Flint 2 (GL-MT6000 ) - bug reports - collective thread

I have been playing around lately by adding 256 qam patches into mt76 driver in my own build, the speed did increase for 2.4ghz.

But the real issue is not that, the range is still heavily affected on very abnormal levels, like the link tracking is in desync with the wireless throughput, to test it further i was able to have full 180mbs speed, i walked 15 meter away and the connection got stuck, then i walked back to the same location and wasn’t able to reach above 60mb/s.

And the weird thing is, on the OpenWrt forums there is also a group of people noticing zero 2.4ghz issues with just openwrt snapshots (not mine) so its likely also very device specific to xiaomis and pocos, if i want full stability i almost need to sit on the router with my phone and not point directly my phone antenna to the flint.

I also tried some illegal channel things knowing it would not make a difference from 21dbm to 31dbm, and there it was very noticeable it does not seem to be related to through put, for my NSS 2 device, maybe false reported/ignored(?), but i think it is some type of desynchronization with the link type from ether to wifi and or wifi to ether, it did exactly the same thing after 15 meters.

Ive used patches from here:

and this one:

But no avail on kernel 6.6, only that the speed increased but that abnormal range issue is not fixed :wink:

The half wavelength of 2.4GHz is 6.2cm
Could you change the antenna position like this below?

Antenna 1: vertical
Antenna 2: horizontal
Antenna 3: vertical
Antenna 4: horizontal

1 Like

I did a quick test, in one room i noticed significant change, but when i walked back to the kitchen 15 meters away there was no connectivity on speedtest.

Walked back and then the link speed became also dramatically from 186mb/s to 60mb/s.

Though you may be on something since the router is vertical mounted with the antennas away, maybe i should try it in the opposite angle.

That is interesting…
I guess GL-iNet should investigate if all four antennas are syncronized in the time domain.
This can explain why we are finding low performance, as one antenna can destroy the beamwidth completely if they are not syncronized. Just a brainstorming :joy:

4 Likes

Bug Area: Wireguard VPN Server

Bug Specifics: When exporting a VPN profile for a client, the downloaded file is a .json, but the Wireguard mobile client requires a .conf or .zip. The downloaded file must be manually renamed to be imported.

F/W Version 4.5.7

Severity: High

1 Like

Second release of 4.5.8 beta is listed for download.

kmod-mt_wifi - 5.4.238+TEST-5 → 5.4.238+TEST-6

1 Like

Installed!
I still have problems with the 2.4GHz wifi with very slow downloads and with the 2.5G LAN port which seems to be at 1G instead

@JerryZhao @alzhao would be interesting if you list the Known Bugs on every Release Notes. This would help us know what your team is working on, and avoid us reporting the same bugs on every beta release.

6 Likes

Is it advisable to upgrade from 4.5.2 to 4.5.6?

Thanks for the heads up. I updated to 4.5.8 release2 from release1. This time around, I kept my settings. It’s only been an hour and a half so far but it seems to be working without any issues.

I’m getting 1.1 Gbps wirelessly which means the Ethernet port is 2.5 Gbps. My current connection is 1.5 Gbps (closer to 1.8 Gbps with over-provisioning).

I just did another test without usb stick connected, it seems this fixed the range problem i had, but im not there yet :+1:

Now i should test bigger ranges + if the desync is still there.

usb in question has product code: SDCZ430-128G-G46 it is a sandisk 3.2 usb stick.

if people are interested in my repo to lookup patches its here and patches are here the QAM patch is from lean’s openwrt and I also used a few patches from the PR here.

though i would not advise to use my images, I made them more like a SDK to spin up a new image with pretty much my own lab topology pre-installed and argon without the full toolchain (that take days for me to compile).

1 Like

I tested an USB Stick on Flint2 and I had interference on 2.4GHz network, even resulting in my ISP Router becoming useless due to the strong interference.

Putting the same USB Stick on Brume2 doesn’t interfere on my ISP Router even when placing it in physical contact all around my ISP Router.

GL-iNet bought the same USB Stick and they reproduced the problem and it’s under investigation, but I don’t believe they will fix this, because it’s a shielding problem.

1 Like

I have updated today from 4.5.7 to the second release of 4.5.8 beta
Definitely better then before at least for the 2.4Ghz WiFi: Speedtest by Ookla - The Global Broadband Speed Test
(1 mt far from the router)

I’m really confused. I got the ASUS TUF AX6000 which uses the same MTK chipset as the Flint 2 and decided to compare both.

I flashed the TUF AX6000 with OpenWRT 23.05.3 (stable) to see if the router also has problems with the 2,4 GHz band speed.

This are the test results (WiFi 6):

20mhz

The upload speed is probably also 1:1, it maxed out my internet upload speed.

Can someone from gl-inet explain the hardware difference?

Hardware i used : AX210 - 50cm away

3 Likes

on my PC with ax200 on mt6000 it also maxes out at 2.4, the problem is related to mobile devices and old machines.

Please redo the test on an Android device or iPhone and post the results here.

1 Like

The only difference i happen to learn was this conversation:

How OpenWrt Vanquishes Bufferbloat - #30 by richb-hanover-priv - Site Feedback and Other Questions - OpenWrt Forum you can also scroll a bit up for more infos.

Not to spread fud or anything but for some reason the small difference is also not listed on the mediatek driver wiki for kernel.org, so its not official supported via them (as how i understand) but that does not mean OpenWrt does, the asus one for 2.4ghz is listed on the kernel.org wiki.

Between asus ax6000 and mt6000 this seem to differ on the table of hardware.

Though i’d still think its maintained by OpenWrt and other contributors, GL-iNet aswell but it can explain the difference.

1 Like

Pixel 7, 20 MHz, it never connected with 40MHz to the 40MHz network, Asus TUF AX6000, OpenWrt 23 stable.

Have you enabled the force 40 MHz setting on the 2.4 GHz settings?

Yes I did, and the AX210 was also using 40 MHz, my Pixel 7 only used 20 MHz.