2

Hardware Setup

I have a custom build with a Gigabyte Z270X-UD5 and an Intel i7-6700K Skylake installed into the LGA1151 socket. My system is in a custom-cut BPP crystal link hard-line liquid cooled setup with an XSPC Raystorm Pro, 380mm radiator, and Photon 170 pump. My M_BIOS version is F4 (most recent).

Problem

It seems like I have been having temperature sensor issues on my motherboard. When the machine powers on, and I enter the BIOS, the processor is incidentally under-clocking itself to around 21000mhz due the CPU temp reading a buggy value of 100c, never increasing or decreasing after power on (usually, buggy flip-flopping between 99c and 100c). Due to this low clockspeed, it causes a BSOD in a very short time in Windows, as the OS runs very very slowly and the cache of my nvand nvme Samsung 950 pro m.2 SSD cannot keep trying to push through the underclock.

Research and Tests

I have heard tell of thermal sensor failure causing these same types of failures with Apple devices, due to safety under-clocking even if this value is bad.

  1. Touching my tubing reveals an only mild warmth. I got gutsy and gave the underside of my copper Raystorm Pro a little tap and found it only mildly warm as well. The block is screwed down tightly and mounted correctly, the correct amount of Noctua thermal compound installed underneath. So my loop is most certainly dissipating the heat. Putting my hand over my radiator fans reveals the air is getting warm as the machine runs.
  2. I was thinking my lack of a flow meter and water temp readout could have caused another problem, so I went out and got a Koolance flow meter and a temperature sensor plug and attached them, the Koolance RPM readout to a frequency modulation logic board that converts a 2 pin signal to a 4 pin signal, and the temp plug installed into the free port on my pump. My water temp on the return from my processor reads a mild 51c under load, the flow meter returning a value of 47.5-48 ml/m to the CPU_FAN header as the motherboard manual states to connect a frequency modulator for flow rate to when the CPU_FAN4_PUMP header is populated for pump speed control. I have the dip switch on the logic board set for 10mm ID tubing properly. My pump speed header reads a fluctuating value of around 4600-4800 RPMs. I noticed the motherboard able to correct the pump speed , sometimes only needing it to run at 3700-4000 RPM because my water temp sensor is returning a correct value.
  3. I did a different test as I was thinking the liquid cooler was the problem, installing a Noctua air cooler and experiencing the same temperature readout on the processor. This convinced me its either some kind of bad BIOS setting or sensor problem.

What should I do?

  • How old is the motherboard? – user725131 Jul 20 '17 at 12:05
  • Brand new, purchased 2 weeks ago. – Robert Smith Jul 20 '17 at 12:12
  • 1
    Have you considered using your warranty to get a new board, as the one you have might be defective? – user725131 Jul 20 '17 at 12:13
  • I will contact gigabyte, yes. That's upsetting to hear, though. I've already been through ASUS' RMA disputes. shudder – Robert Smith Jul 20 '17 at 12:15
  • Well while that might be a headache, using the warranty would have been my first choice, as the information you've written, i my personal opinion, would most likely point to a defective card. – user725131 Jul 20 '17 at 12:17
2

Solution

After draining my loop the second time, I dismantled the tubing to take a look at the actual way I had the XSPC Raystorm Pro mounted. Turns out there was a gap of about 1mm between the bottom of the copper waterblock and the CPU because I had assembled the processor stand incorrectly, resulting in an air gap and some heat transfer, but an air pocket in which the processor could heat up. The only thing solidifying the transfer at that point was the tiny 0.5mm thickness of the original thermal compound I'd had in place.

I disassembled the XSPC Raystorm Pro, refreshed the thermal compound (just for safety reasons), and installed four black plastic standoff spacers AFTER I put the block down, and THEN the thumbscrews. Doing this enabled me to get adiquite, tight spring compression and the copper underside of the water-block to secure down to the top of the IHS.

After a prime95 benchmark running for 10 minutes I now rest easily in the range of 48.2-49.1c.

  • 1
    Good thing you investigated. Processors used to burn out quickly if they weren't cooled effectively. – Christopher Hostage Jul 20 '17 at 16:36
  • 1
    Nice work, good thing everything worked out. :-) – user725131 Jul 21 '17 at 16:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.