MikroTik hAP ax3 poor WiFi performance

skocdopolej · Wed Jan 18, 2023 6:24 pm

I just set up hAP ax3 and noticed significantly worse WiFi performance compared to hAP ac3.

I did some tests using iperf3 and hAP ax3 had 30% worse performance. Indicated signal strength (on Android phone) was on average 20% worse on ax3. Test details are in attached file.

Does anyone have the same experience?

ac3 (ROS 7.7):

/interface wireless
set [ find default-name=wlan1 ] band=2ghz-onlyn channel-width=20/40mhz-XX \
    country="czech republic" disable-running-check=yes disabled=no distance=\
    indoors frequency=auto installation=indoor mode=ap-bridge ssid=2G \
    wireless-protocol=802.11 wps-mode=disabled
set [ find default-name=wlan2 ] band=5ghz-onlyac channel-width=\
    20/40/80mhz-XXXX country="czech republic" disable-running-check=yes \
    disabled=no distance=indoors frequency=auto installation=indoor mode=\
    ap-bridge skip-dfs-channels=all ssid=5G wireless-protocol=802.11 \
    wps-mode=disabled

ax3:

interface/wifiwave2/export 
# jan/18/2023 17:22:52 by RouterOS 7.7
# software id = HGT8-C7QW
#
# model = C53UiG+5HPaxD2HPaxD
/interface wifiwave2 security
add authentication-types=wpa2-psk,wpa3-psk encryption=ccmp,gcmp,ccmp-256,gcmp-256 name=default wps=disable
/interface wifiwave2 configuration
add country=Czech mode=ap name=default security=default ssid=W
/interface wifiwave2
set [ find default-name=wifi1 ] channel.band=5ghz-ac .skip-dfs-channels=all .width=20/40/80mhz comment="WiFi 5 GHz" configuration=default configuration.mode=ap disabled=no
set [ find default-name=wifi2 ] channel.band=2ghz-n .skip-dfs-channels=all .width=20/40mhz comment="WiFi 2.4 GHz" configuration=default configuration.mode=ap disabled=no

kowal · Wed Jan 18, 2023 7:05 pm

I have ac3 and ax3, both on ROS7.7 and wifiwave2( ac3 in diffrent location),

Ax3 performance on 5GHz:
Client : Xiaomi Mi10T Pro : ~600-700Mb/s,
Client : HP Probook 450 G9 with Realtek 8852BE, Ubuntu 22.04 LTS : 800-900Mb/s

But I will swap this Realtek card to Intel AX211 because of driver issues on Linux.

Ac3 performance :

Client : Xiaomi Mi10T Pro : ~350-400Mb/s
Client : HP Probook 450 G9 with Realtek 8852BE, Ubuntu 22.04 LTS : ~500-560 Mb/s

skocdopolej · Wed Jan 18, 2023 8:35 pm

Hi kowal,

thank you for your response.

Are we talking about link speed or real speed?

miononno · Wed Jan 18, 2023 10:23 pm

Talking about range and not speed..

I'm comparing the new AX3 and an old Asus 86U(wifi5) and the mikrotik has lower range..

when there's only 2.4ghz network (because i'm too far from AP) the Asus still gives me a little signal, while the AX3 has zero signal.

Is there any way i can raise the power? I'm using Italy as country and all the other settings stock.
Thank you

Rox169 · Thu Jan 19, 2023 9:45 am

Hi,

probably the old ASUS do not care about new wifi regulation and use maximum power of radio. I have very good experience with ASUS and with they MESH system I do recomend asus rt-ax56u.

You can change the country for better signal, but do not expect miracles. And be careful router can use specific 5 ghz frequency and some clinets/device can have problem when they are set to czešch country.

miononno · Thu Jan 19, 2023 10:12 am

What the whole preceding post was quoted for? Is it a problem to just use "Post Reply" button?

I see.
I will try Czešch country and then i will decide what to do.
I love routerboard inside, so maybe i will disable wifi6 and leave wifi to be managed by the old one as a dumb AP only.

Thu Jan 19, 2023 11:52 am

Be careful when changing country settings to something which may be considered illegal for your country.
Just saying ...

Thu Jan 19, 2023 11:59 am

Ac3 performance :

Client : Xiaomi Mi10T Pro : ~350-400Mb/s
Client : HP Probook 450 G9 with Realtek 8852BE, Ubuntu 22.04 LTS : ~500-560 Mb/s

Adding my AC3-performance using iperf3 on internal server:
Client: Samsung S20 -> 450 - 480Mb/s
Client: Custom Laptop using Realtek PCI GbE Family Controller Win11 -> 970Mb/s

skocdopolej · Thu Jan 19, 2023 12:34 pm

Hi holvoetn,
which app are you using for iperf3 client on Android?

Thu Jan 19, 2023 12:58 pm

HE.NET
Works pretty well. Has quite some other tools embedded as well.

skocdopolej · Thu Jan 19, 2023 1:41 pm

Great app indeed.

Thank you!

mkx · Thu Jan 19, 2023 3:14 pm

Client: Samsung S20 -> 450 - 480Mb/s

Does S20 support more than 1 chain (MIMO)? Did you ever observe it using 2S (in wireless registration table) while performing throughput test? I have a few android devices (phones and tablets) and none actually support 2 chains (2S) ... which logically reduces throughput quite severely.

miononno · Thu Jan 19, 2023 3:16 pm

I did some other test.. but it's clearly less powerfull, talking about RF emitted

GREEN MARKED 0_MAIN is the SSID of ASUS 86U (wifi5)
BLUE MARKED 1_MAIN and 1_MAIN5 are SSID of HAP AX3

i tried moving around and Asus signal is always better.
I hope i'm doing something wrong..

Thu Jan 19, 2023 3:23 pm

Does S20 support more than 1 chain (MIMO)? Did you ever observe it using 2S (in wireless registration table) while performing throughput test? I have a few android devices (phones and tablets) and none actually support 2 chains (2S) ... which logically reduces throughput quite severely.

Don't know for sure. I will have to test again somewhere this weekend.
But from the spec-pages I get this:

Wi-Fi
802.11 a/b/g/n/ac/ax 2.4G+5GHz, HE80, MIMO, 1024-QAM

On some places I also saw the Wifi6 symbol already

erlinden · Thu Jan 19, 2023 3:29 pm

You are using 40MHz bandwidth on the 2.4GHz radio of the MikroTik while the Asus is 20MHz. That will make a huge difference, especially looking at the saturation of the 2.4GHz band. Next, if you want to compare you should use the same settings (bands) on both devices (ofcourse not at the same time).

Rox169 · Thu Jan 19, 2023 3:47 pm

As I wrote you asus probably do not care about new regulation. I fully agree with erlinden if you are comparing 20Mhz with 40Mhz you probably do not know what are you doing

You must compare router on the axact place, with the axact same Mhz and on exact same band and of course one by one...not both at the same time...

Signal -37 is better/stronger than -47 on the 2,4Ghz and the signal -59 is bit worse than -58 on 5,4Hgz. You yould also try the throughput if you have something like ipref or openspeedtest server on your local network...Worse signal on wifi 5,6Ghz AX will have better throughput than better signal on 5,4Ghz AC

mkx · Thu Jan 19, 2023 3:56 pm

Wi-Fi
802.11 a/b/g/n/ac/ax 2.4G+5GHz, HE80, MIMO, 1024-QAM
On some places I also saw the Wifi6 symbol already

Yeah .. but ax can work with single chain as well ... and will perform slightly better than ac with same single chain due to better modulation scheme available (1024QAM vs. 256QAM). Device manufacturers tend to pick lowest possible denominator when displaying indicators of "advanced" functionality.

miononno · Thu Jan 19, 2023 4:20 pm

Signal -37 is better/stronger than -47 on the 2,4Ghz and the signal -59 is bit worse than -58 on 5,4Hgz. You yould also try the throughput if you have something like ipref or openspeedtest server on your local network...Worse signal on wifi 5,6Ghz AX will have better throughput than better signal on 5,4Ghz AC

yes i know that -37 is higher..

just look at the graph, not at a specific moment.

I don't care about throughput. When i need throught i use Ethernet, not wifi.
What i need is coverage for all the devices. ( an simple IOT device need just to connect)
I simply cannot unplug the Asus and plug the HAP, because some device will be completely offline.

However, maybe it's because of new EU laws about emissions..
My posts are not intended to just say the HAP is not good. I'm just saying it gives less coverage or maybe i'm doing a mistake, so i appreciate any comment toward the solution.

I will try to put them in same conditions..
Thank you

erlinden · Thu Jan 19, 2023 4:31 pm

I'm just saying it gives less coverage or maybe i'm doing a mistake, so i appreciate any comment toward the solution.

Yes...it's a mistake on your site. Please read my previous post and change config accordingly.

miononno · Thu Jan 19, 2023 4:40 pm

take on your site. Please read my previous post and change config accordingly.

Let's talk about 2.4ghz..

I try to put channel width to 20Mhz.. ok?
Thank you for your help

mkx · Thu Jan 19, 2023 5:58 pm

20MHz channel is good for coverage.

What country regulations limit is effective radiated power per device. Effective means it takes into account antenna gain so high antenna gain on transmitter will not present as higher signal strength[*]. Per device means that it'll sum over all transmit antennae and using many antennae (e.g. 4x4 MIMO or higher) will present as lower signal strength[**]. It will also mean that it will sum power per whole channel width. With wider channels power is dispersed more and signal strength will again be lower. So for best coverage use 1-chain AP with 20MHz channel width.

[*] higher gain antenna does help to improve reception. Remember that obtainable throughput is related to SINR of receiver. Signal in receiver is sum of transmitter power (power amplifier), transmitter antenna gain, path loss and receiver antenna gain. Sum of first two is more or less EIRP and thus limited by country regulation, path loss is property of physical space between transmitter and receiver and receiver antenna gain is something that can be improved on certain devices. Higher gain antenna, by being directional (more or less), can also help to reduce noise part of SINR, but amount of reduction highly depends on overall situation around receiver.
When using higher gain antenna on AP downlink doesn't change much (if at all): coverage is the same, service is the same (or slightly better; power amplifiers in low cost WiFi APs have "power backoff" meaning that when transmitting higher MCS/VHT, the power capability is lower. If country regulations limit Tx power lower than PA capability, then real power backoff happens at higher VHT or even doesn't happen). However higher gain antenna helps in uplink, upload will be better because AP "hears better". In case of PtP links, using higher fain antennae on both ends help improve both directions.
The power backoff explains why maximum throughput can be same on two APs with different signal strengths, specially if Tx power is set for maximum coverage area.

[**] using 4x4 (and higher order) MIMO offers some features which give better service outside best area of AP coverage (beam forming, space diversity). However, beacons are transmitted without using those tricks because they are intended also towards "about to register" clients (which are thus not yet known to AP) and towards dormant clients (which might have moved relatively to AP's location but without gaving active two-way connection AP can not determine direction towards such client). So improvement is only seen during actual data transfers allowing for higher datarates at the same (marginal) signal strength.
The same applies to broadcast/multicast traffic if multicast helper is not enabled.

miononno · Thu Jan 19, 2023 6:50 pm

Thank you very much for this comprehensive answer.
I'll try as soon as possibile.

Thu Jan 19, 2023 6:57 pm

Does S20 support more than 1 chain (MIMO)? Did you ever observe it using 2S (in wireless registration table) while performing throughput test?

Small problem ... in wave2 wireless registration table nothing is to be found anymore regarding 1S or 2S.
It does indicate 866/650 for tx/rx ?

mkx · Thu Jan 19, 2023 8:09 pm

866 can be result of different connection properties ... single chain, 80MHz and 256QAM. Or dual chain and 40MHz and 256QAM. Probably @bpwl could deduct more.

pabelo76 · Mon Jan 23, 2023 11:10 pm

I have ac3 and ax3, both on ROS7.7 and wifiwave2( ac3 in diffrent location),

Ax3 performance on 5GHz:
Client : Xiaomi Mi10T Pro : ~600-700Mb/s,
Client : HP Probook 450 G9 with Realtek 8852BE, Ubuntu 22.04 LTS : 800-900Mb/s

But I will swap this Realtek card to Intel AX211 because of driver issues on Linux.

Ac3 performance :

Client : Xiaomi Mi10T Pro : ~350-400Mb/s
Client : HP Probook 450 G9 with Realtek 8852BE, Ubuntu 22.04 LTS : ~500-560 Mb/s

Can you show me your configuration? I have transfers at 379 Mbits/sec (AX210 card)

theprojectgroup · Sat Jan 28, 2023 11:58 pm

Same here with hAP ax3.

Client MacBook Air M2 (AX) iperf3 sending > WiFi1 (5G) + ether1 in one bridge > ether1 > nuc8 350-750 Mbit/s (quiet unstable...)

Client MacBook Air M2 (AX) iperf3 receiving < WiFi1 (5G) + ether1 in one bridge < ether1 < nuc8 70-120 Mbit/s (quiet unstable...)

Client MacBook Air M2 (AX) iperf3 receiving < WiFi1 (5G) < ether5 pppoe < fibre nuc8 35 Mbit/s (stable, WAN 100% saturated)

What?

If the traffic is routed and natted its fast(er) and stable

If the traffic is bridged it's unstable and it's really slow if WiFi of ax3 is sending

Sun Jan 29, 2023 12:37 am

ax3
Same samsung s20.
Reliably around 600mbs

EgidijusL · Mon Jan 30, 2023 1:46 am

with v7.8 speed sucks... ant wifi stability... Downgrade to v7.6 and speed return.
v7.6 iphone 12 pro max ~800Mbs
v7.7 iphone 12 pro max ~400Mbs

Mon Jan 30, 2023 6:16 am

Especially for you I downgraded to 7.6, also went over 7.7 again.
I keep seeing the same figures with Samsung S20.
Time to ditch Apple, then ?

