hAP ax3 limits 5GHz TX Power to 8

maciekn · Mon Aug 28, 2023 9:32 pm

Hello,
Recently I've noticed some weird behavior with my hAP ax3 (Router OS 7.11).
Topologically, I have two units, one configured as router, the other works essentially as AP with CAPSMAN (the one which casues this problems).

After a while (hours) _something_ trims down TX Power on device which is managed by CAPSMAN to 8 dBm for 5GHz interface. Other TX powers according to status info seem to be fine. Changing TX power value does not help. Helps only restart, or changing country to US. Once I changed country to US, it stopped to degrade tx power. (not that wifi got much more stable but oh well...)

On daily basis I select country which is inside EU regulatory domain. Here's `actual-config` extract for this interface:

2 name="roof_5g" mtu=1500 mac-address=[MAC] arp-timeout=auto radio-mac=[MAC] 
   configuration.mode=ap .ssid=[SSID] .country=[Conuntry] .tx-power=24 .antenna-gain=3 
   security.authentication-types=wpa2-psk,wpa3-psk .passphrase=[Passphrase] .wps=disable .ft=yes 
   steering.neighbor-group=my-group .rrm=yes .wnm=yes

What magic am I missing? Any hint would be helpful because I'm slowly loosing my mind
Thanks!

İmposss · Mon Aug 28, 2023 9:50 pm

Hello, I have a similar error. As a solution, I choose the wrong radio in the radios tab and fix it by pressing the provision button. Sometimes it is powered by Cap poe, and the system may stop working due to low voltage. If there is autosupout.rtf in the files, the poe voltage is insufficient.

my old post:viewtopic.php?p=1017316#p1017316

Wed Aug 30, 2023 10:20 am

How / where are you monitoring the TX power?
Are you using the wireless link when the power is decreased? There should be something downloading when you check

İmposss · Wed Aug 30, 2023 4:00 pm

How / where are you monitoring the TX power?
Are you using the wireless link when the power is decreased? There should be something downloading when you check

Hi you can see it in this post : viewtopic.php?p=1017316#p1017316

On wifiwave2 tab, two tx power selected on coloumns. One from configuration, one from status on double click.

my config:viewtopic.php?p=1019404#p1019404

Wed Aug 30, 2023 4:43 pm

why make so many independent threads?

İmposss · Wed Aug 30, 2023 6:23 pm

Hello. I don't open new threads. I am writing about my experiences with similar problems. To support myself or to support someone else. There was a return from the support ticket I opened today. My problem was due to choosing different countries for 5 and 2.4 ghz. It used the lowest values on both radios. I confirmed this. I am trying to solve other problems, I will share as I find a solution. Thanks for the response.

maciekn · Thu Aug 31, 2023 7:48 pm

Thank you both for your contributions!

How / where are you monitoring the TX power?
Are you using the wireless link when the power is decreased? There should be something downloading when you check

I monitor TX power by observing TX Power value in status tab in WinBox
I have not observed value increasing while using the interface though I have to say that the interface is not used by many endpoints

Let me share a bit of background story why I even asked this question. I struggle with general wifi instability. There are hours, sometimes days when everything is stable, then *magic* happens and WiFi stability goes to shit - video calls got delays, ping fluctuates between 10ms and 2000ms (most cases 10ms, but once per few pings it gets 2000ms), RDP connection are broken and so on. There are two repeatable patterns I found:

frequency being changed
TX power being trimmed

Fixing channels did not help much, so I started to investigate TX power issue. So far my theory seems consistent - I have not observed problems with WiFi having bigger TX power.

