Migrating a mesh of RB952Ui-5ac2nD to single C53UiG+5HPaxD2HPaxD

RouterOS Wireless Networking
1

Got myself an early Christmas present and currently planning the transition.

I have a few modern (ax-capable) laptops, mobile devices and a bunch of low-power (n) IoT devices. There is no need for much bandwidth and I’m happy to trade it to reduce lag, such as when low-powered devices join and force AP to adjust channel and/or rates.

Current configuration:

Main
Runs at full power, taking the best channels. Most devices are connected to it.

/interface wireless channels
add band=2ghz-onlyn frequency=2427 list="2.4" name=ch4 width=20
add band=2ghz-onlyn frequency=2452 list="2.4" name=ch9 width=20
add band=5ghz-onlyac extension-channel=eeCe frequency=5700 list="5 DFS-80" name=ch140-80 width=20
add band=5ghz-onlyac extension-channel=eeCe frequency=5620 list="5 DFS-80" name=ch124-80 width=20
add band=5ghz-onlyac extension-channel=eeCe frequency=5540 list="5 DFS-80" name=ch108-80 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5700 list="5 DFS-40" name=ch140-40 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5660 list="5 DFS-40" name=ch132-40 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5620 list="5 DFS-40" name=ch124-40 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5580 list="5 DFS-40" name=ch116-40 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5540 list="5 DFS-40" name=ch108-40 width=20
add band=5ghz-onlyac extension-channel=Ce frequency=5300 list="5 DFS-40" name=ch60-40 width=20

/interface wireless
set [ find default-name=wlan1 ] adaptive-noise-immunity=ap-and-client-mode ampdu-priorities=0,1,2,3,4,5 antenna-gain=1 arp=disabled band=2ghz-onlyn \
    basic-rates-b="" country="united states" disabled=no disconnect-timeout=5s distance=indoors frequency=auto hw-protection-threshold=512 mode=ap-bridge \
    multicast-helper=full name=livingroom-2.4 rate-set=configured scan-list="2.4" security-profile=... ssid=... station-roaming=\
    enabled supported-rates-a/g=6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps supported-rates-b="" vlan-id=2 vlan-mode=use-tag wds-default-bridge=bridge \
    wireless-protocol=802.11 wmm-support=enabled wps-mode=disabled
set [ find default-name=wlan2 ] adaptive-noise-immunity=ap-and-client-mode ampdu-priorities=0,1,2,3,4,5 arp=disabled band=5ghz-onlyac basic-rates-a/g=9Mbps \
    channel-width=20/40/80mhz-XXXX country="united states" disabled=no disconnect-timeout=10s distance=indoors frequency=auto hw-protection-threshold=512 \
    mode=ap-bridge multicast-helper=full name=livingroom-5 rate-set=configured scan-list="5 DFS-80,5 DFS-40,5" \
    security-profile=... ssid=... station-roaming=enabled supported-rates-a/g=9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps,54Mbps vlan-id=2 \
    vlan-mode=use-tag wireless-protocol=802.11 wmm-support=enabled wps-mode=disabled

Supplemental
Runs with reduced power in hopes to minimize AP switching on devices. Connects all IoT devices and covers devices that happen to get too far from Main

/interface wireless channels
add band=2ghz-onlyn frequency=2412 list="2.4" name=ch1 width=20
add band=2ghz-onlyn frequency=2417 list="2.4" name=ch2 width=20
add band=2ghz-onlyn frequency=2437 list="2.4" name=ch6 width=20
add band=2ghz-onlyn frequency=2442 list="2.4" name=ch7 width=20
add band=2ghz-onlyn frequency=2462 list="2.4" name=ch11 width=20