In all seriousness, it could be device specific.
There have been brand-related issues in the past with wifi (was related to security IIRC)

wiseelf · Wed Feb 01, 2023 1:01 pm

I have the same issue after upgrading to 7.7. TX Rate on 5Ghz went down from 1200Mbps to 800-400. Later the same day any of my devices weren't able to see 5Ghz network at all.
Downgraded to 7.6 and everything works well now.
Opened ticket to support, let's see what they will answer.

Device is hap ax3.

kenyloveg · Sat Feb 04, 2023 7:54 am

I had poor speed (35MB/s files copy from lan NAS) when 2.5Gbps was bridged with wireless.
Speed did go up to download=85MB/S upload=35MB/s after remove 2.5G port from bridge. I suspect this may be about default TX RX flow control (all off on interface).
My laptop wireless card is AX200, 1 wall between laptop and ax3, distance is about 4 meters.

bpwl · Sun Feb 05, 2023 1:12 am

866 can be result of different connection properties ... single chain, 80MHz and 256QAM. Or dual chain and 40MHz and 256QAM. Probably @bpwl could deduct more.

Just reverse lookup is my tool. It is silly MT is not giving the input (bandwith or #tone, MCS HT/VHT/HE index, GI, #S), and so one has to match the rate with possible executions.

866.7 is 6 times in this table https://mcsindex.com/ (using ctrl-F text search) , some pre-802.11ax, some 802.11ax (ax has different Guard interval), not all settings are possible with MT (160MHz channel? , 8 antenna )

Mikrotik is adjusting wifiwave2 to the old wlan interface features. Hopefully one day also the registration table is the same (would love this for DUDE "Routeros info" global overview of the registration table, even as it misses MCS index)

derict · Thu Feb 09, 2023 9:19 am

with v7.8 speed sucks... ant wifi stability... Downgrade to v7.6 and speed return.
v7.6 iphone 12 pro max ~800Mbs
v7.7 iphone 12 pro max ~400Mbs

same problem, downgrading hap ax3 to v7.6 after tasting v7.7 and v7.8b3 wifi stability (range and connectivity) gone wrong

and hap ax3 with external HGO (range) is lower than hap ac2 with internal antenna

even run normally with v7.6

is it problem with wifiwave2 package? the old wireless package seems more stable

massinia · Fri Feb 10, 2023 9:02 pm

I just ordered one hAP ax3 me too, hopefully it will be fixed with 7.8

maigonis · Mon Feb 13, 2023 12:35 pm

Pay attention what country and channel is used. Not all channels are allowed to transmit whit hi power. Lower TX power will reduce range and modulation.

theprojectgroup · Wed Feb 15, 2023 10:42 am

Still the same issue on latest RC.

Wed Feb 15, 2023 11:00 am

You may want to take a look at this thread.
viewtopic.php?t=193578&hilit=macbook

User reported Wifi issues on AX2 (same wifi as AX3) using ROS 7.7 yet the problem turned out to be related to Macbook.
Other test device, problem went away.

So once again my suggestion to ditch Apple

In all seriousness, to further pinpoint this problem, do you have the option to use another device for your testing ?

theprojectgroup · Wed Feb 15, 2023 11:29 am

Will Try other Windows AX devices but I don't see why this is mac related.

If the traffic is bridged = bad
If the traffic is NATed = good
So this seems a bridging issue, right?

The AX Macbook Air M2 and MacBook Pro M1Max work great with UniFi AX APs btw.

Wed Feb 15, 2023 12:06 pm

Will Try other Windows AX devices but I don't see why this is mac related.

Different OS, different network stack, different way of handling traffic.

anav · Wed Feb 15, 2023 2:48 pm

I suspect unifi developers all use macs and mt developers all use linux LOL

Wed Feb 15, 2023 3:35 pm

MacOS and Linux are not thát different IMHO.
MacOS is BSD based (closed source), Linux is ... well, Linux (open source). Both are Unix derivates, however.
Problem will be mostly in difference between Mac/Linux and Windows, I guess.

theprojectgroup · Wed Feb 15, 2023 6:13 pm

The issue doesn't exist when doing NAT, only when WLAN < Bridge > ethernert > server.

stark82 · Wed Feb 15, 2023 6:17 pm

I'm facing the very same issue.... running version is 7.7

I have tested the setup by drop in replacement of the existing cheap AP - place, obstacles, distance and devices operating are exactly the same. Environment is the same.
I'm so much disappointed with AX3 performance... jaw dropped... testing stations never reaches more than 300 Mbit/s real speed to NAS or Internet (AP-> 1G LAN ->virtual pfSense router on i5-12600 server, so fast as it can be-> 1G internet) and it never ever negotiates more link speed than 400 Mbit.... it is also very unstable, drops to very low MCS and low speeds (80-108 Mbps)
Interesting thing that this 300 Mbit seem to be some cap as it mostly equal 300 sharp.

My testing laptop and phone (MacBook Pro 2019, 802.11ac 3/MIMO, iPhone 11 which is 802.11ax 2/MIMO) are about 2.5 meters away from the AP in a straight line. Last machine tested was win11 PC with Intel ax 210 nic. This one stands closest to the AP - 50 cm, antena to antena.

Usually, when stations moved away even slightly from straight line, the connection drops parameters even harder, then stations decide to move to ... further AP which is 6-7 meters away behind brick wall and wooden door, it is Asus AX55 (same radio configuration as AX3)
All of this happens in 5Ghz band. When it comes to 2Ghz band, performance is even worse.... devices close to the AX3 like old printer, old AV receiver, some ESP controllers from IoT can't catch the radio and moving to Asus. Unusable even in the same room!

Config - rather simple, but adding channels, channel width etc. gave no gain, so I decided to left it as default as possible but still analogue to other devices I use:

/interface wifiwave2
set [ find default-name=wifi2 ] channel.band=2ghz-ax .skip-dfs-channels=\
    disabled .width=20/40mhz configuration.country=Poland .mode=ap .ssid=\
    hardc0re_2G disabled=no name=wifi-2G security.authentication-types=\
    wpa2-psk,wpa3-psk .encryption=ccmp,gcmp,ccmp-256,gcmp-256
set [ find default-name=wifi1 ] channel.band=5ghz-ax configuration.country=\
    Poland .mode=ap .ssid=hardc0re_5G disabled=no name=wifi-5g \
    security.authentication-types=wpa2-psk,wpa3-psk .encryption=\
    ccmp,gcmp,ccmp-256,gcmp-256 .group-encryption=ccmp .management-encryption=\
    cmac .management-protection=allowed

Channels are different on both APs, for 2 and 5 Ghz bands. There are no other interferences I can notice.

Funny thing that I have replaced .... cheap-of-cheapest Xiaomi AX3000 Mesh system box, which barely have any configuration. Very basic thing. On this one, I got full possible negotiated 802.11ac speeds (866 or 10xx something, but rarely) on MBP, and 1200 on iPhone 11. Xiaomi AX3000 has no external antenas. 2Ghz band was giving full range for older devices too...

It looks like some flaw somewhere - usually, from my experience, the problem might be with hardware radio path eg. some shorting to ground, internal interference, problems with antenas and connectors... just guessing. But this behavior is very strange.
What I have checked seems ok. Received power and noise floor for 5Ghz on the stations seem quite good but 2Ghz are terribly bad and very unstable (jumps from very low to good values constantly)

Any ideas what can be wrong with this device? I'm very very surprised by its poor wifi performance, would never expect that from Mikrotik box.....

theprojectgroup · Thu Feb 16, 2023 3:32 pm

Does this looks like a bridging + TCP + ether1 2.5GbE issue?
UDP works fine.
Still had a chance to access an AX Windows machine but it looks like it's not a platform issue.

Macbook Air M2 < wifi1 < bridge 1 < ether1 (2.5GbE) < Intel NUC8 with Thunderbolt 10 GbE

TCP - iperf3
Up - Good

iperf3 -c 192.168.99.11 -t10  -i1 -P1
Connecting to host 192.168.99.11, port 5201
[  5] local 192.168.99.128 port 63183 connected to 192.168.99.11 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec   104 MBytes   870 Mbits/sec
[  5]   1.00-2.00   sec   104 MBytes   872 Mbits/sec
[  5]   2.00-3.00   sec   102 MBytes   855 Mbits/sec
[  5]   3.00-4.00   sec  94.1 MBytes   791 Mbits/sec
[  5]   4.00-5.00   sec  92.0 MBytes   772 Mbits/sec
[  5]   5.00-6.00   sec  90.6 MBytes   760 Mbits/sec
[  5]   6.00-7.00   sec  90.6 MBytes   760 Mbits/sec
[  5]   7.00-8.00   sec  91.4 MBytes   767 Mbits/sec
[  5]   8.00-9.00   sec  93.0 MBytes   780 Mbits/sec
[  5]   9.00-10.00  sec  91.2 MBytes   766 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec   953 MBytes   799 Mbits/sec                  sender
[  5]   0.00-10.02  sec   950 MBytes   795 Mbits/sec                  receiver

Down - Bad

iperf3 -c 192.168.99.11 -t10  -i1 -P1 -R
Connecting to host 192.168.99.11, port 5201
Reverse mode, remote host 192.168.99.11 is sending
[  5] local 192.168.99.128 port 63209 connected to 192.168.99.11 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  14.8 MBytes   124 Mbits/sec
[  5]   1.00-2.00   sec  15.5 MBytes   130 Mbits/sec
[  5]   2.00-3.00   sec  13.2 MBytes   111 Mbits/sec
[  5]   3.00-4.00   sec  15.6 MBytes   131 Mbits/sec
[  5]   4.00-5.00   sec  15.4 MBytes   129 Mbits/sec
[  5]   5.00-6.00   sec  13.4 MBytes   113 Mbits/sec
[  5]   6.00-7.00   sec  14.7 MBytes   123 Mbits/sec
[  5]   7.00-8.00   sec  14.6 MBytes   123 Mbits/sec
[  5]   8.00-9.00   sec  13.3 MBytes   112 Mbits/sec
[  5]   9.00-10.00  sec  15.3 MBytes   128 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   147 MBytes   123 Mbits/sec  284             sender
[  5]   0.00-10.00  sec   146 MBytes   122 Mbits/sec                  receiver

Down - NAT - Good (ISP: 300Mbit Down)
Macbook Air M2 < wifi1 < NAT< ether5 (1GbE) + PPPOE < FRITZ!Box 7583 G.Fast in house 100 meters < FTTB M-Net ISP

iperf3 -c speedtest.wtnet.de -p 5202 -P 1 -4 -R
Connecting to host speedtest.wtnet.de, port 5202
Reverse mode, remote host speedtest.wtnet.de is sending
[  6] local 192.168.99.128 port 64394 connected to 213.209.106.95 port 5202
[ ID] Interval           Transfer     Bitrate
[  6]   0.00-1.00   sec  33.6 MBytes   282 Mbits/sec
[  6]   1.00-2.00   sec  40.1 MBytes   336 Mbits/sec
[  6]   2.00-3.00   sec  40.0 MBytes   335 Mbits/sec
[  6]   3.00-4.00   sec  40.4 MBytes   339 Mbits/sec
[  6]   4.00-5.00   sec  40.2 MBytes   338 Mbits/sec
[  6]   5.00-6.00   sec  40.2 MBytes   337 Mbits/sec
[  6]   6.00-7.00   sec  40.3 MBytes   338 Mbits/sec
[  6]   7.00-8.00   sec  40.2 MBytes   337 Mbits/sec
[  6]   8.00-9.00   sec  38.7 MBytes   324 Mbits/sec
[  6]   9.00-10.00  sec  38.1 MBytes   319 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  6]   0.00-10.02  sec   424 MBytes   355 Mbits/sec  173             sender
[  6]   0.00-10.00  sec   392 MBytes   329 Mbits/sec                  receiver

UDP - iperf3
Up - Good

iperf3 -c 192.168.99.11 -t10  -i1 -P1 -u -b 1G
Connecting to host 192.168.99.11, port 5201
[  5] local 192.168.99.128 port 56491 connected to 192.168.99.11 port 5201
[ ID] Interval           Transfer     Bitrate         Total Datagrams
[  5]   0.00-1.00   sec   114 MBytes   954 Mbits/sec  82345
[  5]   1.00-2.00   sec   121 MBytes  1.01 Gbits/sec  211465
[  5]   2.00-3.00   sec   108 MBytes   903 Mbits/sec  285654
[  5]   3.00-4.00   sec   104 MBytes   869 Mbits/sec  320205
[  5]   4.00-5.00   sec   104 MBytes   869 Mbits/sec  274020
[  5]   5.00-6.00   sec   103 MBytes   866 Mbits/sec  314495
[  5]   6.00-7.00   sec   103 MBytes   868 Mbits/sec  276878
[  5]   7.00-8.00   sec   104 MBytes   871 Mbits/sec  317805
[  5]   8.00-9.00   sec   104 MBytes   872 Mbits/sec  275742
[  5]   9.00-10.00  sec   104 MBytes   869 Mbits/sec  321306
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-10.00  sec  1.04 GBytes   896 Mbits/sec  0.000 ms  0/2679915 (0%)  sender
[  5]   0.00-10.01  sec  1.03 GBytes   886 Mbits/sec  0.005 ms  1879486/2645129 (71%)  receiver

