USB disconnects problem
nuccio raciti
about this annoying problem, I'm trying to follow two paths:
1) reproduce the problem on Sim1 using the last kernel provided by
Cirrus (2.6.20.4),
2) try to get the problem using the edb9307A board
I'm convinced that the results of these two tests could help us.
I want to be sure that this is realy a hardware problem.
Could you help me create a simple test to reproduce this problem?
What do you think about?
Cheers,
Nuccio
marcus
> 1) reproduce the problem on Sim1 using the last kernel provided by
> Cirrus (2.6.20.4),
> 2) try to get the problem using the edb9307A board
Regards,
Marcus
marcus
> I already have the Cirrus 2.6.20.4 kernel working on sim1,
> I need help in order to understand how reproduce the USB problem.
Oh, I misunderstod. Nevertheless, I will spend the afternoon with SimOne :)
Short instruction to (try to) reproduce the USB disconnects:
---------------------------------------------------------------
0. Attach serial console cable, ethernet cable, 7.5V power adaptor to the SimOne.
1. On a desktop PC: Put a file system on an empty USB stick (USB2.0) with a ext2
file system:
tar zxfpS debian-armel-lenny-rootfs-20091017.tgz
2. Boot the kernel from u-boot-1.1.6-cirrus-edb93xx-platform (Apr 5 2009 -
09:38:43)
3. Log in to the system.
4. For me there are two ways to "force" a USB disconnect. Repeat these commands
until USB disconnect occur:
# dd if=/dev/zero of=junkfile count=262144
# dd if=/dev/zero of=junkfile count=262144
# ...until USB disconnect occur
or
# apt-get update
# apt-get install alpine
# apt-get remove alpine
# apt-get install alpine
# apt-get remove alpine
# apt-get install alpine
# ...until USB disconnect occur
or a combination of the two. Any other package than "alpine" can be used to get
USB disconnect too. A few apt-get autoremove and similar might be needed if your
system seem stable. dd almost always need two or more runs before
USB-disconnect, but sometimes only one run is needed. If the count is low, eg.
dd if=/dev/zero of=junkfile count=32768 there is a greater chanse to not get a
USB disconnect.
What you also can do is to boot as normal, then reset the SimOne (power toggle
is ok too, I think) so that you get a forced file system check the next time you
boot. This fsck sometimes suffer from USB resets, and maybe disconnects too. But
sometimes everything is normal, so this method is the last to try.
Tell me if you want even more detail in the procedures.
Did I tell I ran the USB stick from USB2 on the addon1 port?
The results was the same as with the normal USB ports. Maybe even worse due to
my wiring. I never investigated further.
Regards,
Marcus
nuccio raciti
about the point 1,
>1) reproduce the problem on Sim1 using the last kernel provided by Cirrus (2.6.20.4),
too hard to configure a working environment, the Cirrus build sistem
is a mess. I failed to reproduce... :-(
about the point 2:
> 2) try to get the problem using the edb9307A board
I successfully run our linux on dev board (EDB9307A),
(please note, some drivers are not recognized by the kernel, see the
EDB9307A log.txt file)
and I successfully installed X11....
# apt-get update
# apt-get install xserver-xorg-video-fbdev xfonts-base
so seems it is a Hardware Problem....
I did another (HW) test on the board SIM1,
I powered the device (connected to the USB of SIm1), whith a reliable
external source of 5 volt,
but the problem is still present.
Regards,
Nuccio
sergio.sorrenti
> so seems it is an Hardware Problem....
>
> I did another (HW) test on the board SIM1,
> external source of 5 volt,
> but the problem is still present.
I think that the next step is to change the CPU Crystal Oscillator,
with one more reliable, like a TCXO and see what happens.
Sergio
martinwguy
On Dec 28 2009, 6:08 pm, "sergio.sorrenti" <ser...@4star.it> wrote:
> > so seems it is an Hardware Problem....
> I think that the next step is to change the CPU Crystal Oscillator,
> with one more reliable, like a TCXO and see what happens.
Gaspare thinks it is a problem that he has already seen, to do with
the placement of teh crystal so far from the CPU.
However, if you think that a different type of crystal may solve the
problem, can you send me one and I'll try mounting it.
Cheers
M
sergio.sorrenti
on the Sim.One in order to solve the USB disconnect :
1. We used a more accurate Crystal oscillator
(removed from the latest EP9307A Evaluation board)
1. Measured the capacitance of each of the capacitors,
and the resistor of the crystal circuit. They are correct.
Since the above do not returned valuable results,
we are going further with measurement
with more accurate instrumentation.
Sergio
sergio.sorrenti
> Since the above do not returned valuable results,
> we are going further with measurement
> with more accurate instrumentation.
Well, displaying both of the USB signals (D+ and D-) :
both signals on EVB9307 and Sim.One are without differences
with same ripple and on Sim.One there are not overlap of noise.
All the test was performed with two Sim.One crystal on the two boards
and both boards was powered with two Sim.One power supply.
Some documents:
1- A printscreen of the image with the two signals during access to an
USB pendrive.
http://www.simplemachines.it/doc/BitScope.bmp.zip
2- the signals recording in BitScope format
of the normal access to an USB pendrive on the EVB9307 board
(standard csv files, can open with openoffice)
http://www.simplemachines.it/doc/DevBoard_USB_Signals.csv.zip
3- the signals recording in BitScope format
of the access to an USB pendrive on the Sim.One
With the I/O error...
http://www.simplemachines.it/doc/SIM1_USB_Signals.csv.zip
Guide of the signal recorder here:
http://www.bitscope.com/software/dso/guide/1.3/?p=recorder
Any Idea?
Sergio
sergio.sorrenti
here:
http://groups.google.com/group/sim1/browse_thread/thread/3dd51bb0d9eb9147
Sergio