Page 1 of 1

Using XMOS-based board in the 24/7 mode

Posted: Tue Jan 09, 2018 10:55 am
by ffomich
Hi,
does anyone use XMOS-based board in the 24/7 mode succesfully?

I have self-made board based on the XE216-512-TQ128-C20 with UAC1 firmware. Windows application uses this card for record and play sound continuously for a long time. The card uses standard Windows audio driver.

It worked without interrupts for about a month but after an undefined moment it started to disappear from OS for 1 second several times a day.
When Device Manager is open I see USB device disappears and appears again. After the card power reboot it continue to disappear from OS for 1 second several times a day.
This effect is present on the different PCs with Windows 7 & Windows 10.

Does anyone have the same problem?
What analyzer and test tool can you recommend?

Re: Using XMOS-based board in the 24/7 mode

Posted: Tue Jan 09, 2018 7:19 pm
by mon2
No direct experience with this subject but understand that Microsoft device drivers undergo days of stress testing to ensure the IP is solid for a best end user experience. Our s/w group in Asia does this for us as we threw in the towel after attempting to build a compliant box using certified components for this suite of testing. Wasted 3 weeks and quit. The device driver tests go on for days and keep log files. If you pass these tests without serious faults, then Microsoft will bless you with their signature to state you are worthy of being Microsoft Windows compliant.

You could review the following:

https://docs.microsoft.com/en-us/window ... ting-guide

Any chance the host PC is going to sleep mode and therefore disconnecting your USB device? Try tinkering with the PC power settings of your PC and shut off any sleep mode.

Recommend to send out an email to each such supplier of these tools with your exact requirements that you wish to stress test your USB device - what is the best solution. These options should be more robust than the method you are using now because the tests often will simulate USB hot unplug and dock, etc. - thousands of times to be sure your widget does not choke.

Aside from the above, what is the Vbus capacitance in your design? Cannot exceed 10 uf.

Are you using impedance controlled traces for the USB interface? 4L PCB? In-rush current protection on the USB Vbus line? ESD protection? These are all recommended.

Is your widget "becoming faulty" after a while and remains to be faulty? That is, a fresh board works and then after xx period of time, the board acts up? Can you resolve the quirk by removing and docking again to the USB port?