Help with troubleshooting poor WiFi speed

Hey all! Long time hobby user of Mikrotiks here, no idea why it took me this long to join this forum.

In any case, I recently redid my entire home network and I’ve hit a roadblock that I lack the knowledge to overcome: my WiFi speed is not as fast as I would wish.

What is the best way to troubleshoot WiFi speed, starting with what tools to use to measure it properly?

Here’s a few details about my setup:

  • gigabit FTTH
  • RB5009UPr+S+
  • RBcAPGi-5acD2nD connected to the router with a 10m long CAT7 cable, with PoE
  • MacBook set 1 meter away from the AP
  • wired download speed: 900 Mbps according to speedtest.net
  • wifi download speed: 100 to 200 Mbps according to speedtest.net
  • ping more or less the same around 15 ms

Here’s how the wifi is configured on the AP:

/interface wireless
set [ find default-name=wlan2 ] band=5ghz-a/n/ac channel-width=20/40mhz-Ce \
    country=spain default-authentication=no disabled=no distance=indoors \
    frequency=auto hide-ssid=yes installation=indoor mode=ap-bridge ssid=null \
    wireless-protocol=802.11 wps-mode=disabled
/interface wireless
add disabled=no mac-address=XXX master-interface=wlan1 name=\
    VAP2-HOUSE security-profile=SECPOLICY-HOUSE ssid="XXX" wps-mode=\
    disabled

Hypotheses that I don’t know how to validate/reject:

  • I mucked up something in wifi config. I tried to keep this as out-of-the-box as possible, but maybe that’s not OK?
  • I use VLANs to separate work, family and iot devices. On the AP, I see that the bridge from “ether2” to “BR2-HOUSE” has “hw-offload”. Is this why ether is so much faster? How could I test this?
  • I have another 5 APs around the house, all providing the same SSIDs as the AP under test. Are they conflicting with the test AP? How to test this?

I’m sure there is stuff I need to read to be able to troubleshoot this. Where should I start? What other info should I post here to help you help me?

There are several factors that can impact your performance.

You say you have 5 more APs around the house and if all of them transmit on the same frequency that is one source of interference.

I would never use auto frequency, who knows what frequency AP will choose. Scan your area to see if there is APs around you that could interfere with your APs.

What is link speed when you connect to AP ?

Do you get the same/similar speed on all APs?

SInce you’re asking for pointers:
(some of those are coming from this nice presentation: https://mum.mikrotik.com/presentations/UK18/presentation_5900_1539209343.pdf)

  • Unless really needed, don’t use a/n/ac. Use ac only, worst case n/ac. Try to avoid a.
  • Since you use legacy wireless, you can still play with datarates setting. Uncheck the lower rates and see where clients start to drop. Until where can you go ?
  • As indicated, don’t use auto. Scan your environment, set the APs to the frequency having the least interference for their positioning.
  • Maybe they are interfering with each other ? You can have too much APs, you know… Frequency scan.
  • Since you don’t show full relevant config, how are security settings? Don’t use WPA, it’s slower then WPA2.
  • Since you have cap AC, you can opt to use newer wave2 drivers. Makes a HUGE difference (and you will also be able to move to WPA3 security).
  • For reference: what’s the speed like when you hook laptop to ether2 of that cap ac ?
  • 1m away from AP might be (really !) too close ! Move a bit further away and see what happens.
  • What if you use another wireless client for reference ? (e.g. my former laptop only used AC, my Samsung S20 can use AX. Makes a big difference when testing Wifi speed on wave2 APs).

But what does the rest of the config look like for such a cAP AC ?

If you really have 200mbit/s measured throughput on just 40mhz channel width, then you’re already maxing out what is even possible IMHO.

I seem to recall on AC3 I could reach close to 400Mbps. Same chipset for wifi.
Has been a while since I tested it, though. So probably it was on 80MHz width.

EDIT:
but speaking of which … why not use 20/40/80 Ceee setting ? The chipset can do it.

Adding: you got the right attitude, that’s already a big plus :sunglasses:

set channel-width=20/40/80mhz-XXXX wlan2

Then retest again on MacOS. make sure you’re on 5ghz (https://osxdaily.com/2011/06/15/get-detailed-wifi-info-from-the-menu-bar/)

price question: why wlan2 ssid=null? what does this do?

In no particular order:

  1. I’ve seen many people suggest to “scan you area” but I’m not sure how to do that. Is there a guide/wiki that you can point me to?

What is link speed when you connect to AP ?

I don’t know how to answer this. Is this the speed between the AP and the Router? How do I measure this?

Do you get the same/similar speed on all APs?

Yes.

Since you use legacy wireless, you can still play with datarates setting. Uncheck the lower rates and see where clients start to drop. Until where can you go ?

When doing the testing, the MacBook is the only client on 5GHz wifi.

Since you don’t show full relevant config, how are security settings? Don’t use WPA, it’s slower then WPA2.

