Is there not just some way of saying "hey, USB port, are you alive?"
Sure, but you might need to add it to the driver. Read the following page on
USB testing for more info. There may also an option to ask dbus, hal, or udev
whether the device is still alive (I haven't found one so far, but it might
exist)
I can come up with some kind of check if network is up or down and can
be brought up and whatnot, but it seems like each of these checks I
put in for this problem just keep getting more and more hackish.
Yes, but is there a "clean" way to write a workaround for a hardware bug?
Especially an intermittent one that defies characterisation.
I was under the impression that there was a lot of hope for a software
solution: that's one of the reasons no RMAs are being taken for this
problem. Is that not so?
We have a fix that will work on some of the boards, but not all of the boards. You can place a large cap, 22uF across C97. This has shown to fix a lot of boards, but again, not all of them.
The production fix will come in the next revsion of the board, C4.
I'm not entirely comfortable soldering things onto my board for a
maybe. If you could say definitively it would work, I'd probably risk
it, but I don't trust myself not to screw things up more.
If the fix is coming with C4, what are me and my hiccuping C3 supposed
to do?
I can solder a bit but I am SMD soldering newbie so a more detailed howto would be nice. Should I remove existing C97 (and how) or can I place another capacitor over the existing one (and how)? Can it be non-SMD part? Thanks.
The same question came to mind, C97 seems to be a 0402 part, but I was
thinking a tantalum capacitor (I'm sure I have some) across C97 might do
the job - some circuit tracing would need to be done to ensure which end
of C97 is connected to positive.
Regards
Sid.
The first number is the DigiKey number. The manufacturer is TDK.
Gerald
Thanks, rapidonline.co.uk has them in stock and I even had the stock
number for the 0805 written down since last night.
They are a good source for components over this way.
Regards
Sid.
I got my order from rapidonline.co.uk pretty rapid (excuse the pun). I
used a toothpick with a spot of bluetack on the end which kept melting
and the cap kept falling off, but once I got it seated squarely, it
soldered fine. Why do tools like my fine point tweezers hide when you
need them?
Hoping I am not speaking too soon - it's been up 11+ hours without a hiccup.
Regards
Sid.
We are also seeing good results with this as well. Please keep us posted!
Gerald
Running 22+ hours and still stable, looks a good fix.
Regards
Sid.
It has cured the hub drop out, but occasionally the eth or the wlan
needs to be replugged. Still a better situation than having to reboot.
I don't know how or when, but wlan0 also now gets an IP address from the
router - perhaps something done in the last opkg upgrade.
Regards
Sid.
The same question came to mind, C97 seems to be a 0402 part, but I was
thinking a tantalum capacitor (I'm sure I have some) across C97 might do
the job - some circuit tracing would need to be done to ensure which end
of C97 is connected to positive.
So what side of the capactor is the positive one?
The one facing the SD slot or the one facing the audio output?
Does it have to be a SMD part wouldnt a "normal" 0.22uF do the job?
We use ceramic caps which are non-polarized. The side closest to the SD
card is positive.
Gerald
The 0805 22uF ceramic cap is the one I have used. An improvement, but
not a solid cure, I wonder if another 22uF on top would likely cure the
problem of periodically losing contact with the Wifi and the LAN - the
hub now stays up so I don't have to reboot, re-plugging the devices is
all that's needed.
Another effect I have noticed since installing the cap, the speakers
emit a sound just like if you have a hard drive or DVD drive having
trouble reading media.
Regards
Sid.
The issue that the CAP is intended to solve is the lockup of the PHY. When I say lock up, it can’t be reset without a power cycle. We have not tied any other behavior than this to this issue. I don’t think added capacitance will help but it can’t hurt to try.
Please note that adding capacitance on the output of the switch mode
power supply (placed inside TPS65950), will reduce the phase/gain
margin of the power supply. So adding too muck capacitance on the
output will make the conrtrolloop unstable, and then it will start to
oscillate. This oscillation will generate a lot of noise on your power
rails, so you might end up making things worse!!!
I am not saying that the power supply IS getting unstable when you go
from an output capacitance of 10uF to 10uF+22uF, but it will get less
stable and MAYBE unstable!
