SXT 5 AC - Wireless link issues at 1,8 km

RouterOS Wireless Networking
1

Hello everyone,

I have problems with a link between two SXT 5 AC, I have tried different configurations but the most stable one is with 802.11 in WPA2;

I also tried with NV2 and NSTREME but the link is not successful, also the maximum speed I reach is 50/50 Mbps

With firmware 7.2.3 the link is very unstable, while using version 6.48.6 it is much better.

I am attaching the configuration for firmware 6.48.6 if someone can give me useful advice to improve the link in order to have the maximum possible bandwidth.

6.48.6 - Bridge : https://pastebin.com/xqJwid0q
6.48.6 - Station Bridge : https://pastebin.com/DcD2nNP5

Bandwidth Test:
Schermata 2022-06-24 alle 11.42.39.png
Thank you.

2

What is CPU load during bandwidth test? That is only single core cpu, it cant generate much traffic, use PC to do tests.

3

In idle, 3 -9 %, during the test 98-100%; currently i don’t have a pc on the other side but i can connect it and test via iperf

4

That’s the problem with your testing procedure. So yes, do the test using iperf through the wireless link.

According to radio stats it should be possible to push at least 150Mbps, 300Mbps is probably unreachable.

5

You are measuring the traffic that the SXT processor can generate. You measure it in the most processor-heavy TCP mode in both directions. Measure in UDP separately receive and transmit you get the highest possible speeds. You should not run the traffic generator on the device you are testing, the results will be biased.

6

Change the radio to 5GHz-only-AC.
NV2 is not optimal for these devices.

7

ok I did some tests but honestly the performances are very poor
config-screen2.jpg
config-link.png
tcp-test.png
udp-test.png

8

You are testing both ways and over TCP, for that device and 2km its very good result…

Try testing one way only, and also try test UDP traffic, its not only TCP that will go thru that link…

I got very same APs and at same distance, i get 450mbit UDP in one direction(around 170mbit both ways same time)
Also when i test TCP both ways i get 50/50mbit same as you, but one way its around 150mbit.

9

Have you missed the remarks regarding your test procedure? Don’t use bandwidth test, use iPerf to (only) test network speed.

10

Exactly

and 802.11 is not full duplex, is half duplex… when tx can not rx and vice-versa.

TCP 50+50 = 100 and for a 2km link with CCQ at 60%/70% is very good…

11

okay, I’ll do further tests on this, otherwise I thought I’d replace them with the “LHG XL 5 ac” but only if you can have at least 400Mbps (200 + 200)

According to the scheme, the SXTs reach 1.5 km instead the LHGs at 20km for 802.11ac “866Mbps” but I have seen that the results are much lower so I am perplexed about it
antenas-160404123306.jpeg
lhg_distamce-607fd12378a3e-1570x898.jpeg

12

Probably there are some reason if the dynadish cost more money than that lhg model…

13

the price from the supplier between dynadish 5 and lhg xl 5 ac, is really minimal, only 30 euros of difference

compare all three:

DynaDish 5 - 5GHz, 25dBi, TX Power 31dBm, 1xGLAN, 802.11a / n / ac, RouterOS L3

LHG XL 5 ac - 5GHz, 27dBi, TX Power 25dBm, 1xGLAN, 802.11ac, RouterOS L3

SXT 5 ac - 5GHz, 13dBi, TX Power 31dBm, 1xGLAN, 802.11aс, RouterOS L4


For the moment I will try to make the SXT work better, then eventually I will evaluate the other two alternatives

14

ignoring this (jus for compare only the other two)
SXT 5 ac - 5GHz, 13dBi, TX Power 31dBm, 1xGLAN, 802.11a/n/ac, RouterOS L4

Regardless your country limits:

DynaDish 5 - 5GHz, 25dBi, TX Power 31dBm, 1xGLAN, 802.11a/n/ac, RouterOS L3
25 + 31 = 56

LHG XL 5 ac - 5GHz, 27dBi, TX Power 25dBm, 1xGLAN, 802.11a/n/ac, RouterOS L3
27 + 25 = 52

56 - 52 = 4

and very approx +3 = double tx power on logarithmic scale…
4 = more than the double

DynaDish can transmit more than double power,
LHG can receive (approx) the double better than DynaDish… But also, for this, receive better the interferencies…


But remember: do not use for compare the 6Mbit/s power,
for compare use, for example MCS9…
22 on DynaDish and 18 on LHG

15

I can spare you all the trouble, getting longer range aps won’t get you much benefit, you will have about same speeds at those distance, they only perform better at bigger distances..

16

Try to check the channel noise, maybe lower or higher channel ?

17

I have the same problem, model 922UAGS-5HPacD with mipsbe and atheros ar9888 chipset.

Already in another ptp with the model RBDiscG-5acD with arm and chipset IPQ4019 works perfectly.

I believe that the problem is there, related to the radio architecture or the chipset.