/interface wireless
set [ find default-name=wlan1 ] adaptive-noise-immunity=ap-and-client-mode ampdu-priorities=0,1,2,3,4,5 antenna-gain=5 arp=disabled band=2ghz-onlyn \
    basic-rates-b="" country="united states" default-authentication=no disabled=no disconnect-timeout=10s distance=indoors frequency=auto mode=ap-bridge \
    multicast-helper=full name=bedroom-2.4 rate-set=configured scan-list="2.4" security-profile=... ssid=... station-roaming=\
    enabled supported-rates-b="" vlan-id=2 vlan-mode=use-tag wireless-protocol=802.11 wmm-support=enabled wps-mode=disabled
set [ find default-name=wlan2 ] adaptive-noise-immunity=ap-and-client-mode ampdu-priorities=0,1,2,3,4,5 antenna-gain=7 arp=disabled band=5ghz-onlyac \
    channel-width=20/40mhz-XX country="united states" disabled=no disconnect-timeout=5s distance=indoors frequency=auto mode=ap-bridge multicast-helper=\
    full name=bedroom-5 security-profile=... ssid=... station-roaming=enabled vlan-id=2 vlan-mode=use-tag wireless-protocol=802.11 \
    wmm-support=enabled wps-mode=disabled

My only issue with this config is that when mobile devices switch APs the lag is unbearable as it usually happens during a video call. Otherwise the coverage and performance is acceptable.

Transition plan:


  1. Lock the current APs to one channel, put the new AP to where it’s going to be and run spectral scan on it for a day, at intervals of 5s, saving all to a file
  2. Transition network to the new AP, lock it to one channel, put old APs to locations where I need coverage and run spectral-scan on them, at intervals of 5s, saving all to a file
  3. Build the channels list based on this data

Newbie question: can one separate chains (MIMO) to get a better configuration of channels and rates, e.g. one antenna on “free” channels with high rates, and another on “busy” channels and low rates?

2

No, MIMO chains are part of same radio and can not be used individually.

Even devices with proper dual radios have to be used very carefully not to destroy the other radio’s receivers if they are hardware-wise capable of running in same frequency spectrum.

3

Newbie question #2:

When setting band in the channel profile of wifiwave2 / WiFi, do I set the minimum required or maximum allowed standard? I.e. when configuring a 2.4Ghz chain to accommodate both 802.11n and 802.11ax, what value must I use?

4

My impression is that you’re setting the “newest” standard, older than setting are then supported as well. E.g. if you set band=5ghz-n, AP will support 802.11a and 802.11n but will not support 802.11ac nor 802.11ax.
To support all standards, set highest supported by AP hardware (i.e. 5ghz-ax).

5

I contacted the support (requesting to improve the online doc) and got the following reply:

Band setting specifies what band/wireless standard should be used, “Supported frequency band and wireless standard. Defaults to newest supported standard.”.
Wireless standards are made to be backwards compatible. So if you set 2ghz-n, it means that AP will work as 802.11n AP, forgoing the 802.11ax functionality.
2ghz-ax should be used in most, if not all, cases.
There is no “allowed” clients, no filtering is done by band setting. As an example, if you set AX band, either on 2.4 or 5GHz interface, AX clients will be able to join and use higher rates among other AX features. 802.11AX is backwards compatible, so 802.11n clients will also be able to join MikroTik AP that has 802.11ax band set, they just won’t be able to use 802.11ax feature set, as the 802.11n client itself doesn’t support it.
If you set 802.11n band, 802.11n clients will have mostly the same experience as if they were connecting to 802.11ax, but 802.11ax clients connecting to 802.11n AP will have to limit themselves to 802.11n feature set.

@mkx, your understanding is correct. It’s a bit confusing because previously there was the “onlyn” option. I wonder if it’s still possible to configure channels in a way to “force” 802.11n.

6

No, not to my understanding. With wifi/wave2 we’re back to supporting legacy clients (e.g. 802.11a and 802.11b).

7

I saw elsewhere, that VLAN needs to be configured via /interface/bridge/port, i.e. the wifi interfaces deal with untagged traffic. At the same time I do see that both Access List and Datapath allows some VLAN configuration. How do these options coexist?

