Works great then becomes unresponsive – overheating?

[ntutak]

Hi everyone – I have a OpenSprinkler / Raspberry Pi and it works fantastic. However, I have it set up outdoors in the metal sprinkler box that housed my old automated sprinkler controller. When I leave it powered on for day and try to access it the Pi is unresponsive. I live in Souther California where it gets pretty warm and I think it may be overheating? Does anyone have any experience with this? Would it be worth trying to add a small fan to the sprinkler box or am I just going to have to move the whole thing into the garage or something?

For now I am keeping the OS +RPi inside and taking it out and plugging it in every day when I need to use it, then shutting it down and bringing it back inside when I am done. Not very efficient! However, I know that it works.

Any help is appreciate, thanks!

[Samer]

If it’s the Raspberry Pi that stops responding that might be odd because the FAQ (http://www.raspberrypi.org/faqs) says the temperature for the AP is qualified from -40°C to 85°C and the LAN9512 is specified by the manufacturers being qualified from 0°C to 70°C. Both, directly from the FAQ.

Curiously, if you run it indoors for 24 hours do you experience the same lock up?

Edit: You could try something like this, http://hackaday.com/2012/06/26/adding-heat-sinks-to-a-raspberry-pi/

[kenbob]

I have my Opensprinkler and Raspberry Pi in the garage, it has been as warm as 90F with no issues so far. Since the Pi is a computer, it really should be somewhere where it has some ventilation and out of direct heat. I would also recommend setting static address on both devices if you don’t already.

[Ray]

@ntutak: can you find out which version of OSPi you have? Version 1.0 and 1.1 use different switching regulators. It is possible that when the ambient temperature is very high the switching regulator’s internal circuitry might shut it down. To test whether the issue is caused by switching regulator, you can simply unplug the 3-pin wire, and power RPi separately using a USB cable. See if the problem continues.

[ntutak]

Hi everyone, thanks for the help. My original post was inaccurate. The RPi itself actually seems fine after leaving it on and outside overnight, but the Manual Sprinkler app no longer loads in my browser. It looks like the program itself is still running because when I ssh into my RPi and try to launch the manual sprinkler app I get the following:

./ospi_manual.py:96: RuntimeWarning: This channel is already in use, continuing anyway. Use GPIO.setwarnings(False) to disable warnings.
GPIO.setup(pin_sr_clk, GPIO.OUT)
./ospi_manual.py:97: RuntimeWarning: This channel is already in use, continuing anyway. Use GPIO.setwarnings(False) to disable warnings.
GPIO.setup(pin_sr_noe, GPIO.OUT)
./ospi_manual.py:99: RuntimeWarning: This channel is already in use, continuing anyway. Use GPIO.setwarnings(False) to disable warnings.
GPIO.setup(pin_sr_dat, GPIO.OUT)
./ospi_manual.py:100: RuntimeWarning: This channel is already in use, continuing anyway. Use GPIO.setwarnings(False) to disable warnings.
GPIO.setup(pin_sr_lat, GPIO.OUT)
OpenSprinkler Pi is running…

On I relaunch the manual sprinkler program (even though it is already running) then the web interface is accessible again.

Any thoughts?

[Samer]

I think the program your using was intended as a demo and not a 24/7 production program, I could be wrong though. I would implore you to try the interval program available on Ray’s Github instead and see if the same issues exist. Otherwise, you would need some sort of error logs to try and debug the manual mode program your using.

You could run the manual mode program in the foreground instead of in the background and report any errors it might output.

[ntutak]

Thanks salbahra. I’ll try the interval programs that is in my demo directory, is that the one you are talking about? It is found here on github, in the “demos” directory:

https://github.com/rayshobby/opensprinkler/tree/master/OpenSprinkler%20Pi/software/demos/interval_program

[Samer]

That’s it! It’s the program most user’s use with the Raspberry Pi and is also compatible with my mobile app (http://rayshobby.net/phpBB3/viewtopic.php?f=2&t=154, shameless plug :P).

[ntutak]

Wow, that’s much better! Now I’ll have to check out your mobile app, thanks!

[sco3tt]

It may yet be overheating. I live in a similarly warm environment (Houston, TX) and also have mine mounted outside in a sealed utility box. I added a laptop exhaust fan to the box and a simple controller for active cooling, but for the first two days after installation I ran it with the fan shut off just to see whether it was really necessary. On the second day my Raspberry Pi stopped responding to ssh, RDP and ping, requiring a reboot. I wasn’t logging temps at the time but I assume it was due to the 85°C CPU shutdown limit. After that I configured it so the fan starts any time the CPU temp exceeds 50°C and it has been trouble-free ever since (even this weekend when it was 104°F here).

If you are operating via WiFi and don’t need wired Ethernet, I would suggest that you use a Model A rather than a Model B. The A runs 15-20C cooler than the B under similar load… the LAN controller chip on the B generates a lot of heat, even when it’s not in use.

[Oleengr]

The following is just some info about an outdoor OS installation really sees some hot conditions.

My OpenSprinkler sits in a metal box on the NW side of the house in Phoenix, AZ. It is shielded from direct sun.
I measured the temperature in the box and under the cover of OS some weeks ago when temp was in the low 100s. If I remember correctly, the temp rise under the cover was around 9 degrees F above the temp inside the box, and the inside the box was a couple degrees above ambient max.

Since the max temps were certain to be a lot higher, I removed clear plastic cover from OS. Last week, my Davis weather station recorded a max temp of 118,116, 112, 111,111,111 deg F on consecutive days, and OS seemed to survive OK. The temperature within the box was probably very close to the max temp. I checked OS this AM and noticed the preview function showed that the time was about 5 hours fast. I don’t know if there were any disruptions; we were gone the hottest days. The time error really makes no difference in my installation, and a remote reboot corrected the time. At lower temps, the clock seemed to hold reasonably well.

There are three components dissipating power in the metal enclosure at all times, a small transformer, an X-10 Appliance module, and OS. The X10 module is connected to a powerline carrier adapter which is turned off except for a couple hours in the morning when it energizes the Ethernet adapter. I never measured the steady state power dissipated in the box, but the transformer and X10 module dissipate 1 watt each plus what ever OS dissipates.

I’m going to monitor the clock every few days to see how it holds. There was never a significant error earlier at lower temperatures.

[Author]

Posts