magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Temperature in sealed case

Thu Sep 29, 2016 4:06 am

I'm running my pi in a 100% sealed case (less a couple of 1/8" holes for condensation draining). Inside the enclosure is also a 12V to 5V regulator, and an ethernet hub. The pi is on 1.5" standoffs with a proto-HAT on top.

The enclosure is an 8x8x4 PVC electrical junction box. It will be installed on a 15 foot pole outside.

Right now, i have the enclosure sealed up in the house. When i sealed it, the SoC was at 51.5C / 124F. Within 10 minutes it is at 54.8 / 130.

At what temperature should I become worried about system issues?

I would be interested in adding a heatsink, however all the ones I see are adhesive style. With the proto-HAT on top and the board oriented vertically, i would be afraid of heatsink hitting or becoming unglued and falling off. Would prefer a design that attached to a mounting bolt.

Image
http://i1105.photobucket.com/albums/h35 ... 171521.jpg

mosespi
Posts: 508
Joined: Mon May 12, 2014 3:35 pm
Location: 34,-118
Contact: Website

Re: Temperature in sealed case

Thu Sep 29, 2016 6:31 am

I wouldn't be too worried about it. I've had a Pi 3 in a car all summer with the CPU temp usually over 80c+ all day, sometimes going all the way up to 95c, it worked fine. I did put a small heatsink on the CPU later, it did help with the temperatures, and likely wasn't throttling the CPU down as much. It didn't matter if the Pi throttling or not in my application.

I have a pack of those small copper CPU heatsinks that I like to put on the the Pi.. I've yet to have the adhesive fail and would probably have a difficult time removing them! I haven't seen many non-adhesive ones for the Pi. If you don't find anything else and you are still concerned, maybe a zip tie as backup holding down the heatsink?

Electrolytic capacitors are likely the first to fail at elevated temperatures, but I've had consumer electronics in an 'outdoor' box for a few years without too much trouble. You can drill some vents on the top and bottom (with small covers to keep rain and bugs out) to aid in bringing the box as close to outside ambient as possible, would help. Condensation and moisture are trouble makers outside.. in that case a bit of heat from the boards does help.

Regards,
-Moses
Power problems? MoPower UPS for the Pi
http://www.allspectrum.com/mopower/

User avatar
faramon
Posts: 123
Joined: Sat Jun 11, 2016 8:36 am
Location: Croatia

Re: Temperature in sealed case

Thu Sep 29, 2016 12:09 pm

My Pi CPU in closed raspberry box in sommer (when temperature in room is from 28 to 30 degrees) is about 56, 57 degrees and now is 51, 52, the temperature in my room is now about 24 degrees.
I have heatsink and i do have my python script that read the CPU temperature and start to cool with FAN (5V) when the temperature reaches 60 degrees...

Faramon

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 2:23 pm

mosespi wrote:I wouldn't be too worried about it. I've had a Pi 3 in a car all summer with the CPU temp usually over 80c+ all day, sometimes going all the way up to 95c, it worked fine. I did put a small heatsink on the CPU later, it did help with the temperatures, and likely wasn't throttling the CPU down as much. It didn't matter if the Pi throttling or not in my application.
It does not matter for me, either. All the Pi is doing is running a web server and reading a potentiometer. Occasionally moving a DC motor via H-bridge. There will be 2 IP cameras, the Pi, and a Ubiquiti AREDN node going into the ethernet hub.
I have a pack of those small copper CPU heatsinks that I like to put on the the Pi.. I've yet to have the adhesive fail and would probably have a difficult time removing them! I haven't seen many non-adhesive ones for the Pi. If you don't find anything else and you are still concerned, maybe a zip tie as backup holding down the heatsink?
I did not think the adhesive was that strong. I'm used to the adhesive being primary and a spring clip being secondary (old PC north and south bridges). Granted it has taken me hell and high water sometimes to get even those off with the spring removed.
Electrolytic capacitors are likely the first to fail at elevated temperatures, but I've had consumer electronics in an 'outdoor' box for a few years without too much trouble. You can drill some vents on the top and bottom (with small covers to keep rain and bugs out) to aid in bringing the box as close to outside ambient as possible, would help. Condensation and moisture are trouble makers outside.. in that case a bit of heat from the boards does help.
I've thought of the condensation and drilled some small 1/8" holes in the bottom. If I had some small window screen or other kind of mesh, I suppose I could go to 1/4 or 3/8" and melt the screen into the PVC enclosure.

