USB & Ethernet dead

[fsp]

Hi folks,

I got my RPi an flashed debian & arch.

It all boots up, 5V & 3,3V is stable all the time, but no keyboard (HID) is recognized. Nor does the LNK LED flash, when I plug in a ethernet cable connected to the switch.

Bootup screen (debian) says eth0 is configured within rc.local but not present.

Is there any chance this is a problem I could fix on my own?

I tried different power sources (700mA, 1A, active USB HUB, ...)

Is there any way I check if the LAN9512 is working properly?

Thank you for your advice.

Best Regards

Felix

[Bengaul]

Does your pi simply not recognise your keyboard, or does it matter which USB port it is connected to?

[jbeale]

You mention 5V is "stable" but how close is it to 5.0 V? It's known that low voltage can cause this:

http://elinux.org/R-Pi_Trouble.....plugged_in

If you're above 4.8 V or so, likely it's something else. Do you get ethernet working with no keyboard at all?

[fsp]

Thank you for the quick reply, guys.

Bengaul: It doesn't matter which Port. 5V is available at the USB Ports, but no connected device seems to talk the Pi.

jbeale: I use a calibrated digital multimeter, which reads 5.02V at C6.

When I connect some external USB device or an ethernet plug, this value doesn't change significantly.

Ethernet doesn't show any lifesign, whether USB is connected or not.

[Chrunchstick]

There is the chance that the USB bus is malfunctioning because the Ethernet port is also connected to that same USB bus.

That is what I would think if both my usb ports and ethernet port where'nt working.

[jbeale]

That's an unusual one, we can't blame either power supply or a USB device; sounds like it is something on the board itself. If it's the same firmware and kernel that is working for other folks, sounds like we have to blame a hardware fault.

[fsp]

aargh, I was fearing that answer...
I'm going to check if X1 is generating the 25MHz clock for the LAN9512, tomorrow.
Is there any chance, the there is a faulty Ethernet jack assembled? If so, do you know how to identify, which one's the right one?
Thanks a lot.

[fsp]

Hi folks,

it turned out that my first guess was right. The 25MHz clock for the LAN9512 was missing.
After applying an external clock (using a NI frequency generator), the USB and ethernet worked immediately.
Now for the weird part: After removing the external clock source, the crystal now works on its own, everytime I start the device. So mi Pi is working *cheers*.
Although I have no explanation for that, I hope this might one or another.
Thanks again for your support.

Best Regards
Felix

[RussIT]

I have the same fault both ethernet and the USB ports don't work. Have sent the unit back to RS hope to get a replacement soon.

[shad0w]

Sadly I've got the same problem. Is there an easy way to fix it without deeper knowledge? Do you know how long I would have to wait for a replacement unit?

[jbeale]

Now for the weird part: After removing the external clock source, the crystal now works on its own, everytime I start the device.

Truly strange. Just looking at the package on the board, it appears to be using an external oscillator, not simply an external passive crystal with the active circuit on-chip. If there was a "low activity" or mistuned crystal with the wrong/missing caps that could maybe not startup, but continue on its own once kicked by an external signal. But in any case I cannot think what problem could be fixed in this way and have the fix survive a power cycle. Maybe a contamination layer on the crystal surface (which would detune it) that was knocked off on initial startup but I've never heard of such a thing. Also these were supposedly 100% tested in the factory (?)

Anyway sounds like a manufacturing defect to me, the normal remedy would be just return the board. From the few reports we've had, I believe you can get a replacement in a week or so.

[mmcv2008]

Had the same problem as my Raspberry Pi arrived on Friday, just used the scope to check the oscillator and suddenly it started to work. Looks like a cold solder point. I'm wondering if this has passed QA!?
Should not happen for the targeted audience - school kids - just fustrating and will never be touched again. So there is some room of improvement.

[shad0w]

Mine arrived friday too. Maybe there is a problem with the latest batch? Did you order also from RS?

[rurwin]

I would expect that the distributors will be feeding back these sorts of problems into the test process, so newer boards with the same fault will be caught in production.

Remember that only 20 or so boards were extensively tested. The sorts of problems that only happen 1% of the time would not have been seen.

[mahjongg]

One thing that can cause these sorts of problems is a supply voltage that raises too slowly, so the crystal doesn't get the "push" it needs to start (and yes, I'm simplifying what really happens). In such cases a tiny amount of injected noise on one of the crystal pins might be all that is needed to start it.
What is what might explain what happens when you put a probe on one of the pins.

This also points to a too small gain in the crystal oscillators inverting amplifier (inside the LAN9512 chip).
Increasing the feedback resistor, and or lowering the parallel capacitors might help fix this.
Its one of the problems one might find in early production batches.

Sometimes it might help the oscillator to start by making sure the supply voltage rises faster, in practice this can be accomplished by not plugging in (or switching on) the adapter, but by powering the adapter first, and then plugging in the micro USB plug, obviously this isn't a permanent fix, so I would also advise returning the device with a good explanation what is wrong with it.