Down - Good

iperf3 -c 192.168.99.11 -t10  -i1 -P1 -u -b 1G -R
Connecting to host 192.168.99.11, port 5201
Reverse mode, remote host 192.168.99.11 is sending
[  5] local 192.168.99.128 port 57426 connected to 192.168.99.11 port 5201
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-1.00   sec   116 MBytes   975 Mbits/sec  0.052 ms  1008/85201 (1.2%)
[  5]   1.00-2.00   sec   116 MBytes   971 Mbits/sec  0.002 ms  2395/86278 (2.8%)
[  5]   2.00-3.00   sec   116 MBytes   975 Mbits/sec  0.006 ms  2097/86250 (2.4%)
[  5]   3.00-4.00   sec   114 MBytes   954 Mbits/sec  0.004 ms  3627/85953 (4.2%)
[  5]   4.00-5.00   sec   117 MBytes   981 Mbits/sec  0.008 ms  2186/86877 (2.5%)
[  5]   5.00-6.00   sec   116 MBytes   975 Mbits/sec  0.006 ms  1985/86180 (2.3%)
[  5]   6.00-7.00   sec   117 MBytes   985 Mbits/sec  0.029 ms  1149/86211 (1.3%)
[  5]   7.00-8.00   sec   114 MBytes   960 Mbits/sec  0.101 ms  3402/86297 (3.9%)
[  5]   8.00-9.00   sec   114 MBytes   954 Mbits/sec  0.004 ms  4082/86380 (4.7%)
[  5]   9.00-10.00  sec   116 MBytes   975 Mbits/sec  0.013 ms  2495/86648 (2.9%)
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-10.00  sec  1.16 GBytes  1000 Mbits/sec  0.000 ms  0/863269 (0%)  sender
[  5]   0.00-10.00  sec  1.13 GBytes   971 Mbits/sec  0.013 ms  24426/862275 (2.8%)  receiver

Macbook Air M2 < wifi1 < bridge 1 < ether4 (1GbE) < Intel NUC8 with Thunderbolt 10 GbE

TCP - iperf3 -

iperf3 -c 192.168.99.11 -t10  -i1 -P1  -R
Connecting to host 192.168.99.11, port 5201
Reverse mode, remote host 192.168.99.11 is sending
[  5] local 192.168.99.128 port 64768 connected to 192.168.99.11 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  89.0 MBytes   747 Mbits/sec
[  5]   1.00-2.00   sec  83.3 MBytes   698 Mbits/sec
[  5]   2.00-3.00   sec  86.4 MBytes   725 Mbits/sec
[  5]   3.00-4.00   sec  86.9 MBytes   729 Mbits/sec
[  5]   4.00-5.00   sec  89.3 MBytes   749 Mbits/sec
[  5]   5.00-6.00   sec  77.3 MBytes   648 Mbits/sec
[  5]   6.00-7.00   sec  75.3 MBytes   632 Mbits/sec
[  5]   7.00-8.00   sec  85.7 MBytes   719 Mbits/sec
[  5]   8.00-9.00   sec  87.8 MBytes   736 Mbits/sec
[  5]   9.00-10.00  sec  78.2 MBytes   656 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.01  sec   842 MBytes   706 Mbits/sec  106             sender
[  5]   0.00-10.00  sec   839 MBytes   704 Mbits/sec                  receiver

Thu Feb 16, 2023 4:23 pm

Thanks for testing.

In that case I guess the remaining action will be to create a supout.rif file and create a ticket towards support.
Make also sure to point them to this thread since you already did quite some testing.

theprojectgroup · Thu Feb 16, 2023 4:24 pm

yep, already did both!

esj · Fri Feb 17, 2023 11:41 pm

Hi theprojectgroup

Would you share your Wi-Fi settings please?

I’m getting bad Wi-Fi speeds and would like a config to get me up and running.

For comparison my

ISP router (a 6E unit), but only using 6 on my iPhone 12 Pro Max is 814 down/230up
hAP AX3 in same spot is 639 down/228up

From further distance, it's even more prevalent
ISP 6E unit 400 down/200 up
hAP 108 down/78up

Thank you

Rox169 · Sun Feb 19, 2023 10:47 pm

Hi,

what unit 6E do you have?

PeterXC · Fri Feb 24, 2023 6:32 pm

Does this looks like a bridging + TCP + ether1 2.5GbE issue?
UDP works fine.

What is your range during tests?

PeterXC · Fri Feb 24, 2023 6:36 pm

Hi theprojectgroup

Would you share your Wi-Fi settings please?

I’m getting bad Wi-Fi speeds and would like a config to get me up and running.

For comparison my

ISP router (a 6E unit), but only using 6 on my iPhone 12 Pro Max is 814 down/230up
hAP AX3 in same spot is 639 down/228up

From further distance, it's even more prevalent
ISP 6E unit 400 down/200 up
hAP 108 down/78up

Thank you

I'm afraid you will not fix it. So far, on the current and rc software, ax3 has a worse range than most other modems, including cable. I have the same issue, and I'm really annoyed.

theprojectgroup · Fri Feb 24, 2023 8:43 pm

Does this looks like a bridging + TCP + ether1 2.5GbE issue?
UDP works fine.

What is your range during tests?

60-100 cm

PeterXC · Fri Feb 24, 2023 11:00 pm

I have the same issue after upgrading to 7.7. TX Rate on 5Ghz went down from 1200Mbps to 800-400. Later the same day any of my devices weren't able to see 5Ghz network at all.
Downgraded to 7.6 and everything works well now.
Opened ticket to support, let's see what they will answer.

Device is hap ax3.

Thanks! I've just downgraded mine. Looks a bit better. Will stay with 7.6 until positive messages from the forum

esj · Sat Feb 25, 2023 1:26 pm

Hi,

what unit 6E do you have?

Hello. I have the Vodafone Pro II Ultrahub 6E - Vantiva produced unit for Vodafone (used to be called Technicolor)

F-Pro2-ProductFamilySide-2000x1000.jpeg

Can't get anything anywhere near this to match the speeds and range. (Happy for suggestions, anyone

...)

https://www.vodafone.co.uk/broadband/pro-ii-devices
https://www.benzinga.com/pressreleases/ ... 6e-gateway

I've tried
1/ Mikrotik hAP ax3 - poor, as described above (even with 7.6 FW re-installed )
2/ ASUS RT-AX86U Pro, which is supposed to be excellent, but I've found is about 70% as good as the Vodafone unit (same location), same conditions.
3/ Next steps will be separate Aruba or Ubiquiti WAPs and use with the Mikrotik hAP ax3 (I love the software and control/granularity of the Mikrotik - RouterOS - only just learned this, but it's great) - I've hard wired Cat6A everywhere in the house, so it has been recommended to me that I buy multiple AP's for each zone and turn them down a bit. To be honest, if I could have the Mikrotik functionality with the WiFi of the Vodafone unit (+ a booster for the yard) I'd be a very happy person.

The bizarre thing is the the Vodafone Pro II Ultrahub reaches everywhere in my house, even 2 levels up I'm getting 600 down/300up and this is 12 meters away through blockwork walls. Best WiFi i've ever had for a single unit!

I'm also contemplating using the Mikrotik hAP ax3 with the Vodafone Ultrahub router as a dumb access point (not sure how this will work though - i'll just give it a fixed IP, turn off DHCP and let the MT do it's stuff). I just wish we were able to get custom firmware for the Ultrahub and flash it and remove the crippling Vodafone software. Can't find anything about how to do this though as the device is locked down. Also there are separate boosters for the Vodafone unit, but the ethernet backhaul isn't supported by Vodafone (again locked down by their custom firmware) so no control of these.

mozerd · Sat Feb 25, 2023 3:36 pm

I have the Vodafone Pro II Ultrahub 6E

MikroTik does not have any WiFi device that can compete with the Vodafone Pro II Ultrahub 6E.
For excellent Router I suggest the RB5009 and for WiFi I suggest the TP-Link EAP660 HD ... this combination will provide excellent Routing and EXCELLENT WiFi equivalent to the Vodafone Pro II Ultrahub 6E

esj · Sat Feb 25, 2023 7:36 pm

I have the Vodafone Pro II Ultrahub 6E
MikroTik does not have any WiFi device that can compete with the Vodafone Pro II Ultrahub 6E.
For excellent Router I suggest the RB5009 and for WiFi I suggest the TP-Link EAP660 HD ... this combination will provide excellent Routing and EXCELLENT WiFi equivalent to the Vodafone Pro II Ultrahub 6E

Thanks Mozerd.

I had looked at the RB5009, but discounted it due to availability. I might just hang out for this. FYI I'm running a 1gig up/ 1gig down fibre service. The hAP AX3 was my foray into Mikrotik gear and I have been very impressed with the versatility of it.... just the WiFi that's let it down for me.

Appreciate the pointer towards the EAP660HD, however will probably hold out for the EAP770 if it's due soon. I swapped out 2 x TP LInk EAP225's and a Cisco 240ac when I installed the Vodafone unit and it really blew them away. Now I've had a taste for what's available, I'm going to try to future-proof and a separate router, with replaceable AP's seems to be the way forward, as advised!

Rox169 · Sat Feb 25, 2023 8:26 pm

you should get approx 800Mbit from AX3, your setting is wrong.... try to do not select any country...it should boost you up...

liviu2004 · Sat Feb 25, 2023 8:28 pm

Is that a joke, not select a country? Why is there a setting to begin with?

Rox169 · Sat Feb 25, 2023 8:31 pm

do not cry if you do not understand....

liviu2004 · Sat Feb 25, 2023 8:42 pm

I cry. You happy now?

PeterXC · Sat Feb 25, 2023 9:07 pm

you should get approx 800Mbit from AX3, your setting is wrong.... try to do not select any country...it should boost you up...

Yeah, when you put your device next to the antenna, maybe. In real-life usage, it is more like 300-600 max.

You can play with it and check your negotiated rates and how it changes with the location in the terminal:

interface/wifiwave2/registration-table/print detail stats interval=1 where interface ="wifi1"

An interesting thing, as antennas are in theory omnidirectional. in reality turning it 180 deg without adjusting the position can increase the performance. At least it worked for me. Still, I'm far from expectation

Rox169 · Sat Feb 25, 2023 9:37 pm

What is your expectation?

PeterXC · Sat Feb 25, 2023 9:48 pm

What is your expectation?

stable 500-600Mbps at my desk, around 5m from ap, with a very light wall/door in LOS

I will not expect 2Gbps, available from my cable provider's AP with 160Mhz channel

Rox169 · Sat Feb 25, 2023 11:01 pm

What is your expectation?
stable 500-600Mbps at my desk, around 5m from ap, with a very light wall/door in LOS

I will not expect 2Gbps, available from my cable provider's AP with 160Mhz channel

stable 500-600Mbps at my desk, around 5m from ap, with a very light wall/door in LOS

you should be fine then with AX3....just tweak settings....

I do not understand why do you need more than 500Mbps...half of the wired cable....

Sat Feb 25, 2023 11:03 pm

I do not understand why do you need more than 500Mbps...half of the wired cable....

Finally someone with a sensible view on things.

PeterXC · Sat Feb 25, 2023 11:16 pm

I do not understand why do you need more than 500Mbps...half of the wired cable....

Well, we live in a wireless world, and new laptops, not only Apple, HP as well, are, unfortunately, without a physical Ethernet port. We have WiFi6, and 160Mhz to (in theory) compensate it.

Yes, I don't need 500Mbps+ every day. Still, it would be nice to have, especially when I had this kind of performance on AC-only Asus or Amplify. I have a better performance on a 'stupid' cable modem, with 160Mbps support, delivered 'free' with my 1Gbps 'fibre' internet.

Ax3 is announced as a super powerful and great home AP. It is great, I can run Pihole in a container, and the CPU is at 1%. But for a HW claimed to be primarily an AP (hAP), WiFi performance is not the best. It makes me sad.

Sat Feb 25, 2023 11:31 pm

I grew up in a time internet didn't even exist.
I still remember the sound of a 1200 baud modem.

That's 1200 bits per second.

People get addicted too fast to speed... And I'm not taking about that other thing.

PeterXC · Sat Feb 25, 2023 11:47 pm

I grew up in a time internet didn't even exist.
I still remember the sound of a 1200 baud modem.

That's 1200 bits per second.

People get addicted too fast to speed... And I'm not taking about that other thing.

So do I. I understand your view. Still, why can’t I have the best performace? Especially when it was promised by MikroTik.

mozerd · Sat Feb 25, 2023 11:49 pm

People get addicted too fast to speed... And I'm not taking about that other thing.

My clients expects 3 features …. And that is what I deliver
1. Stability
2. Performance
3. PERFORMANCE

And that is in every aspect …

In my market 90% have the very same expectation

The mainframe/Terminal days of horrible performance but OUTSTANDING stability only linger in the minds of high priests and Dinosaurs…

Rox169 · Sun Feb 26, 2023 4:53 am

So .what gear do you sell then?

Znevna · Sun Feb 26, 2023 9:52 am

Usualy the wrong gear, based on past experience viewtopic.php?p=849183#p849231

mozerd · Sun Feb 26, 2023 1:56 pm

So .what gear do you sell then?

My primary focus in gear is as follows all depending on circumstances:
Routing + Firewall : MikroTik, UBNT, Custom Build with OPNsense or Arista [untangle]
Wireless AP's : UBNT, TP-Link, Ruckus [CommScope]
Switches : Tik, UBNT, Cisco

anav · Sun Feb 26, 2023 4:47 pm

On TPlink gear I can create one SSID that includes both 2.4ghz and 5ghz chains........
Is that possible on AX3???

I would be happy if I could put all chains on one 5ghz SSID if that makes any sense.......
Lost in ax3 lala land

Sun Feb 26, 2023 4:59 pm

Yes, this is possible.
Why wouldn't it be ?

theprojectgroup · Sun Feb 26, 2023 5:10 pm

you could set both interfaces (one for 2,4 and one for 5ghz) to the same ssid name.

anav · Sun Feb 26, 2023 5:15 pm

Is that the same as band steering, I dont think so!! You guys are clowns if you thought I meant simply assigning the same SSID to both chains....... egads... WTFUP

Sun Feb 26, 2023 5:28 pm

Then your question was not clear and unambiguous enough.
A network diagram, perhaps ?

