3 Replies Latest reply on May 12, 2014 9:55 AM by sylvia_intel

    NUC fan problem-high temperatures-simple mechanical fix-pictures included

    hgm

      Hi all,

       

      I recieved a DC3217IYE today. After assembly I noticed that the CPU temp at idle in the BIOS was 71C. I updated to BIOS version GKPPT10H.86A.0036.2012.1213.1503 but this did not solve the problem. I also observed the CPU fan speed to be 1600-1700RPM maximum.

       

      Fortunately I picked up the NUC while it was running Ubuntu to show a co-worker. As soon as I put pressure on the top of the chassis, I could hear a mechanical noise from the fan (perhaps best described as "growling"). I rebooted to the BIOS screen and found that I could reduce fan speed and increase the level of the noise by pressing on the top of the chassis, even to the point where the fan would stop.

       

      I removed power and disassembled the unit (only two screws hold the board in the chassis, at least one reviewer has gotten this wrong). I found that the 4-wire lead to the PWM fan was folded back over itself and crimped between the chassis top and the top of the fan. This caused it to put pressure on the fan case, distorting it and interfering with the blades.

       

      The fix was simple: unplug connector, use fingertip to smooth out the wire bundle into the valley present for cable management on top of fan, rearrange wire routing to prevent the problem from reoccurring on assembly, plug fan back in, and reassemble the unit. In retrospect, the ESD safe cloth tape from the WLAN wires could have been used to fix them in place.

       

      At extended idle in BIOS the temperature is now approximately 64C with fan speed of 2240 to 2280 RPM and the system is noticably cooler to the touch. This is with the cooling settings at Custom - 78C temp, 30% minimum duty cycle, and 8% duty cycle increment (the default settings from the BIOS flash, I believe).

       

      Pictures below.

       

      Before:

      fanwire_before.jpg

       

      After:

      fanwire_after.jpg

       

      Thought I'd post this simple fix for all who might encounter the issue, and perhaps an assembly line process improvement could be suggested from this.