Hi,
i am very long time trying to get best from new 802.11n technology using Mikrotik. There is very noticeable improvement in version 5.4 generaly. It looks like it has better CCQ and holds higher data rates. Looks like it better understands what other site sends.

Using 2x433AH + R52Ncards + very perfect 24dB and good isolated dual polarisation antenas.

All tests done on simple way :
local WWW server <-> 433AH AP<-> 433AH Client<-> PC just using multiple http downloads
Signal 47/48dB
Run short ping 100x second before each test to get the best wirelles rate, than stoped and run the download test

The best what i can achieve is with 40MHz channel, using nv2 and single chain. It makes 100/100CCQ and holds rock solid 150/150Mbit (before 5.4 just 130/130 in nv2). The download speed comes to 68-83Mbit and looks to be good.

Any other combination was not able to run as fast. Dual chain 40Mbit channel has poor CCQ just over 40/40, and even at modulation speeds 300/300Mbit is not able to deliver more than very unstable 28-48Mbit.

I realy dont know what iam doing bad, because i was not able to achieve good resluts on dual chain outdoors. The single chain is always more stable and has more troughput. Many locations even 50km away, and always the same result.

Anybody an ideas ?

what isolation you have between the polarizations?
Try to lower the tx-power a little bit.

Used antenas http://en.jirous.com/anteny-5ghz/jrc-24-duplex

Technical parameters
Frequency range 5,45 - 5,9 GHz
Gain 23,6 ±0,6dBi
VSWR ≤ 1,5
Isolation between connectors 53dB (≥50dB)
Front to back ratio ≥ 38 dB
Beamwith-3 dB 9°
Connector N - Female - JRC-24DuplEX
R-SMA - JRC-24DuplEX-SMA
Polarization linear, horizontal and vertical
Parabola ø 38 cm
aluminium alloy with baking colour
Cover UV steady plastic ABS
Instalation for mast ø 27 - 74 mm


I already watched videos from MUM Hungaria, and tryed lower down the output power

10dB fixed results in lost off cca 5% troughput
5dB fixed results in lost off cca 10% troughput
0dB fixed results in lost off cca 20% troughput

The signal level for both chains is simetrical. (nearly the same)

Just an idea: Is it possible it want work right with extra 40Mhz wide channel? Ill try it with 20Mhz tonight. Is there some difference on R52N cars usinf MMCX or UL-F output connector? We use MMCX

But like i said, i was never able to get stable link speed using dual chain.

I am using dual chain and dual antennas and I get 100mbps udp half duplex at 5 miles and 50mbps at 10miles.

2 dallas: And how do you test it ? If you use the BTest on Mikrotik 2 Mikrotik makes 120MBit udp simplex on single chain on my setup - this is irelevant. If i use BTest on dual chain it makes over 240Mbit siplex or 117Mbit duplex but again UDP BTest is irelevant. The troughput is different from this.

Tcp is over 80mbps at 5 miles

i use 2x2 mode, but only with 20MHz channel 70Mb/s TCP outdoor, JRC 24dBi duplex are great

i use 2x2 mode, but only with 20MHz channel 70Mb/s TCP outdoor, JRC 24dBi duplex are great

Cool I use the pacwireless 2ft dishes HD version. Little cheaper than what pay. At 23 miles 2x2 I get 150mbps udp.

2 Dallas: You are still talking about theretical troughput readed from some testengine. Try to download some real files trough this link. You will see the difference.

I am talking about actual. I don't go by over the air rates.

I am talking about actual. I don't go by over the air rates.

Actual what? Actual BTest on Mikrotik or real data download using this link ?
How you achieve 150Mbit on 100Mbit ETH interface ?

I am talking about actual. I don't go by over the air rates.

Actual what? Actual BTest on Mikrotik or real data download using this link ?
How you achieve 150Mbit on 100Mbit ETH interface ?

I did bandwidth test udp half duplex via the wireless card. We are only talking about 802.11n, not the throughput of the board going through the ethernet port.

I am talking about actual. I don't go by over the air rates.

Actual what? Actual BTest on Mikrotik or real data download using this link ?
How you achieve 150Mbit on 100Mbit ETH interface ?

I did bandwidth test udp half duplex via the wireless card. We are only talking about 802.11n, not the throughput of the board going through the ethernet port.

I think what he is getting at is that the built in BW test isn't the most accurate test method. You really need to set up two PC's on either end of the link and run an IPperf test across it to get accurate link capacity.

Yeah I understood. I only tested what the forum topic is about.

This topic is talking ebout difference between real troughput using chan0 or chain0+1.

I was already talking with Microtiks representatives about big difference between BTest on Wireless cards, and the real troughput.

So again:

40MHz - chain0
BTest Ncard - Ncard BTest
i get 127Mbit UDP

40MHz - chain0+1
BTest Ncard - Ncard BTest
i get 212Mbit UDP


40MHz - chain0
WWW - Mikrotik+Ncard - Ncard+Mikrotik - PC
i get 68-86Mbit TCP

40MHz - chain0+1
WWW - Mikrotik+Ncard - Ncard+Mikrotik - PC
i get 38-65Mbit TCP


So i see better perfomance using chain0+1 just on the unrealistic Btest, but the real performace is lacking speed against just chain0 alone.

Sorry, I can't help you. I use nothing but routerboards.