If you want to do true band steering, you should set separate SSIDs, in my opinion.
If you set both chains using the same SSID, it's up to the client device to decide which radio will be used.
But you can "help" the client a bit with ACL rules (especially kick them off when below a certain threshold since client will try to stick to the same radio as long as possible) and/or lower TX power of 2GHz radio.
Or are you still talking about something else ??

theprojectgroup · Sun Feb 26, 2023 5:57 pm

Nice manners, please.

mkx · Sun Feb 26, 2023 6:31 pm

There's no such thing as band steering in WiFiland. But I guess that devices have some preference built in and one can help them to select AP in their favourite band by enabling the whole suite of 802.11 r+k+v ... As far, as I know, MT implemented 802.11r (a.k.a. BSS fast transition) but not the other two, for those a working and full-featured capsman2 is probably necessary.
So seat back and relax

anav · Sun Feb 26, 2023 7:06 pm

No band steering............ , by the way, maybe you dont know but there is something called flush toilets too. '=PP

Band Steering
Automatically moves dual band devices onto the wider 5GHz band for faster connections.

doesnt-exist-eh.jpg

mkx · Sun Feb 26, 2023 7:52 pm

Well, it is possible to make car, driving on a highway lane, to change to another lane by hitting it from a side, but I wouldn't call that "steering".

After you configure all the bells of 802.11r (the whistles of k+v are missing), you can configure access-list to kick devices off 2.4GHz radio at rather good signal strength. Devices tend to take it personally and veer off to another AP, hopefully a 5GHz one. Contrary to what many would think of first, you actually want to kick clients with excellent signal strength (because they likely have at least very good 5GHz signal they can work with). Then you want to allow clients with mediocre signal to save them from loosing wireless connection. And then you want to kick clients with bad signal because they won't be able to get any decent throughputs (and they are far away, probably neighbours

).

A more considerate implementation would remember client's past behaviour. E.g. if a client with good signal, after being kicked, doesn't register back or keeps trying to register to 2.4GHz, then it might be a 2.4GHz-only client. Or some vendor might compile a device list grouped according to wireless capabilities - this option got expelled with "privacy enhanced" wireless faking their MAC addresses. But you can partially implement it for your own 2.4GHz-only devices by setting them always allowed in access-list. Probably some vendors came up with something even more sophisticated.

Sun Feb 26, 2023 8:19 pm

Long explanation for what I already said.
But maybe he will accept it coming from you

anav · Sun Feb 26, 2023 8:29 pm

I am quite aware of using Access Rules for kicking folks off.
I tried to implement that once on some cAPACs, I was savaged by family members for screwing things up, and
it also buggered my own connectivity as we have a complex setup. So I am gun shy of that functionality.

mozerd · Sun Feb 26, 2023 8:35 pm

So I am gun shy of that functionality.

Anav my friend

you are wasting your valuable retirement time with this … I know that you want to conquer this bugger but tell me something …. Is it really worth the struggle ??? Poke poke POKE

Rox169 · Sun Feb 26, 2023 10:51 pm

Hi,

just to be clear 800.11 r,k,w has been already implemented in ROS only 800.11v is missing. on the another hand it is implemented only on one AP, not on goup of AP, mybe in the new caps man,,,

bpwl · Sun Feb 26, 2023 11:47 pm

"Band steering" should be better than "Access list signal limit kicking" and TX power level difference between 5GHz and 2.4GHz radio's.

Can MT wifiwave2 do this, what Engenius descibes as bandsteering ?
https://helpcenter.engeniustech.com/hc/ ... available-
How do you select the "Steering mode" in MT then???

"Band Steering" seems to be still something additional to 802.11 r,k,v https://www.mist.com/documentation/802- ... r-802-11v/

(My favorite) Fortinet has a strong control over client access, which only see one BSSID (MAC address of AP radio's are identical), as only some AP (radio) respond to a connection request, as steered by the controller.
Experienced: Used it with Fortigate cluster (HA controller) and 37 FortiAP

mkx · Mon Feb 27, 2023 11:48 am

Can MT wifiwave2 do this, what Engenius descibes as bandsteering ?

I don't think it can ... in current state. Apart from ACLs (which are static) there's no mechanism in ROS to ignore wireless clients when they try to associate ... which is, by the way, the gist of Engenius' implementation: ignore association requests on non-preferred band from devices which are known to support the preferred band (and the client support for different bands is built dynamically by letting client try to associate to any of APs serving same SSID, hopefully client will reveal support for 5GHz band by trying to associate to it).

Poor man's solution in ROS would be scheduled job, skimming through registration table of preferred band and set ACLs preventing devices to associate to non-preferred band. It's slow, clumsy and might break some clients.

Of course, all of that (proper solution and poor man's solution) work fine if there's a central wireless controller which correlates signalling from all APs serving same SSID.

pe1chl · Mon Feb 27, 2023 12:02 pm

(My favorite) Fortinet has a strong control over client access, which only see one BSSID (MAC address of AP radio's are identical), as only some AP (radio) respond to a connection request, as steered by the controller.

That is basically a co-channel diversity network where all the APs are receiving and reporting the data and signal strength to the controller, so the controller can decide which AP will send data to that client. It is one big "virtual AP". The advantage is that there is no roaming issue, the disadvantage is that you use only a single channel so you limit the capacity of the network.
However, such a system can also run that same method on different channels (with different BSSID probably) in parallel, to increase capacity a little.
Of course not on all channels at the same time, as the transmitters would overwhelm the receivers, but maybe on 2 or 3 channels spread over the 5GHz band, with good filters.

MikroTik has no offerings in this area, but several other enterprise WiFi suppliers do.

pe1chl · Mon Feb 27, 2023 12:13 pm

I don't think it can ... in current state. Apart from ACLs (which are static) there's no mechanism in ROS to ignore wireless clients when they try to associate

This weekend I finally implemented RADIUS-based access on my own home network. I had ACLs in both my main RB4011 and my hAP ac2 additional AP in another room, and merged all SSIDs into one to have several separate networks (VLANs) for different types of clients. Instead of having several virtual wireless interfaces with different SSID, which wastes airtime, I first put "use vlan" and vlan# into the ACL rules, then realized I needed to do that twice every time so I finally installed user-manager on the RB4011 to do it.
Now I only need to add the MAC of a client and put it in a group that sends the proper VLAN number, and it works on both APs.

During experiments with this, I found that the default of user-manager to allow only 1 connection per user breaks roaming. When the client tries to roam, it connects to the nearer AP, that sends a RADIUS request to user-manager, which sees "a second connection", rejects it, and the client gets refused. It immediately goes back to the other AP, but as it sees the weak connection and a stronger signal nearby, it keeps flipping. Only removing that single-connection setting per user resolved that.
(it is understandable that the user-manager thinks there are 2 connections because the original is not yet broken)

As the access request to RADIUS also includes the WiFi interface name, it seems it should be possible to reject requests coming in on 2 GHz and allow on 5 GHz.
However, it seems that user-manager is unable to do that, and neither can it have a "default user action", so you still need another RADIUS server.

anav · Mon Feb 27, 2023 3:05 pm

My questions are simple.
If we are supposed to leave things AUTO, and by that I mean dont select frequency. how does the MT select frequency and Tx power??
They provide scan and frequency tools, does this mean we are supposed to use those and forget AUTO, or does AUTO use that information for optimal frequency selection.

Here is what I was successful in getting working, minimally, I basically decided to avoid ax for 2ghz and
selected very basic stuff. The only delving into menus I did was to disable pmkid which I thought should be defaulted to disable??? and WPS disable.
Took me a while to get even this setup. Then realized I had no encryption, used to AES, didnt have a clue what CCMP was.

Still not working then deciding to look at everything, interfaces checked out............. bridge ports OOPSIE frigging with the wifi settings means whatever was in bridge ports is gone.......or unknown.
Thus put the right interfaces there and then started working. The bridge ports is a gotcha as we are used to the MT replacing interfaces with the new name as you do the work..........

/interface wifiwave2 channel
add band=5ghz-ax disabled=no name=channel1 width=20/40/80mhz
add band=2ghz-g disabled=no name=channel2 width=20mhz
/interface wifiwave2 configuration
add channel=channel1 channel.band=5ghz-ax .width=20/40/80mhz country="Canada" disabled=no mode=ap name=cfg1 ssid=MT1
add channel=channel2 channel.band=2ghz-g .width=20mhz country="Canada" disabled=no mode=ap name=cfg2 ssid=MT2
/interface wifiwave2 security
add authentication-types=wpa2-psk comment="Trusted Password" disable-pmkid=yes disabled=no name=HOME wps=disable
/interface wifiwave2
set [ find default-name=wifi1 ] channel=channel1 comment="5GH - WIFI" configuration.mode=ap .ssid=MT1 disabled=no security=HOME
set [ find default-name=wifi2 ] channel=channel2 comment="2GHz WIFI" configuration=cfg2 configuration.country="Canada" .mode=ap .ssid=MT2 disabled=no security=HOME

bpwl · Mon Feb 27, 2023 3:20 pm

"a second connection"

As expected, no ?
We want to have fast roaming. And some roaming enhancements connect before they disconnect.
The client device will not continu with 2 connections.

On the other hand, I did not limit the number of connections per named user, for other reasons.
In a holiday resort each vacation home tenant gets a username/password for the family. Anyway they come with multiple devices (PC,tablet,smartphone,smartwatch ...) and multiple family members.
It is all one and the same account. Number of connections is not the problem, data volume is the quantity to be limited, even more than speed.
I split the authenticating RADIUS server (3th party) from the accounting RADIUS server (e.g. MT Usermanager, ROS 6), by blocking accounting packets to the authenticating/authorizing server, and using "Accounting Backup" for the accounting server. (Accounting server has usernames, not passwords).

With RADIUS and username/password, one does AAA on the user , not on the device. Random private administered MAC addresses is no problem in this.
Roaming is smooth and fast. (30 AP's, 40 named users, +- 280 connections and more)
RADIUS should be able to set the VLAN for that RADIUS user (same WLAN-wifi mechanism as with the local ACL, but on username not MAC and ACL). Didn't test that yet.
3th party RADIUS and accounting RADIUS to be replaced by hAP ax3 if user based VLAN setting works , actually not for it's wifi, but for the ROS licence level 6, and User manager ROS7.
[ RB5009 is only licence level 5 = max 50 wifi sessions via Usermanager with accounting

]

pe1chl · Mon Feb 27, 2023 4:00 pm

My questions are simple.
If we are supposed to leave things AUTO, and by that I mean dont select frequency. how does the MT select frequency and Tx power??
They provide scan and frequency tools, does this mean we are supposed to use those and forget AUTO, or does AUTO use that information for optimal frequency selection.

AUTO does a one-time scan at each boot to find the "least used channel" and selects that. It then configures the highest allowed TX power according to local regulations for that channel.
As usual this is far from optimal. You need to make at least a channel list with channels 1, 6 and 11 (2412, 2437, 2462 MHz) for 2 GHz and apply that, or else it will choose a random channel like 3 or 8 and foul up the carrier detection in the area. Really substandard that MikroTik does not use 1-6-11 channel selection by default.

Also (like many other manufacturers) it does not do a re-scan of the spectrum at any time after boot. So changed circumstances on the RF spectrum are never taken into account.

pe1chl · Mon Feb 27, 2023 4:03 pm

"a second connection"
As expected, no ?
We want to have fast roaming. And some roaming enhancements connect before they disconnect.
The client device will not continu with 2 connections.

I "expected" this problem only after having encountered it. As I wrote, it is obvious why it happens but not so obvious to me why the default in user-manager is 1 connection, thus causing problems when roaming. Limiting to 1 connection should be left to the user, and only when connections are used for things like PPPoE with fixed address.

anav · Mon Feb 27, 2023 5:59 pm

As usual this is far from optimal. You need to make at least a channel list with channels 1, 6 and 11 (2412, 2437, 2462 MHz) for 2 GHz and apply that, or else it will choose a random channel like 3 or
Why would I not want it to select channel 3 if everyone else is hogging 1 and 6 LOL, it would be better than choosing 1 or 6 NO. At least at peak channel 3 freq I am flying solo

Mon Feb 27, 2023 7:59 pm

No, you're not alone.
You will have interference from 1 and 6 which is worse then co-existence.

pe1chl · Mon Feb 27, 2023 9:07 pm

Indeed. Your AP on channel 3 will not detect the transmissions from others on 1 and 6, so it will transmit at the same time as them, instead of waiting for the channel to be clear.
And due to the overlap, it will have interference from (and cause interference to) both channel 1 and 6.

bpwl · Mon Feb 27, 2023 9:43 pm

SImple wifi AP devices, select the lowest occupied or lowest energy intense wifi channel, and as such select the typical signal dip for channel 3 and 8.
Seen too often. Fully agree with this text for years: https://metis.fi/en/2018/10/rrm-en/
Keeps us busy in this formum ....on "poor wifi performance" topics.

pe1chl · Mon Feb 27, 2023 10:03 pm

SImple wifi AP devices, select the lowest occupied or lowest energy intense wifi channel, and as such select the typical signal dip for channel 3 and 8.

