Fenvi FU-AX1800 performance

[kinhsman]

The performance of this USB dongle is acceptable, but it falls short of being ideal. I consistently achieve just a bit over 100Mbps for both upload and download speeds. I've tested it on both Windows and Linux (Debian and OpenWRT), and also both in Client mode and AP mode. I've come across instances where users report achieving much better results, around 300Mbps. If anyone has experience using this specific dongle, could you please share your setup details, including hardware, operating system, configurations, etc.? Your insights would be greatly appreciated. Thank you!
Edit: the speed tests were ran using iperf3, and openspeedtest server

[morrownr]

Hi @kinhsman

The performance of this USB dongle is acceptable, but it falls short of being ideal.

Let's see what we can figure out. I have a speed comparison up on the Main Menu. Here is the direct link:

https://github.com/morrownr/USB-WiFi/blob/main/home/Speed_Comparison_Test.m

Note that the fastest adapter uses the same chipset that your Fenvi adapter uses. It has internal antennas which give a slight advantage to your Fenvi adapter. Let's see what we can do to optimize things.

I consistently achieve just a bit over 100Mbps for both upload and download speeds.

I assume this is managed mode operation. Questions:

• How are you measuring that speed?
• What band are you using?
• What channel are you using?
• How congested is the channel?
• What is the capability of your AP/WiFi router? WiFi 5? WiFi 6?
• What does wavemon show for signel level and link quality?

You can install wavemon on Debian with:

$ sudo apt install wavemon
or
# apt install wavemon

Run it:

$ wavemon

[kinhsman]

@morrownr
I just set it up from scratch. Below are the details:

How are you measuring that speed?

via iperf3:

root@DietPi:~# iperf3 -c 10.1.30.50
Connecting to host 10.1.30.50, port 5201
[  5] local 10.1.30.92 port 38224 connected to 10.1.30.50 port 5201
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.01   sec  22.0 MBytes   183 Mbits/sec    0   1007 KBytes       
[  5]   1.01-2.00   sec  20.0 MBytes   169 Mbits/sec    0   1.61 MBytes       
[  5]   2.00-3.00   sec  20.0 MBytes   168 Mbits/sec    0   2.05 MBytes       
[  5]   3.00-4.02   sec  18.8 MBytes   155 Mbits/sec    0   2.27 MBytes       
[  5]   4.02-5.00   sec  21.2 MBytes   181 Mbits/sec    0   2.40 MBytes       
[  5]   5.00-6.00   sec  20.0 MBytes   168 Mbits/sec    0   2.65 MBytes       
[  5]   6.00-7.05   sec  21.2 MBytes   170 Mbits/sec    0   3.01 MBytes       
[  5]   7.05-8.04   sec  17.5 MBytes   149 Mbits/sec    0   3.01 MBytes       
[  5]   8.04-9.00   sec  21.2 MBytes   185 Mbits/sec    0   3.01 MBytes       
[  5]   9.00-10.12  sec  21.2 MBytes   160 Mbits/sec    0   3.01 MBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.12  sec   203 MBytes   169 Mbits/sec    0             sender
[  5]   0.00-10.14  sec   202 MBytes   167 Mbits/sec                  receiver

iperf Done.
root@DietPi:~# 

What band are you using?

5GHz

What channel are you using?

Channel 44, 80MHz

How congested is the channel?

My AP is showing around 25%. Other devices can connect and get 700Mbps easily

What is the capability of your AP/WiFi router? WiFi 5? WiFi 6?

802.11AX WiFi 6. Its a Meraki MR56
https://documentation.meraki.com/MR/MR_Overview_and_Specifications/MR56_Datasheet

What does wavemon show for signel level and link quality?

link quality: 97% (68/70)
signal level: -42 dBm (0.06 uW)

