NO usb device works

[tangmonster]

Hi there. Got my raspberry pi yesterday. Since one of my work colleges got his on Friday i was extra excited.

sadly absolutely no usb devices works on mine.

I take my friends working keyboard and sd card and powersupply and swap in my board , boots just fine but no keyboard or mouse at all.

I contacted RS south africa. But they just give me a link to a form that i must fill in and send it all the way back to the UK!!

Anybody got any ideas?

[tangmonster]

ethernet also did not work , so i started to suspect the crystal.

after i puled out the oscilloscope it was obvious that the cpu 19.2Mhz crystal was running just fine but no go on the 25mhz.

I reflowed the 25mhz with the soldering iron. And measured again and then the crystal started clocking.

I have USB AND ETHERNET!!!

YAY!!

[jbeale]

I take my friends working keyboard and sd card and powersupply and swap in my board , boots just fine but no keyboard or mouse at all.?

Seems to me a convincing demonstration of a hardware fault on your board, with known good KB, SD Card and Power Supply, what else could it be? (as long as nothing else is connected to the board- did you also use his mouse?).

A few people have reported a dead clock chip going to the USB/Ethernet controller, which could cause your symptoms. Might be a bad chip or bad solder joint but unless you want to debug or rework the hardware yourself, a return is the only thing I can suggest.

[tangmonster]

thanks

like i said in my previous post.

I re-soldered the 25mhz crystal pins and rechecked with the oscilloscope and it started to clock so all is working now.

[mahjongg]

I take my friends working keyboard and sd card and powersupply and swap in my board , boots just fine but no keyboard or mouse at all.?

Seems to me a convincing demonstration of a hardware fault on your board, with known good KB, SD Card and Power Supply, what else could it be? (as long as nothing else is connected to the board- did you also use his mouse?).

A few people have reported a dead clock chip going to the USB/Ethernet controller, which could cause your symptoms. Might be a bad chip or bad solder joint but unless you want to debug or rework the hardware yourself, a return is the only thing I can suggest.

Not a dead clock chip, but a non oscillating quartz crystal!
Quartz crystals are temperamental beasts, whether they start up can depend on many factors, such a as temperature, but also on how quickly the voltage on the driver part of the crystal oscillator rises.
Generally a quartz oscillator uses a very high gain linear loopback (resonator) circuit, and amplifies the little bit of "start signal" that is available on its input, the crystal only "lets its resonating frequency through", so that ultimately the resonator oscillates at the quartz crystals resonant frequency (or an uneven harmonic of it). To start oscillating there should be that little "start signal" (or ringing) which in systems such as this is provided by the initial rising signal of the output of the driver because it suddenly receives power from the power supply. But if this rising signal raises too slow the crystal doesn't get the "kick" it needs to start ringing, and thus the oscillator never starts. Sometimes its possible to put a bit of random signal (noise) on the oscillators input to start it up, that is why the oscillator sometimes starts when you put an oscilloscope probe on the crystal.

Now not all crystal + analog loopback circuits are equally sensitive, so start up behavior might vary, but it also depends on how the system is powered up, slowly or quickly. If you first connect the adapter to the PI, then plug in the adapter in the mains, the voltage might raise slowly. If you plug the adapter in the mains, so that it fully started up, then plug the mini USB in, the voltage on the board should raise abruptly, and the oscillator might start. Due to the physical nature of a crystal, once it has oscillated it might start up much easier afterward. So it is worth to try to get the quartz to start an way you can.

Cases like this can come about by rough handling during transport, affecting the crystal, but it can also point to a too critically designed circuit, with not enough gain in the analog feedback circuit, this can be tuned by increasing the resistor (that sets the gain) mounted parallel with the crystal, and/or decreasing the "tuning capacitors" of a few pF on each side of the crystal to ground.

If you cannot get the crystal to oscillate no matter what you should conclude that the crystal has been damaged during transport.

P.S. I just read you have the xtal working.. great! I suspect that just the act of measuring the signal got the crystal going, its unlikely the soldering was bad, unless one of the pins was actually disconnected.