I am using WPA2, here’s the config:

/interface wireless security-profiles
set [ find default=yes ] authentication-types=wpa2-psk supplicant-identity=\
    MikroTik
add authentication-types=wpa2-psk mode=dynamic-keys name=SECPOLICY-HOUSE \
    supplicant-identity=MikroTik

what’s the speed like when you hook laptop to ether2 of that cap ac ?

The speed is 900 Mbps. Sorry for not being clear in the first post, but the laptop is connected to ether2 on the AP.

what does the rest of the config look like

Attached.
cap.rsc (1.44 KB)
8.

If you really have 200mbit/s measured throughput on just 40mhz channel width, then you’re already maxing out what is even possible IMHO.

I don’t understand enough about channel width to comment on this. I’ve read on the interwebs that one shall not use large channel width, but I lack the intuition on what the consequences of that advice are. Wikipedia says 5Ghz should go to “at least gigabit speeds” so that’s what I’m aiming for.

Should I increase the channel width?

why not use 20/40/80 Ceee

Because I’ve read multiple times that I should not. Should I?


10. Things I’ll try tomorrow and report back, it’s getting late :

  • “don’t use a/n/ac. Use ac only”
  • “use newer wave2 drivers”
  • “Move a bit further away”
  • “use another wireless client for reference”

Thanks, will do!

make sure you’re on 5ghz (> https://osxdaily.com/2011/06/15/get-detailed-wifi-info-from-the-menu-bar/> )

100% I am on 5ghz as I turn of 2.4ghz interface on the AP when doing the test. Triple checked the menu bar, just in case.

price question: why wlan2 ssid=null? what does this do?

I believe this comes from the fact I am using virtual APs to have multiple SSIDs on one AP, so that each SSID goes to its dedicated VLAN. I do this for work/family/iot device separation. So the ssid=null line comes from the real interface configuration, and a bit below there is the ssid=XXX which is the virtual interface configuration.

Last part
Doesn’t make sense.
Set the master interface to a real SSID, the one most used or the one most secure, whatever.
Then virtual for the rest you need.

Winbox, wifi/wireless (depending on what drivers you use), first tab.
There are 2 buttons labelled Scanner and Frequency Usage.
Hit Scanner or Frequency Usage and then select which radio you want to use for scan.
While doing this, that radio will become unusable for logical reasons (it’s doing something else for a while) so don’t do this when connected to that router via wifi using that same radio.
More info:
https://help.mikrotik.com/docs/display/ROS/Wireless+Interface#WirelessInterface-Scan

Your client device should give you an indication. I know my Samsung S20 and Windows Laptop show this without any problems when I go to Wifi details.
You can also see this on your router: Wifi / registrations, MAC address of your client device.

Channel width: yes.
at least gigabitspeeds in 5GHz: don’t believe everything they say on interweb without grasping the whole context. Not with AC technology and 2x2 streams. Around 700MBps would be the max if you move to wave2 drivers.


See 8.

Where exactly did you read that? In general this is not true.


If the signal is strong enough, noise is low enough and the spectrum is free: Of course, go as high as feasible. Channel width directly correlates to throughout.

So I found “Tx Rate” in the macOS Wifi UI:
Screenshot 2024-01-18 at 10.02.59.png
When I change the channel width to “20/40/80mhz-XXXX” I see the “Tx Rate” increases:
Screenshot 2024-01-18 at 10.08.16.png
And speedtest.net goes up to 400+ mbps. This is quite good!

But after AP reboot (just to confirm it still works), the speed goes back to ~100 mbps. I notice that the channel is different and the Tx Rate dropped. I guess I should really do the frequency scan next.
Screenshot 2024-01-18 at 10.56.53.png

Related question: what’s a better alternative to running speedtest in a browser? Is there a way to test throughput from the macOS via the AP to the router, and make the test more replicable by removing the FTTH connection speed noise?

When I’m doing the tests, I disable the 5ghzs interfaces on all other APs in the house. I am assuming that I can keep the 2.4ghzs interfaces running, as they don’t interfere. Is this assumption correct?

Couple of things:

  • on the test AP, set the frequency manually to 5260 (=channel 52 but it is the first 80Mhz wide block)
  • speedtest: I use an in-house iperf3 server for such testing. Any computer can be setup to act as iperf3 server.
  • keeping 2.4GHz enabled: does not impact 5GHz testing.

Scanner does not find anything in the 5ghz range.

Freq. Usage reports the following:
Screenshot 2024-01-18 at 11.19.16.png
Suburban location, detached house.

So your basically alone there :laughing:

See how it goes setting that frequency fixed and using 20/40/80

This helps make the speed more stable. It goes above 200 mbps even after multiple tests, reboots, yay!

Ah nice, I’ll set it up on the NAS and report back!

Thanks!

So your basically alone there > :laughing:

Yes :smiley: