2

I have the following working conditions

  • Raspberry Pi 2
  • Raspbian Jessie with PREEMPT-RT patch
  • connecting remotely using ssh
  • Wifi adapter is connected (edimax)
  • setting max_usb_current=1
  • running htop
  • running a loop that executes 'vcgencmd measure_temp' every 10 minutes

I got then the following problem

  • Raspberry has run 25 days continuously (temperature increases 10 degrees every 7 days).
  • Temperature increased to 72.4°C then raspberry switched off.

does any one have idea, why temperature is increasing to that extend?

EDIT: Here is htop output before shutting down enter image description here

5
  • Is your raspberry overclocked? Can you tell some more about your power supply?
    – Huczu
    Jun 6, 2016 at 8:06
  • I haven't used overclocking. my power supply output is 5V-3A
    – moibrahim
    Jun 6, 2016 at 8:09
  • 2
    How's the ambient temperature where you are? How does it compare to the previous 25 days?
    – goobering
    Jun 6, 2016 at 8:37
  • Can you track the load average of the system?
    – user236012
    Jun 6, 2016 at 9:33
  • ambient temperature is between 25 and 30 °C. for load average it is 0.04. I added screenshot of htop
    – moibrahim
    Jun 6, 2016 at 10:42

2 Answers 2

1

does any one have idea, why temperature is increasing to that extend?

Either something else in the environment is heating up very slowly this way or something is wrong; the third load average average number (presuming htop doesn't do something bizarre, which I don't think it does) is over 15 minutes, and it is higher than the first one, which is the past minute, meaning this pi has not been very busy for that period1 -- which should be plenty of time for heat to dissipate from the SoC unless, e.g., it is encased it with some kind of spray on insulation. In fact, I am sure I have read about people sealing the whole board up with thick plastic film for use underwater and not report overheating problems (the water may help with that, but I really think the manufacturer is telling the plain truth when they say it does not require any cooling or ventilation).

Even if it were working 100%, I think a Pi 2 can sustain that indefinitely (presuming again it is not in a very constrained, airtight enclosure), without heatsinks or much ventilation, in 20-25 C air without ever going over 60 C; when I've run one that way it plateaus in the low - mid fifties after a few minutes then stays there. Of course, if you overclock it sufficiently, at some point that behaviour will start to change.

max_usb_current=1

This doesn't seem necessary just for a wifi dongle, so you could try setting that back to 0. I do have an TP-Link RTL8188CUS based adapter which I used to use with pis but stopped because it had some unusual disconnection glitches and also got ridiculously hot (although this did not affect the SoC). So if you can physically inspect the pi, you might check that, since it is probably indicative of an excessive amount of current flowing. Other adapters I've used did not have this problem.

For comparision, the B+ I run 24-7 in a very poorly ventilated (plastic case inside a wooden box) 20-25 C environment currently has a load average of 0.0 0.1 0.3, an uptime of ~16 days, and a core temp of 42.7 C. It also has max_usb_current set in order to power a USB drive, with a total 2.1 A supply. Sometimes I've used a 2.5 A supply, I've been doing this for years, and I occasionally have uptimes of 5-6 weeks without noticing anything like this (I do like to prod and check things). I've done the exact same thing with a B, although I haven't used a 2 in that role.

In short, this is not what should happen.


1. You also have a very small cache (the red bars at the end of the memory line), another clue probably nothing serious has happened for a long time. By comparison the B+ in the last paragraph is currently using less RAM (73 MB) but the rest of the 1/2 GB is all red bars (i.e., occupied by cached data) which is what usually happens with long running systems unless they are dedicated to a very small number of specific, light duty tasks. Not an absolute, but a significant sign.

3
  • actually I am not using any packing, moreover I am using heat sink. but this is really weird. When I restart the board again, it starts with temperature 69 and increases dramatically (1 degree every 2 seconds). after few minutes vcgencmd measure_temp reported that core temperature reached 100.3°c !!!
    – moibrahim
    Jun 6, 2016 at 12:38
  • 1
    I would say the SoC has been damaged/was defective. It sounds like it is garbage now...or else it is just reporting the temperature wrong. If you can physically access the board that is easy to tell. 40-50 C is noticeably warm, 80 C uncomfortable to touch, and 100 C will burn skin within seconds.
    – goldilocks
    Jun 6, 2016 at 12:40
  • yes it burns my skin in seconds
    – moibrahim
    Jun 6, 2016 at 12:51
1

I would dare say that either your RPi board has a dead short on the board (somewhere), or your power supply or cable is having an issue. Try a different power supply and mirco usb cable for test reasons. If you should have one.

A dead short will cause a high amperage draw. Causing the overheating.

Good Luck

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.