The issue is being identified and a work around is being worked on. It is not a Beagle issue per se, but an OMAP issue. Some boards are worse than others and some boards do not have the issue. The root cause has not been totally nailed down yet and TI is working the issue aggressively.
Gerald
I have 2 rev C3 beagleboards. 1 of them have this issue, the other not.
do keep us posted if they managed to find a workaround.
KP
then this may be my problem ...
I have a BB Rev C3 the video works but is slow and can not run video wmv
Will do. What we have found is that if you lower the ARM clock to 250MHZ the issues goes away. We are looking into the clock tree settings inside the OMAP. We are also working on a way to recover from the condition.
Gerald
This is an EHCI issue that will force a disconnect of the USB PHY requiring you to power cycle the board. Is this what you are seeing?
Gerald
my problem is that I am not able to run the video in a BB C3 wmv HD video
Then I don’t think this is an EHCI issue but some other issue you are having.
Gerald
config ethernet do not work out in the 2009 angstrom
Then I suggest that you post a sperate email thread to the group concerning this to get some assistance.
Gerald
I found how to lower the ARM clock to 250MHZ.
The source u-boot file that have to be modified is include/asm-arm/
arch-omap3/clocks_omap3.h.
Here I modified the MPU_M_13_ES2 register.
For more informations go to http://git.mansr.com/?p=u-boot;a=commitdiff;h=045149ea1076575f773079677a3d1b01ff71757c
Dario
I found how to lower the ARM clock to 250MHZ.
This solves the instability with EHCI?
Dario,
I found how to lower the ARM clock to 250MHZ.
Have your problem gone away???
This is VERY PRELIMINARY information. It appears that lowering the VDD2 voltage solves the EHCI issue that some people may be having. What we don’t know is what impact, if any, that lowering the voltage will have in other areas of the system. We also don’t know what the exact voltage setting needs to be. We think that a seting of between 1V and 1.05V will work, but that is not definite. We are working to do more testing to determine where it needs to be set to. The problem is that the issue is only seen in about 40% of the units. 60% of the units work just fine at the current default settings.
If anyone that is having this issue wants to play around with the VDD2 voltage settings feel free to do so and please let us know your results. We hope to have an official solution in a couple of weeks at the latest after more data is collected from our testing.
Gerald
Yes. It finally resolves the problem about EHCI.
The BB becomes really stable and no any more instability behaviour
appears.
I have found a strange phenomena jet: when you boot the BB sometimes
(not predictable) the usb adaptor is not detect correctly and eth0
will not come up. In order to avoid this I compile the kernel and I
have applied the patch described here:
http://article.gmane.org/gmane.linux.usb.general/19647
Best regards
Dario
Gerald Coley wrote:
If anyone that is having this issue wants to play around with the VDD2 voltage settings feel free to do so and please let us know your results.
Is VDD2 software controlled? If I ask such question does it automatically mean I am not qualified to play around with this? The best would be prebuild kernel with tunable VDD2 voltage via some knob in /sys. Or maybe it can be changed via some uboot command before booting kernel?
I think I do have this issue and I would be willing to play with it. It is rev C2 board and no matter what power supply I attach or what usb hub I attach the hub is disconnected when I attach usb keyboard and the EHCI port doesn't work (re-plugging hub, trying other usb 2.0 device) until I boot the board again. I tried 2 hubs, 3 power supplies (5V/1A and two 5V/2.5A), 2 keyboards.
Also usb harddisk connected directly to EHCI port mostly works but fails later when reading more data (using dd, testing longer movie in mplayer).
All this happens when using images from
http://code.google.com/p/beagleboard/wiki/BeagleboardRevCValidation
We hope to have an official solution in a couple of weeks at the latest after more data is collected from our testing.
Great. I hope this can be fixed only in software. I noticed this issue only recently and it is more than 90 days from buying the board so according to Support BeagleBoard Foundation now on Amazon Smile - BeagleBoard I cannot do RMA (and anyway I am in Europe and bought it via friend in US so the digi-key invoice has different name and address).
Frantisek
Gerald, i too have this problem, at this moment im running a test
using the OMAP3 processor at 250 Mhz in two beagleboards that have
this problem :(,
at this moment it is stable, if your project is viable using the board
at a lower performance i can send you the uboot image with that little
mod in it.... good luck 
Nuno
There is another fix that we are working on. We hope to have more information later this week on that. It does not require slowing down the processor.
Gerald
After 2 days one of two beagleboards i was running at half the speed
(250Mhz) to attempt to solve the EHCI issue have dropped the ECHI
port :(, im sorry but i must give up at this moment my attempts to
work around this problems .... I will RMA the board that failed and
use for the moment another OMAP3 board that have an ethernet port on
board...
here is another fix that we are working on. We hope to have more
information later this week on that. It does not require slowing down the
processor.
I hope this solves the issue .... the beagleboard have many
potentialities but it need stable connectivity....
Good Luck!!! 
The RMA for the EHCI issue will not be accepted. The solution is based on a SW fix that we are working on which involves adjsuting the VDD2 volatge level. It is nothing that can be “repaired” via an RMA. You can attempt to do this yourself or you cna wait until the SW fix has been released.
Gerald
Gerald, thank you for the reply
The RMA for the EHCI issue will not be accepted.
I have 3 RevC boards, all of them have this problem, i need to
complete a project for my company, its a critical project and for that
i bought another board so that i have hardware that worked well.
The solution is based on a
SW fix that we are working on which involves adjsuting the VDD2 volatge
level.
Yes, you have said that already, but HOW ? in uboot? in kernel ? using
APM ? in what range ? Up or Down ? Much or little ? Maybe an
experimental patch.. something.....
You can attempt to
do this yourself
I know a little of embedded work using linux and ARM arch but.... for
example to discover howto to slow down the OMAP i had to search in
uboot, to use tome tips from another users, nothing is officially
explained how to deal with this issue.
> you cna wait until the SW fix has been released.
No i cant, im sorry but my project is not an hobby project.
I understand the complexity of the problem, but you must understand
also that some of the buyers of the beagleboard are doing serious
work, and if there are problems with the EHCI port ( a critical
component for me and im sure, to many others) i think that a word of
warning to potential buyers should be said... There are many people
have been wanting for an working usb Host port to buy an
beagleboard... and now this...
thanks again and good luck
Nuno