RB922 Excellent speeds of 535 Mbps! Can you go faster?

RouterOS Wireless Networking
1

I just installed some Mikrotik RB922s with 2.4 Mini PCI and 5 ac built in as wireless APs. I had a play about with bandwidth tester and got excellent speeds of 535 Mbps. See image.

Running on CAPsMANv2 but for test switched to standalone.

Key Info:

80 wide channel - Ceee
FW:3.22
ROS 6.28
3 metres apart
256 QAM
802.11ac

Anybody gone faster? :sunglasses:
802.11ac speed.png

2

How can them be installed if are “3 metres apart” ?

3

test lab. Just playing about to see what i get!

4

good luck outside :smiley:

5

Yep! in the real world and with TCP I will get nothing like it! Buts it’s interesting and fun to have a mess about.

6

Yep I can :slight_smile:
MT_AC_test_8_11.jpg

7

Fake.
Not Routerboard.

8

It;s not a fake. It;s 3x3 connection in lab.
With 2x2 you can get:
hdx 550Mb.png

9
10

Battling the elements is another matter. It is fair to do lab comparison between equipments to see their theoretical maximum.

People test cars for their acceleration speed, but nobody says “good luck in traffic jam”. Everybody knows the situation on the road.

11

About how much TCP throughput can someone expect with 2x2 802.11ac in real-world outdoor installations with NV2? Could someone give me a raw figure?

12

Agree.

At least 200mbps TCP but I don’t think more than 300. In an 24km link I had 185mbps. It’s not much but is more than RocketM5 that had ~50mbps. It is a heavy loaded tower and interference.

Here is an comparison but note that is made by an UBNT “fanboy”. Some of them sucks hard but not quite all.
http://community.ubnt.com/t5/airMAX-Stories/Radio-Shootout-Pt-2-let-s-try-a-whole-bunch-of-them/cns-p/1232309#comments

13

Thanks! Let’s hope to see some improvement (if there is any room for it in the software department)…

14

Today I installed my two new NetMetal 5 on my rather short p2p link to interconnect two houses on different ends of a road. About 100 m distance with clear LoS. Each tiny tower is equipped with 3x 19 dBi panel antennas (Elbox TetraAnt). The polarization of the antennas is H/V/H. I intend to mainly use 2 chains. Enabling the third chain does not give any improvement at all. The separation of 30 cm between the two horizontal antennas is not enough to create separate spartial streams. I setup ac with 80 MHz (Ceee) and according to MikroTik Freq. Usage all four used 20 MHz channels are clean (max. 0.2 usage). Tx power is reduced to 6 dBm. Unfortunately RB922 does not allow lower values.

My results: UDP throughput ~ 450 Mbit/s, TCP throughput ~ 150 Mbit/s.
iperf only gives ~ 100 Mbit/s with spikes between 70 and 170 Mbit/s.

Honestly I expected a bit more from the new ac hardware.
A net throughput of 100 - 150 Mbit/s should have also been possible with n standard.
So where is the practical benefit of doubling the channel witdh and 256 qam?
In theory both should be able to also increase (more than double) our ac bandwidth compared to n.

Any hints what I can improve?

Screenshots of status and benchmark are attached.
Here is my wifi config:

/interface wireless
set [ find default-name=wlan1 ] antenna-gain=17 band=5ghz-onlyac bridge-mode=\
    disabled channel-width=20/40/80mhz-Ceee country=germany \
    default-authentication=no default-forwarding=no dfs-mode=no-radar-detect \
    disabled=no distance=1 frequency=5ghz-80mhz-outdoor-ch116 frequency-mode=\
    regulatory-domain hide-ssid=yes hw-retries=15 l2mtu=1600 mac-address=\
    [...] max-station-count=1 mode=ap-bridge nv2-cell-radius=10 \
    nv2-preshared-key=[...] \
    nv2-security=enabled radio-name=[...] rx-chains=0,1 scan-list=\
    5ghz-80mhz-outdoor security-profile=[...] ssid=[...] tx-chains=0,1 \
    tx-power=6 tx-power-mode=all-rates-fixed wireless-protocol=nv2
/interface wireless channels
add band=5ghz-onlyac extension-channel=Ceee frequency=5580 list=\
    5ghz-80mhz-outdoor name=5ghz-80mhz-outdoor-ch116 width=20
15

And here is a screenshot of the iperf test. (I wasn’t able to attach more than 3 pics in one post.)

16

And last but not least, here is a test with the same equipment but just using 802.11n instead of 802.11ac.

bwtest: UDP 200 Mbit/s, TCP 140 Mbit/s.
iperf: 110-140 Mbit/s.
As you can see in the attached iperf, this run looks much cleaner and more stable. Average iperf throughput is also ~20% better.

So it looks like there is something broken with 802.11ac implementation at the moment (at least for me).
UDP bwtest looks promising. But more practical tests with TCP (bwtest and iperf) are not a bit better with ac; and often even worse compared to n.

17

I have been trying to set my routerboard 951G according to the specs posted by dfroe but its saying “no support channel” and “scan-list does not contain valid channels”.

18

This topic was about 802.11ac with RB922 boards.
You mentioned a 951G which has a 2.4 GHz 802.11n radio.
This is completely different hardware, different standards, different frequencies, and so on.
Of course my 5 GHz 802.11ac configuration will not work on your 2.4 GHz 802.11n device.
So I guess you’ll better open a new topic regarding your 951G / 802.11n issues.