8

IMO things around (new) wifi are very murky right now. 7.13beta bringing wifi separated into different packages is IMO proof that MT is in the middle of serious reworking of wave2/wifi … so we’ll have to be a bit patient and wait to see what will come out of this process. I’m hoping to see vlan-handling return to wifi drivers (if nothing else, it’ll make @pe1chl stop whining about it :wink:).

9

The same settings exist for wifiwave2.

10

If the existing setting gets ignored, what good is it then? Let’s be patient and see what 7.13 stable brings us, shall we?

11

Please bear with me, because I do struggle to get a coherent picture of the wifiwave2 (and very wifi, which is very similar) configuration.

Are you saying that VLAN-related settings do not work at all, they do work but only under CAPsMAN or it’s a mix of that?

12

I’m saying that wifiwave2 (and its successor wifi) is under active development. Some config options might exist, but the functionality is yet to come (doesn’t happen often, but can happen), options might still exist but functionality is getting deprecated … or options exist, but the way they affect functionality is changing. So it’s probably not worth to invest too much time or energy into building mental picture of how things should work until things stabilize.

Alas: the old wireless driver had capability to deal with VLAN tags. Settings under Access List affect it (per-client VLAN ID). The new driver (and both old and new capsman) has Datapath setting which currently affects the way wifi interface is made bridge port (i.e. it sets wifi port PVID), but actual VLAN tag handling is left to bridge, which has to be set up with vlan-filtering=yes. And one can set these things manually (for local wifi interfaces) …
But things might change (me and @pe1chl hope to see the wlan driver handling VLAN tags again).

13

For the love god, where do I see the channel my radios are roaming in wifiwave2?

/interface/wifiwave2/monitor

14

How does the antenna-gain property work on AX3 and wifiwave2? The spec says the antenna gain is 5.5dBm, but I don’t see any reduction of the tx power in /interface/wifiwave2 monitor until I set configuration.antenna-gain to 11dBm (-1dBm in tx power).

My goal is to reduce power for 5Ghz roaming, so that far away devices stay connected to 2.4Ghz instead of going back and forth.

15

It’s a calculation: EIRP (which is Tx power + antenna gain) must not exceed country limitations.

hAP ax3 on 5GHz has Tx power anywhere between 20dBm and 28dBm (depending on radio symbol rate). If talking about fastest rates, where lowest Tx power can be used (20dBm), and when using channels without any specific power limitations, then it’ll probably limit Tx pwoer to keep EIRP lower than 30dBm. Which gives 10dB as headroom for gain.
At lower rates (and I guess that becons fall into this lower rate category) the difference should be visible at once since with default setting (it’s 6 dB, it’s only possible to enter integer numbers and 5.5 rounds up to 6) already means Tx power has to be limited to 24dBm to keep EIRP below 30dBm and setting antenna gain to higher values (e.g. to 7) means actual Tx power reduced by appropriate amount. Since actual antenna gain doesn’t change (as Normis like to point out), it’s actual EIRP which reduces below used limit.

The logic above should apply both to legacy wireless driver and new wave2/wifi driver. The only difference is that new driver supports TPC so it can use 3dB higher power than legacy wireless driver when using channels requiring TPC.

16

I don’t think this will solve the ping-pong problem … it’ll just move it closer to AP (or even prompt brain-dead stations to remain on 2.4GHz even when really close to AP). Many people solved this problem by decreasing 2.4GHz Tx power so that dual-band stations won’t connect to 2.4GHz at all. Hence 2.4 GHz will only be used by stations not capable of 5GHz. Perhaps it also helps roaming back to 5GHz because it becomes stronger when getting closer to AP. But yes, you do loose wifi coverage, offered by 2.4GHz when transmitting at full power. So you have to accept some trade-off (or decide to sacrifice one function to have another one). 802.11r/k/v should help with this problem, but only for stations fully supporting these standard.

If you are going to go with reduction of Tx power, do it by setting configuration.tx-power to some value … it’s taken into account when determining needed Tx power (to stay below EIRP) and you don’t have to play with antenna-gain.

