we are currently developing a small batch product where the VIM3 would be a very good compute board for our needs. But we need USB3 and WiFi working in parallel without issues.
In this review of the VIM3, the reviewer mentions that WiFi stops working if USB3 is used:
I know that USB3 and WiFi are prone to interference, seems to be the case on the VIM3
Any chance to get this resolved before launch?
Can you check especially with a high throughput device such as a high(er)-res camera with USB-3 interface?
I am asking as we evaluated the Up Core Board from another manufacturer with such a camera (1280x720 RAW stream), and the WiFi became almost unusable. Even ping times went up to a few seconds some times. Especially if a few other wifi router are in range (high noise floor). This is also why the findings from NicoD sounded unfortunately familiar to me …
Not only the usb3/wifi issue but also the thermal issues in NIcoD ‘s review should be evaluated how serious it impact on normal use before launching. Nothing good comes easy.
We’ve done a very detailed thermal testing, the VIM3 with quite good thermal performance due to the 12nm SoC process, you can also check CNX’s post for the further details:
I use the VIM3 now daily.
I can say I almost always have this with USB3 devices that wifi stops sending packages.
It did happen a few times that USB3 was working and wifi together. But this is rare.
For that I use a wifi dongle in the USB2 port. Then there’s no problem using the USB3 port.
Also when using a 2.4Ghz dongle in the USB3 port the on-board wifi slowed down a lot.
Almost every SBC with USB3 has this problem. The N2 has it when using 2.4Ghz devices in the USB3 ports, the NanoPi M4 has it when using 2 ports next to each other(not crossed)
Thermals are not really an issue. It does throttle without a fan when maxed out for long.
But it is very useable. You can clock it lower if you don’t want it to throttle. It’ll still be a beast.
A bigger heatsink would be great.
But even with a fan at 3V it’s sufficient to keep it cool. Compared to any raspberry pi that’s very good.
Overall I’m very happy with the board. I’ve also got the Odroid N2 too. That one has a lot more issues, and the support from Hardkernel has become a lot less. Last image is still from March, full with issues.
The NanoPi M4 is all-round very good. Good heatsink, best wifi of all SBC’s, and good USB3 ports.
But the Khadas VIM3 is a lot more powerful and uses a lot less energy. So it’s all about what your use case is.
Greetings.
I checked a SSD mobile hard disk at the USB3.0 port.
test method : Use iozone command to test the speed of mobile hard disk to keep the hard disk and USB3.0 port working at high speed.At the same time, I tested the WiFi rate of the board on the server.
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.0.250 port 5001 connected with 192.168.0.133 port 57326
[ ID] Interval Transfer Bandwidth
[ 4] 0.0-30.1 sec 733 MBytes 204 Mbits/sec
[ 5] local 192.168.0.250 port 5001 connected with 192.168.0.133 port 57328
[ 5] 0.0-30.0 sec 750 MBytes 209 Mbits/sec
And then, I plugged in wireless mouse and keyboard on USB 3.0 port.
There was no change in the test results.
Can you tell me about your specific testing tools and procedures? I can restore your steps to test.
I’ve got this with my USB3 SD-card reader plugged in to USB3, and also with my SSD with SATA to USB3 adapter.
Wifi is still connected. But surfing isn’t possible. Packages are getting lost. When not connected to wifi I can see the available wifi networks. But when trying to connect to it, it asks for the password(normally it should automatically log in without asking this). But after entering this, it tells me it’s the wrong password. (it for sure isn’t)
I had this with Ubuntu Server without and with a desktop installed, and also with Ubuntu XFCE.
I haven’t been able to test in another OS since I’m unable to connect wifi in Android (missing connect button).
And as far as I know none of the other available OS’es have the wifi drivers integrated yet.
I use Armbian with a wifi dongle. So I don’t find it a problem personally.
With my 2.4Ghz mouse/keyboard dongle plugged into the USB3 port I only got about 1Mbps download speed. (normally +40Mbps)
That was the first thing I noticed. After a while I thought about using a USB hub in the USB2 port for everything 2.4Ghz wifi and since I haven’t had any issues at all.
I’ve just gotten a few NVMe drives. I’ll connect one. Then the USB3 port should work as USB2. I’ll see if on-board wifi then works with something plugged into that port.
Greetings.
Thanks für the test, the results are looking good on a first look!
Did you test with a 2.4 or 5 GHz WiFi access point?
And what distance between access point and the VIM3? Was there a wall in between?
I had the problems on the Up Core with a 2.4 GHz access point about 5 meters away.
5 GHz was not available on that board, but I would need 2.4 GHz anyhow due to the much better range when walls etc are present. I also read that USB3 is especially prone to 2.4 GHz interference: https://en.wikipedia.org/wiki/USB_3.0
I did not test a USB3 hard disk. It was a USB3 camera. But I suspect that should not make any difference.
At least this very interesting white paper from Intel states a hard drive also as problematic: https://www.intel.com/content/dam/www/public/us/en/documents/white-papers/usb3-frequency-interference-paper.pdf
They also measured a ~ 25 dBm Noise floor increase, from -115 going up to -90 dBm, that’s a lot. Still, WiFi needs a signal with about 8-10 dBm signal-to-noise-ratio for proper decoding. So at small distances and with line of sight there might be no drop in transmission rate even in case of such a strong USB3 interference. A -80 dBm Signal is easily possible in such conditions. That’s why I asked about the testing conditions
Go on with your good work, really appreciate your support!
@NicoD@thm82 Can you tell me you firmware version which you used ? This may be related to firmware versions. Early firmware versions of BUG were more common.
Which tool you used to test ? I passed the LAN test.Because passing the extranet test will be affected by the speed of the extranet.My test tool is iperf.
@NicoD@thm82
Maybe I found the problem.
The reason why our test results are different is that we use different networks.
I’m using a 5GHZ wifi, and you’re using a 2.4 GHz wifi.
With the USB 3.0 port using the device, 5 GHz WiFi can still be used normally, but 2.4 GHz WiFi will be affected.We will try to find out the reason and solve it in the next step.It may take some time.
Glad you’ve found it.
Indeed only with 2.4Ghz. USB3 has an interference on the 2.4Ghz frequency. It should be well known, but clearly not known enough.
I can’t connect to 5Ghz wih it(in Ubuntu). It only shows my 2.4Ghz wifi. But it does work very fast for 2.4Ghz wifi.
It never shows my 5Ghz equivalent.
I’ve asked @Gouwa multiple times if this chip was able to do 5Ghz wifi but did not get a reply so I assumed it’s only 2.4Ghz.
I also say in my conclusison that with 5Gz wifi this would not be a problem.
So another thing is, why can’t I see any 5Ghz wifi connect point.