But at work I have installed a system from the other well-known competitor in the "cheap" market segment, and without any configuration it selects only channel 1, 6 or 11 when setting to auto. So it can be done.
On my MikroTiks at home I use this channel list:

/interface wireless channels
add band=2ghz-b/g/n frequency=2412 list=ch1_6_11 name=ch1 width=20
add band=2ghz-b/g/n frequency=2437 list=ch1_6_11 name=ch6 width=20
add band=2ghz-b/g/n frequency=2462 list=ch1_6_11 name=ch11 width=20

bpwl · Mon Feb 27, 2023 11:49 pm

Good channel list.
I don't trust "auto", not even the XXXX in 20/40/80 XXXX. Check the 80 MHz channel alignment !
viewtopic.php?t=183744&hilit=Interestin ... nt#p918062

mkx · Tue Feb 28, 2023 9:04 am

Too bad wifiwave2 doesn't allow constructing frequency lists ... or does it? Entries under /interface/wifiwave2/channel don't seem to have property list available?

Tue Feb 28, 2023 10:35 am

But you can

Tab Channel, frequency, use the drop down arrow to add frequencies.

2023-02-28_09-36-46.jpg

PeterXC · Tue Feb 28, 2023 11:21 am

Good channel list.
I don't trust "auto", not even the XXXX in 20/40/80 XXXX. Check the 80 MHz channel alignment !
viewtopic.php?t=183744&hilit=Interestin ... nt#p918062

What is your recommendation for 5GHz channel list? Would you go for all Ceee or eCee?

marsbeetle · Tue Feb 28, 2023 11:27 am

I have set my hAP ax3 to the following which for me works well:

/interface/wifiwave2/channel
add band=2ghz-ax frequency=2412,2437,2462 name=2ghz-1_6_11 width=20mhz skip-dfs-channels=disabled
add band=5ghz-ax frequency=5500,5520,5540,5560 name=5ghz-no_dfs width=20/40/80mhz skip-dfs-channels=disabled

I took the non-restricted channels my country(south africa) is supporting for 5ghz from here - https://en.wikipedia.org/wiki/List_of_WLAN_channels

So far so good and I can maintain 750+mbit on 5ghz and with 2.4ghz i can reach to end of my house which is all I wanted to begin with. This wasn't the case with the defaults.

Edit: just to add - I pointed the top round edge of the antenna in the direction I wanted the best signal(perpendicular). sounds crazy but it made a difference for me.

mkx · Tue Feb 28, 2023 12:27 pm

The way @pe1chl showed actually allows to have different channel parameters depending on selected centre frequency. When simply listing frequencies in single statement all the rest of properties have to be the same. So the new way is not as flexible as the old one ... sigh, I guess we'll have to get used to new reality with much reduced flexibility.

Re. 5GHz channel list: I guess the most straight forward (and proper) way of doing them is to select some channel layout (perhaps @bpwl knows how other vendors do it, but I guess Ceee would be a good starting point) and then set frequencies to those which should according to standard. If one takes channel list from wikipedia ... then proper frequency for Ceee channel (with maximum backward compatibility) would be e.g. 5260 ... because 5260 is fine for single 20MHz channel (number 52) and for Ce 40MHz channel (number 54) and for Ceee 80MHz channel (number 58). If one throws in 160MHz (or 80+80), then it should be e.g. Ceeeeeee with frequency 5180.
I guess that when setting channel layout to XXXX it should be possible to use any of (5260, 5280, 5300, 5320) as frequency and ROS will (hopefully) select right channel layout which goes with selected frequency. But there's room for bugs here and in worst case some wireless clients won't like it.

pe1chl · Tue Feb 28, 2023 12:30 pm

In the "old" wifi driver there is a bug that means that your channel list will fail to work after export/import when its name starts with a digit.
The export fails to put the list name in quotes, and the import assumes it is a literal frequency list (like 2412,2437,2462) when it starts with a digit, then errors out with a syntax error.
I don't know if this has ever been fixed, but after having been bitten by this I use the name ch1_6_11 instead.

Tue Feb 28, 2023 12:42 pm

I don't know if this has ever been fixed, but after having been bitten by this I use the name ch1_6_11 instead.

Good tip !

marsbeetle · Tue Feb 28, 2023 12:58 pm

In the "old" wifi driver there is a bug that means that your channel list will fail to work after export/import when its name starts with a digit.

Thanks for that. I will change the name just in case.

oliviu56 · Wed Mar 01, 2023 10:58 am

I have the same issue after upgrading to 7.7. TX Rate on 5Ghz went down from 1200Mbps to 800-400. Later the same day any of my devices weren't able to see 5Ghz network at all.
Downgraded to 7.6 and everything works well now.
Opened ticket to support, let's see what they will answer.

Device is hap ax3.

Hello!
I also want to do a firmware downgrade on a hap ax2 from 7.8 to 7.6 (also, I have signal problems). Please tell me, step by step, the downgrade steps (because there are two firmware files and wifiwave2). I don't want to use Netinstall!! Only downgrade.
Thank you very much.

maigonis · Wed Mar 01, 2023 11:52 am

A side tip. On 2.4ghz band you should use ch 1,6,11, as they don't overlap, but in real life, your neighbors, most likely, will not follow this practice. As result, it does not matter if you use 1,6,11, if you neighbor is using ch2, 40mhz wide. You should use the cleanest channel you have in your environment. At work I have full control over APs, so there I can use 1.6.11.

pe1chl · Wed Mar 01, 2023 12:14 pm

It is like saying "you should not dump trash in your public park. but in real life, your neighbors will do it anyway. so you can just as well dump your trash in the park!".

EgidijusL · Wed Mar 01, 2023 3:02 pm

I have the same issue after upgrading to 7.7. TX Rate on 5Ghz went down from 1200Mbps to 800-400. Later the same day any of my devices weren't able to see 5Ghz network at all.
Downgraded to 7.6 and everything works well now.
Opened ticket to support, let's see what they will answer.

Device is hap ax3.
Hello!
I also want to do a firmware downgrade on a hap ax2 from 7.8 to 7.6 (also, I have signal problems). Please tell me, step by step, the downgrade steps (because there are two firmware files and wifiwave2). I don't want to use Netinstall!! Only downgrade.
Thank you very much.

1. Make backup
2. Download RouterOS https://download.mikrotik.com/routeros/ ... -arm64.npk and https://download.mikrotik.com/routeros/ ... 64-7.6.zip
3. Unzip all_packages-arm64-7.6.zip
4. Upload routeros-7.6-arm64.npk and wifiwave2-7.6-arm64.npk to File list (Winbox menu: Files)
5. System > Packages > Click Downgrade

erlinden · Wed Mar 01, 2023 4:44 pm

Instead/besides a backup, you can also perform an export which will save your configuration:
https://help.mikrotik.com/docs/display/ ... tandExport

gigabyte091 · Wed Mar 01, 2023 6:08 pm

On Ubiquiti there is a band steering option and it's working excellent, on Mikrotik when 2.4 and 5 GHz radios are given same SSID phone reports this network as dual band (it says 2.4/5 after SSID) but for me, this just mean that device will connect to the 2.4 GHz and even ACL will not force it to stay connected to 5GHz on ax2.

When I tried to force clients to stay connected to 5 GHz with ACL devices just keep trying to connect to 2.4 GHz.

I reduced channel width to 20/40 for 5GHz as i really don't have need for 1200 Mbps speed (internet is 170/150 so even 40 MHz channel is never fully saturated)

And 2.4 GHz channel is 20 MHz. Frequency is fixed, i scanned my area and set WiFi to free channels. ( I live in a house and there really is no other networks other than my that could interfere with my WiFi, only one very weak 2.4 from neighbour, and no 5 GHz)

I ended up simply separating 2.4 and 5 GHz bands but on ax2 i experience wifi "crapout", less now that 7.8 is out... And i have few other issues as upload is at max spees but download is half that speed...

oliviu56 · Wed Mar 01, 2023 6:15 pm

Hello!
I also want to do a firmware downgrade on a hap ax2 from 7.8 to 7.6 (also, I have signal problems). Please tell me, step by step, the downgrade steps (because there are two firmware files and wifiwave2). I don't want to use Netinstall!! Only downgrade.
Thank you very much.
1. Make backup
2. Download RouterOS https://download.mikrotik.com/routeros/ ... -arm64.npk and https://download.mikrotik.com/routeros/ ... 64-7.6.zip
3. Unzip all_packages-arm64-7.6.zip
4. Upload routeros-7.6-arm64.npk and wifiwave2-7.6-arm64.npk to File list (Winbox menu: Files)
5. System > Packages > Click Downgrade

Successfully downgraded to 7.6!
Thank you.

marsbeetle · Wed Mar 01, 2023 10:07 pm

Is ROS 7.6 still better than 7.8 in terms of wifi stability and signal strength? …surely not.

PeterXC · Wed Mar 01, 2023 10:13 pm

On Ubiquiti there is a band steering option and it's working excellent, on Mikrotik when 2.4 and 5 GHz radios are given same SSID phone reports this network as dual band (it says 2.4/5 after SSID) but for me, this just mean that device will connect to the 2.4 GHz and even ACL will not force it to stay connected to 5GHz on ax2.

When I tried to force clients to stay connected to 5 GHz with ACL devices just keep trying to connect to 2.4 GHz.

For me, acl on Wireless rejecting mac from 2.4 interface worked without issues.

oliviu56 · Thu Mar 02, 2023 5:22 am

Is ROS 7.6 still better than 7.8 in terms of wifi stability and signal strength? …surely not.

In my case, as a hap ax2 user, in ros 7.6 the signal strength is about 20% better than 7.7 and 7.8. As for stability, in 7.6 I had two situations where I found all devices disconnected from the wireless router. I don't know the reason for the disconnection. I am a basic user of these Mikrotik routers.

LSan83 · Thu Mar 02, 2023 7:06 am

As for stability, in 7.6 I had two situations where I found all devices disconnected from the wireless router. I don't know the reason for the disconnection. I am a basic user of these Mikrotik routers.

From 5Ghz radio only or from both? A radar detection events trigger a disconnection on 5Ghz DFS channels... Next time check the log.

EgidijusL · Thu Mar 02, 2023 11:23 am

I have the same issue after upgrading to 7.7. TX Rate on 5Ghz went down from 1200Mbps to 800-400. Later the same day any of my devices weren't able to see 5Ghz network at all.
Downgraded to 7.6 and everything works well now.
Opened ticket to support, let's see what they will answer.

Device is hap ax3.

What does the support say?

Is ROS 7.6 still better than 7.8 in terms of wifi stability and signal strength? …surely not.

7.8Beta and 7.8RC there was no improvement

oliviu56 · Thu Mar 09, 2023 5:53 am

As for stability, in 7.6 I had two situations where I found all devices disconnected from the wireless router. I don't know the reason for the disconnection. I am a basic user of these Mikrotik routers.
From 5Ghz radio only or from both? A radar detection events trigger a disconnection on 5Ghz DFS channels... Next time check the log.

The decrease in signal strength is manifested in both 2.4 and 5Ghz channels

pe1chl · Thu Mar 09, 2023 11:22 am

Decreasing signal strength after installing an update (any manufacturer!) can sometimes be caused by the restrictions in the country where you live.
When the law specifies a lower max power than the usual defaults over the world, and the manufacturer did not know about that, they may change the country table once they hear about it, and from that time on you will have less power in the same access point.
Another issue is that for some bands (this should not affect 2 GHz but it does affect 5 GHz) the max power is not the same for all channels.
When you have set the frequency to "auto" the access point selects a channel based on current occupation and it may be that after an upgrade it selects a different channel (either randomly or because of changed detection rules), and now it has a channel with lower power limits. Then it will reduce the power.
You can work around this by changing the installation "indoor/outdoor/any" settings and/or by setting a fixed channel frequency instead of "auto".

marsbeetle · Thu Mar 09, 2023 12:40 pm

The problem I'm seeing on my hAP ax3 is that the signal strength(reach) is not much better on 2ghz than it is on 5ghz. Normally this wouldn't be a problem but when your main seating location falls within this range you get disconnects and constant flapping between the two radio's. I still have my old hAP ac2 as a backup and this never gave any issues with the same wifi configuration(other than the driver)

I've seen too many complaints with the same hardware to assume it is configuration based and I've tried so many different things that I'm now just going to leave it in the hope it improves with future updates.

alibloke · Thu Mar 09, 2023 1:11 pm

I dropped tx-power on the 2GHz radio to 12 so that clients prefer the 5GHz radio. It doesn't seem to affect 2.4GHz connectivity much when you're far enough away for the 5GHz signal to be attenuated.

mkx · Thu Mar 09, 2023 1:22 pm

I dropped tx-power on the 2GHz radio to 12 so that clients prefer the 5GHz radio.

You could try to "push" clients to 5GHz using access-list rules. I constructed a set of rules on my Audience (ac device running wifiwave2 in ROS v7.

:

add action=reject allow-signal-out-of-range=10s comment="kick clients with too good signal, hopefully they'll reconnect on 5GHz AP" interface=wifi-2G signal-range=-65..120
add action=accept interface=wifi-2G
add action=reject allow-signal-out-of-range=10s interface=wifi-5G signal-range=-120..-80
add action=accept interface=wifi-5G

Before I did these rules, some of wireless clients sticked to 2.4GHz band no matter what. With these settings, they are kicked to 5GHz band. The cut-off value (-65dBm in my example) has to be set according to Tx power of both bands and the point where 5GHz signal drops below usable level, my device is operating at default (i.e. maximum allowed) Tx powers.

