Community discussions

MikroTik App
 
offbyone
just joined
Topic Author
Posts: 3
Joined: Sat Jan 04, 2025 4:01 pm

Unstable connection with pair of nRAYG60ad

Fri Jan 17, 2025 3:33 pm

I'm new to wireless links and installed this one partially for learning purposes.
I bought a pair of nRAYG60ad dishes and installed them in 20 meters distance to connect two homes with another and share the internet connection.
It works fine with non-realtime applications like web browsing and watching videos, speedtests with iperf3 indicate gigabit speeds between the houses.

Just online gaming seems to not work properly all the time. At some point the suddenly
When using the built-in bandwidth test of the dishes, the output shows lost packets between 3000 and 15000 each second at around 1 gigabit.
When running the aiming tool the TX sector is constantly changing.
I thought about interference with the dish for satellite TV, mounted on the same pole in a more or less 90 degrees angle. Moving the nRAY so the TV dish is now "behind" and not "beside" did not help.
Is 20 meters too close for stable selection of TX/RX sectors?

Do you have any idea how to find the cause(s) of the instability? I read somewhere that a person "forced the TX sector to center", how would I forcibly set a specific sector to prevent it from constantly changing?
Thank you in advance!
 
ConnyMercier
Forum Veteran
Forum Veteran
Posts: 777
Joined: Tue Dec 17, 2019 1:08 pm

Re: Unstable connection with pair of nRAYG60ad

Sat Jan 18, 2025 11:04 pm

Not a Wireless expert, but I did some Tests in my Lab
I used a Mirkotik Wireless-Wire between two Mikrotik hEX S that generate Traffic on the LINK.
Changed the Distance between the Access-Points and monitored the LINK

At a distance smaller then ~20cm, the connection was horrible (tx-packet-error-rate up to 95% and fluctuate a lot)
Tested distances up to 8 Meters, connection was O.K with a tx-packet-error-rate up to ~ 10%
But I was able to reduce down to ~ 1% , by no placing the Access-Point directly in front of a Hard-Surface (e.g. Wall,Windows)

Again, not an Expert and only did a few test
Maybe somebody else on the Forum can comment on this..


Regarding your Question to Disable beamforming and locks to selected radiation pattern
https://wiki.mikrotik.com/Manual:Interface/W60G
/interface/w60g/set tx-sector=(integer [0..63])
 
jaclaz
Forum Guru
Forum Guru
Posts: 2260
Joined: Tue Oct 03, 2023 4:21 pm

Re: Unstable connection with pair of nRAYG60ad

Sun Jan 19, 2025 12:41 am

A device that Is intended for links up to 1500 m or more cannot be very good at short distances, generally speaking.

I like to imagine that the two devices, which have very good hearing, are shouting at each other at the top of their lungs.

At 20 meters alignment Is a non-issue, you turn the device in the general direction of the other and that is enough, compare with:
viewtopic.php?t=205283

There may be settings that can be adjusted, which Ros version are they running on?
 
User avatar
sirbryan
Member
Member
Posts: 421
Joined: Fri May 29, 2020 6:40 pm
Location: Utah
Contact:

Re: Unstable connection with pair of nRAYG60ad

Sun Jan 19, 2025 5:11 pm

I bought a pair of nRAYG60ad dishes and installed them in 20 meters distance
That could be contributing to the problem, but I doubt it. The Qualcomm radios (in my experience) seem to adjust power automatically towards the -60's (you can't control that at all). What's the signal level? Closer to the -40's, -50's or -60s? If it's in the -70's, you've got a problem.

I have 60GHz links working in the low -30's and -40's and work just fine. And lock the channel down to one of them instead of setting "auto". 60480 and 62640 have the most oxygen absorption loss, but at 20m that doesn't come into play.

That said, these particular radios are best for a couple hundred to a thousand meters or so. If you were to do it again, you would be better off with a pair of Cube60's or even wAP 60's (which might be discontinued, but could be had as distributors clear out their inventory).
When using the built-in bandwidth test of the dishes, the output shows lost packets between 3000 and 15000 each second at around 1 gigabit.
MikroTik's UDP tests always show packet loss. The CPU is cramming as much as it can towards the other end. A TCP test would be a bit more accurate. 1Gbps sounds correct.
When running the aiming tool the TX sector is constantly changing.
These radios are always beam forming, and at such a close distance, they're probably aligning to account for reflections.
I thought about interference with the dish for satellite TV, mounted on the same pole in a more or less 90 degrees angle. Moving the nRAY so the TV dish is now "behind" and not "beside" did not help.
Satellite signals are much lower on the band; they're not going to interfere with 60GHz.
Is 20 meters too close for stable selection of TX/RX sectors?
No. I would try aligning them perfectly at each other first, and report the signal levels. If that's not stable, purposely align them to be a little bit off center, until the signal is in the -50's. You'll have to experiment.
Do you have any idea how to find the cause(s) of the instability? I read somewhere that a person "forced the TX sector to center", how would I forcibly set a specific sector to prevent it from constantly changing?
That's generally useful for alignment, but you don't want that locked all the time to the center. 60GHz is like a laser, and the beam forming helps correct for imperfect alignment.
 