[erul]

Not a dead clock chip, but a non oscillating quartz crystal! Quartz crystals are temperamental beasts, whether they start up can depend on many factors, such a as temperature, but also on how quickly the voltage on the driver part of the crystal oscillator rises.

I put my Pi in the fridge for 20 minutes and both Ethernet and USB worked for 2 minutes.
No need to tell me how stupid I am, I'm fully aware of it.

[mahjongg]

Not a dead clock chip, but a non oscillating quartz crystal! Quartz crystals are temperamental beasts, whether they start up can depend on many factors, such a as temperature, but also on how quickly the voltage on the driver part of the crystal oscillator rises.

I put my Pi in the fridge for 20 minutes and both Ethernet and USB worked for 2 minutes.
No need to tell me how stupid I am, I'm fully aware of it.

Not so stupid! There is a direct relation to temperature and the switching speed of silicon, for each extra degree (kelvin) the speed drops a part of a percentage point, (don't know how much exactly from the top of my head, I have to look it up) that is why PC builder try to keep the chips cool.
But for a crystal analog loop-back amplifier the switching speeds also translates to gain, so cooling it down indeed helps to increase the gain of the loop-back circuit.
This experiment proved that the crystal wasn't broken, only that the gain is very critical.

[erul]

Not a dead clock chip, but a non oscillating quartz crystal! Quartz crystals are temperamental beasts, whether they start up can depend on many factors, such a as temperature, but also on how quickly the voltage on the driver part of the crystal oscillator rises.

I put my Pi in the fridge for 20 minutes and both Ethernet and USB worked for 2 minutes.
No need to tell me how stupid I am, I'm fully aware of it.

Not so stupid! There is a direct relation to temperature and the switching speed of silicon, for each extra degree (kelvin) the speed drops a part of a percentage point, (don't know how much exactly from the top of my head, I have to look it up) that is why PC builder try to keep the chips cool.
But for a crystal analog loop-back amplifier the switching speeds also translates to gain, so cooling it down indeed helps to increase the gain of the loop-back circuit.
This experiment proved that the crystal wasn't broken, only that the gain is very critical.

Even a broken Pi is a great "educational device" .
Is there a more permanent solution to this? Keeping the crystal below 15c is not so easy. I really don't want wait for RS slow support.

[johnbeetem]

Even a broken Pi is a great "educational device" .
Is there a more permanent solution to this? Keeping the crystal below 15c is not so easy. I really don't want wait for RS slow support.

Some suggestions if you're handy with a "heater" (so much more poetic than "soldering iron" )

1. Resolder X1. It appears to be hand soldered and quality is variable according to reports. While you're at it, touch up C41, C51, and R39 around X1. It's really important that all of these components have sound connections.
2. Replace X1 itself. Maybe you've just got a defective crystal.
3. Replace X1 with a 3.3V crystal oscillator. Crystal circuits are very hard to get right and the extra circuit elements in the PC board can screw things up. For example, every via hole is an inductor and can affect crystal operation. An oscillator chip is more expensive than a crystal, but they're pretty fool-proof. You do want good bypass on the 3.3V supply. See the RasPi schematics, PC board layout, and LAN9512 data sheet.

[jbeale]

thanks
like i said in my previous post.
I re-soldered the 25mhz crystal pins and rechecked with the oscilloscope and it started to clock so all is working now.

Glad you got it working! Yes now I see your 2nd post. Despite the posting times shown, when I wrote my reply, there was only just the first post visible in my browser.

I see from the schematic that X1 is a simple 25 MHz crystal, and not a packaged oscillator circuit. I was confused by just looking at a photo, as I am using to seeing 4-pin packages like that hold a complete oscillator circuit.

By the way, Section 4.6 "Clock Circuit" (page 48) of the LAN9512 datasheet shows the recommended crystal spec http://www.smsc.com/media/Downloads_Pub ... s/9512.pdf and the reference design schematic http://www.smsc.com/media/Downloads_Pub ... 12_sch.pdf shows the same part values (2x 33 pF to ground, and 1M in parallel) for the LAN9512 xtal osc. that the R-Pi schematic does.

