Hitting 100 C on Custom EKWB Water Cooling Loop??

zacharyphill

Honorable
May 12, 2017
5
1
10,510
I've built my first custom water cooling loop using EK components, and I can't figure out why my CPU temperatures are so high. It's currently idling at 33 C, but it's been around 45 C idle at times. While running a Prime95 stress test, temperature climbs to ~75 C within a few seconds and eventually settles around 98-100 C when the CPU starts throttling. When I stop the stress test, the CPU comes back down to 50 C in just a few seconds. The stock cooler performed better! No overclocking has been done. The GPU is in the loop after the CPU, and it idles around 32 C and peaks around 50 C while running FurMark. No issues there. Here's what's in the loop:

- EK-SBAY DDC 3.2 PWM pump/reservoir combo
- EK-Coolstream PE 240 dual radiator
- EK-FB ASUS Z270E Strix RGB Monoblock
- EK-FC1080 GTX Strix
- 16 mm hard lines
- EK HDC fittings
- Clear EK Cryofuel concentrate and distilled water

All components are arranged in a series loop: radiator first, CPU second, GPU third, and back to the pump/reservoir. Here's some photos to better describe the loop:

qeklTSu.jpg

System overview

fV9F8hM.jpg

Monoblock detail

1QDxhr2.jpg

Pump/reservoir/radiator detail. The pump suction is toward the back and the discharge is toward the front, where the 180 degree bend is.

I believe there's an issue only at the CPU, since the graphics card is being properly cooled. I've disassembled and reassembled the loop three times (kind of a pain with hard lines) to check the thermal paste on the CPU. At first I thought I put too little, so I added some. I then thought maybe there was too much and I went back down to less paste. No significant changes were seen with each disassembly/reassembly and reapplication of TIM.

There are a few small bubbles visible in the CPU area of the monoblock, but I haven't been able to free them. I don't think they're big enough to cause any issues but they're worth mentioning.

What could the problem be? I thought for sure I was using too much thermal grease, but now I'm down to a small amount (the size of a lentil) and the symptoms remain. I haven't contacted EK yet but I might if anyone thinks it could be a defective monoblock. Before I do that, any other ideas?

Thanks in advance for your input.
 
Solution
If you're slowly climbing to 100C, good chance the monoblock design and your cooling capacity are the culprit. The monoblock cools the VRMs as well as CPU. These get quite toasty, hotter than CPU even, especially under stress. With a single 240 radiator, you're likely to overwhelm cooling capacity. Even more of a possibly if you run both CPU and GPU load simulatanously.


I may misunderstand airtight, but yes, the system is completely sealed. The reservoir does have a max fill line though, so I kept the coolant level below there.

Flow rates have made no difference; I saw the same results running the pump and fans at 100% as I did when running them on a curve, just with more noise.

Testing the loop with the pump at the bottom would be impossible right now, but if I had some soft lines and different fittings I could move things around.
 


I was running Prime95 v28.10, but I downloaded v26.6 on your recommendation. The symptoms look mostly the same, but I'm running a few degrees cooler now. I just started the stress test a few minutes ago and I'm up to 77 C now. Could be the older version or because I burped some air out of the CPU block.

Thanks for telling me about the temp spike bug on these processors, but I'm not seeing a spike. Temperature rises quickly to ~70 C and then slowly rises and hangs out around 100 C. Maybe a little cooler now that I got some more air out. At any rate, reading about that issue I found the Intel Processor Diagnostic Tool; I'm running it now to make sure there's no defects on the processor.
 
If you're slowly climbing to 100C, good chance the monoblock design and your cooling capacity are the culprit. The monoblock cools the VRMs as well as CPU. These get quite toasty, hotter than CPU even, especially under stress. With a single 240 radiator, you're likely to overwhelm cooling capacity. Even more of a possibly if you run both CPU and GPU load simulatanously.
 
Solution
few details are missing:
1. your room temperature also called ambient.
2. what is your GPU temperature under load ? i mean running Heaven benchmark for 30-40 minutes - should be well below 50
3. What is the CPU voltage settings/values during load ?
4. what fans you use and what is their speed under load ?

Are you sure you got all air out of the monoblock and radiator ? large air pockets significantly reduce liquid flow withing the loop and reduce cooling capacity.

which torture test do you use ? follow this guide for more accurate results:
http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

As already mentioned above, the 240 rad is not enough to keep the system both cool and quiet. you need at least to double the rad surface.
 

TRENDING THREADS