R1CH · Mon Sep 04, 2023 2:07 am

You're probably getting your frequency changed due to radar detect, and it's selecting a TPC DFS channel which is designed for short range devices. Don't pick DFS channels or use "auto" frequency to avoid this.

vtsvetkov · Fri Oct 20, 2023 7:06 pm

I also have tx-power=8 once I set country=Poland. If I unset country or select UK, it increases to 24 or 28 dbi. No radar logged. That is too low signal. I have tried all DFS values, the result is the same. I also reset configuration, simply setting country limits tx-power. What the reason may be?
v.7.11.2

RoadkillX · Sun Feb 04, 2024 1:33 pm

I bought an ax3 recently and wifi was great 800mbps next to the Router and ~400mbps two rooms away. I upgraded to 7.13.3 about a week ago and it went down to 300mbps and -70 Next to the Router and -85 and 10mbps at best two rooms away and lagging and loosing connection. Long story short after reading this thread i changed the country on the interfaces, if i set US tx power on 5ghz goes to 28 and on 2.4ghz goes to 27 if i set another country or none It goes down to 8 on the 5ghz and 16 on the 5ghz and even if modified does not set the new value. I tried downgrading to 7.13.2 and even lower with no luck.

robmaltsystems · Sun Feb 04, 2024 5:01 pm

I bought an ax3 recently and wifi was great 800mbps next to the Router and ~400mbps two rooms away. I upgraded to 7.13.3 about a week ago and it went down to 300mbps and -70 Next to the Router and -85 and 10mbps at best two rooms away and lagging and loosing connection. Long story short after reading this thread i changed the country on the interfaces, if i set US tx power on 5ghz goes to 28 and on 2.4ghz goes to 27 if i set another country or none It goes down to 8 on the 5ghz and 16 on the 5ghz and even if modified does not set the new value. I tried downgrading to 7.13.2 and even lower with no luck.

Maybe connected but I suspect the version before 7.13.3 wasn't setting the regulatory Tx power levels correctly. I certainly spotted a reduction in Tx power in the UK with this version because I happened to be doing some lab work on Tx power when I upgraded. The version notes state that it assumes country = Latvia which possibly has lower Tx power limits than before.

mke · Tue Mar 12, 2024 11:47 am

I am playing with a new hAP ax2, which I bought to specifically be able to run 5ghz on channel 149. The reason for 149 is I am annoyed by my Apple devices constantly splitting wifi airtime between my WLAN and AWDL.

If I set the 5ghz channel to 149 my TX Power drops to 9 and signal on a brand new MacBook Pro is -70 dBm about 4 meters away with line of sight.

I am in Australia and channel 149 has a permitted EIRP of 36 dBm inside and out.

Am I missing something here, perhaps I don't understand how these values get calculated.

Edit: I originally stated TX Power drops to 8, just double checked and it dropped to 9 which align with results in chart below.

robmaltsystems · Tue Mar 12, 2024 12:11 pm

I've got a hAP ax2 as well and yes, almost same in UK - Tx power on channel 149 is 9. Sad as I am, I documented the powers on various devices in the lab. The cAP ac and hAP ac lite have much higher power on that channel. Must be a mistake in the internal configuration table for country vs frequency vs regulatory power. BTW the reason there is a yellow line there is that some cheaper interfaces can't connect at channel 149 or above.

mke · Tue Mar 12, 2024 12:29 pm

I've got a hAP ax2 as well and yes, almost same in UK - Tx power on channel 149 is 9. Sad as I am, I documented the powers on various devices in the lab. The cAP ac and hAP ac lite have much higher power on that channel. Must be a mistake in the internal configuration table for country vs frequency vs regulatory power.

Wow amazing, great work! So it definitely looks like a bug then.

mke · Tue Mar 12, 2024 2:29 pm

Here is reg-info for Australia:

/interface/wifi/radio/reg-info country="Australia" number=0    
ranges: 2402-2482/20
          5170-5250/23/indoor
          5250-5330/23/indoor/dfs
          5490-5590/30/dfs
          5650-5730/30/dfs
          5735-5835/36

I was on channel 36 before trying 149 and TX Power was 18. This makes sense from what I understand: 18 + 4.5 = 22.5.

Before switching to 149 I did some very quick tests comparing ax2 wifi to my bridged Eero 6+. Worth noting the Eero likes running 40MHz on 2.4 band and 160 on 5. This is with iperf3 server on NAS, iPerf app on iPhone 12, 2 streams, 30s duration:

tests.png

Signal on my devices was from memory 5 to 10 dBm better on the Eero, but as you can see speeds are roughly comparable. Note that the double bathroom speeds are probably due to Eero's 40Mhz width on 2.4. Very impressive for such a tiny device, I would think the overall volume is smaller even than the Eero!