The box is getting mounted on the vertical metal 12" pipe, below the rotator.

Image
http://i1105.photobucket.com/albums/h35 ... 122926.jpg

JimmyN
Posts: 1109
Joined: Wed Mar 18, 2015 7:05 pm
Location: Virginia, USA

Re: Temperature in sealed case

Thu Sep 29, 2016 3:05 pm

magnethead wrote: I did not think the adhesive was that strong. I'm used to the adhesive being primary and a spring clip being secondary (old PC north and south bridges). Granted it has taken me hell and high water sometimes to get even those off with the spring removed.
I use the Artic Silver thermal adhesive when I need to glue a heatsink to something. It's an epoxy based (two part) compound, and they also make Artic Alumina adhesive which contains ceramic particles rather than silver. Either one is permanent though, once applied you'll never remove the heatsink again without destroying whatever it's glued to, but it has a lot better thermal transfer than the adhesive pads.

User avatar
DavidS
Posts: 4334
Joined: Thu Dec 15, 2011 6:39 am
Location: USA
Contact: Website

Re: Temperature in sealed case

Thu Sep 29, 2016 3:16 pm

I wold put a peltier pile going through the case wall to pull heat out of the case. That would still allow it to be seeled, while providing cooling.

Use a larger 5v peltier pile, with a 330 ohm resister inline, and use a seperate output from your power supply to power it.
RPi = The best ARM based RISC OS computer around
More than 95% of posts made from RISC OS on RPi 1B/1B+ computers. Most of the rest from RISC OS on RPi 2B/3B/3B+ computers

MarkTF
Posts: 300
Joined: Tue Mar 03, 2015 4:59 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 5:20 pm

magnethead wrote:I'm running my pi in a 100% sealed case (less a couple of 1/8" holes for condensation draining). Inside the enclosure is also a 12V to 5V regulator, and an ethernet hub. The pi is on 1.5" standoffs with a proto-HAT on top.

The enclosure is an 8x8x4 PVC electrical junction box. It will be installed on a 15 foot pole outside.

Right now, i have the enclosure sealed up in the house. When i sealed it, the SoC was at 51.5C / 124F. Within 10 minutes it is at 54.8 / 130.

At what temperature should I become worried about system issues?

I would be interested in adding a heatsink, however all the ones I see are adhesive style. With the proto-HAT on top and the board oriented vertically, i would be afraid of heatsink hitting or becoming unglued and falling off. Would prefer a design that attached to a mounting bolt.
I'd be concerned if the SoC routinely got over 80 C in the deployed environment. The SoC itself is ok at that temperature, but there are probably other components on the board (electrolytic capacitors) that aren't going to be so happy over the long run.

You could just build the thing and see what happens, but you can estimate deployed thermal performance as follows:

1) Evaluate the temperature rise over ambient. To do this you'd seal the box and run it with a worst case activity profile over a few hours until the SOC temperature stabilizes. That minus the room air temperature is the SOC rise over ambient.

2) Evaluate sun loading on your enclosure by putting a thermometer inside and leave it out in full sun on a still day. The difference between that and ambient temperature gives sun loading.

3) Add (1) plus (2) plus the highest ambient temperature expected in your area to get a reasonable estimate of worst case SoC temperature deployed.

From a thermal design perspective, you have two things working against you. First is getting heat off the board to the air. Still air is a lousy thermal conductor. A fan would help but isn't very elegant and is potentially unreliable in an outdoor environment. The second problem is the air to pvc to air junction which is also a lousy thermal conduction path.

