Alfa Network AWUS036AXML: TX power limited to 3 dBm (not sure if for real or just innacurate report), UL speed much worse than AWUS036ACM

[patrakov]

I bought a new USB WiFi adapter, Alfa Network AWUS036AXML, and tried to use it on my desktop with Arch Linux. The kernel version is 6.3.3, the linux-firmware package version is 20230404.2e92a49f-1.

The adapter is recognized, but in the 5 GHz band, the TX power is limited to 3 dBm. I am in the Philippines, and the regulatory domain is set correctly. The bad received signal strength is also confirmed by my router.

As this is a USB card, there is no such thing as a "factory partition" from where the calibration data can be obtained.

# iw dev
phy#2
	Interface wlp0s20u4i3
		ifindex 34
		wdev 0x200000001
		addr 00:c0:ca:b3:c7:3c
		ssid Archer
		type managed
		channel 153 (5765 MHz), width: 80 MHz, center1: 5775 MHz
		txpower 3.00 dBm
		multicast TXQ:
			qsz-byt	qsz-pkt	flows	drops	marks	overlmt	hashcol	tx-bytes	tx-packets
			0	0	0	0	0	0	0	0		0
phy#0  <-- onboard WiFi, please ignore
	Interface wlp4s0
		ifindex 4
		wdev 0x1
		addr f2:b5:c2:8a:89:b9
		type managed
		txpower 20.00 dBm
		multicast TXQ:
			qsz-byt	qsz-pkt	flows	drops	marks	overlmt	hashcol	tx-bytes	tx-packets
			0	0	0	0	0	0	0	0		0

[patrakov]

[389839.043749] usb 3-4: new high-speed USB device number 14 using xhci_hcd
[389839.899850] usb 4-4: new SuperSpeed USB device number 4 using xhci_hcd
[389839.913461] usb 4-4: New USB device found, idVendor=0e8d, idProduct=7961, bcdDevice= 1.00
[389839.913467] usb 4-4: New USB device strings: Mfr=6, Product=7, SerialNumber=8
[389839.913469] usb 4-4: Product: Wireless_Device
[389839.913471] usb 4-4: Manufacturer: MediaTek Inc.
[389839.913473] usb 4-4: SerialNumber: 000000000
[389843.461174] Bluetooth: hci0: Device setup in 3458344 usecs
[389843.461188] Bluetooth: hci0: HCI Enhanced Setup Synchronous Connection command is advertised, but not supported.
[389845.522024] Bluetooth: hci0: Opcode 0x c03 failed: -110
[389847.570104] Bluetooth: hci0: Failed to read MSFT supported features (-110)
[389849.618187] Bluetooth: hci0: AOSP get vendor capabilities (-110)
[389849.732595] usb 4-4: reset SuperSpeed USB device number 4 using xhci_hcd
[389849.783694] mt7921u 4-4:1.3: HW/SW Version: 0x8a108a10, Build Time: 20230302150916a

[389849.900223] Bluetooth: hci0: Device setup in 149656 usecs
[389849.900230] Bluetooth: hci0: HCI Enhanced Setup Synchronous Connection command is advertised, but not supported.
[389849.970167] Bluetooth: hci0: AOSP extensions version v1.00
[389849.970171] Bluetooth: hci0: AOSP quality report is supported
[389850.060740] mt7921u 4-4:1.3: WM Firmware Version: ____010000, Build Time: 20230302150956
[389851.953094] mt7921u 4-4:1.3 wlp0s20u4i3: renamed from wlan0
[389904.364191] wlp0s20u4i3: authenticate with e8:9f:80:d4:9e:c6
[389904.922280] wlp0s20u4i3: send auth to e8:9f:80:d4:9e:c6 (try 1/3)
[389904.927584] wlp0s20u4i3: authenticate with e8:9f:80:d4:9e:c6
[389904.936713] wlp0s20u4i3: send auth to e8:9f:80:d4:9e:c6 (try 1/3)
[389904.939690] wlp0s20u4i3: authenticated
[389904.940446] wlp0s20u4i3: associate with e8:9f:80:d4:9e:c6 (try 1/3)
[389904.953527] wlp0s20u4i3: RX AssocResp from e8:9f:80:d4:9e:c6 (capab=0x1011 status=0 aid=3)
[389904.987585] wlp0s20u4i3: associated
[389904.987719] wlp0s20u4i3: Limiting TX power to 30 (30 - 0) dBm as advertised by e8:9f:80:d4:9e:c6
[389905.990537] IPv6: ADDRCONF(NETDEV_CHANGE): wlp0s20u4i3: link becomes ready

But still the actual TX power is 3 dBm.

[rany2]

I wouldn't expect it to be able to connect with 3dBm as TX power. Maybe it's an issue with it reporting incorrect TX power?

[patrakov]

There is also an upload speed issue.

I can connect this to a Lenovo 2-in-1 tablet that is currently set up for dual-booting and test it this way. Please wait.

[patrakov]

I have deliberately placed the adapter in a bad spot. Well, sorry for the misleading report - Linux connects at a higher speed than Windows.

Arch Linux:

The AP (Linksys E8450) reports a signal strength of -79 or -80 dBm. The adapter itself reports -69 dBm as the signal strength coming from the AP. iperf3 speed test yields 334 Mbits/sec download (from the tablet's perspective) and 109 Mbits/sec upload.

Windows 11:

The AP reports a signal strength of -87 dBm. iperf3 speed test yields 244 Mbits/sec download (from the tablet's perspective) and 33.7 Mbits/sec upload.

[rany2]

I think it's worth keeping that issue open but with "inaccurate TX power report" as the title.

[patrakov]

I still think that the TX power is wrong. I have placed an Alfa Network AWUS036ACM adapter into the same bad spot, and the AP now reports a much higher signal level, and the speed test is returning better results.

Linux:

The AP reports a signal strength of -69 dBm, iperf3 reports 237 Mbits/sec download and 179 Mbits/sec upload

Windows 11:

The AP reports a signal strength of -69 dBm. iperf3 reports 221 Mbits/sec download and 193 Mbits/sec upload.

Now compare this with a -80 or -87 dBm from AWUS036AXML.

[patrakov]

Another hypothesis is that the AX mode on Linksys E8450 (with OpenWrt) is somehow cursed. There were reports of horrible throughput with iPhones, maybe it's the same problem? Let's test by disabling AX on the AP and retesting.

Linux:

AP reports -77 dBm, the adapter itself reports -67 dBm, iperf3 reports 353 Mbits/sec download and 131 Mbits/sec upload (from the tablet viewpoint)

Windows:

AP reports -82 dBm, iperf3 reports 297 Mbits/sec download and 54.4 Mbits/sec upload (from the tablet viewpoint)

So I think the bug is real, and not Linux-specific. Might be just that I should not recommend this AWUS036AXML hardware to anyone.

[rany2]

@patrakov if it's mt7915 try to disable bw160 with this patch: rany2/mt76@2e31de4 (simply using VHT80/HE80 is not enough, you need this patch for it to work smoothly)

[patrakov]

It is mt7921u, not mt7915.

[rany2]

pretty sure there are no consumer routers with mt7921? I'm referring to Linksys E8450..

[patrakov]

It is MT7915E then. OK, I will try your patch tomorrow.