The problem: as it seems now, rules work fine when device does initial connection to the AP. However, it doesn't get kicked off 2.4GHz channel if the signal level raises above the threshold. I haven't done thorough testing yet, but with default logging config nothing is in logs that would indicate that AP is trying to kick off the "offending" device.
So I wonder if failing to kick off the device is a bug in wifiwave2 driver or error in my config?

bpwl · Thu Mar 09, 2023 2:10 pm

There is no signal range on your accept ACL rules. So the device will stick under all signal conditions on that accept rule, once it gets there,
That is when it did not match (even just once) the reject rule in your list.

reject signal range = -120..-66
accept signal range = -65..120
... the accept rule will now stop when below -65 dBm for longer than the 'allow-signal-out-of-range' time, And the rules will be re-evaluated, top down when reconnecting.

mkx · Thu Mar 09, 2023 3:57 pm

That is when it did not match (even just once) the reject rule in your list.

Right. I adjusted the rules that did not include signal-range to have one. Will test and report back.

Edit:

Here's report: I had to roll back the ACL changes.

My daughter has a Nokia 7.1 android phone which seemingly doesn't like being kicked off. After being kicked off it started to see two wireless networks with same SSID (one formed by Audience and one formed by other, older, APs without ACL set up) and wouldn't connect to either of them. After phone was restarted it connected Audience (its signal was strongest) and would drop very soon. After ACL setup roll-back and phone restart, everything returned to normality (it took a while longer for daughter's mood to pick up again).

TomSF · Thu Mar 09, 2023 7:38 pm

I just got an ax3 as a drop-in replacement for a hap ac and quickly discovered the poor wifi bandwidth issue discussed in this thread. Using the same test devices (Samsung Galaxy s10e, Windows 11 PC with Intel Killer wifi 6E ax1675x adapter, and Microsoft Surface w/ Intel wireless) from the same locations, using Speedtest to the same server, the ax3 has 1/4 the wifi bandwidth of the ac (51 mb/s vs 222 mb/s). I am just starting the process of figuring out why, but my first question/issue is chains. The Mikrotik specs say the ax3 has 2 chains for 2G and 2 for 5G. Winbox-wireless-radios show 0-1 for both tx and rx for both 2G and 5G. In addition, I cannot change the chains configuration. When I try, the interface gets removed and an error "chains not supported" is displayed in its place. So, my first question is why the radios show 0-1 chains. It implies that there could be two chains which would probably be shown as 1-1. On the other hand, 1 chain for rx and one chain for tx might meet the spec for 2 chains (with a little deceptive marketing).

As background, the ac and ax3 are/were configured as access points only. Both are/were directly connected to a CCR-1009-7G which is directly connected to my cable modem. Everything Mikrotik is running v7.8. I have tried to match the ac and ax3 configurations as closely as possible.

Gripe paragraph: Given the wifi bandwidth issues, the discovery that old capsman does not support the ax3 and the new capsman does not support my other 7 ap's, I am really disappointed with the ax3. I can live with having to separately configure the ax3 with no capsman, but cannot live with 1/4 the wifi bandwidth.

mkx · Thu Mar 09, 2023 8:47 pm

What does /interface/wifiwave2/monitor 0 (and 1) say about used Tx power? It should be somewhere above 20[dBm] ...

I don't think you have to set chains ... it'll just use all chains available in hardware.

TomSF · Thu Mar 09, 2023 9:43 pm

It says 26. 2G tx power is 27.

PeterXC · Thu Mar 09, 2023 11:31 pm

I just got an ax3 as a drop-in replacement for a hap ac and quickly discovered the poor wifi bandwidth issue discussed in this thread. Using the same test devices (Samsung Galaxy s10e, Windows 11 PC with Intel Killer wifi 6E ax1675x adapter, and Microsoft Surface w/ Intel wireless) from the same locations, using Speedtest to the same server, the ax3 has 1/4 the wifi bandwidth of the ac (51 mb/s vs 222 mb/s).

What is your cpu temperature on idle and on load?

TomSF · Fri Mar 10, 2023 12:08 am

54C. I do not know how to put it under load but typically there are only about 6 wifi clients at a time and three clients connected via ethernet ports. In other words, it is never highly loaded.

TomSF · Fri Mar 10, 2023 12:27 am

I thought I posted the following but do not see it so here I go again.

I ran a bunch of btest's involving 2 hap ac's and the ax3. All of them flow through the CCR router. The results are mysterious but maybe related to the wifi issues.

AC1 <-> AC2
- sending @ 755Mb/s
- receiving @ 790Mb/s
- it does not matter which AC runs the client

Any AC <-> AX3
- client on AC
-- sending @ 95Mb/s
-- receiving @ 945Mb/s
- client on AX3
-- sending @ 945Mb/s
-- receiving @ 95Mb/s

Notice the order of magnitude difference between sending and receiving when the ax3 is involved. The ax3 has been connected to the ccr via ether1 (2.5G) or one of the 1G ethernet ports with no change in results. Any explanation or suggestions are appreciated.

Rox169 · Fri Mar 10, 2023 8:29 am

Hi, your tests make no sence..how you can get on AC 945Mb/s? Iguess you mean thorough LAN.

mkx · Fri Mar 10, 2023 8:53 am

Hi, your tests make no sence..how you can get on AC 945Mb/s?

The old hAP ac (not the new ac2 variant) has 3 chains on 5GHz radio and has thus up to 1300Mbps interface rate. hAP ax3, OTOH, has only 2 chains and thus supports up to 866Mbps interface rate (if running in ac compatibility)or 1200Mbps interface rate (if running in ax mode). All numbers are for 80MHz channel, according to specs hAP ac doesn't support 160MHz channels.

So yes, one does wonder how on Earth can AC<->AX work at such a high rate.

PeterXC · Fri Mar 10, 2023 9:04 am

Hi, your tests make no sence..how you can get on AC 945Mb/s?

The old hAP ac (not the new ac2 variant) has 3 chains on 5GHz radio and has thus up to 1300Mbps interface rate. hAP ax3, OTOH, has only 2 chains and thus supports up to 866Mbps interface rate (if running in ac compatibility)or 1200Mbps interface rate (if running in ax mode). All numbers are for 80MHz channel, according to specs hAP ac doesn't support 160MHz channels.

Looking at specs you shared, if you have a mixed ac/ax devices you should not go for ax3 as it will perform worse than ac2, right?

maigonis · Fri Mar 10, 2023 9:25 am

Those who have poor throughput performance, what PHY speed do you see? I use ax3 whit capsman v2 and performance is great. Real wold speeds are around 500-600mbps in the same room/open space. ac2 performance was x2 slower as capsman v1 throughput is around 350mbps (80mhz channel).

bpwl · Fri Mar 10, 2023 1:33 pm

Just some technical things ....

https://mcsindex.com/

- AC with VHT goes up to 866Mbps interface rate with 2-stream on 80 MHz bandwidth (using 256QAM). The well known limit.
- AX with HE goes up to 1024 QAM encoding and has wider channels (see mcsindex table for interface rates
- Other stream combinations are in the mcsindex table.
- What is the AC 80+80 (primary + secondary) doing, in performance ?

In practice, and that practical limit is strong with MT classic drivers, the net data transfer rate is reduced because the air-time is shared among all transmitters in the channel, and the channel-contention process requires gaps between successive transmissions.
With small transmission packets, the gap loss for every transmission in VHT becomes very important with higher rates. Classic MT drivers have small A-MSDU (MPDU) and small A-MPDU, and don't use the standard large transmissions.
viewtopic.php?p=951438&#p951138. The impact of this on the data rate can be calculated

What is AX , with HE instead of VHT doing with this? https://www.extremenetworks.com/wifi6/what-is-80211ax/

TomSF · Fri Mar 10, 2023 5:37 pm

Hi, your tests make no sence..how you can get on AC 945Mb/s? Iguess you mean thorough LAN.

Yes, I meant LAN only. For now, I have taken wifi out of my investigation, although I wonder if the LAN-only weirdness is related to my poor wifi performance. Sorry for all the confusion.

TomSF · Fri Mar 10, 2023 5:52 pm

(snip) ...hAP ax3, OTOH, has only 2 chains and thus supports up to 866Mbps interface rate (if running in ac compatibility)or 1200Mbps interface rate (if running in ax mode). All numbers are for 80MHz channel, according to specs hAP ac doesn't support 160MHz channels. ... (snip)

This relates to my earlier post where I said /interfaces/wireless/radios reports there is one tx chain and one rx chain for both the 5G and 2G radios. Is that what you mean when you said the ax3 has only 2 chains (1+1)?

bpwl · Fri Mar 10, 2023 6:13 pm

according to specs hAP ac doesn't support 160MHz channels.

But AFAIK it does support 80+80MHz. 80+80 = Primary+secondary.
The "secondary" option appears in Winbox only when 20/40/80 bandwidth is choosen.
The secondary channel is 80MHz only. (No Ceee notation, so maybe they use the mid-freq notation only for the secondary (as specified in HELP for wifiwave2)

WLAN:

secondary-channel (integer; Default: "")	

Specifies secondary channel, required to enable 80+80MHz transmission. To disable 80+80MHz functionality, set secondary-channel to "" or unset the value via CLI/GUI.

wifiwave2:

secondary-frequency (list of integers | 'disabled')
	
Frequency (in MHz) to use for the center of the secondary part of a split 80+80MHz channel.
Only official 80MHz channels (5210, 5290, 5530, 5610, 5690, 5775) are supported.
Leave unset (default) for automatic selection of secondary channel frequency.

.
.

What is the throughput? Well it looks promising : https://www.qnap.com/en/how-to/faq/arti ... or-wi-fi-5

TomSF · Fri Mar 10, 2023 6:37 pm

Those who have poor throughput performance, what PHY speed do you see? I use ax3 whit capsman v2 and performance is great. Real wold speeds are around 500-600mbps in the same room/open space. ac2 performance was x2 slower as capsman v1 throughput is around 350mbps (80mhz channel).

With my Surface I see tx-516, rx=344. Running a Speedtest, I get 94Mb/s down, up is capped at 18Mb/s. Running Speedtest connected to a hap ac I get 134Mb/s down. Last week I was at a location with wifi6 (I do not know the equipment) and I got 392Mb/s down.

bpwl · Fri Mar 10, 2023 6:47 pm

radios show 0-1 chains

Or shows/means chains 0-1 , that is 2 chains, number 0 and number 1. This is confirmed by the documentation.
2.4GHz has 2 chains (named chain 0 and 1), 5 GHz has 2 chains (named chain 0 and 1).
The chains on radio 2.4GHz and 5 GHz are independent, they can use the same antenna.

AFAIK chain 0 must always be selected for receiving, chain 1 is optional, is selected or not for RX.
TX could be on 1 or 0, or 0 and 1 (default)

anav · Fri Mar 10, 2023 6:54 pm

bpwl can you tag a WLAN??

typical MT setup - access port
/interface bridge port
add bridge=bridge interface=WLAN1 pvid=5
/interface bridge vlan
add bridge=bridge tagged=bridge untagged=WLAN1 vlan-ids=5

Possible?
trunk port
/interface bridge port
add bridge=bridge interface=WLAN1
/interface bridge vlan
add bridge=bridge tagged=bridge,WLAN1 vlan-ids=5

TomSF · Fri Mar 10, 2023 7:12 pm

radios show 0-1 chains
Or shows/means chains 0-1 , that is 2 chains, number 0 and number 1. This is confirmed by the documentation.
2.4GHz has 2 chains (named chain 0 and 1), 5 GHz has 2 chains (named chain 0 and 1).
The chains on radio 2.4GHz and 5 GHz are independent, they can use the same antenna.

AFAIK chain 0 must always be selected for receiving, chain 1 is optional, is selected or not for RX.
TX could be on 1 or 0, or 0 and 1 (default)

Thanks. That clears up my chain confusion.

TomSF · Fri Mar 10, 2023 8:17 pm

Hi, your tests make no sence..how you can get on AC 945Mb/s? Iguess you mean thorough LAN.
Yes, I meant LAN only. For now, I have taken wifi out of my investigation, although I wonder if the LAN-only weirdness is related to my poor wifi performance. Sorry for all the confusion.

I realized I had not tested one btest scenario; ax3<->ccr.
With client on ax3
- receive=95.4
- send=955
With client on ccr
- receive=957
- send=95.6
Those numbers match the btest results between a hap ac and the ax3, which makes sense since the packets are flowing through the ccr to and from the ax3. So, the issue seems to be between the ax3 and ccr. That would also explain wifi issues with the ax3 since all packets have to flow through the ccr. Any suggestions/solutions to the ethernet issues between the ax3 and the ccr? Again, all btest'ing is LAN only!

bpwl · Fri Mar 10, 2023 8:51 pm

bpwl can you tag a WLAN??

Yes

What happens with the WLAN1 wireless settings ..... when set to

(case 1) If "Vlan mode = use tag", "VLAN ID = 5" , the packets from the bridge will be filtered on VLAN ID, will be tagged when sent to the bridge, untagged when sent to wifi
WLAN1 now acts a a VLAN aware switch, where the wifi side is an access port for PVID=5, and the bridge side is a trunk port
(*) !

(case 2) If "Vlan mode = no tag", "VLAN ID = 1" , no VLAN filtering done, the packet tags will not be touched when sent out to wifi, or when received from wifi
WLAN1 now acts as a dump switch, totally unaware of VLANs, and just forwards everything as is.

