If the problem is HW it cannot become better anymore.

In case you plan to do a new board, find below some rules for your HW design (not fully respected by us in our board). The evaluation board respects them....:

-Differential signals length less than 1inch
-Same trace length and width
-Ground planes on solder and component screen around the traces
-Avoid vias
-don't route under oscillators

Spyros



On 23/3/2012 5:33 μμ, Conor O'Gorman wrote:
On Fri, 2012-03-23 at 17:24 +0100, Spyridon Tompros wrote:
Some additional info:
Do you have a proprietary board or an EASY50712 evaluation board?
We have both and on the EASY board we do not see any errors any more. On
the proprietary board we encounter some at high rates, that's why I
believe it is a matter of traces routing.
Eventually the USB I/F is the most high bit rate of all, therefore
routing of its differential signals must be done very carefully.
Proprietary board also.

If you build the driver in debug mode I think it reports these error
stats. The error reported for us, is Transaction Error (XactErr). Of all
the errors this looks like a bit/data/crc type error.

I intended to do an eye-diagram test, but it hasn't been done yet.

As you say, high speed usb is the highest speed signal. I did turn on
the option to operate at 12 Mbps. But after the fix, I didn't see much
difference, and we need the higher speed for newest 3G modems.

Conor


_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel



--
Best Regards,
Spyridon Tompros

----------------------------------------
Dr. Ing. Spyridon Tompros
36 GE Lebon, 1160 Brussels
tel:+322.5133076
www.qualtek.eu
----------------------------------------

_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to