I looked in the manual, and searched the forums here, but could not find the maximum allowable power for the fan connector. This bit of information should be added to the manual, given the wide possible variety of cooling solutions people might choose to use for the UDOOx86. I am currently using a quiet 120mm fan that draws around 2 to 3 watts, or up to about 0.3 amps. It is working fine, and can easily keep up with Furmark + CPUburn when running at 50% speed as set in BIOS. As expected, it starts at 100% while BIOS is still in control, and leaving it running for a long time while exploring the BIOS settings did not burn anything out. I salvaged the tiny fan connector from a dead laptop CPU blower/fan, and I am guessing this is the standard connector used for most 3 pin laptop fans, albeit with a different pin-out as inferred by the wire colors (which I swapped around to match the UDOO because I'm fussy). In case anyone is wondering why such a large fan, it was simple: it was the quietest fan I had on hand, and I was looking for effective silence for my mother's TV computer. The UDOO lives in a decorative bamboo slatted box that is about 11" long, 8" wide, and 5" tall. The box actually has a wire frame made of steel wire about like typical coat hanger wire, with 1/2" bamboo slats bound to the wire to cover the box, with about 1/4" air gaps in between. We mounted the fan inside the box with small tie wraps around the slats, with the fan on the inside of the hinged lid, where it opens out of the way with the lid for access. The fan blows down on the UDOO, with the heat sink sitting under one side of the blades to avoid the dead zone in the middle of the fan. There are no extra air holes, but the fan sucks plenty of air through the slats, and air blows out all around the sides. The UDOO, an SSD and the stock 12V power supply all just sit on the bamboo bottom of the box, with cables for HDMI, LAN and a 120v extension cord leaving the box through a handle-hole at one end, and a power+reset button on a small plastic panel (salvaged from an old case) tie-wrapped into the handle hole at the other end. We are delighted, and the UDOO is running perfectly on Win7.
Do you see the strange fan behavior (changing frequently rpm during operation) in your configuration like we see on the Udoo Fan? https://www.udoo.org/forum/threads/cpu-fan-strange-behaviour.7696/
FAN 12V DC control's MOSFET ( IRLML5203TRPBF ) per datasheet can handle up to 3A. Naturally it is subject to PCB trace thickness and heat dissipation. It is a tiny SOT-23 package. I would derate it by 50% thus 1.5A probably would be a fairly safe driving spec.
Did not notice any strange behavior. The fan seemed to switch from OFF to 50% to 100%, and then back down, all in sensible time periods and without any back and forth. Benchmarked with Furmark+CPUburn, and the fan pointed away from the CPU. It took maybe 1m to heat up enough to turn the fan on at 50%, then a couple more minutes to finally jump to 100%. I gave it maybe 30s with no air, then aimed the fan at the heat sink from maybe 6" away, and it dropped back to 50% in about 30s, and stayed there for as long as the burn programs were running, then turned off again maybe 30s after I stopped the burn. Repeated a couple of times, all consistent, nothing unstable happening.
I think that seems like a sane answer, even though it's way overkill for the job. Given how very much air can be moved with far less than 18W of fan power, I can't see any compelling reason to even go over 1A, and I have to note the irony of anyone using a fan that draws as much or more power as the whole computer it's cooling. A silly person like me ends up wondering if you would then need a second fan to cool the first fan, and if that could lead to a fan-ception? Thank you for looking up the answer and making a sensible estimate for the de-rating. I should have done it myself and posted the answer instead of a question. I have to get used to a platform where the information is actually public (I like this), and start thinking like a computer electronics technician again (my diploma from the 1980's is worth something like toilet paper by now). Too many years of unapproachable surface mount and effectively disposable computing, I guess I'm just rusty and gave up even thinking about tinkering at board level. I think my last motherboard repair was cutting off couple of shorted electrolytic filter caps to save a friend's computer, and then we finally got all solid caps, and there's been nothing more than a few wiggly connections and swapping dead boards. This UDOO has me excited again.
Thanks! From your first post it is not clear (for me ) if you connected it with 2 wires or 3? The 3th wire is an rpm sensor that gives feedback to the controller, perhaps that is giving strange behavior.
Yes, I connected with all 3 wires. I think the fan speed controller must use the RPM wire, in order to measure what 100% RPM is, so that it can reduce the power until it measures 75% or 50% RPM. I note that when the computer boots, BIOS always runs the fan at 100% for a little while, and that would be the time that the full speed RPM measurement would be taken. This method would be necessary because different fans will slow down by different amounts for a given reduction in power delivered to them. You can't just assume you can give a fan 50% power to reach 50% speed, in fact it would almost always take a lot more than 50% power to maintain 50% speed, and some fans might not even keep spinning at only 50% power. I learned from years ago, playing with giving fans less than 12V to make them run slow and quiet, that many fans will not even start spinning at 5V, and a few won't even start spinning at 7V. And different fans will run at very different speeds when fed reduced power. The only way to reliably run a fan at 50% speed is to actually measure what 100% speed is, then monitor the RPM while reducing the power, until it slows down to 50% RPM. And that obviously can't be done without the RPM wire connected.
I don't think this is correct. I am using a three wire fan my ebm-papst, http://uk.rs-online.com/web/p/products/0482106/ and this exhibits exactly the same behaviour. I will try disconnecting the RPM control wire and see what that does.