17

I have a place in my home where overall signal drops sharply, with about 10dBm difference between 5Ghz and 2.4Ghz. Apparently for this very specific case 802.11r/k/v does nothing useful. The station clearly supports it (iPhone 13) but in the logs, in that very specific spot, I see a ping pong rather than graceful roaming.


My understanding is that Tx Power is rate-specific and I tried to avoid setting a single constant value by providing a negative difference instead. I.e. instead of a constant 17dBm I want -3dBm of the maximum allowed Tx Power per given rate, regulatory domain (US) and channel (DFS). But further thinking about it, constant Tx Power probably makes more sense in this case.


Does each of the two antennas propagate in 2.4Ghz and 5Ghz bands or is it a split where one antenna is 2.4Ghz and the other is 5Ghz?

18

Increasing antenna gain and decreasing Tx power has same end effect. Normis explained in a post a few years ago (I can’t find a reference, but I remember the contents very well) that wifi radio chooses actual Tx power to be the lowest of these values:

  • country EIRP limitation less antenna gain
  • Tx power setting
  • chipset capability

Country limits usually don’t take symbol rates into account, neitger does antenna gain. Tx power setting is nowdays again constant for all rates (Tx power table is history long gone). The only variable limitation is chipset capability, which decreases with increased rate.

Let’s consider two scenarios:

  1. Varying antenna gain
    I think that default Tx power setting is 30dBm. Let’s say that country limit is 30dBm and we start with default antenna gain of 5.5dBi rounded to 6dBi. So limitations say that radio Tx power should not exceed 24dBm. Chipset capability for low rate is 28dBm, so actual Tx power used will be 24dBm … At MCS7, chipset capability drops to 25dBm, so anything slower will be transmitted at constant power of 24dBm. With max speed (MCS11) chipset capability drops to 20dBm and actual Tx power will follow chipset capability figures.
    If you increase antenna gain to, say, 10dBi, then what changes is limit due to country limit … it drops to 20dBm. So essentially chipset capability is always (but at highest rates) higher than limit that is result of limited EIRP and AP radio will transmit at 20dBm at all rates.
  2. Varying Tx power
    Similar scenario, but this time instead of increasing antenna (from default 5.5dBm) gain we set Tx power to, say, 23dBm (which is lower than country EIRP limit less antenna gain which is 24dBm). So at rates, slower than MCS9, the limiting value will be Tx power setting. At faster rates, the limit will be chipset capability (because it’s lower than 23dBm). If you set Tx power setting to 20dBm, then AP radio will again transmit all rates at same power.

So in both cases you’ll get same behaviour. With possible one difference: if country regulations are not constant for all chanbels and you let device auto-select channel, then setting higher antenna gain instead of lower Tx power can yield in lower actual Tx power used on channels with lower permitted EIRP than with Tx power setting.
Note that coverage edge will not be a hard one (i.e. station transitions from full rate to signal drop) because receive sensitivity drops with increased rate … when station is getting closer to coverage edge, the receive sensitivity prevents it to work with high rates and forces it to downswitch to slower rates.

I don’t know how exactly ax3 uses the two antennas, but I would expect them both be dual-band and hence used by both 2.4GHz and 5GHz radios. Images of device internals, found on internet, don’t reveal any internal antennas. And both bands require two antennas for their 2x2MIMO (dual-chain) operation.

19

Thank you for clarifying the Tx Power selection. I see now that the idea of “a negative difference from the auto selection” is not viable as Tx Power reflects a transmission rate and the transmission rate is reflected in the auto selection.

Perhaps I’m pushing my luck, but I have more question about the AX3 antennas :slight_smile: The are somewhat flattened and I wonder how it affects radiation profile, if at all. In particular which side should face the most penetrable direction (open space, thinner walls or doors etc): wide or narrow.

20

I’ve no idea about exact radiation patterns of those sticks.

I have a suggestion: cut it to half (both along the long axis and perpendicular), take a few good photos and post them here. Then we’ll start to guess about possible radiation patterns :wink:

Without that, my guess is that the only orientation that actually matters is “along the long axis”. In other directions radiation patterns are probably not homogenous, but far from having some systematic shape.