If you have a thermal problem (and you might not) one approach would be to use a sealed metal box instead of PVC and to using thermally conductive tape to bond the backside of the Pi printed circuit board to the case. This would give a decent conductive path of SOC to circuit board to thermal tape to metal enclosure to ambient air.

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 6:01 pm

I liked the peltier idea until i saw the power consumption.

The unit shut off last night, my soldered recycled broken micro usb connector re-broke.

I just sealed the box from a fresh boot. 42C/108F was the starting temp. In a matter of a few minutes it's already up to 51 / 124. I'm going to do some chores and come back to it in a bit and see where/if it stabilizes.

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 6:02 pm

JimmyN wrote: I use the Artic Silver thermal adhesive when I need to glue a heatsink to something. It's an epoxy based (two part) compound, and they also make Artic Alumina adhesive which contains ceramic particles rather than silver.
I use AS5 all the time on CPUs, i did not know they make an adhesive form?

Graymalk
Posts: 55
Joined: Wed Nov 11, 2015 8:33 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 7:35 pm

magnethead wrote:I just sealed the box from a fresh boot. 42C/108F was the starting temp. In a matter of a few minutes it's already up to 51 / 124. I'm going to do some chores and come back to it in a bit and see where/if it stabilizes.
That's what mine all do and they never leave the 50s. They're in much more airy cases though so maybe add 10 for yours to be safe? I leave mine running indefinitely, over clocked, full throttle, and have cron tabbed a temp checker to be sure. I've also monitored externally with a thermocouple. All the worry over heat with these things has so far turned out to be a waste of my time lol.

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Thu Sep 29, 2016 7:55 pm

Graymalk wrote:
That's what mine all do and they never leave the 50s. They're in much more airy cases though so maybe add 10 for yours to be safe? I leave mine running indefinitely, over clocked, full throttle, and have cron tabbed a temp checker to be sure. I've also monitored externally with a thermocouple. All the worry over heat with these things has so far turned out to be a waste of my time lol.
It seems to have stabilized at:

RasPi SoC Temp: 55.8 °C / 132.44 °F

I'm running a command line request via PHP every 5 minutes (http refresh) and putting it on the Ras-Pi's webpage. So that's how i'm monitoring it.

So figure it gained about 15C just running in the box. Outside in Texas gets to around 105 (40C) max compared to 76 (25C) in the house right now. I know in the car I've seen around 130-140F (60C).

So right around the 75-80C range i guess would be a valid prediction?

ejolson
Posts: 3568
Joined: Tue Mar 18, 2014 11:47 am

Re: Temperature in sealed case

Thu Sep 29, 2016 10:59 pm

magnethead wrote:Outside in Texas gets to around 105 (40C) max compared to 76 (25C) in the house right now. I know in the car I've seen around 130-140F (60C).
If you paint the box black and put it in direct sunlight it may get too hot. Maybe a white box shaded by an umbrella is better. Metal boxes can dissipate heat to the outside if you've got a thermal pipe between the CPU and the enclosure. This is the idea behind a cell phone with an aluminum chassis. It will be interesting how the temperature of the enclosed Pi evolves over time and what the final solution turns out to be. Please continue to update this thread.

mosespi
Posts: 508
Joined: Mon May 12, 2014 3:35 pm
Location: 34,-118
Contact: Website

Re: Temperature in sealed case

Thu Sep 29, 2016 11:24 pm

ejolson wrote:
magnethead wrote:It will be interesting how the temperature of the enclosed Pi evolves over time and what the final solution turns out to be. Please continue to update this thread.
I think the only temperature of much concern is of the beer OP will be holding as he admires his boom mounted Pi in the sky!

Cool looking home made antenna rotator! Looks like its made from erector set parts.

