Overheating problem with new 512MB model


33 posts   Page 1 of 2   1, 2
by falk0069 » Wed Nov 07, 2012 12:23 am
I've been struggling to get my Raspberry Pi to stay booted for more than a minute. I've tried various OSes and various power supplies. I still have a 'special' power supply in the mail adafruit which I don't think will help. All the FAQ seem to point to a bad power supply that can't maintain the 5V. The thing that was unique for me is after it shuts down I needed to wait at least a minute before trying again (cool down needed). At which I could try again but only get another minute of operation.

I finally made a breakthrough today, but it doesn't bold well. I placed the raspberry pi on an ice pad (with moisture protection). The Raspberry Pi has now been running for 15 minutes.

This is one of the newer 512MB models which I was happy about since I've been waiting 6 months for this, but it looks like I'm going to have to figure out how to return this now. Does anyone have any potential solutions that might avoid having to send back or to get more information on what is overheating?

Thanks
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by redhawk » Wed Nov 07, 2012 12:57 am
Where on the board is your PI overheating??

If you PI isn't lasting very long then it may be drawing in excessive current and tripping out the F3 polyfuse.

If you have a digital volt meter check TP1/TP2 and when the PI stops working check across F3 polyfuse, what kind of readings are you getting??

Richard S.
User avatar
Posts: 3519
Joined: Sun Mar 04, 2012 2:13 pm
Location: ::1
by malakai » Wed Nov 07, 2012 1:03 am
Define the Power Supplies you are using. 5v ?amp It would seem to me less of overheating and more of underpower. What got you to believe it's overheating? It is very likely if your power supply is not powerful enough it won't overheat the unit it will get a draw from the keyboard or mouse and then can't provide the power needed for the Pi.
http://www.raspians.com - always looking for content feel free to ask to have it posted. Or sign up and message me to become a contributor to the site. Raspians is not affiliated with the Raspberry Pi Foundation. (RPi's + You = Raspians)
User avatar
Posts: 1383
Joined: Sat Sep 15, 2012 10:35 am
by drgeoff » Wed Nov 07, 2012 1:16 am
Looks like you may have the same problem as several other people posting in viewtopic.php?f=28&t=20657. :(

My advice would be to ask the supplier (RS?) for a replacement.
Posts: 3142
Joined: Wed Jan 25, 2012 6:39 pm
by falk0069 » Wed Nov 07, 2012 2:31 am
@drgeoff - I did read that posting earlier and I don't doubt it might be a micro-crack on the board that expands just enough with just a little heat. My supplier is RS-Components, unfortunately they are on the other side of the ocean from me, so shipping is expensive and time consuming.

@malakai - I tried a USB charger for a DroidX, Evo4G and a Droid Incredible. I also tried with a Car changer for a DroidX. I also tried with just the USB cord from multiple computers. All the charger are rate for 5V. When I measured across TP1 and TP2, with the different chargers, it would read between 4.6V - 5V. Most of my testing has been with the DroidX changer. With RCA video, Ethernet, and Keyboard plugged in, the voltage hovers at 4.72V - 4.78V during the boot up. When it dies, it goes up to 5.0V. So, the voltage is on the lower end, but it doesn't explain why it works longer on a block off ice or why it needs to 'cool' down before getting it to boot again. I do have a Adafruit charger on its way which might help.