┌─Interface─eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee┐
│wlan0 (IEEE 802.11), phy 0, reg: US (DFS-FCC), SSID: Mom Click Here For Wifi
├─Levels─sssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss┤
│                                                                           │
│link quality: 97%  (68/70)                                                 │
│========================================================================   │
│                                                                           │
│                                                                           │
│signal level: -42 dBm (0.06 uW)                                            │
│================================================                           │
│                                                                           │
├─Statistics─sssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss┤
│RX: 64k (4.49 MiB)                                                         │
│TX: 299k (419.32 MiB)                                                      │
├─Info─ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo┤
│mode: Managed, connected to: E6:55:B8:12:E5:66, time: 6:53m, inactive: 0.0s│
│freq: 5220 MHz, ctr1: 5210 MHz, channel: 44 (width: 80 MHz)                │
│rx rate: 1080.6 Mbit/s 80MHz, tx rate: 1200.9 Mbit/s 80MHz                 │
│station flags: WME, preamble: short, slot: short                           │
│power mgt: off,  tx-power: 3 dBm (2.00 mW)                                 │
│retry: short limit 7,  rts/cts: off,  frag: off                            │
├─Network─kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk┤
│wlan0 (UP RUNNING BROADCAST MULTICAST)                                     │
│mac: 90:DE:80:FC:03:0F, qlen: 1000                                         │
│ip: 10.1.30.92/24                                                          │
└─44444444444444444444444444444444444444444444444444444444444444444444444444┘

One thing I notice is the CPU utilization is maxing out every time I run speed tests

[morrownr]

@kinhsman

You signal level and link quality are excellent. This should allow for maximum speeds, give or take the congestion.

About a month ago, I was helping another person. We did a lot of digging and finally discovered that the problem was the slow CPU in his old RasPi. I think the RasPi cpu had a single core running at 500 MHz. There was the problem. When you have a very old cpu, it may not have the power to push packets faster than what you are seeing. What hardware are you using?

@morrownr

[kinhsman]

@morrownr
I was running a DietPi VM (debian-based) on proxmox hardware (Intel N100 CPU). Then I moved the VM to WMware workstation on Windows with 8 CPUs (Intel i9-12900K) with about the same results.
However, when I ran it bare metal on a NanoPi RS6, the speed is optimal:

root@DietPi:~# iperf3 -c 10.1.30.50
Connecting to host 10.1.30.50, port 5201
[  5] local 10.1.30.95 port 59716 connected to 10.1.30.50 port 5201
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  70.4 MBytes   590 Mbits/sec    0   2.84 MBytes
[  5]   1.00-2.00   sec  85.0 MBytes   713 Mbits/sec    0   3.16 MBytes
[  5]   2.00-3.00   sec  81.2 MBytes   682 Mbits/sec    0   3.16 MBytes
[  5]   3.00-4.00   sec  85.0 MBytes   713 Mbits/sec    0   3.16 MBytes
[  5]   4.00-5.00   sec  78.8 MBytes   661 Mbits/sec    0   3.16 MBytes
[  5]   5.00-6.00   sec  83.8 MBytes   702 Mbits/sec    0   3.16 MBytes
[  5]   6.00-7.00   sec  78.8 MBytes   661 Mbits/sec    0   3.16 MBytes
[  5]   7.00-8.00   sec  83.8 MBytes   703 Mbits/sec    0   3.16 MBytes
[  5]   8.00-9.00   sec  82.5 MBytes   692 Mbits/sec    0   3.16 MBytes
[  5]   9.00-10.00  sec  82.5 MBytes   692 Mbits/sec    0   3.16 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   812 MBytes   681 Mbits/sec    0             sender
[  5]   0.00-10.01  sec   809 MBytes   678 Mbits/sec                  receiver

iperf Done.
root@DietPi:~#

I am guessing it has something to do with the virtualization aspect of it, or maybe the USB passthough was not ideal for this type of adapter.

[morrownr]

However, when I ran it bare metal...

Yeah. VMs can be a challenge. I don't run VMs so I won't be any help optimizing things. I have been told by others that this chipset and driver do work well in VMs.

681 Mbps is not bad at all.

What you might do is start another issue with a title something like: Need help with mt7921au based adapter in a XXX VM

And see if anyone can help.

@morrownr

[kinhsman]

@morrownr Thank you! I will do that

[DelanyLove]

Today I ordered this adapter after returning Alfa AWUS036ACS. I hope it will work better with no connectivity issue.

[DelanyLove]

Yesterday I received the order. I am happy it worked (plug and play), You just need to switch usb from 2.0 to 3.1 in vm.

@morrownr I need you help with following

When i enter "iwconfig"

wlan0 IEEE 802.11 ESSID:off/any
Mode:Managed Access Point: Not-Associated Tx-Power =3dBm
Retry short limit:7 RTS thr:off Fragment thr:off
Power Management:on

Why ESSID: off/any
Looks like TX-Power is too low - How can i increase it ?
Not sure about RTS and Fragement
How can i turn on/off Power Management ? Any benefit if i turn it off ?

[morrownr]

When i enter "iwconfig"...

Please don't use iwconfig as it has long been depreciated and is not maintained. Use iw and ip.