Regards,
-Moses
Power problems? MoPower UPS for the Pi
http://www.allspectrum.com/mopower/

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Fri Sep 30, 2016 2:40 am

mosespi wrote:
ejolson wrote:If you paint the box black and put it in direct sunlight it may get too hot. Maybe a white box shaded by an umbrella is better. Metal boxes can dissipate heat to the outside if you've got a thermal pipe between the CPU and the enclosure. This is the idea behind a cell phone with an aluminum chassis. It will be interesting how the temperature of the enclosed Pi evolves over time and what the final solution turns out to be. Please continue to update this thread.
I think the only temperature of much concern is of the beer OP will be holding as he admires his boom mounted Pi in the sky!

Cool looking home made antenna rotator! Looks like its made from erector set parts.
The parts are made by ServoCity in Kansas. It's all extruded CNC aluminum. The prices aren't bad for what it is.

The enclosure is standard electrical PVC gray right now. You have me wondering now if spray painting it gloss white will help the thermal situation any, or if the paint will become yet another thermal barrier.

It's still at a nice and steady 55.8 C / 132.4 F . Sometimes it will flash 56.4C but it goes right back to 55.8 . I'm sure adding a set of heatsinks will help that out just a little.

I'm going to let it run inside the house over the weekend and see what it says on Sunday night.

Here's a couple more pictures of the rotator assembly.

Image
http://i1105.photobucket.com/albums/h35 ... 122926.jpg

Image
http://i1105.photobucket.com/albums/h35 ... 122912.jpg

Image
http://i1105.photobucket.com/albums/h35 ... 095423.jpg

Image
http://i1105.photobucket.com/albums/h35 ... 095457.jpg

Image
http://i1105.photobucket.com/albums/h35 ... 095603.jpg

Image
http://i1105.photobucket.com/albums/h35 ... 123311.jpg

magnethead
Posts: 30
Joined: Mon Jun 02, 2014 6:50 pm

Re: Temperature in sealed case

Tue Oct 04, 2016 4:40 am

It stayed up over the weekend, and my POE injector/splitter came in, so now I have the whole thing running in my server closet with an ambient temperature of 96.8 degrees. (I have one of those wireless indoor/outdoor weather deals with the "outside" transponder in the closet and the display on my desk)

RasPi SoC Temp: 62.3 °C / 144.14 °F

With the heat of direct sunlight, I'm thinking that the 80/176 degree range is entirely possible.

User avatar
jasonclark
Posts: 55
Joined: Sun May 13, 2012 3:51 pm
Location: Hertfordshire, UK
Contact: Website

Re: Temperature in sealed case

Tue Oct 04, 2016 9:10 pm

In my experience, conduction to the walls of the enclosure (if it's a metal enclosure) works really well. Also trying to put the enclosure in the shade.

On a RPI, conduction is a problem if you're using hats as they are exactly where your heat spreader will go (use an extension cable?).
Heat sinks will dissipate within the enclosure, removing heat directly from the SOC but raising the ambient temperature within the enclosure. A larger enclosure will help too..

If you do get problems, put the enclosure in the shade, and if possible increase the volume of the enclosure.
As suggested, a Peltier device will help... but will take more power and complicate the project when some practical means are usually cheaper and easier.

In my industry, we've installed rugged systems in the Desert (middle east) and the main option is to place it under a solar canopy or bury the box not suitable for your project... but does work great. When it's 45C in the shade and you want water, a bottle in a deep hole is a lot cooler than one that's been sitting in the sun!

Pithagoros
Posts: 580
Joined: Wed Nov 12, 2014 8:16 pm

Re: Temperature in sealed case

Wed Oct 05, 2016 7:56 am

I have one on the top of a mast (though it's not a 3). It's in a case that is sealed with silicon, my idea is that the cable run from antenna to bandpass and SDR was minimised. I use a PoE to power it.
Our temperature extremes here are from about -10C to +33C, 14F to 93F. I've never checked its temperature but it has always been 100% functional.

If a Pi 3 is only doing lightweight work then consider lowering the clock speed (effectively pre-emptive throttling) to avoid approaching the high temperatures within the enclosure.

Return to “General discussion”