For now however I am disabling wifi and running the Eero again. This bug for TX Power on channels >= 149 kind of defeats the purpose for me, will wait for it to be fixed.

ToTheFull · Tue Mar 12, 2024 3:01 pm

Ahh yes, that old chestnut. Mikrotik seem terrified to give an opinion!

interface/wifi/monitor 0,1,2,3
                 state: running        running   running        running
               channel: 5500/ax/Ceee   2412/ax   5745/ax/Ceee   2462/ax
      registered-peers: 2              2         2              0
      authorized-peers: 2              2         2              0
              tx-power: 22             14        9              15
    channel-priorities: 0:5500/ax/Ceee 0:2412/ax 0:5745/ax/Ceee 0:2462/ax

mke · Tue Mar 12, 2024 3:07 pm

Interesting. Here's mine just for proof of claims above. ax2 but seems to be the same issue across both devices:

interface/wifi/monitor 0,1    
                 state: disabled     disabled
               channel: 5745/ax/Ceee 2427/ax
      registered-peers: 4            5
      authorized-peers: 4            5
              tx-power: 9            15

robmaltsystems · Tue Mar 12, 2024 6:00 pm

I was on channel 36 before trying 149 and TX Power was 18. This makes sense from what I understand: 18 + 4.5 = 22.5.

Yes, that's my understanding too. tx power + antenna gain <= regulatory limit. With some rounding down going on I assume. So you can drop tx power but you can't increase it so total would be over regulatory limit.

Actually for UK, the power might be correct *if* this table is correct. If antenna gain is 4.5, looks like it's rounded up to 5.

> /interface/wifi/radio/reg-info country="United Kingdom" number=0         
  ranges: 2402-2482/20
          5170-5250/23/indoor
          5250-5330/23/indoor/dfs
          5490-5730/30/dfs
          5735-5875/14

Off I trundle to check UK limits...

Later - found this useful link: https://www.solwise.co.uk/downloads/fil ... uk5ghz.pdf

In band B, maximum should be 1W which ChatGPT tells me is 30dBm so 5490-5730 is correct. Channel C is 4W which is 36dBm so why RouterOS has a limit of 14 is perplexing. I'm clearly missing something.

Sorry for quoting UK but looks like same problem although in our case, the table in RouterOS looks wrong. Should be same as Australia (4W/36dBm). This is the formula BTW for Watts to dBm:

robmaltsystems · Tue Mar 12, 2024 6:21 pm

Aside but interesting comments about attenuation through walls, I can certainly attest to the ability of Kingspan to kill Wi-Fi signals.

Scatter
One of the classic problems with any WiFi radio is ‘it don’t go through walls!’. It will, to some extent, penetrate thin wood or plasterboard (assuming the plasterboard is not foil backed) but not bricks-n-mortar  Ofcom have done some rough tests which show that half of the wifi signal for 2.4GHz is lost with a single brick wall. So a double brick wall would half it again so that's something like three quarters of the signal lost going through a normal house outside wall. For 5GHz it's actually worse by about 40%! So in order to get the wifi signal around a property, to reach the various rooms, WiFi largely needs to reflect off the walls and ceilings and get into the rooms via the door ways (and the windows). Therefore, the better the WiFi reflects or scatters then the better the propagation around a house.

ToTheFull · Tue Mar 12, 2024 6:21 pm

https://www.ofcom.org.uk/__data/assets/ ... r-2030.pdf
Page 42 SRD devices

bpwl · Tue Mar 12, 2024 6:36 pm

You're probably getting your frequency changed due to radar detect, and it's selecting a TPC DFS channel which is designed for short range devices. Don't pick DFS channels or use "auto" frequency to avoid this.

I would never use "auto" in 5GHz with hAP ax2 and hAP ax3 since ROS 7.13 in Europe
The low permitted power frequencies (14dBm EIRP= radio + antenna gain rounded upward) are non-DFS in most regions/countries.
The hAPax2/3 likes that range a lot since 7.13.1 or .2 , and picks the default country Latvia if not specified, and then allows 14dBm power level.