[johnbeetem]

it turned out that my first guess was right. The 25MHz clock for the LAN9512 was missing.
After applying an external clock (using a NI frequency generator), the USB and ethernet worked immediately.
Now for the weird part: After removing the external clock source, the crystal now works on its own, everytime I start the device. So mi Pi is working *cheers*.
Although I have no explanation for that, I hope this might one or another.

My guess is that when you pressed down on a component with a probe you made an electrical contact that was missing before. Or if you soldered a lead onto one of the components to attach a wire for your frequency generator you cleaned up a cold solder joint. In the former case, the bug could very well come back after RasPi has thermal cycled a few time. Aren't Heisenbugs fun? (A Heisenbug goes away when you try to measure it.)

[mmcv2008]

Hi johonbeetem your right, but I have been lucky so far - switched off over night and switched on again and still works. So I do not agree to mahjongg's arguments and suggest the boards should be automatically tested better, as nurwin mentioned. Yes, I got my board from RS.

[rwhite226]

I have a quick question. I got by rp on Friday and it worked (except for the repeating key problem) but today it has the problems you said (no usb for lan). So I tried all the troubleshooting things you listed and none of them worked. By chance when testing the voltage I hit the first two gpio pins (ones closest to the sd card) and the lan and usb came on, so i grabbed a jumper and hooked them together and now all the problems stopped. My question is is it fast to keep these pins connected?

[rwhite226]

i mean safe not fast.

[mahjongg]

Hi johonbeetem your right, but I have been lucky so far - switched off over night and switched on again and still works. So I do not agree to mahjongg's arguments and suggest the boards should be automatically tested better, as nurwin mentioned. Yes, I got my board from RS.

Which "arguments" are those, I'm not aware I made any arguments against better testing.

Now I think of it, there is a logical explanation for the 5V to come up unusually slow, and that is (again) the polyfuse, which together with capacities on the board forms an RC circuit,

[somers]

I have a quick question. I got by rp on Friday and it worked (except for the repeating key problem) but today it has the problems you said (no usb for lan). So I tried all the troubleshooting things you listed and none of them worked. By chance when testing the voltage I hit the first two gpio pins (ones closest to the sd card) and the lan and usb came on, so i grabbed a jumper and hooked them together and now all the problems stopped. My question is is it fast to keep these pins connected?

I had the same problem but i've send the board back.

But on what pins did you put the jumper on? Lets say this is the GPIO header:

: : : : : : : : : : : : :

Did you put it on the first two pins of the bottom line (horizontal) or on first pin of top line and first pin of bottom line (vertical).

[somers]

I have a quick question. I got by rp on Friday and it worked (except for the repeating key problem) but today it has the problems you said (no usb for lan). So I tried all the troubleshooting things you listed and none of them worked. By chance when testing the voltage I hit the first two gpio pins (ones closest to the sd card) and the lan and usb came on, so i grabbed a jumper and hooked them together and now all the problems stopped. My question is is it fast to keep these pins connected?

So when you remove the jumper the both the usb and ethernet port stop working?

[Burngate]

... By chance when testing the voltage I hit the first two gpio pins (ones closest to the sd card) and the lan and usb came on, so i grabbed a jumper and hooked them together and now all the problems stopped. My question is is it fast to keep these pins connected?

If "the first two gpio pins" are top row (edge of board) 1 & 2, that's 5v and Do-Not-Connect, so who knows what that is! but it's not a good idea to connect them!
Bottom row is 3v3 and the data line of I2C - shouldn't do much, but I still wouldn't leave them shorted.
If you mean the two right at the end, then that's 5v and 3v3 - shorting those will destroy the Pi.
At the other end of the connector we have SPI signals and another Do-Not-Connect - unlikely to do much, and I wouldn't leave them shorted.
The only thought I have is that you connected 5v to 3v3, but were lucky and have just blown some track. But in doing so you put enough current through C11 (47uF, tantalum on the 3v3 line) to reform it.

[on74]

Hi!

I think my Raspberry is broken to.

Recived my board today from RS.
USB-keyboard and ethernet do not work. The keyboard sometimes work for a short time until I get a kernel panic (when It works it feels sluggish and some keystrokes do not get registrated).
First I thought it was the keyboard so I tested several, also tried to use a powered USB-hub.
Buy the way, I use a powersupply from RS so I don't think that is a problem.

The ethernet also sometimes works for a couple of seconds (sometimes it manage to register a IP-adress from the DHCP-server) but the connection always drops dead a moment later (never last for more than some seconds).

I tried to measure the crystal with a oscilloscope but the frequency is random?
I am no expert with a oscilloscope but if i measure on pin8 (and ground) shouldn't it show me a steady 25MHz reading? (I tried to measure the crystal on my Arduino board and it worked - but the crystal has only 2 pins)

Pinout
1 7
14 8

Sorry for my bad english.
/Olof

[on74]

Now I definity know something is wrong with my board.
Measured voltage on TP1 and TP2.

TP1 = 5v
TP2 = 1.95v

/Olof