Looks like TX-Power is too low - How can i increase it ?

Disregard the txpower reading. It is in error. The txpower is actually working fine.

How can i turn on/off Power Management ?

Use iw.

Any benefit if i turn it off ?

Not that I am aware of but you can try.

[morrownr]

Hopefully by this time next year the Alfa ACS will have good in-kernel driver support but for right now, the in-kernel driver for this Fenvi adapter is already in place as long as your kernel is at least 5.19. I've read reports for several users that have bought the Fenvi and they seem to get along with it. Let us know how it works out.

It is kinda interesting that you went from an AC600 adapter to an AXE3000 adapter for what is likely less money.

[mobeicanyue]

Does the Linux support FU-AX1800 in kernel? Where is the info source? Thank you.

[morrownr]

Hi @mobeicanyue

Does the Linux support FU-AX1800 in kernel?

Yes, as long as your distro is running kernel 5.19 or later. The chip is uses is the Mediatek mt7921au which is an AXE3000 WiFi 6 capable chip. The driver is mt7921u.

Where is the info source?

I guess the best info source would be the Plug and Play List. The Fenvi adapter is shown in the mt7921au chip section. Look on the Main Menu, item 2 for the Plug and Play List:

https://github.com/morrownr/USB-WiFi

[mobeicanyue]

Hello, thanks for reply.
I learn about the info about FU-AX1800 from this repo, but I wanna know where does the info from? Like offical kernel release list?

[morrownr]

@mobeicanyue

I wanna know where does the info from?

The following is a link to the Mediatek information webpage on kernel.org:

https://wireless.wiki.kernel.org/en/users/drivers/mediatek

This list shows which Mediatek chips are supported in the Linux kernel. The chips that show USB are the chips for USB adapters and modules. The chip that the Fenvi FU-AX1800 uses is the mt7921 (USB). It also appears, but is not confirmed yet that Fenvi has another adapter that uses the mt7921 chip:

#455

Fenvi makes several adapters that use Realtek chips as well, so you have to be aware of that and not accidently order the wrong adapter. My opinion is that Realtek USB support is inferior to that of Mediatek. Users have real challenges using Realtek chip based wifi adapters and modules so it is best to avoid the Realtek based adapters, at least for now. Maybe in 5 years it will be different.

Does this information help?

[mobeicanyue]

Yes, it helps a lot, thank you.

[mobeicanyue]

I could use it normally, thanks for your share!
One more question:
Does the wlp1s0 is laptop's wifi device,
and the wlp3s0f3u2 is FU-AX1800?

[mobeicanyue]

I running nmcli device command
It printed

DEVICE              TYPE      STATE          CONNECTION 
wlp3s0f3u2          wifi      connected      999        
lo                  loopback  external       lo         
wlp1s0              wifi      disconnected   --         
p2p-dev-wlp3s0f3u2  wifi-p2p  disconnected   --         
enp3s0f4u1u4        ethernet  unavailable    --         

wlp3s0f3u2, p2p-dev-wlp3s0f3u2, enp3s0f4u1u4
the 3 looks like new device

[morrownr]

Does the wlp1s0 is laptop's wifi device, and the wlp3s0f3u2 is FU-AX1800?

That is likely the case but if you pull the usb wifi adapter out and look again, you will know for sure.

[keynet]

I just got one of these and disappointed in its overheating in AP mode.
Yes it runs as AP, my pixel phone even reported a 300Mbit connection, but after about 30 mins as AP @25c ambient it overheats - even if idle - and I get a USB disconnect, then of course the AP is dead.
I'm using it as a 5GHz AP on Hostapd 2.10 on SuSe 15.6 (default kernel 6.4.0).
Firmware reported as "mt7921u 1-1.1:1.0: WM Firmware Version: ____010000, Build Time: 20240219111038", this from kernel-firmware-mediatek Version 20240712-150600.3.3.1
On a USB3 port it seems to overheat slightly quicker - presumably the extra heat of running the higher speed USB? Also possibly quicker if using higher speed modes VHT etc.
After cooling down it's OK again for another few mins, but it really isn't usable as a reliable AP at the moment.

[morrownr]

@keynet

its overheating in AP mode.

I don't have this specific adapter but I do have 3 usb wifi adapters that use the mt7921au chip. My testing on the mt7921au chip shows low power usage and cool operation even with heavy throughput. The first thing that comes to mind is possibly a manufacturing problem. You might want to contact the adapter maker to see what they have to say. Or maybe the seller as they might send you a replacement.