If you get kicked out of a DFS channel you can also end up here. 8dBm for radio power is indeed the low power level that is legally permitted and is what you will get.
Check the values for the Country set !
At least set a frequency range that excludes that SRD range , eg 5180-5720 if it is low power (SRD, short range devices)
"Auto" then is not a good choice !
DFS will not limit the power, but might detect some (fake) radar signals, what makes the AP change the freq used.
TPC is handled well by the newer wifiwave2 drivers. The legacy WLAN drivers only used non-TPC power levels (3 dB lower than TPC)

robmaltsystems · Tue Mar 12, 2024 6:47 pm

https://www.ofcom.org.uk/__data/assets/ ... r-2030.pdf
Page 42 SRD devices

Ahh, was the table I found outside powers? Are they higher or lower typically? Also, 112 pages! Typical

bpwl · Tue Mar 12, 2024 6:56 pm

Check the values in RouterOS in use.
It varies over (even minor) versions. It might vary over devices/platforms.
Just check it!

For wifi drivers use "/interface/wifi/radio/reg-info country="XXXXXXXXX" number=0 "

If spelling of country name is unkbnown ... type "a" then <TAB> key ....

/interface/wifi/radio/reg-info country=a                             
"Aland Islands                         "New Zealand                                      "United States Minor Outlying Islands     Bulgaria       Guyana         Malta           Qatar         
"American Samoa                        "Norfolk Island                                   "Viet Nam                                 Cambodia       Haiti          Martinique      Romania       
"Antigua and Barbuda                   "North Macedonia                                  "Virgin Islands                           Cameroon       Honduras       Mauritania      Russia        
"Bosnia and Herzegovina                "Northern Mariana Islands                         "Wallis and Futuna                        Canada         Hungary        Mauritius       Rwanda        
"British Virgin Islands                "Papua New Guinea                                 Afghanistan                               Chad           Iceland        Mayotte         Samoa         
"Brunei Darussalam                     "Saint Barthelemy                                 Albania                                   China          India          Micronesia      Senegal

robmaltsystems · Tue Mar 12, 2024 7:09 pm

If spelling of country name is unkbnown ... type "a" then <TAB> key ....

With the caveat that this doesn't work with countries with space in the name.

bpwl · Tue Mar 12, 2024 7:11 pm

With spaces, "" are needed
Or just enter county value on the next line .. without ""

 /interface/wifi/radio/reg-info country=           
country: New Zealand
number: 0
  ranges: 2402-2482/20
          5170-5250/23/indoor
          5250-5330/23/indoor/dfs
          5490-5730/30/dfs
          5735-5835/36

robmaltsystems · Tue Mar 12, 2024 7:16 pm

I would never use "auto" in 5GHz with hAP ax2 and hAP ax3 since ROS 7.13 in Europe

Agreed but not using auto isn't black & white either. I've got 5500, 5180 specified for frequency.

Using 5500 alone is problematic as radar events appear to disable the interface entirely, i.e. no devices can connect. One would kind of expect it to switch to "auto" if a radar event was triggered and there aren't any other frequencies specified but it doesn't.
I've had problems on one site with 5500,5180 specified - it keeps instantly selecting 5180 I've been told because it does some kind of instant noise check on 5500 and decides to use 5180. NOTE: this is not the radar check. Not right IMO - if I've specified 5500 first, it should use it. Only CAG radar checks should make it choose 5180.
I get IMO fake radar events on 5500 so I've got an hourly scheduled script that turns the interface on/off which then goes back to 5500. Shouldn't need fudges like this. I say they're fake because if radar was operating in my house, it would occur more often. It's like once every few days.
Auto is problematic as it can select channel 149 and above which some Realtek crap wireless adapters can't see (gets country wrong). If the access point is in free space (e.g. ceiling in open plan office), then it's probably okay but if you're in a house and want to squeeze every dBm out of it, it's not ideal.

robmaltsystems · Tue Mar 12, 2024 7:19 pm

With spaces, "" are needed
Or just enter county value on the next line .. without ""

Another quirk

[admin@ROUTER001] > /interface/wifi/radio/reg-info
country: "United Kingdom"
Script Error: action cancelled

ToTheFull · Tue Mar 12, 2024 7:21 pm

interface/wifi/radio/reg-info country="United Kingdom" 0

jaclaz · Tue Mar 12, 2024 7:28 pm

I suspect that in a next release we will have United_Kingdom and New_Zealand ...

bpwl · Tue Mar 12, 2024 8:13 pm

've had problems on one site with 5500,5180 specified - it keeps instantly selecting 5180 I've been told because ...