It seems some boards have bad solder joints, and perhaps others a sluggish "low activity" crystal?
For those interested, here is a summary of reasons why crystal oscillators sometimes do not work: http://www.freescale.com/files/microcon ... AN3208.pdf The LAN9512 chip on the R-Pi uses a parallel resonant (Pierce Oscillator) configuration.

[tangmonster]

I was probing at that crystal with the oscilloscope like crazy and it did not run at all.

Under the microscope it was obvious that the crystal was soldered well enough. The reason why i was soldering was that we had some 25Mhz crystals here at the office we used for an ethernet circuit we designed and built a few years back. I was going to replace the crystal but after reflowing i figured let's check it one last time.

And that is when it started working. So for me I think it was actually the crystal heating up a bit that got it started running.

It is still working today though.

[erul]

Even a broken Pi is a great "educational device" .
Is there a more permanent solution to this? Keeping the crystal below 15c is not so easy. I really don't want wait for RS slow support.

Some suggestions if you're handy with a "heater" (so much more poetic than "soldering iron" )

1. Resolder X1. It appears to be hand soldered and quality is variable according to reports. While you're at it, touch up C41, C51, and R39 around X1. It's really important that all of these components have sound connections.
2. Replace X1 itself. Maybe you've just got a defective crystal.
3. Replace X1 with a 3.3V crystal oscillator. Crystal circuits are very hard to get right and the extra circuit elements in the PC board can screw things up. For example, every via hole is an inductor and can affect crystal operation. An oscillator chip is more expensive than a crystal, but they're pretty fool-proof. You do want good bypass on the 3.3V supply. See the RasPi schematics, PC board layout, and LAN9512 data sheet.

I used "heater" on the X1 chip and now it works like a charm . (for now)

I'll report back if something changes

[Terahnee]

Desperate as i was, being an owner of a Pi but without a working USB/Ethernet-Port, i put my Pi into the fridge for at least 1 hour. Now, 3 hours later it still works!

[baneat]

After you stuck it in the fridge, still working even later?

[matalo]

I am experiencing this problem. After putting in the fridge for a number of hours I got back USB and ethernet for about 5 minutes before they both disconnected while the RPi was still switched on and working.

Gotta be honest, I thought this thread was a joke at first and I got some stick from my housemates when they saw me putting the RPi in the fridge

I'm no electronics expert, and whilst I grasp the basics of the information in this thread I have no hope of being able to fix this myself.

I guess I need to go to RS and tell them I have a faulty unit?

[boley]

I have the same symptom on one of the two Rpi's that I have access to. We'll call the problem Rpi my Dad's and the working one mine. Using the same SD card, with Debian Squeeze, power supply, Ethernet cable and USB hub to ensure that the keyboard and mice weren't drawing too much power, Dad's Rpi boots up and waits for me to login, but I can't. None of the FDX, LNK, or 10M LEDs flash, on Dad's. My LEDs flash, and of course I can log in and have a great time.

I'm trying the fridge trick first, then I may pull out the old O'scope to see what I can see. The solder joints on X1 and the surrounding components look good under a magnifying glass.

BTW, these both came from element14 within a few days of each other.

[churchill]

Hello,

I have the same symptom on one of the two Rpi's that I have access to.

Almost the same for me !
When (yesterday) I received my Rpi, the board started working like a charm on Debian. I played 2 hours with, launching lxde, working with usb keyboard/mouse and ssh.
Then, I unplugged all except power and ethernet link... and now more ethernet nor usb (not even the leds).
I'll try the fridge trick. I'm not particulary happy to send back my Rpi : I think the cost of shipping from Switzerland would be more or less the same cost as for a new card (and I've waited soo looongtime) !

PS: sorry for my english...