DenisPDA
Frequent Visitor
Frequent Visitor
Posts: 77
Joined: Tue Sep 04, 2018 5:42 pm
Contact:

Re: Unstable connection with pair of nRAYG60ad

Mon Jan 20, 2025 12:27 pm

When running the aiming tool the TX sector is constantly changing.
I thought about interference with the dish for satellite TV, mounted on the same pole in a more or less 90 degrees angle. Moving the nRAY so the TV dish is now "behind" and not "beside" did not help.
Try locking on the center sector and aligning (27, 28, 35, 36)
viewtopic.php?t=199599
 
offbyone
just joined
Topic Author
Posts: 3
Joined: Sat Jan 04, 2025 4:01 pm

Re: Unstable connection with pair of nRAYG60ad

Mon Jan 20, 2025 8:03 pm

Thank you for your answers!

https://wiki.mikrotik.com/Manual:Interface/W60G
/interface/w60g/set tx-sector=(integer [0..63])
Found the command here too: https://help.mikrotik.com/docs/spaces/R ... 59501/W60G (The docs seem to be more recently updated than the wiki pages)

There may be settings that can be adjusted, which Ros version are they running on?
They run Version 6.49.17, which seems to be the latest on stable channel. They got license level 3.

What's the signal level? Closer to the -40's, -50's or -60s? If it's in the -70's, you've got a problem.
I have 60GHz links working in the low -30's and -40's and work just fine. And lock the channel down to one of them instead of setting "auto". 60480 and 62640 have the most oxygen absorption loss, but at 20m that doesn't come into play.
I did
interface w60g monitor wlan60-1
on both ends with the following output:
From the antenna on the roof:
             connected: yes
             frequency: 58320
        remote-address: F4:1E:__:__:__:__
                tx-mcs: 8
           tx-phy-rate: 2.3Gbps
                signal: 95
                  rssi: -51
             tx-sector: 42
        tx-sector-info: right 0.4 degrees, up 0.2 degrees
              distance: 21.97m
  tx-packet-error-rate: 0%
For the antenna on the wall:
             connected: yes
             frequency: 58320
        remote-address: F4:1E:__:__:__:__
                tx-mcs: 8
           tx-phy-rate: 2.3Gbps
                signal: 95
                  rssi: -55
             tx-sector: 62
        tx-sector-info: right 0.8 degrees, up 0.6 degrees
              distance: 21.88m
  tx-packet-error-rate: 0%
They are using the lowest channel and don't seem to switch channels at any time. May changing the frequency help with stability?
The signal strength seems to be okay too.

MikroTik's UDP tests always show packet loss. The CPU is cramming as much as it can towards the other end. A TCP test would be a bit more accurate. 1Gbps sounds correct.
Okay, TCP tests show around 700Mbit/s steady when only sending or receiving and around 250/250 fluctuating when sending in both directions. I assume that is to be expected due to the TCP overhead?

Satellite signals are much lower on the band; they're not going to interfere with 60GHz.
I didn't mean interference between the signals, but the possible interference with reflections of the nRAYs own signal off the back of the TV dish.

But I was able to reduce down to ~ 1% , by no placing the Access-Point directly in front of a Hard-Surface (e.g. Wall,Windows)
Well, as one dish is mounted on the wall that could introduce interference with reflections I guess. But I cannot exactly change the location of it anymore.

While playing with frequencies, I observed that lower MCS correlates with lower bandwidth but also lower TX error rates. I do not understand the term MCS deeply yet, will take a look at this tomorrow. I will try to align the antennas more precisely later this week and report back then.
 
User avatar
sirbryan
Member
Member
Posts: 421
Joined: Fri May 29, 2020 6:40 pm
Location: Utah
Contact:

Re: Unstable connection with pair of nRAYG60ad

Mon Jan 27, 2025 5:54 pm

They run Version 6.49.17, which seems to be the latest on stable channel. They got license level 3.

There are some incremental fixes in various versions of 7, including more in 7.18b2. I wouldn't necessarily recommend 7.18b2 yet, but up to 7.12.1 should work.

They are using the lowest channel and don't seem to switch channels at any time. May changing the frequency help with stability?
The signal strength seems to be okay too.

Yes, those are excellent numbers. 2.3Gbps on the PHY rate, MCS 8, 95% signal (quality I believe). With only -55 at 21m on those larger dishes, they are backing off TX power automatically. The decimal degrees on alignment (they use beam forming) means you're pretty much spot on. No alignment improvements should be necessary, since they'll do what they can to correct any changes you make at this point.

That said, if you're seeing instability on traffic traversing the links, it is likely due to reflections (multipath). Since they are so close, and you're on a wall, there's not a lot of space for the off-axis waves to dissipate and not get detected. In other threads on the forums about 60GHz gear it has been discussed that placing the radios too close to reflective surfaces can have unexpected (often undesired) effects on operation.

Okay, TCP tests show around 700Mbit/s steady when only sending or receiving and around 250/250 fluctuating when sending in both directions. I assume that is to be expected due to the TCP overhead?

If that's radio-to-radio, you're running into CPU limitations. If you run the tests on the networking gear connected to the radios, that will give you a more accurate picture of what to expect. But since these are still WiFi-based radios, you're going to have some reduced throughput trying to send + receive decent amounts at the same time.

While playing with frequencies, I observed that lower MCS correlates with lower bandwidth but also lower TX error rates. I do not understand the term MCS deeply yet...

MCS means Modulation and Coding scheme for 802.11-based gear. As the radios adjust to conditions, they'll generally start at a lower rate and work their way up. Too many errors or not enough traffic and they automatically back off.
 
offbyone
just joined
Topic Author
Posts: 3
Joined: Sat Jan 04, 2025 4:01 pm

Re: Unstable connection with pair of nRAYG60ad

Mon Feb 03, 2025 1:22 pm

A bit later than I expected, but I will share my findings now.
The problem with unstable connections was probably caused by a router in between the gamer's PC and the antenna. After omitting the router and using a simple switch instead, the instabilities were essentially gone. Maybe another cause were server side problems of the game itself.

But to be clear, having these antennas on 20 meters distance works pretty decently. Home office with remote desktop also works.
I am still happy to have acquired more in-depth knowledge due to a pseudo-problem.

Thank you again for your answers!