I have no control over DFS workings ... It indeed selects in a strange pattern., list is not used as a priority list, nor does it ever take the next in the list.

My wAP ac's generate fake radar patterns, eg for SXT sq 5ac, signal seen is 230MHz (or was it 320MHz?) off the used wAP ac frequency. (Snooper sees those phantom frequencies as very weak extra AP's.)

mke · Wed Mar 13, 2024 9:07 am

Ok some more testing.

Firstly, I posted this up thread:

/interface/wifi/radio/reg-info country="Australia" number=1  
  ranges: 2402-2482/20
          5170-5250/23/indoor
          5250-5330/23/indoor/dfs
          5490-5590/30/dfs
          5650-5730/30/dfs
          5735-5835/36

But I didn't notice then that 2402-2482/20 for Australia is incorrect. Max power limit is 36dBm in Australia for 2.4 band. Should be 2402-2482/36

Secondly, I switched between Australia and United States (for indoors US is the same as Australia but with more limits, there is nothing illegal I can do) and United States sets 2.4ghz channels and channel 149 correctly:

# United States
interface/wifi/monitor 0,1
                 state: running      running
               channel: 5745/ax/Ceee 2447/ax
      registered-peers: 5            4
      authorized-peers: 5            4
              tx-power: 26           25
# Australia              
interface/wifi/monitor 0,1
                 state: running      running
               channel: 5745/ax/Ceee 2467/ax
      registered-peers: 4            5
      authorized-peers: 4            5
              tx-power: 9            15

Conclusion: if you run an AX device in Australia your 2.4 band is handicapped via error in baked in config, and UNII-3 channels are handicapped via bug in applying correct config. Just use US as your country for now.

What a mess.

robmaltsystems · Wed Mar 13, 2024 11:13 am

What a mess.

It does seem rather confused. Along with vague notions of how frequency/channel selection works with DFS/radar collision. I've asked for a whitepaper type article on this entire topic which I feel would remove a lot of the guesswork.

mke · Wed Mar 13, 2024 2:53 pm

for indoors US is the same as Australia but with more limits, there is nothing illegal I can do

Just want to clarify I was wrong on the above, the UNII band limit is higher in America (30 dBm vs 23 dBm in Australia). This doesn't affect me right now (using United States profile) as I am testing fixed on 149 which has a higher limit here.

On this stronger high channel my device signal strength and iperf3 tests are now is equivalent to or better than the eero 6+ at all distances.

Besides the weirdness with wrong regional channel limits I am super impressed, what an amazing little router.

ToTheFull · Wed Mar 13, 2024 3:42 pm

On this stronger high channel my device signal strength and iperf3 tests are now is equivalent to or better than the eero 6+ at all distances.

yup...

iperf3 -v
iperf 3.16 (cJSON 1.7.15)

iperf3 -c 192.168.88.10 -u -P 20 -b 40M -R -O 5

[SUM] 0.00-10.00 sec 954 MBytes 800 Mbits/sec 0.000 ms 0/690866 (0%) sender
[SUM] 0.00-10.00 sec 954 MBytes 800 Mbits/sec 0.102 ms 0/690756 (0%) receiver

I don't want to post my Mikrotik tests, people will loose their minds. hahahahaha

robmaltsystems · Wed Mar 13, 2024 4:45 pm

Doing some comparisons of power between UK and USA. I note that USA has less severe DFS/CAC checks. 55800 in UK is 10 minute CAC but only 1 minute in USA.

ToTheFull · Wed Mar 13, 2024 6:09 pm

I tried US settings earlier balanced for the UK, wasn't worth the bother!