[grumpyoldgit]

It is worth pointing out that Liz has mentioned more than once that most of the Pis being returned to RS and Farnell have nothing wrong with them. While I am sure that there are faulty Pis out there, it is worth going through all the alternatives first. I would also say that because one user has put his Pi in the fridge and it now works, I am not entirely sure that this is going to be the golden bullet. The majority of issues are going to be down to power or the SD card.

It is also true to say that getting information out of RS or Farnell is like getting blood out of a stone so I doubt whether we are ever likely to get any official confirmation on this.

[boley]

I'm trying the fridge trick first, ...

I can confirm that the fridge trick did not work for me, but it was worth a shot. I'm glad that I have access to two units with at least one working. I would have wasted a lot more time thinking there was something wrong with my configuration. (I did for several hours, trying every USB device in my office.)

Here is a tip if you only have one Pi, which is 99.999% of the folks.

If you suspect you are having this problem, and aren't sure if it is due to your usb devices or SD-Card, only plug in your ethernet cable. Leave the USB devices unattached. Power it up. If the ethernet LED never flashes, and you get to the point where you are asked to login, you most likely have this problem. You can also use a multimeter and test from the shield of the USB to F1 and F2 to verify your power supply is delivering 5.00 volts or there about.

[jbsteam]

I'm having the same trouble with my Pi that was delivered from RS yesterday.
Going to try the "fridge" for an hour before i get onto RS.
All solder points look ok. No network or USB.
4.8v present between TP1 and TP2.
When i plug a card reader into any USB the green light on the reader comes on.

Def sounds like its going to be a big issue with Pi's to me.

[churchill]

Almost the same for me !
When (yesterday) I received my Rpi, the board started working like a charm (...) and now more ethernet nor usb

Some news :
I made 2 changes :
- the board went to the fridge for a couple of hours
- I changed my 2 differents double-usb chargers rated at 1A for a single phone charger at 0.7A
And the Rpi worked ! I don't know if is the fridge or the charger, but my Rpi works (for the moment).

[grumpyoldgit]

Can I suggest that your double usb charger rated at 1A may not be providing 1A from each usb socket but a potential 1A overall. It could be that each socket is rated at 500mA. I am still very dubious about the fridge being a solution for all Pi issues.
Most people tend to have issues associated with power. You have swapped power units and your Pi is now working so I am inclined to suggest the issue was with the original power block.

[churchill]

Can I suggest that your double usb charger rated at 1A may not be providing 1A from each usb socket but a potential 1A overall. It could be that each socket is rated at 500mA. I am still very dubious about the fridge being a solution for all Pi issues.
Most people tend to have issues associated with power. You have swapped power units and your Pi is now working so I am inclined to suggest the issue was with the original power block.

Yes, I think you're right, my 2 doubles 1A usb chargers are more probably two 2x0.5 usb chargers. The fridge trick seems too misterious for me (but I don't have big electronic knoweldge)

[jbsteam]

GOT IT WORKING!!!!

Tried the fridge overnight, DID NOT WORK.
Tonight i remembered that i had a can of FREEZER spray in the shed (i'm a telecom engineer and used to use it years ago).
I gave it a 1 second blast on the underside of the pi, freezing BOTH crystals and most of the pie solid (white over with ice).
It thawed after about 2mins and worked perfectly straight away, usb and network port as they should be.
I've since tried a 500ma psu and even that works perfectly.

In my opinion (not that its worth much) it must be a problem relating to the 25mhz crystal and freezing it kick starts it into action.
Anyone with better knowledge back this theory up?
JB.

[boley]

Yes, I think you're right, my 2 doubles 1A usb chargers are more probably two 2x0.5 usb chargers.

My thought is that unless you have a really sophisticated charger/adaptor, you have a 1 Amp charger that has two ports. I doubt they bother to current limit each port to 500mA. But when in doubt change it out.

My own plans to drag out the old O'scope and see what I could see, went south when I realized it is only a 15MHz scope. Yes it is that old. Believe it or not, it was state of the art, back when I was trouble shooting 1 MHz boards.