pi cooling system.


10 posts
by paradoxwh » Sun Sep 01, 2013 10:14 am
Hi everyone.
I'm building my pi home server, with nginx(php,cgi-bin), ssh, btsync, iptables, cron, ramlog. I'm running the pi as the DMZ of my network, so I can access my pi services from anywhere with a dynamic dns update.

I've installed a small single heat sink cutted off with my dremel from an old video card bigger one.
I'm planning to make 2 more heat sinks for eth controller and current regulator. I thought it was enough.

While I was installing the software, I ended up writing a simple cgi-bin to monitor my pi-status.
At this point I realized that the temperature of the cpu of my pi inside the enclosure was 51.5°C! :shock: I'd like to run my pi 24/7, so I think I'll need a sort of cooling system.
Without the top enclosure the pi is exposed to dust and the temp is 45°C. :?

I took an old cpu fan. It's a 12v x 0.12A = 1.44W fan. 50mm x 50mm
I've connected it to pin 4-6 -> 5v-ground and it' spinning enough to keep my pi to 31°C. 8-)
1.44W / 5V = 288mA is it too much to run 24/7? the current regulator is cool right now.

In case I would connect some usb devices without an external powered hub, would it be a problem? :mrgreen:
I think I'll have to use a second cellphone psu, and conltrol the fan status with gpio.
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am
by ShiftPlusOne » Sun Sep 01, 2013 10:16 am
Just wondering, why do you think 51.5°C is hot for a pi?
Forum Moderator
Forum Moderator
Posts: 1544
Joined: Fri Jul 29, 2011 5:36 pm
Location: The unfashionable end of the western spiral arm of the Galaxy
by paradoxwh » Sun Sep 01, 2013 10:21 am
I think that if I can keep it cool hardware will last longer. a couple of years more maybe :lol:
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am
by ShiftPlusOne » Sun Sep 01, 2013 10:35 am
Up to you, but I would be very surprised if the CPU was the first thing to fail. It's hard to say what the lifetime of these things is, but will you really care in 10 years time whether your pi will last 1 more year?

The pi will run happily up to 85 degrees, then it will turn off over-clocking and continue to run happily until the temperature goes up to about 95 degrees, IIRC. Even 95 degrees seems like they are trying to stick to the safe side by shutting it off much earlier that strictly necessary. Not to mention that it would be a challenge to reach 80 degrees in the first place. This is a cpu designed to run without any cooling (when was the last time you saw a mobile phone with a fan?).

Anyway, if you still think cooling makes sense, go for it.

If you are sure your fan is drawing about 288mA, yes you will have trouble connecting USB devices. Let's say the pi alone uses about 350mA. All up, you'd be using about 650mA. The pi has a polyfuse which will start to trip at around 750mA, leaving you 100mA to play with. Many devices, even ones that claim to only use 50mA, will trip your polyfuse at turn off your pi. This is all just a rough estimate though.
Forum Moderator
Forum Moderator
Posts: 1544
Joined: Fri Jul 29, 2011 5:36 pm
Location: The unfashionable end of the western spiral arm of the Galaxy
by paradoxwh » Sun Sep 01, 2013 10:49 am
Thanks, maybe I'll go just with the heat sinks for now.
I think I'll try to find a small 5v fan, and enable it only if temp go up to 50°C.

I'm not an expert in electronics, but if I install a resistor. Would it drain less current or would it simply make the resistor a little "hot" but still drain 288mA? Is there a way to build a safe current regulator (an active component like a transistor) to make it drain only 100mA?
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am
by ShiftPlusOne » Sun Sep 01, 2013 11:07 am
Take this with a grain of salt, since I may get it wrong. If the fan draws 288mA, and you add a resistor in series, the same current will flow through the resistor, causing extra power dissipation through it. I think the correct way to control the power usage of the fan is to use PWM. You can use a small 555 timer circuit and a mosfet for that, but is it worth the hassle?
Forum Moderator
Forum Moderator
Posts: 1544
Joined: Fri Jul 29, 2011 5:36 pm
Location: The unfashionable end of the western spiral arm of the Galaxy
by paradoxwh » Sun Sep 01, 2013 11:29 am
yeah I thought the same thing.. thanks again :)
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am
by Burngate » Sun Sep 01, 2013 11:37 am
So much depends on the design of the fan.

For a simple fan, when it starts up before it gets moving, there'll be no back-EMF and the current - and therefore the torque - will be large.
As it speeds up, the back-EMF grows, the current goes down, and so does the torque, until the mechanical load equals the torque so it gets no faster.

If you take a simple DC fan and strip off all the blades, so there's (almost) no mechanical load, it'll spin fast enough that the back-EMF = applied voltage, and (almost) no current will flow. No current means no torque.
If you take the same fan, and fill its bearings with Araldite, it won't move. All the voltage will be across its winding resistance, and the current will be limited by that. Plenty of torque, but no movement.
It will have been designed so that, in normal air, it will spin fast enough that the back-EMF plus the voltage across its resistance equals the applied voltage.

Putting a resistor in series with it means that the voltage seen by the fan will be lower, but how much lower depends on the current, which depends on the back-EMF which depends on the speed which depends on the (square of the) air-speed which depends on the torque which depends on the current ...

Then you could have a more sophisticated fan with some electronics in it to control the speed, so ...
Brain surgery is easier than psychoanalysis
User avatar
Posts: 2669
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK
by paradoxwh » Sun Sep 01, 2013 12:39 pm
I tried actually to measure the current with my multimeter.
but when I do so the fan won't spin.
The multimeter is set to cc 2000u in series with the fan. 5v ----- (+ A -)---- (+ FAN -)------ ground.
maybe the mult is broken :P I'll try another one :\

EDIT...
2000u is 2mA ... yeah I'm going mad ahaha... but the max measured is 200m it means 200mA?

EDIT2:
wou guys, the multimeter amp was broken. with another one it drains 40mA.
It makes sense now. the fan is not spinning as fast as it makes on a cpu 12v.
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am
by paradoxwh » Sun Sep 01, 2013 1:11 pm
two images from my enterprise :P
it's a good fan, anotherone woudn't even start to turn by itself.
Attachments
web.png
web.png (33.32 KiB) Viewed 453 times
mess.jpg
mess.jpg (55.8 KiB) Viewed 453 times
Posts: 7
Joined: Sun Sep 01, 2013 9:44 am