@redhawk - I tried figuring out 'where' it is overheating, but nothing really feels all that warm. While the Pi is booting and at the login prompt, the F3 polyfuse has between .19V - .21V across it. When the Pi stops working, it drops to .07V and stays steady. Does that mean anything to you?
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by falk0069 » Wed Nov 07, 2012 2:47 am
When I measure ohms across the F3 polyfuse I see this:
167 ohms during boot up and while sitting at the login prompt. However there were a few spikes to 180 ohms. When it stops working the ohms drop to 60 for a few secs and then 53. When I unplug the Pi, it drops to 1.5 ohms.
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by malakai » Wed Nov 07, 2012 4:43 am
The point with power supplies isn't the voltage as much as the amps. A 5v 500mA supply isn't going to work as far as stability goes. Are you using 1000mA or 1amp rated chargers? Unfortunately if you do have a crack I don't think it's something you could ever find. It is strange that the cold keeps it alive longer but I would think if your using the correct amp charger it would most likely be a defective board. Ultimately the chargers that suppliers offer that claim to be compatible with the Pi. These are 5v 1amp stepping power supplies. At this point if you can't plug in just video and power nothing else and it dies within a minute you are having a serious issue. The keyboard is plug and play you can plug it in whenever you want and it should be detected.
http://www.raspians.com - always looking for content feel free to ask to have it posted. Or sign up and message me to become a contributor to the site. Raspians is not affiliated with the Raspberry Pi Foundation. (RPi's + You = Raspians)
User avatar
Posts: 1383
Joined: Sat Sep 15, 2012 10:35 am
by tass2001 » Wed Nov 07, 2012 8:29 am
I have the same problem - I can keep my Pi powered on as long as there's a fan blowing on it. If the room temp goes over 80F it will shut down regardless of the fan. I also notice that when the Pi dies out the voltage from TP1 to TP2 goes up to 5V and my supplier is also RS Components :/.

I think the board is defective. Perhaps a bad batch?
Posts: 3
Joined: Fri Jun 22, 2012 11:24 am
by drgeoff » Wed Nov 07, 2012 10:16 am
@falk0069. I surmise that you are trying to measure the resistance of F3 while the RPi is powered up. You cannot use a multimeter to do that - any readings you take are completely meaningless.

My guess is that there is a batch problem with the soldering and (one or more) of the connections of the SoC BGA package is just a mechanical contact (rather than a solder weld) and thermal expansion is causing it to open. (I have seen examples of this problem myself on other equipment with a BGA device and I believe it has been encountered widely on some games consoles.) When the RPi SoC stops working properly, its current consumption is very likely to fall and hence the rise in measured voltages.

The more people that return these boards to RS, the sooner this will be investigated and the the problem, if it exists, acknowledged.
Posts: 3142
Joined: Wed Jan 25, 2012 6:39 pm
by Burngate » Wed Nov 07, 2012 11:25 am
falk0069 wrote:When I measure ohms across the F3 polyfuse I see this:
167 ohms during boot up and while sitting at the login prompt. However there were a few spikes to 180 ohms. When it stops working the ohms drop to 60 for a few secs and then 53. When I unplug the Pi, it drops to 1.5 ohms.

The measurements of 167 and 180 ohms are meaningless. However the 1.5 ohms after it's crashed doesn't bode well - assuming your meter is ok and you're making good contact, it says the polyfuse is half-blown. 0.1 ohms would be what I would expect for a good one.
Wyszkowski's Second Law: Anything can be made to work if you fiddle with it long enough.
Brain surgery is easier than psychoanalysis
User avatar
Posts: 3051
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK
by pluggy » Wed Nov 07, 2012 11:36 am
If you're not too worried by such things as warranties and sending it back to the supplier is a pain. You could try baking it at 200C / 385F for 10 minutes in an oven and then letting it cool down in the oven for an hour before moving it.

No - seriously. There are lots of success stories about cooking PCBs in a domestic oven on Youtube. Its a crude available version of what they used to make it in the first place - a reflow oven. Theres evidence that a small number of Pis have BGA soldering issues.
Don't judge Linux by the Pi.......
User avatar
Posts: 2864
Joined: Thu May 31, 2012 3:52 pm
Location: Barnoldswick, Lancashire,UK
by falk0069 » Thu Nov 08, 2012 2:20 am
Thanks for all the suggestions. It is sounding like the polyfuse is probably bad and heating it in an oven isn't going to help that. I'll give my adafruit power supply a shot once I receive it and if it doesn't help then I'll just return it.
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by obcd » Thu Nov 08, 2012 8:14 am
I have simply soldered another 1.1Amp over the already existing one. The Pi is still short circuit protected. TP1 - TP2 voltage went up from 4.85V to 5.05V. Both usb ports should be able to provide 500mA to connected devices, still leaving a safety barrier for the Pi.
Posts: 890
Joined: Sun Jul 29, 2012 9:06 pm
by drgeoff » Thu Nov 08, 2012 10:19 am
falk0069 wrote:Thanks for all the suggestions. It is sounding like the polyfuse is probably bad and heating it in an oven isn't going to help that. I'll give my adafruit power supply a shot once I receive it and if it doesn't help then I'll just return it.