Your Access port
case 1: WLAN wifi will send packets it receives as tagged with the correct VLAN ID from the bridge as untagged wifi , and delivers wifi received untagged as VLAN tagged to the bridge : not as expected for a bridge access port
case 2: the VLAN 5 packets will be untagged by the bridge (access port) and as such sent and received as untagged by WLAN wifi (as is / untouched / dumb / transparent )

Your Trunk port
case 1: WLAN wifi will send and receive VLAN tagged packets from and to the bridge with the correct VLAN ID, and wifi send or receive them as untagged wifi.
case 2: the VLAN packets if delivered as VLAN tagged by the bridge are sent and received untouched as VLAN tagged wifi by WLAN (VLAN over wifi is no problem, even hybrid)

(*) access lists overrule the VLAN ID and VLAN Mode "use tag" or "no tag" setting per connection , if set in the access list rule !!! I expect RADIUS attributes to do similar things.

case 1 : WLAN1 bridge side is tagged, wifi side is untagged
case 2 : WLAN1 unaware of VLAN, just forwards as is

Double VLAN tags to be studied. (VLAN tunneling, Q-in-Q, Tag stacking, ..... ) .... pfffff

TomSF · Fri Mar 10, 2023 11:49 pm

Yes, I meant LAN only. For now, I have taken wifi out of my investigation, although I wonder if the LAN-only weirdness is related to my poor wifi performance. Sorry for all the confusion.
I realized I had not tested one btest scenario; ax3<->ccr.
With client on ax3
- receive=95.4
- send=955
With client on ccr
- receive=957
- send=95.6
Those numbers match the btest results between a hap ac and the ax3, which makes sense since the packets are flowing through the ccr to and from the ax3. So, the issue seems to be between the ax3 and ccr. That would also explain wifi issues with the ax3 since all packets have to flow through the ccr. Any suggestions/solutions to the ethernet issues between the ax3 and the ccr? Again, all btest'ing is LAN only!

For what it might be worth, I did one more test. I put a hap ac between the ax3 and the ccr and did a btest between the ac and the ax3. The numbers were the same as the test between the ax3 and the ccr. Is there an ax3 user out there that would perform a similar test between their ax3 and a router it is connected to? Is it unreasonable to expect ethernet tx and rx symmetry with the ax3, given that there is symmetry with a hap ac?

Fri Mar 10, 2023 11:51 pm

Between ax3 and rb5009, close to 2.3g using 2.5 ether port.
Up and down.
Both devices on 7.8

TomSF · Sat Mar 11, 2023 12:04 am

Between ax3 and rb5009, close to 2.3g using 2.5 ether port.
Up and down.
Both devices on 7.8

Thanks for the info. Just to be sure, you were using btest reported numbers?

Sat Mar 11, 2023 7:14 am

Yes.
Btest results from AX3 to rB5009, first send, then receive (other way around performing test on RB5009 towards AX3 is same)
On send cpu on AX3 was close to 100%, on RB around 75-80%.
On receive (oddly enough) it became around AX3 70% / RB 50%. Not clear to me why there is such a difference but as we all know, btest is not to be run on the device itself you are testing. Those are the only ones I have with a port capable of 2.5Gb so nothing else available to fill that pipe

2023-03-11_06-20-51.jpg

Someone will surely make a remark about the packet loss. I redid the test using TCP, results as follows:
AX3 -> RB5009:
RX: 2.1Gbps
TX: 940Mbps (CPU local on 34%, remote on 53%)
0 packet loss

And now some will be triggered, so redid the same test from RB to AX3:
TX 2.2 Gbps
RX 2.3Gbps (CPU on AX3 was not 0%, that's a bug I've seen appearing from time to time in that tool. I saw it was around 91%)
0 packet loss

The connection between AX3 and RB5009 is as far as I can conclude perfectly capable of delivering those speeds in both directions.
But it depends where you start.
Again, btest is NOT a tool you should run on the device under test.

2023-03-11_06-36-18.jpg

TomSF · Sat Mar 11, 2023 5:13 pm

Great info, thanks. Yes, I know not to run btest on the router but desperate times call for desperate measures. In my testing I only experienced packet loss when I had the direction set to "both". I have done so many tests that I have gotten totally confused. I am going to rerun tests and record the results methodically in preparation for submitting a support ticket to MT. I'll post a condensed version of the results when I am done. One thing I discovered is that I can get even weirder results by playing with the btest local and remote transfer speeds.

TomSF · Sun Mar 12, 2023 3:00 am

My problem has been solved, and it only took me most of the day!! Once I turned off hardware offload for the ax3 port connected to the ccr, the order of magnitude difference between Tx and Rx went away. My Speetest results went from 94.7 to 233 (not quite as good as the 241 when I was associated with a hAP ac). During my tests I noticed that my low Speedtest result was exactly the same as btest send speed to the ax3 or the receive speed when btest was running on the ax3, so I was convinced that something was throttling wifi speeds. What is odd is that my PC is wired to a port on the ax3 and Speedtest run on the PC got 608 which is probably all my ISP is giving me. I cannot explain why the PC speed was not affected by the hardware offload issue.

Another unexplained weirdness was that when btesting to a btest server on my PC, with the default 20 connections, I only can send at 11.7 Kbs. Yes, Kbs! When I change the connections to 1, I send at 710 Mb/s. Also, I cannot connect from the PC btest to any btest server running on a Mikrotik device. At this point, I don't really need to figure out those issues.

TomSF · Sun Mar 12, 2023 8:27 pm

A brief update. I found Mikrotik documentation that said hardware offload for the IPQ PPE chip used in the ax3 is a work in progress and to not use it. Instead, turn on rstp for the bridge. I turned off hardware offload for all the ports and turned on rstp. Speedtest now has 764 download speed! That is better than my PC wired to the ax3.

alibloke · Sun Mar 12, 2023 10:08 pm

I turned on rstp, but not sure where you disable hardware offloading. Any clues?

TomSF · Sun Mar 12, 2023 10:14 pm

I turned on rstp, but not sure where you disable hardware offloading. Any clues?

I found two places that look relevant. In the switch settings, uncheck the box that says, "Switch all ports". Then for each bridge port (I have all ports in the bridge), uncheck the box that says, "Hardware offload".

alibloke · Sun Mar 12, 2023 10:18 pm

Thanks, that didn't seem to do anything for me, still stuck at 500Mbps with capsman enabled compared to 7-800Mbps without.

TomSF · Sun Mar 12, 2023 10:23 pm

Thanks, that didn't seem to do anything for me, still stuck at 500Mbps with capsman enabled compared to 7-800Mbps without.

I am not using capsman, since I only have 1 wifiwave2 device and its capsman cannot configure my other 7 access points.

alibloke · Sun Mar 12, 2023 10:42 pm

Interesting, enabling rstp on the capsman device (rb5009) has increased speeds to ~670Mbps.

mkx · Mon Mar 13, 2023 9:00 am

Thanks, that didn't seem to do anything for me, still stuck at 500Mbps with capsman enabled compared to 7-800Mbps without.

Do you have local-forwarding set to no in CAPsMAN datapath configuration (it's default so this property might not be shown in export)? This setting is known to reduce throughput for wireless clients due to much increased processing overhead (CAP has to push all traffic in an encrypted tunnel towards CAPsMAN, CAPsMAN has to do it as well, additionally CAPsMAN has to bridge/switch all the traffic, even traffic which is targeting LAN hosts).

alibloke · Mon Mar 13, 2023 9:28 am

Wifiwave2 capsman only supports local forwarding.

natman · Sat Mar 18, 2023 6:23 pm

Just got ax3 couple of days before and having very poor wifi performance. Sitting next to router(~100cm), 5G band shows -60dBM, and 2.4G band shows -30dBM. And ISP router too measures -30 dBM in 5G band. My laptop having intel 210(Precision 7560) sitting about 10ft with one concrete wall having very poor performance. Laptop achieves > 100Mbps in speedtest.net results consistently in 2.4G/AX mode, but in 5G/AX it gives around 60Mbps. ISP router's 5G/AC gives more than 200Mbps. My internet speed is 500 Mbps symmetrical. My country is set to India and when i tried setting to US, 5G band given above 100 Mbps, but still far less than AC router given by ISP.
I'm new to mikrotik, so far using the default settings, have raised support too. All the configuration done in wifiwave2.

Terminal output for command /interface/wifiwave2/monitor shows tx power 18 for 5G and 16 for 2.4G.
Is there anyway to enhance the range and speed in 5G AX?

Rox169 · Sat Mar 18, 2023 9:58 pm

try 80Mhz

bpwl · Sat Mar 18, 2023 10:01 pm

ISP router too measures -30 dBM in 5G band.

-30 dBm That is strong, very strong! Might overload your receivers. Move away from the ISP router or reduce it's power.

TomSF · Sat Mar 18, 2023 10:14 pm

I turned on rstp, but not sure where you disable hardware offloading. Any clues?
I found two places that look relevant. In the switch settings, uncheck the box that says, "Switch all ports". Then for each bridge port (I have all ports in the bridge), uncheck the box that says, "Hardware offload".

Earlier in this thread are my posts regarding the poor performance I experienced. My solution was to turn off hardware offloading for the port connected to your router and turn on rstp. Give it a try and let us know if it helped.

natman · Sun Mar 19, 2023 7:32 am

try 80Mhz

Its already set to 20/40/80(same in ISP router too), there is no option of setting 80 alone.

-30 dBm That is strong, very strong!

Agree, its kept in adjacent room which is 10-12ft away and planning to replace with this ax3. -30dB is just with in 50-60cm distance from the router. In fact ax3 too has similar intensity for 2.4G band in that distance. Poor performance measured only for 5G band.

In the switch settings, uncheck the box that says, "Switch all ports". Then for each bridge port (I have all ports in the bridge), uncheck the box that says, "Hardware offload".
My solution was to turn off hardware offloading for the port connected to your router and turn on rstp.

In my case, 'Switch all ports' was already unchecked and rstp protocol was selected too. I cannot find 'Hardware offload' option anywhere and terminal too does not support l3-hw-offloading commands. Not sure, whether these options were removed in 7.8?

What im confused is, why 2.4G AX gives me > 100Mbps speed, but 5G AX gives only 60M just sitting 10ft away from the router

TomSF · Sun Mar 19, 2023 5:18 pm

I turned off hardware offload individually for each of the 5 ax3 bridge ports. I had bridged all the ports. I just noticed that my wireless interface has it turned on, but it was dynamically added to the bridge, and I do not see a way to turn it off.

While I was troubleshooting the performance issue, I discovered it was not really a wireless issue. I used btest to test the ethernet speed between the ax3 and the router. Running a btest server on one and a btest client on the other and then running a send test followed by a recieve test, I observed an order of magnitude difference between the sending and receiving speeds. The slower speed was what I was seeing from wifi. Turning off hardware offload on all the bridge ports made the speeds equal and the faster of the two. You might try doing a similar btest briefly since btest will bog down the router. If you see that order of magnitude difference then it is probably not a wifi issue, rather an ethernet issue.

I think you said your router was an ISP router so you might not be able to run btest on it. If so, then maybe you can test from something connected to the ax3 via a cable to something not connected to the ax3, thus passing through the ax3-router connection. I have 8 Mikrotik access points, all connected to the router, so I was able to use one of them for the btest. I also used btest.exe on a Windows machine but for some reason, I was only able to run a btest server on the Windows machine. The Windows machine would never connect to a btest server on the LAN. My Windows machine is wired to the ax3 so I was able to test to the MT router and other access points.

BTW, I am running 7.8 too.

Grant · Mon Mar 20, 2023 4:47 pm

I got ~900 Mb/s over wifi:
ROS 7.8
skip-dfs-channels=10min-cac
width=20/40/80mhz
country=empty (or usa) - (at other settings the speed is 30% slower)
frequency=auto (or 5745)
tx-power can be reduced to 20, because the default is 27dBm
maximum speed was obtained in unii-3,

and different bands have different output power (https://help.mikrotik.com/docs/pages/vi ... =150601820)

hap ax3 should support unii-4, but it does not.

photo_2023-03-20_16-46-48.jpg

Mon Mar 20, 2023 5:04 pm

hap ax3 should support unii-4, but it does not.

Yup, noticed it as well.
viewtopic.php?p=990865#p990682

I already created a support ticket.
SUP-111009

bpwl · Mon Mar 20, 2023 10:31 pm

I got ~900 Mb/s over wifi:
ROS 7.8
skip-dfs-channels=10min-cac
width=20/40/80mhz

Any tests with 80+80 MHz bandwidth? (80MHz primary, 80 MHz secondary bandwidth)

As documented in wifiwave2: "frequency" in XXXX notation (C channel center freq), and "secondary frequency" with 80MHz center freq notation
In classic WLAN driver this should be the parameters labeled "frequency" and "secondary channel"

Grant · Mon Mar 20, 2023 11:10 pm

Any tests with 80+80 MHz bandwidth? (80MHz primary, 80 MHz secondary bandwidth)

there is only one option here

photo_2023-03-20_23-10-44.jpg

Rox169 · Tue Mar 21, 2023 12:32 am

AX2 and AX3 has maximum 80Mhz.

bpwl · Tue Mar 21, 2023 1:06 am

If you select 20/40/80MHz, only then the secondary frequency field will appear in Winbox.
Then you can enter secondary frequency for another 80MHz bandwith, eg work with 80+80MHz bandwidth operation in 802.11ac and ax.

It is called Secondary Channel in the WLAN classic setup.

Klembord-2.jpg

viewtopic.php?t=193805#p987667
The primary channel seems to be "dynamic bandwidth" , one C channel (20MHz wide) , plus 1 or 3 adjacent e-channel extensions
The secondary channel is usually in another range, as the 80 MHz wide channel is used (and here indicated by the 80 MHz center frequency)

anav · Tue Mar 21, 2023 1:58 am

Hoping you get your MT wifi6 soon bpwl, cannot wait for the 'blessed' configuration that works!!

Grant · Tue Mar 21, 2023 5:56 am

If you select 20/40/80MHz, only then the secondary frequency field will appear in Winbox.
Then you can enter secondary frequency for another 80MHz bandwith, eg work with 80+80MHz bandwidth operation in 802.11ac and ax.

It is called Secondary Channel in the WLAN classic setup.