Just to make sure: This is the Fenvi FU-AX1800 with the mt7921au chip?

[keynet]

Yes it's a Fenvi FU-AX1800, the chip actually says mt7921AUN, I don't know if the N on the end is the issue. Once into AP mode, it gets baking hot after a few minutes, so I'm assuming that's what causes it to do a USB reset, then all sorts of failures, the kernel says "dazed and confused!"

Kernel firmware is the same version as listed in the kernel git link, though SuSe compresses it as xz.
If there's a later firmware version please let me know. I ordered a stick-on heatsink!

Here's dmesg

[18907.911798] mt7921u 1-1.3:1.0: vendor request req:63 off:d02c failed:-71
[18908.000849] usb 1-1.3: USB disconnect, device number 39
[18913.343746] usb 1-1.3: new high-speed USB device number 40 using ehci-pci
[18913.457825] usb 1-1.3: New USB device found, idVendor=0e8d, idProduct=7961, bcdDevice= 1.00
[18913.457838] usb 1-1.3: New USB device strings: Mfr=2, Product=3, SerialNumber=4
[18913.457851] usb 1-1.3: Product: Wireless_Device
[18913.457854] usb 1-1.3: Manufacturer: MediaTek Inc.
[18913.457857] usb 1-1.3: SerialNumber: 000000000
[18913.583800] usb 1-1.3: reset high-speed USB device number 40 using ehci-pci
[18913.739097] mt7921u 1-1.3:1.0: HW/SW Version: 0x8a108a10, Build Time: 20240219110958a

[morrownr]

@keynet

Are you certain this is an overheating problem? We have seen several wifi stack problems in recent kernel as the wifi stack is modernized. The problems are being fixed but it can take time for the fixes to make it into the kernel. What kernel are you using?

New mt7921 firmware was released, I think it was last week. Do you know how to update the firmware?

[keynet]

...I bought a self-adhesive heatsink designed for a Raspberry Pi I think - as big as possible. I'd have to cut a hole in the cover if I want to put that back on.
Now it works as an AP continuously - the heatsink still gets very hot (50C+), but seems to save it from overheating and crashing/USB disconnecting. Photo below.
There's also been an automatic firmware update, unclear what effect that has had if any, but I'd say the heatsink is making the difference, given how hot it runs.
"WM Firmware Version: ____010000, Build Time: 20240716151027"
I tried AX mode, but got "kernel reports: NLA_F_NESTED is missing" which looks like an old issue. I maybe need to build latest hostapd from source.
AC is giving me 133 Mb/s throughput on iperf with some competing stations taking the whole band, and a block wall in between, which seems about right.

[morrownr]

@keynet

Thanks for continuing to report. Since I can't own and test all adapters, I have to depend on users and other sources of information to maintain the Plug and Play List. The list is a living document where adapters are added and removed based on information as it comes in.

Your experience is different than others have reported with this adapter. Manufacturing issues happen and I am wondering your adapter has an issue that is causing the heat build up. You might try to contact Fenvi to see if this is a known issue or it unusual and they will send a replacement.

I'm going to dig out my usb power meter and retest a couple of adapters for power use and heat buildup. I'll let you know.

[morrownr]

@keynet

I did some testing with my EDUP EP-AX1672, mt7921au chip, this morning. I am pretty good at abusing computer hardware. You can see the Edup adapter in the Plug and Play List.

I use my usb power meter in line with the Edup adapter.

I used the following command line several times so 2 minutes full speed followed by a very brief pause and then again. I did this 5 times:

$ iperf3 -c 192.168.1.1 -t120

[  5]   0.00-1.00   sec  86.8 MBytes   728 Mbits/sec    0    976 KBytes       
[  5]   1.00-2.00   sec  88.8 MBytes   744 Mbits/sec    0    976 KBytes       
[  5]   2.00-3.00   sec  88.8 MBytes   744 Mbits/sec    0    976 KBytes       
[  5]   3.00-4.00   sec  88.8 MBytes   744 Mbits/sec    0    976 KBytes       
[  5]   4.00-5.00   sec  88.8 MBytes   745 Mbits/sec    0    976 KBytes
...

The adapter was in WiFi 6 mode on channel 100 (DFS, clean air) 80 MHz channel width.

Result: The adapter idled at about 90 mA and when running was using just over 500 mA. I did not remove the case but rather just used my fingers to judge if there was excessive heating. It was slightly warm to the touch after the lengthy abuse but nothing that got my attention. This test was in managed mode but I have used this adapter in AP mode a lot with a RasPi4B but heat never seemed to be an issue. I never saw any stoppage or fallouts during the abuse.

