24x7x365
Posts: 4
Joined: Mon Jul 02, 2012 10:18 pm
Location: Staffordshire, UK

Re: Pi = busted, yay.

Tue Jul 03, 2012 9:44 pm

jamesh wrote:At about 0.05% that's not a widespread problem.
Thanks for your wise words jamesh - but I was actually looking for some support/advice on what I could do with my bricked Raspberry Pi. Glad to hear you're not having any issues with yours and that my case is merely an insignificant statistic.

In the meantime, if anyone has any constructive advice, please feel free to contact me. I've had my Pi for 4 days - and it's game over. I've sent an email to RS support - no response. Looks like I have wasted my money! (plus the money I spent on the case, SD card, PSU, HDMI cable, etc etc...
So disappointing. - Surely this is a manufacturing fault and I should be eligible for a replacement??

24x7x365
Posts: 4
Joined: Mon Jul 02, 2012 10:18 pm
Location: Staffordshire, UK

Re: Pi = busted, yay.

Tue Jul 03, 2012 9:50 pm

mahjongg wrote:My advice, try another power-supply, and/or another sd-card.
Thanks mahjongg. I have purchased another SD card and created a new card from the image file. Same issue - the Pi boots fine, but there's no USB or ethernet support, so no way to interface with the device at all (and no LAN LED's either).

Have tried 3 different USB PSU's - including the one supplied with the Pi by RS - all produce the same result. If I put the Pi in the freezer for an hour or so, it works fine again for up to around 30 minutes.

What's a polyfuse? Am I missing something?

Dilligaf
Posts: 283
Joined: Wed May 23, 2012 6:48 pm

Re: Pi = busted, yay.

Tue Jul 03, 2012 11:58 pm

24x7x365 wrote:
mahjongg wrote:My advice, try another power-supply, and/or another sd-card.
Thanks mahjongg. I have purchased another SD card and created a new card from the image file. Same issue - the Pi boots fine, but there's no USB or ethernet support, so no way to interface with the device at all (and no LAN LED's either).

Have tried 3 different USB PSU's - including the one supplied with the Pi by RS - all produce the same result. If I put the Pi in the freezer for an hour or so, it works fine again for up to around 30 minutes.

What's a polyfuse? Am I missing something?
If you can't be bothered to RMA it then preheat the oven to 380 and bake it for 10 minutes to reflow the solder. Search Google for "circuit board baking". I have fixed countless electronics this way.

Mike

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12982
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Pi = busted, yay.

Wed Jul 04, 2012 12:19 am

A "polyfuse" is a fuse made from a polymere. Its a fuse that after it has "blown" will self heal, that is after several days it will (almost) behave as if nothing has happened.

The USB specifications say that all USB ports must be protected, but the protection may not be permanent, that is may not require human intervention to recover from. There are two ways to do that, the first is to use a polyfuse, and the second is a chip, called a current monitor and protection chip, the latter are quite expensive so the RPI uses polyfuses.

There are three polyfuses in a RPI, one to protect the whole board, this one is rated 1.1 Amperes, and is the device marked "F3" on the back of the board. There are also two smaller polyfuses, one to protect each USB port. These have a very low rating of just 140mA.
A a polyfuse "blows", just like any regular fuse, because there is running so much current through it that it heats up, in a regular fuse that means so much current that a thin wire simply melt, in a polyfuse it means that a small piece of conductive (polymere) plastic get so hot that suddenly its resistance increases dramatically, so that it gets hotter still, and things escalate to a point that the resistance goes so high that most of the current stops flowing, after the current is turned off, a long time after the fuse has cooled down the fuse slowly regains it original state and in the end becomes conductive again.

With the USB polyfuses something special is going on, because these are designed to "blow" very quickly, with just a few 100mA, because they blow with very low currents does not mean they need less energy (heat) to blow, therefore they need more voltage over them, as P = U x I (Power equals voltage times current) (Ohms law). If they need more voltage across them, it means that with the same current running through them they need to have more resistance (U = I x R) Voltage equals current times resistance (yes also Ohms law). Thus it means that the lower the "blow current" (given the same mechanical size of the fuse) the fuse needs to have more resistance, so much resistance in fact that for normal currents the resistance becomes noticeable.

If a regular current of something like 100mA runs through it the voltage across the fuse can be something like half a volt, that means that the USB device gets half a volt less, so if the board gets 5.0 Volt, the USB device (when it uses 100mA) gets just 4.5 Volt.

Now normal 5V logic doesn't work then, as it can tolerate no more than 5% undervoltage, so at just 4.75 Volt it begins to stop working. Luckily the USB specification orders the manufactures of simple USB devices, such as keyboards and mice, devices that typically work behind a passive HUB (that gets its power from the computer, not from a separate PSU) that they should design their device so that it can work with the voltage behind a hub. Now it so happens that the design of such a passive HUB describes that the 5V should pass through a diode, so that deices connected to such a hub cannot feed back energy back into the PC, which can happen if they are USB devices with their own powersupply.

Hope you can follow this so far.

So elementary USB devices, so called "single current unit devices", where "1 current unit = 100mA". Are required to still work if their power is one diode drop (0.6 Volt typically) lower than 5V. That means keyboards and mice should work when they only get 4.4 Volt, not 4.75 Volt!

This is ONLY true for these devices, NOT for devices like sound cards, or WiFi adapters or external hard drives, that can use up to 5 current units (or 500mA).