Klembord-2.jpg

viewtopic.php?t=193805#p987667
The primary channel seems to be "dynamic bandwidth" , one C channel (20MHz wide) , plus 1 or 3 adjacent e-channel extensions
The secondary channel is usually in another range, as the 80 MHz wide channel is used (and here indicated by the 80 MHz center frequency)

so which secondary channel should i choose? what happens if you leave it empty?

Tue Mar 21, 2023 6:03 am

AX2 and AX3 has maximum 80Mhz.

.

there is only one option here
photo_2023-03-20_23-10-44.jpg

.

Strange. I see something else.

2023-03-21_05-10-37.jpg

But it does give me "unsupported channels" when trying to use it.

Grant · Tue Mar 21, 2023 6:15 am

AX2 and AX3 has maximum 80Mhz.
.
there is only one option here
photo_2023-03-20_23-10-44.jpg
.

Strange. I see something else.
2023-03-21_05-10-37.jpg

But it does give me "unsupported channels" when trying to use it.

it's in the profiles, but it doesn't work

Rox169 · Tue Mar 21, 2023 7:18 am

Hi,

[moderator removed abusing part] did you checked HW info before buying this device? Im teling you AX2 and AX3 has maximum 80 Mhz.Yes, there is option 80+80 but it is for the future HW.

Tue Mar 21, 2023 8:09 am

No need to be rude.

In all seriousness...
Seems you are correct though.
IPQ6010 does not support more then 80MHz.
Just checked on AC3 ( same wifiwave2 package), same options are there as well (160 and 80+80).
Pretty misleading to offer options which are not supported by underlying HW.

Tue Mar 21, 2023 8:36 am

hap ax3 should support unii-4, but it does not.
Yup, noticed it as well.
viewtopic.php?p=990865#p990682

I already created a support ticket.
SUP-111009

Response received.
Regardless what the release notes state, currently ONLY supported on AX2 (because of internal antenna restriction).

Grant · Tue Mar 21, 2023 8:41 am

Here's what support told me

UNII-4 is supported only for boards that have inner antennas, for now, it's only hAP ax2 sorry for the inconvenience caused.

Tue Mar 21, 2023 9:48 am

Just received an additional response: the wording of release notes is wrong.
It should be AX2 for US and Europe.
But AX3 / Chateau AX only for Europe is supported as well. (because of the FCC limitation on external antenna, they are excluded for US, I understand that)

bpwl · Tue Mar 21, 2023 11:54 am

so which secondary channel should i choose? what happens if you leave it empty?

What the DOCUMENTATION/HELP says about wifiwave2 ROS7:

Klembord-2.jpg

This is about the software , hardware limitations for specific models are not mentioned.

And indeed hAP ac³ has the same limitation in classic driver .... even with the correct value

Klembord-3.jpg

Same limitation on wAP ac , hAP ac², Omnitik AC, .... probably all MT devices.

Grant · Tue Mar 21, 2023 4:45 pm

I was able to turn on the 5865 channel on the hap ax3, but only on 20MHz and any European country, because the channels above are not available and 5885 does not turn on
I can also turn on channel 5845 only on 20/40MHz eC and 5825 on 20/40MHz Ce respectively

Channels 5865 and 5885 are hardware disabled due to external antennas so 20/40/80MHz can't be turned on with 5825 and above

anav · Tue Mar 21, 2023 4:52 pm

Just received an additional response: the wording of release notes is wrong.
It should be AX2 for US and Europe.
But AX3 / Chateau AX only for Europe is supported as well. (because of the FCC limitation on external antenna, they are excluded for US, I understand that)

What If I choose some backwards european country like Belgium for my country on the AX3?? or is the limitation in hardware (where sold).

bpwl · Tue Mar 21, 2023 5:06 pm

Hoping you get your MT wifi6 soon bpwl, cannot wait for the 'blessed' configuration that works!!

Well still reading here, and hoping MT wifi6 gets to a good operational level first, with wifi6 features, before investing in that hardware.
Well not having 4-address mode, even lacking WDS AFAIK, and now even 80+80 or 160MHz still missing, is QBSS already in the beacon? band steering? and what about BSS coloring?
Many MT good features are poorly documented and stay unused, and some documented cannot be used.
(80+80 was in the release notes in 2017 (viewtopic.php?f=7&t=121883&p=626033#p625962) , usable now in RB4011 and Audience??? Any other ? viewtopic.php?t=109764#p672207

Tue Mar 21, 2023 5:08 pm

What If I choose some backwards european country like Belgium for my country on the AX3?? or is the limitation in hardware (where sold).

It could be HW locked so impossible to do.
Regardless, it is most likely illegal where you live

In all seriousness, in some countries there are really HEAVY fines for circumventing those limitations.
So proceed carefully if you want to explore that option ...

EgidijusL · Wed Mar 22, 2023 1:03 pm

And here is my practice after the updates:
https://drive.google.com/file/d/1OvOir0 ... sp=sharing
As you can see in the speedtest.net graph, the download speed is not only low, but unstable, compared to v7.6.

v7.6 and v7.8 settings:

default-name="wifi1" name="wifi1" mac-address=xx:xx:xx:xx:xx:xx arp-timeout=auto radio-mac=xx:xx:xx:xx:xx:xx
        configuration.mode=ap .ssid="EE-Wifi-5G" .country=Lithuania 
        security.authentication-types=wpa2-psk,wpa3-psk .passphrase="PASSOWRD" 
        channel.band=5ghz-ax .width=20/40/80mhz .skip-dfs-channels=10min-cac

with v7.6

channel: 5640/ax/eeeC
tx-power: 28

with v7.8

channel: 5680/ax/eCee
tx-power: 23

When set no Country, get 5745/ax/Ceee channel

Grant · Wed Mar 22, 2023 3:36 pm

When set no Country, get 5745/ax/Ceee channel

no Country=USA and 27dBm or 28dBm, 5745 because it is limited DFS

How to find out the current signal strength in wifi?

Rox169 · Wed Mar 22, 2023 4:13 pm

how do you know that no country is USA?

EgidijusL · Wed Mar 22, 2023 4:33 pm

When set no Country, get 5745/ax/Ceee channel
no Country=USA and 27dBm or 28dBm, 5745 because it is limited DFS

How to find out the current signal strength in wifi?

I use the command: /interface/wifiwave2/monitor 0

Grant · Wed Mar 22, 2023 6:24 pm

how do you know that no country is USA?

because this is the default value and it is written in the documentation

metricmoose · Sat Mar 25, 2023 1:07 am

I also seem to be having issues achieving above 300 Mbps download from my ax3. I've confirmed it with both an ax-capable Dell laptop and Pixel 6 Pro phone to an LibreSpeed HTML5 speedtest server on my LAN and also doing a local file copy from a NAS, both connected to the same CRS328 switch. TCP btests between my RB5009 and hAP ax3 reach over 900 Mbps.

Going by this thread, I've tried disabling hardware offloading on the bridge ports, enabling/disabling flow control, and trying one of the 1G ports after removing the 2.5G port from the bridge with no increase above 300 Mbps. The TX and RX rates in the registration tab show 900-1000 Mbps with the phone and laptop.

/interface wifiwave2 channel
add disabled=no frequency=2412,2437,2462 name=ch-2ghz width=20mhz
add disabled=no frequency=5725-5850,5180-5240 name=ch-5ghz width=20/40/80mhz
/interface wifiwave2 security
add authentication-types=wpa2-psk,wpa3-psk disabled=no management-protection=\
    allowed name=sec1 wps=disable
/interface wifiwave2 configuration
add country=Canada disabled=no name="SSID" security=sec1 ssid=\
    "SSID"
/interface wifiwave2
set [ find default-name=wifi2 ] channel=ch-2ghz configuration="SSID" \
    configuration.mode=ap datapath.vlan-id=100 disabled=no name=wifi-2ghz \
    security=sec1
set [ find default-name=wifi1 ] channel=ch-5ghz configuration="SSID" \
    configuration.chains=0,1 .mode=ap .tx-chains=0,1 datapath.vlan-id=100 \
    disabled=no name=wifi-5ghz security=sec1
    
  
  admin@Hallway AX3 AP] > /interface/wifiwave2/monitor wifi-5ghz 
               state: running
             channel: 5805/ax/eeeC
    registered-peers: 7
    authorized-peers: 7
            tx-power: 28
  available-channels: 5180/ax/Ceee,5200/ax/eCee,5220/ax/eeCe,5240/ax/eeeC,
                      5745/ax/Ceee,5765/ax/eCee,5785/ax/eeCe,5805/ax/eeeC,
                      5180/ac/Ceee,5200/ac/eCee,5220/ac/eeCe,5240/ac/eeeC,
                      5745/ac/Ceee,5765/ac/eCee,5785/ac/eeCe,5805/ac/eeeC

TomSF · Sat Mar 25, 2023 1:52 am

Your configuration looks pretty close to mine. I didn't specify any frequencies, chains or vlan-id though. When I run monitor, I do not see the available channels listed, as your output shows. I have seen weirdness testing with my android phone. At times, usually after turning off wireguard, a speedtest will get something like 40 Mbs. If I turn off wifi and then turn it back on again, causing it to reassociate with the ax3, the speedtest will get 551 Mbs. I have no idea why, but I think it is a phone issue, not an ax3 issue. I did a test with librespeed.org and got nowhere near the numbers I got with speedtest. I only used btest to test on my lan. What do you see if you run btest between the RB5009 and the ax3? Do a send, and then a receive. It doesn't matter which is the server or client. Are the send and receive speeds close to each other?

metricmoose · Sat Mar 25, 2023 9:36 pm

Your configuration looks pretty close to mine. I didn't specify any frequencies, chains or vlan-id though. When I run monitor, I do not see the available channels listed, as your output shows. I have seen weirdness testing with my android phone. At times, usually after turning off wireguard, a speedtest will get something like 40 Mbs. If I turn off wifi and then turn it back on again, causing it to reassociate with the ax3, the speedtest will get 551 Mbs. I have no idea why, but I think it is a phone issue, not an ax3 issue. I did a test with librespeed.org and got nowhere near the numbers I got with speedtest. I only used btest to test on my lan. What do you see if you run btest between the RB5009 and the ax3? Do a send, and then a receive. It doesn't matter which is the server or client. Are the send and receive speeds close to each other?

My Librespeed tests aren't run using their website, I'm using a locally-hosted version on my home LAN. It makes it convenient to test both on my phone and other devices and get somewhat comparable results.

Doing a TCP btest from the RB5009 to the ax3 (RB5009 -> CRS328-24P-4S+RM -> ax3) gets 930 Mbps in each direction. Doing a TCP btest to the laptop gets a new high score of about 400 Mbps, though file copies from my NAS and Librespeed tests get about the same 250-300 Mbps as I get from the phone, while tests run while connected to the same switch as the ax3 show full gigabit file copy and Librespeed tests.

theprojectgroup · Mon Mar 27, 2023 8:21 pm

v7.8 [stable] is released: viewtopic.php?t=193986
- Update:
- still the same issue - ~100Mbits/sec TCP download on client / 780 Mbits/sec upload
- UDP is fine
v7.9beta [testing] is released: viewtopic.php?t=194781
- Update:
- still the same issue - ~100Mbits/sec TCP download on client / 780 Mbits/sec upload
- UDP is fine

EgidijusL · Thu Mar 30, 2023 3:17 pm

@theprojectgroup
Using automatic frequency?

theprojectgroup · Sun Apr 02, 2023 8:11 pm

No, it was:

name="ch-5ghz-march" frequency=5180,5260,5500 width=20/40/80mhz skip-dfs-channels=all

I just tested with fully auto. Does it mean to leave everything empty? Seems to work but TCP is still very slow!
Other none AX devices like 802.11ac don't show this behaviour...

name="ch-5ghz-april"

CleanShot 2023-04-02 at 19.07.15@2x.png

CleanShot 2023-04-02 at 19.07.39@2x.png

iperf3 -c 192.168.99.11 -t10  -i1 -P1 -u -b 1G -R

Is the WiFi interface somehow hardware offloaded or why do I not see the traffic on the bridge?

CleanShot 2023-04-02 at 19.05.09@2x.png

iperf3 -c 192.168.99.11 -t10  -i1 -P1  -R

CleanShot 2023-04-02 at 19.09.53@2x.png

CleanShot 2023-04-02 at 19.17.08@2x.png

mkx · Mon Apr 03, 2023 3:14 pm

About different personalities of bridge. Interface stats are about bridge interface, which is not on the way between bridge ports (either wired or wireless, doesn't matter).

Regarding throughput results: the UDP test shows that wireless was able to transfer data at around 900Mbps (one has to observe the receiver side stats). It's TCP that struggles. And TCP has its own quirks, and when wireless is involved (can be broadband, e.g. 4G, as well), it's delay jitter which kills its performance. You might get much better overall performance figures if running iperf3 with multiple parallel streams ("-P 8" or something).

Sad thing, which most wireless zealots fail to acknowledge, is that wireless, even if technically it provides high throughput, will never be as good as wired (when promising same throughput) due to systematically longer delay, larger delay jitter and susceptibility to external interference. Add also duplex delay/issues if Rx and Tx use same frequency channel (which is the case with WiFi).
Wireless will never be as good as wires.

anav · Mon Apr 03, 2023 3:37 pm

So how do you propose to string that comm line to the moon??

Mon Apr 03, 2023 3:40 pm

Correct observations.
All too often when toying with setups, I notice iperf3 tests using TCP are quite a bit lower then when using UDP (albeit some minor packet loss can be observed, which does usually not happen with TCP).
Multiple streams also helps (I already have -P 10 embedded in my muscle memory

)

Wireless will never be as good as wires.

Let's repeat that once more.

Mon Apr 03, 2023 3:41 pm

So how do you propose to string that comm line to the moon??

And what do you think will happen over time ?
Earth rotates, that string will pull the moon closer !!