I certainly believe what you are reporting but I do not have that specific adapter and my only recommendation would be to tell the story to Fenvi.

[keynet]

Perhaps we have different views on heat! The only active component of significance in the adapter is the mt7921 chip, 11 x 9 mm approx. If you stick ~2.5W into a chip that size it is going to get pretty hot in my experience (30 years of design engineering says...)
No matter, the heatsink is working for me so far.

Good news on hostapd, I compiled latest git, and got AX to work with your config info (distro version only supported AC, previous results were with 802.11N client on a laptop).
My pixel 6 reports 1200Mbit max. Here's a test with the phone 2m from the AP running Pingtools as iperf3 client to check optimum performance - looks v. similar to your results - this is channels 36-48 with a few external stations competing. I can't (yet) get DFS to work to move to clear channels, but IIRC somewhere I read DFS isn't supported on this chip.

ha-server:~> iperf3 -c 10.0.0.75
Connecting to host 10.0.0.75, port 5201
[ 5] local 10.0.0.100 port 44610 connected to 10.0.0.75 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 87.6 MBytes 735 Mbits/sec 0 2.02 MBytes
[ 5] 1.00-2.00 sec 88.0 MBytes 737 Mbits/sec 0 2.02 MBytes
[ 5] 2.00-3.00 sec 88.6 MBytes 744 Mbits/sec 0 2.02 MBytes
[ 5] 3.00-4.00 sec 88.8 MBytes 744 Mbits/sec 0 2.02 MBytes
[ 5] 4.00-5.00 sec 90.0 MBytes 755 Mbits/sec 0 2.02 MBytes
[ 5] 5.00-6.00 sec 91.1 MBytes 765 Mbits/sec 0 2.02 MBytes
[ 5] 6.00-7.00 sec 89.9 MBytes 753 Mbits/sec 0 2.02 MBytes
[ 5] 7.00-8.00 sec 89.6 MBytes 752 Mbits/sec 0 2.02 MBytes
[ 5] 8.00-9.00 sec 88.0 MBytes 738 Mbits/sec 0 2.02 MBytes
[ 5] 9.00-10.00 sec 90.1 MBytes 756 Mbits/sec 0 2.02 MBytes

---

[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 892 MBytes 748 Mbits/sec 0 sender
[ 5] 0.00-10.04 sec 890 MBytes 743 Mbits/sec receiver

[morrownr]

Perhaps we have different views on heat!

No doubt about that. I am not an EE but do have years of experience dealing with electronics in specific situations. I have over the years tested adapters for power usage and temp. I have an inline usb power meter and a laser temp device. In this case, since I did not want to take the case off, I just the good ole fingers. The biggest single factor in engineering for heat is to allow for dissipation. The Edup adapter I was testing has vent holes all over the back and on both sides and it has removable antennas with some of the metal exposed externally. There is no substitute for testing but it seems the Edup engineers did a good job of accounting for heat. I have not had other reports of as much heat as you indicate with your adapter and folks don't seem to be shy about reporting things like this. I'll keep an eye open for this.

Good news on hostapd, I compiled latest git, and got AX to work with your config info ...

That is good news. It can take a while for distros to catch up with newer versions of hostapd.

similar to your results ...

Yes, around 730-750 is about the maximum you should expect. That is pretty damn fast. I am not seeing much variation or retries so your setup is stable.

I read DFS isn't supported on this chip.

DFS is supported for managed mode but not for AP mode. The limitation is in the driver, not the chip. AP mode DFS support requires additional FCC certification and us usb users, being considered a niche market, don't always get support for some things. The only driver that I am aware of for AP mode DFS support is the rtl8812au out-of-kernel driver that I have up here at this site. We are currently working to get through the process of getting a new standards compliant rtl8812au driver into the kernel and it does not support DFS so the driver I have up here may be an accident. Is it possible for you to add DFS support if you want to compile your own driver? Probably. I know how with the mt7612u driver but it would be against the law so I don't say anything about it.

[keynet]

Thanks for the detailed info. It's still working well with the heatsink, fingers crossed that continues. The EIRP level is good too, way better than my previous ath9k based pcie card, probably +6dB or more, and one of my significant motivations for trying it - I was getting swamped by neighbours even on 5GHz bands. I live close to a 10 storey block, and at last count was competing with 19 either AP's. Now I'm fighting back and I see a couple of rival AP's have shifted their channels, so it's working!