US Power levels
interface/wifi/monitor 0,1,2,3                         
                 state: running        running   running        running
               channel: 5500/ax/Ceee   2412/ax   5745/ax/Ceee   2462/ax
      registered-peers: 0              1         1              0
      authorized-peers: 0              1         1              0
              tx-power: 21             23        26             25
    channel-priorities: 0:5500/ax/Ceee 0:2412/ax 0:5745/ax/Ceee 0:2462/ax

US Power Balanced for the UK
interface/wifi/monitor 0,1,2,3                         
                 state: running        running   running        running
               channel: 5500/ax/Ceee   2412/ax   5745/ax/Ceee   2462/ax
      registered-peers: 0              1         2              0
      authorized-peers: 0              1         2              0
              tx-power: 21             14        18             15
    channel-priorities: 0:5500/ax/Ceee 0:2412/ax 0:5745/ax/Ceee 0:2462/ax
    
 UK Levels
   interface/wifi/monitor 0,1,2,3
                 state: running        running   running        running
               channel: 5500/ax/Ceee   2412/ax   5745/ax/Ceee   2462/ax
      registered-peers: 2              2         2              0
      authorized-peers: 2              2         2              0
              tx-power: 22             14        9              15
    channel-priorities: 0:5500/ax/Ceee 0:2412/ax 0:5745/ax/Ceee 0:2462/ax

robmaltsystems · Wed Mar 13, 2024 6:18 pm

Updated my table for hAP ax2 in UK and USA out of interest. The interesting one is channel 36, the most common non-DFS channel. USA is significantly higher power than UK.

mke · Sat Mar 16, 2024 2:18 am

I stumbled across the linux wireless regulatory database:

https://wireless.wiki.kernel.org/en/dev ... less-regdb
https://git.kernel.org/pub/scm/linux/ke ... ree/db.txt

It links to most up to date legislation for Australia, and confirms reg-info for Australia is incorrect / out of date for 2.4 band in Mikrotik:

linux-reg-aus.png

legal-reg-aus.png

I have checked a few other countries with similar limits to Australia and they are equally out of date, eg Hong Kong:

linux-reg-hk.png

/interface/wifi/radio/reg-info country="Hong Kong" number=1            
  ranges: 2402-2482/20
          5170-5250/23/indoor
          5250-5330/23/indoor/dfs
          5490-5730/30/dfs
          5735-5835/33

I tested Hong Kong and it also has the issue with Channel 149 being limited to TX Power 9 despite Mikrotik's reg-info stating an EIRP limit of 33 (also out of date, should be 36 as shown in screenshot of linux wireless regulation database above).

robmaltsystems · Sun Mar 17, 2024 1:20 pm

Very useful database. Will check the UK entry later.

ToTheFull · Sun Mar 17, 2024 4:24 pm

# GB as part of EU/CEPT accepted decisions 2005/513/EC (5GHz RLAN, EN 301 893)
# and 2006/771/EC (amended by 2008/432/EC, Short-Range Devices, EN 300 440)
#  EU decision 2005/513/EC: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02005D0513-20070213
#  EU decision 2006/771/EC: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02008D0432-20080611
# Harmonized CEPT countries (July 2019): https://www.ecodocdb.dk/download/25c41779-cd6e/Rec7003e.pdf
# GB: https://www.ofcom.org.uk/__data/assets/pdf_file/0019/136009/Ofcom-Information-Sheet-5-GHz-RLANs.pdf
# GB: https://www.ofcom.org.uk/__data/assets/pdf_file/0028/84970/ir-2030.pdf
# GB: https://www.ofcom.org.uk/__data/assets/pdf_file/0013/126121/Statement_Implementing-Ofcoms-decision-on-the-57-71GHz-band.pdf
country GB: DFS-ETSI
	(2400 - 2483.5 @ 40), (100 mW)
	(5150 - 5250 @ 80), (200 mW), NO-OUTDOOR, AUTO-BW, wmmrule=ETSI
	(5250 - 5350 @ 80), (100 mW), NO-OUTDOOR, DFS, AUTO-BW, wmmrule=ETSI
	(5470 - 5730 @ 160), (500 mW), DFS, wmmrule=ETSI
	# short range devices (ETSI EN 300 440-1)
	(5725 - 5850 @ 80), (200 mW), NO-OUTDOOR
	(5925 - 6425 @ 160), (250 mW), NO-OUTDOOR, wmmrule=ETSI
	# 60 GHz band channels 1-6
	(57000 - 71000 @ 2160), (40)

Who is online