, in NV2 packet buffering)

I have a RB600A and x86 machine connected via wireless (nv2, 2 chains). I get TCP speeds of 185 mbit/s. So it is possible

I have a RB600A and x86 machine connected via wireless (nv2, 2 chains). I get TCP speeds of 185 mbit/s. So it is possible

good one
I push TCP speed of 220mbit with advanced channel (30Mhz,nv2,2km)

Guys,
you - who pushed more than 100mbps - all use full GigE path, don't you?

Anyone, who has GigE router, but with FastE transmit device, and reaches 90mbps with NV2?
please raise your hands!

check this out This is UDP Test now I don have good equipment to do TCP test but u will get clear picture


/I have to correct my test

Guys,
you - who pushed more than 100mbps - all use full GigE path, don't you?

Anyone, who has GigE router, but with FastE transmit device, and reaches 90mbps with NV2?
please raise your hands!

Yes.. full GigE path.
And when connecting a 100 mbit end-device I reach speeds of 90+ mbit/s.

onnoossendrijver:
can You explain please the structure when using FastE device?

In my experience, when traffic starts from GE and going thru FE wireless with NV2 the thruput is much less than expected.
If the traffic is start from FE and going thru FE wireless with NV2 thruput is almost OK.

In my experience, when traffic starts from GE and going thru FE wireless with NV2 the thruput is much less than expected.
If the traffic is start from FE and going thru FE wireless with NV2 thruput is almost OK.

I tried both situations. They both perform equally well (90+ mbit/s).

My situation:
Linux machine -->
(this link is tested as 100mb and 1gb)
Netgear managed switch -->
*Juniper SRX210 router -->
Netgear managed switch -->
*x86 machine with RouterOS 6.0rcX and AR9380 wifi -->
(wireless)
*RB600A with RouterOS 5.2x and AR9220 wifi -->
(this link is tested as 100mb and 1gb)
*RB450G with RouterOS 5.1x-->
RB250GS -->
Linux server -->
Linux VM

*=routing

hipro5 wrote:
Lakis wrote:
Signal to noise is a bit high


65dB STN is too high?.... It's almost "perfect" you mean.

whatever is best for u
in best practice SNR should be around 30-40db - never the less this does not mean in all conditions should be correct


Lakis, you are posting incorrect data and not understanding SNR.. You have it back to front... Higher the value the better, it is the difference between signal and noise.

10dB = 10
20dB = 100
30dB = 1000
40dB = 10000
50dB = 100000
60dB = 1000000
63dB = 2000000

another words 2 million times more signal than noise

hipro5 wrote:
Lakis wrote:
Signal to noise is a bit high


65dB STN is too high?.... It's almost "perfect" you mean.

whatever is best for u
in best practice SNR should be around 30-40db - never the less this does not mean in all conditions should be correct


Lakis, you are posting incorrect data and not understanding SNR.. You have it back to front... Higher the value the better, it is the difference between signal and noise.

10dB = 10
20dB = 100
30dB = 1000
40dB = 10000
50dB = 100000
60dB = 1000000
63dB = 2000000

another words 2 million times more signal than noise

I read wiki its opposite
U and hipro5 are correct

i don't want to make things confusing BUT
what can be too high is Signal and Noise but NOT SNR.

If signal is too high, that means radio is too loud.
To imagine this, think radio transmission like precisional shooting. 64QAM split the target to 64pieces. If sent from 1, hitting 1 (and only 1) is a perfect hit.
If sent from 1 and hitting 1 (and 2 and 3) it not so good. If you have proper signal, is like using sniper rifle. If you have too much signal, is like using shotgun
on a nearby target.

If noise so too much, than addig extra power is not helping too much. You should avoid the source of noise (interference) somehow.

The bigger the difference in SNR the better...

Yes you are correct to some degree..

I can only quote relating to Radio Transmitters and Receivers at the lower frequencies, my former trades were in HF communications and Band II broadcast.

So I have no idea how the various stages of signal amplification work on the atheros chip sets and even little interest.. Microwaves just do not rock my boat like good old fashion radio.

However lets assume that the principles are the same

If the signal is too loud then the RF amplification stages within the chipset or on the pcb card will have difficulties coping with it.
Certainly at lower frequencies it would cause overload in the detector circuits and leave the radio unable to discriminate clearly the wanted signal from signals on adjacent channels.

Another words, its "selectivity" or its ability to differentiate between channels would be impaired.

Not to be confused with "sensitivity" which is another subject ie a radio card that is "deaf"

Yes wirelesswaves and hipro like u said most of my good p2p links have SNR >55


But I have something opposite in my network This is 20km p2p link running on 20Mhz A, radio cards are XR5 as AP R52Hn as client
throughput is about 34mbit TCP almost max that A-standart non turbo can achieve

But look at SNR its about 30dBi

HI all
thnx For Reply to this post

SNR = Signal - noise
More signal = More Tx power + FLOS

I think reliable SNR is >>> 45 < SNR < 65

High SNR ? > Reduce tx power
low SNR ? > increase Tx Power , Try to FLOS , Change Freq to find better signal

--------
Seanet

high TX power = destruction and Reflection ==> Low throughput
Try to reduce or increase tx power to have better throughput (MT UDP bwtest) ! Not higher SNR

The most important item is transmit ccq ... try to highest CCQ ( 100% ) in RX / TX


With my experience
better throughput when :
ُ50 < signal < 60
45< SNR < 65
tx power @ 1.2 km = 17dBm
UDP throughput (sextant to sextant) = 160 - 205 mbps

I have a RB600A and x86 machine connected via wireless (nv2, 2 chains). I get TCP speeds of 185 mbit/s. So it is possible

good one
I push TCP speed of 220mbit with advanced channel (30Mhz,nv2,2km)

Did you use the extra channel on sextant?

If you want try this configuration.

http://www.wirelessinfo.be/index.php/mi ... /omnitikpp

It's with omnitik but is the same configuration.
This is a example