CPU-340 thermal resistance and core temps

CPU water cooling blocks...
Water Pistol Grunt
Water Pistol Grunt
Posts: 4
Joined: Wed Dec 31, 2008 5:08 pm

CPU-340 thermal resistance and core temps

Postby lutorm » Fri Jan 16, 2009 6:44 pm


I just installed a CPU-340 on my E6400 (overclocked to 2.8GHz), and I'm seeing higher temperatures than seems expected, maybe someone knowledgeable can comment on this.

The flow rate in the system is about 2.3 l/min, the coolant temp is 27C regulated to within half a degree. Water block is mounted with AS5 and tightened down quite a bit (can't rotate it in the mount with a reasonable amount of torque).

When maxing out the 2 cores with P95, the core temps (measured by coretemp) are 22C higher than the coolant temp, ie 49C. This seems higher than expected. Using the graph of CPU340 thermal resistance and the TDP of the E6400 of 65W, that amounts to a delta-T of about 4.5C, ie less than 25% of the observed. Even if the power draw is higher due to the overclock, it's not *that* much higher. (The question is if the graph includes the thermal resistance of the block/heatspreader interface?)

Another way to look at it: At load, the system draws 239W AC, ie about 191W DC assuming 80% PSU efficiency, and deltaT core-coolant is 22C. At idle, the system draws 178W AC ie 142W DC and deltaT is 8.5C. Using these two points and assuming that all the extra draw when starting P95 comes from the CPU (an overestimate) we can estimate the thermal resistance: (22C-8.5C)/(191W-142W) = 0.28C/W. That's again about 4x as much as the graph indicates for the CPU-340, so the two estimates come out roughly consistent. (The fact that not all the extra power is from the CPU only makes that number smaller than it should be.)

So the question is: what is causing that extra 75% thermal resistance? The block/heatspreader interface, if that's not included in the Koolance graphs? Internal resistance in the CPU? Do I have a warped CPU? (It's very slightly convex in one direction when compared to a straightedge.) I have reseated the block on the CPU and reapplied AS5, and the temps did not significantly.

Or maybe it's all normal? ;)



User avatar
[Koolance] Tim
Posts: 451
Joined: Tue Sep 16, 2008 9:34 am
Location: WA, USA

Re: CPU-340 thermal resistance and core temps

Postby [Koolance] Tim » Fri Feb 06, 2009 1:01 pm

The temp difference between core and liquid does seem too high. The graphs on our site incorporate the included thermal paste . Most likely, Coretemp is not calibrated properly for your chipset and therefore misreporting the CPU, or there is a contact problem between the CPU and water block.

If it's a contact issue:
    Try reseating the CPU and check the impression in the thermal paste for adequate contact.
    Use the bare minimum amount of thermal paste necessary to achieve the above.
    Tighten thumb nuts as required. Higher pressure directly equates to improved heat transfer, but don't damage the motherboard.
    Make sure to remove the protective plastic film from the CPU cold plate before installation.
    Do not place the Koolance temperature sensor or adhesive metal tape anywhere between the water block and processor.

If it's a software calibration issue (which we see a lot of), you would need to adjust the compensation range for your given processor and chipset. Many customers install programs like Motherboard Monitor, etc., pick their motherboard chipset, and expect the temperatures to be correct. There is normally a lot of calibration necessary for these programs.

[Koolance] Dean

Re: CPU-340 thermal resistance and core temps

Postby [Koolance] Dean » Fri Feb 06, 2009 2:25 pm

I would strongly recommend using Real Temp, because it allows you to input a TJmax Calibration value. You will also want to read the Calibration page, as it explains the generally accepted method of TJmax calibration.

Return to “Water Blocks - CPU”

Who is online

Users browsing this forum: No registered users and 1 guest