Your original post mentions that your RPi will work again after a matter of minutes. Polyfuses usually don't recover that quickly.
If you power the board via the GPIO connector (thereby bypassing the polyfuse) and it still fails then you know the polyfuse is not causing it.
Posts: 3142
Joined: Wed Jan 25, 2012 6:39 pm
by phoenix » Thu Nov 08, 2012 11:23 am
Here is quite a novel solution to overheating boards. It also makes them more recyclable !
http://ht.ly/f5nZo
Image
Posts: 13
Joined: Wed Aug 08, 2012 10:56 am
by falk0069 » Sat Nov 10, 2012 2:29 am
I'm not exactly sure how to power it via the GPIO, but I wouldn't doubt it isn't the polyfuse that is the problem. I did a little experimenting and it turned out the IC2 chip in the middle of the board is what 'seems' to be overheating. I was able to place a focused amount of cold on various parts of the board and when cooling that chip, the Pi would stay up forever.

Also, I got my adafruit 5.25V power supply today and it did not make a difference.

On a positive note, RS said they would reimburse the return shipping if I provide a receipt.
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by dknute » Sat Nov 10, 2012 8:35 am
IC2 is the RPi I'm afraid. It's the multi-layer chip that has CPU/GPU on the bottom and SDRAM on top. If it overheats while sitting there and doing nothing after boot, your RPi is inedeed faulty.

But keep in mind that IC2 is allowed to get warm, even a bit hot during normal operation. It has an auto-shutdown feature but it kicks in at 85C - that would burn your fingers after you touch it. Does it get this hot? As a rule of thumb (pun intented) as long as you can keep your finger on IC2 and not get burnt it's not "too hot" yet. If your board stays powered long enough for the system to boot, you can execute a command to measure the temperature of BCM2835:

/opt/vc/bin/vcgencmd measure_temp

Also, what config.txt settings do you have? Did you enable overvoltage or overclocking?
Posts: 56
Joined: Thu Nov 08, 2012 9:05 pm
by falk0069 » Sat Nov 10, 2012 5:42 pm
The IC2 chip is maybe getting up 40C. I can barely notice it being hot. I wonder if the auto-shutdown feature is just kicking in too soon.

As far as the config.txt--I did not modify it. All I've tried are these two OS images:
Raspbian “wheezy”
Arch Linux ARM

I've been using the Arch Linux one as of late since it boots much faster and I can issue a few commands before it shutdown.

Inside the config.txt I do see an option I can uncomment for the frequency. As a test I tried lowering the frequency to 600--it made no difference.

I also just tried the vcgencmd to get the temp but it fails due to missing libraries in the Arch Linux version.

Ok, I went back to the Raspbian image to run that vcgencmd command. I was able to issue the command a few times before the shutdown. The temp only got to 31.5C.

Oh well, it was worth validating this. I think I'm off to the post office now to return it.
Posts: 11
Joined: Thu Jun 14, 2012 3:05 am
by Sueno » Thu Nov 15, 2012 12:13 pm
My new RS supplied 512MB RPi is also cutting out consistently at between 5 and 15 mins after switch on.

I used a temperature probe to measure (several times) the Samsung RAM chip and this shows a surface temperature of 34.5 - 34.7C at cut-out. The room air temperature was 22.1C.
I then applied a hair dryer to blow room air at the chip. This was very effective as the Pi did not shut down. After ten minutes the chip surface temperature was at 25.5C where it stayed until I switched the fan off after an hour. Switching off the fan then allowed the chip temperature to rise in two minutes to 34.7C when the PI cut-out.

Although the temperature is not high, something does not like the heat. By logic that still looks like overheating to me. I'll try a passive heat sink next. If that does not work, it looks like a box with a fan is required. Pity.....
Posts: 5
Joined: Fri Jan 27, 2012 5:18 pm
by drgeoff » Thu Nov 15, 2012 1:03 pm
Sueno wrote:My new RS supplied 512MB RPi is also cutting out consistently at between 5 and 15 mins after switch on.

I used a temperature probe to measure (several times) the Samsung RAM chip and this shows a surface temperature of 34.5 - 34.7C at cut-out. The room air temperature was 22.1C.
I then applied a hair dryer to blow room air at the chip. This was very effective as the Pi did not shut down. After ten minutes the chip surface temperature was at 25.5C where it stayed until I switched the fan off after an hour. Switching off the fan then allowed the chip temperature to rise in two minutes to 34.7C when the PI cut-out.