Now coincidentally this means that they also work with the voltage available behind a USB polyfuse, which is typically 4.5 Volt, as we saw before, but other devices need 4.75 Volt, and wont work with 4.5 Volt. Powered HUB's work with 4.5 Volt, because they do not use it for power.
However many powered hubs have their input 5V line, connected to their own 5V from the PSU, so that they could also work from the power given by a PC, other (more strictly designed according to USB specifications) do not connect the 5V from their input cable, but might "look at it", and only turn themselves on if they see it, so that they are off when the PC is off, to preserve energy.

The choice of these high resistance polyfuse has been the cause of many problems, because many people do use the PI for something other than its intended purpose, that is only with a wired keyboard and a mouse, they try to connect WiFi adapters and such , which may fail.

That is the story about the USB polyfuses, if they have triggered, their resistance can increase so much (for days) that even 4.5 volt isn't reached, and the keyboard only gets say 3.5 Volt, and it will fail, start repeating keys etc.

The main power input fuse is much less problematic, but it too can misbehave in similar way.
Especially if for whatever reason it has gotten hot, such as prolonged soldering to it, or as the previous poster suggested by "baking it" (this is a bad idea, only to be done if you expect the soldering has gone wrong on the main chip). So do not heat any of the polyfuses unnecessary, but cooling them down also doesn't do any permanent good, its not how cool the polyfuses are that helps them recover, as its a chemical process, not a mechanical process. letting them have a rest in a cool place might help a "bad" polyfuse recover, but if they get hot again, the effect is undone again.

Some people who insist on using unsuitable USB devices on their RPI have thought of shorting the polyfuse, a process nicknamed a PiPass after a "bypass operation", simply bridging the fuse will lead to another problem, that is that the PI will reset when you connect a USB device, the correct solution is to leave a little bit of resistance in place, say 0.5 or 1 Ohm. Soldering a 1 Ohm resistor over the offending USB polyfuse will usually fix the voltage drop problem.

24x7x365
Posts: 4
Joined: Mon Jul 02, 2012 10:18 pm
Location: Staffordshire, UK

Re: Pi = busted, yay.

Wed Jul 04, 2012 5:57 pm

Hi,

Thanks for the reply re the poly fuses.
Would this also explain the issue with the Ethernet interface.

For clarity, I have only every connected a USB keyboard to my Pi. Not even connected a mouse yet!

Here are the resistance measurements across the poly fuses:
F3 = 0.06 ohms
F1 = 4.5 ohms
F2 = 4.5 ohms

Is 4.5 ohms across 1 and 2 correct - as they are both the same I would assume this is normal??

Voltage between TP2 and TP1 = 5.0v after bootup. (nothing plugged in to USB ports)

Thanks for any further info. Still no response from RS in terms of an RMA etc...

Richard

texy
Forum Moderator
Forum Moderator
Posts: 5161
Joined: Sat Mar 03, 2012 10:59 am
Location: Berkshire, England

Re: Pi = busted, yay.

Wed Jul 04, 2012 8:08 pm

Those fuse resistances match my (working) pi. I wouldn't wait for an email response if I expect a quick answer, I would phone them.

Regards,

Texy
Various male/female 40- and 26-way GPIO header for sale here ( IDEAL FOR YOUR PiZero ):
https://www.raspberrypi.org/forums/viewtopic.php?f=93&t=147682#p971555

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12982
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Pi = busted, yay.

Thu Jul 05, 2012 6:34 pm

24x7x365 wrote:Hi,
Would this also explain the issue with the Ethernet interface.
Richard
I don't see any polyfuse related problems with the ethernet interface.
The ethernet chip is powered with 3V3, so even a quite horrendously bad F3 polyfuse wouldn't affect it much, and yours is fine.

The only thing I suspect is that the crystal is either hand-soldered, and doesn't make good contact, or that the sliver of quartz inside the crystal has dislodged, so that much more energy has to be put into the quartz to get it vibrating.

Cooling down electronics makes it somewhat faster, and a faster switching crystal oscillator driver might help the crystal to oscillate.

So you have a bad crystal, possible because it has had a bad shock.

drgeoff
Posts: 10620
Joined: Wed Jan 25, 2012 6:39 pm

Re: Pi = busted, yay.

Thu Jul 05, 2012 11:49 pm

mahjongg wrote: Cooling down electronics makes it somewhat faster, and a faster switching crystal oscillator driver might help the crystal to oscillate.
Not so. Reducing the temperature of electronics makes the semiconductors run more slowly. This is quite different from cooling (= heat extraction) to allow them to be driven faster without damage from self generated heat.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12982
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Pi = busted, yay.

Sat Jul 07, 2012 7:15 pm

True, the derating of most semiconductors means that absolute switching speeds go down with increases of temperature (true up to about 150 degrees Celsius) but the effect is very slight at just 2% per ten degrees Kelvin. Other effects can dominate, for example (diode) leakages double for every 10 degrees Kelvin.

In a practical sense, I've often observed a bothersome crystal starting after a heavy dose of freeze spray. Could be another effect than increased loop gain, for example it might have been a mechanical effect, hard to say.

Towerade
Posts: 1
Joined: Sun Jul 08, 2012 6:06 am

Re: Pi = busted, yay.

Sun Jul 08, 2012 6:07 am

Same problem here :(

Return to “Troubleshooting”