Although the temperature is not high, something does not like the heat. By logic that still looks like overheating to me. I'll try a passive heat sink next. If that does not work, it looks like a box with a fan is required. Pity.....

Your RPi is defective and should be returned for replacement.
Posts: 3142
Joined: Wed Jan 25, 2012 6:39 pm
by dknute » Thu Nov 15, 2012 1:27 pm
This might not be strictly temperature-related. Rather, a micro crack in one of the solder joints could be doing these. There is no common failure mode of so-called "cold solder joints". Even small temperature change can upset them and the electrical connection will be broken.

Usually this is a good base for requesting a replacement, except you first need to be sure it's not something really obvious like power supply not being stable. Try a different SD card as well, RPi seems to be very picky about those. Probably some firmware bug that's yet to be found. It might be hanging in an infinite loop somewhere during a failed write or something.

Out of curiosity, is your RPi "Made in UK" one or not? Some people have reported rather poor soldering effort - C6 is very easy to break off for example, without any damage to PCB or the capacitor itself. I keep wondering if one of the fabs has switched to less aggressive soldering profile to prevent accidental damage to the SoC. Which in turn caues some areas not to heat up properly past melting point and form fragile connections.
Posts: 56
Joined: Thu Nov 08, 2012 9:05 pm
by Sueno » Thu Nov 15, 2012 1:55 pm
Thanks for that, drgeoff and dknute (even more Danes?),

The issue is certainly repeatable and definitely temperature related. I also have a V1 Pi which runs very happily with the same setup (using a Targus, powered USB hub) and the same perpherals and SD cards (Raspian-latest and RISC OS Pi). I think it is unlikely to be software or firmware related as it always occurs about the same time after switch on and at about the same temperature irrrespective of what I am doing on the machine. I was aware of the Ball Grid Array soldering problem but I'm not sure if just keeping it cool is a long-term solution. I guess that drgeoff's succinct answer is the easiest if most disappointing answer. Yes. It has to go back to RS. Yes, it does say "Made in the UK" on the board. Pity...........

Sueno
(A Scot actually)
Posts: 5
Joined: Fri Jan 27, 2012 5:18 pm
by bercar » Mon Dec 03, 2012 12:34 am
I am having the same issue with both of my 512 PI's. All i have loaded is Raspbmc. It will boot to the menu, run normally for a few min, then if I don't do anything to dissipate the heat on IC2, it just shuts off. I do have time to load the system menu, but the temp display updates really slow.
Highest CPU temp shown through SSH is 65.9C
Posts: 3
Joined: Mon Dec 03, 2012 12:13 am
by mahjongg » Mon Dec 03, 2012 2:03 am
66 degrees Celsius itself is no reason for the chip to "give up the ghost", and die on you! It would be different if it said 88 degrees. I am more inclined to believe that bad soldering is the cause than any "overheated chip", but naturally the result is exactly the same for you.

Its sounds a lot like the badly soldered first series of the Xbox 360, with their dreadful "red-light of death" problems., these were also caused by badly controlled soldering processes, but in that case were exacerbated by the badly designed board mounting system that warped the boards when they expanded through the heat
User avatar
Forum Moderator
Forum Moderator
Posts: 6016
Joined: Sun Mar 11, 2012 12:19 am
by drgeoff » Mon Dec 03, 2012 12:02 pm
bercar wrote:I am having the same issue with both of my 512 PI's. All i have loaded is Raspbmc. It will boot to the menu, run normally for a few min, then if I don't do anything to dissipate the heat on IC2, it just shuts off. I do have time to load the system menu, but the temp display updates really slow.
Highest CPU temp shown through SSH is 65.9C

I don't run Raspbmc myself but I believe that the default installation of it does enable a certain amount of overclocking. If that is true you should disable it before concluding that your RPis are faulty and RMAing them.

http://www.raspbmc.com/wiki/user/freque ... questions/ See question on overclocking.

(Perhaps the 512 Mbyte RAM chips are not as good for overclocking as the 256 ones were?)
Posts: 3142
Joined: Wed Jan 25, 2012 6:39 pm