MAZE2 system warmer than expected??

[krp]

Just finished assembling a liquid-cooled system as described below, and have some questions about the cooling performance, as it seems to be running a bit warmer than I expected it would.

Gigabyte 7DX mobo in an FS020 case with Athlon T-bird 1.4gHz chip running at its rated speed, (not overclocked), all 3/8-inch I.D. cooling components: a Dangerden MAZE2 block:
a CaseEtc "rese-pump" with the external pump:
and a Hayden 676 radiator:

with the flow as: Rese/Pump-->Radiator-->waterblock-->Rese/Pump.
The Hayden radiator is one of those 5x11 tranny-cooler types mounted vertically with two 120mm fans blowing through it.

Although I am not using a peltier, it is sufficiently humid here in Florida that I decided to do the anti-condensation work anyway: The CPU socket was prepped by sealing around the outer edges and inner edges with Dow-Corning silicone, dielectric grease in the pinholes, and a neoprene gasket surrounds the socket per the Octools condensation prevention article

The under-cpu socket temp probe on the Gigabyte 7DX is one of the "flat-yellow-tape" kind, I placed a piece of neoprene foam under the business end of it to keep it pressed firmly against the underside of the CPU, which I had also painted with some Arctic-Silver-II before mounting the chip in the socket. I applied the MAZE2 waterblock using a thin film of Arctic-Silver-II on top of the CPU core, applied the block tightened and removed the block until I found the spring tightness required to "print" an image of the core top in AS-II on the bottom of the MAZE2, then reinstalled with the springs about 2mm tighter than that, (which is the most pressure I dared put on it.)

According to MBM 5.08, The 1.4gHz Athlon is running at a core voltage of 1.81 volts, (which would have it radiating 77 watts), with a CPU temp of 43C, (open case, airconditioned room temp=23C) According to Radiate, this equates to a C/W of 0.26 degrees/watt.

Admittedly, (as I lack any other temperature probe equipment at this time) , this is only based on the likely inaccurate under-cpu socket probe, but it seems to be quite a bit warmer than I would expect for a MAZE2 block setup?? I would think that if the under-cpu socket probe was gonna be "off", it would seem to be more likely to read LOW than high? The MBM temperature reading and the BIOS agree perfectly, (as one would expect since they both use the same probe.)

Most of the MAZE2 review's I've seen imply a C/W rating of around 0.15, which should result in a full-load CPU temp of around 33->35C, so I suspect I've messed up somewhere resulting in about 7->10C higher than optimum for the MAZE2?? If I unplug the pump for a few seconds, the MBM CPU temp reading does seem to "react" very quickly to the loss of coolant flow, starting to climb within about 13 seconds, and quickly dropping back to the steady-state within about 5 seconds of restarting the pump. (I never let it go over 50C though), so although a bit higher than expected, the liquid cooling system is definitly having an effect.

Question for the pro's: Does the observed 43C (full-load) temp sound reasonable for these components??

The thing I am mostly concerned with is that this system needs to have a lot of "headroom" for operation in a much warmer ambient environment than the place where I am currently testing it... I tend to run my airconditioning much cooler than the end-user will.

Would upgrading to a higher-rated pump help much, or would I need to go to a 1/2-inch ID system to see much improvement?

Thanks for any tips...

[David]

I'd change out the radiator, but thats me. That could be the source of your problems. Usually though, mobo temps run high.

[resago]

that pump is crap.
a 350 inline would siute you much better.

[GuyBFF]

I agree, the Rese pumps are sorta inline (as they don't sit in the water), but they don't have that much power.

Also, resevoirs tend to slow flow as well, as instead of a smooth circulation, you loose a bit of the pumps ability to suck the water around as well as push (and add restirction).

I have no experiance with your motherboard, but is it possible the calibration of the sensor is off (Asus has done this in BIOS before to simulate core temperatures)?

P.S. condensation isn't possible unless something in your system drops below air temperature. (unless there's over 100% humidity / fog conditions in your room!). The air conditioner is probably already removing a lot of humidity. No need to worry.

[BrianW]

The Hayden radiator is one of those 5x11 tranny-cooler types mounted vertically with two 120mm fans blowing through it.

I would position the fans so they suck as opposed to blow. Should really improve the effectiveness of your setup.

[stevo]

one thing I found on my setup was strange temp readings with the cpu socket probe when I had placed some foam under the chip in the socket cavity.I think it seemed to perhaps insulate heat in, giving higher temp readings.
I removed the foam and filled the cavity with silicon grease instead.
My temps with 72w peltier read 5-6 degC idle and 23degC full load which are correct
stevo

[redleader]

Often times motherboards will overstate the CPU temp a little to compensate for the poor accuracy of measureing though the back of the CPU. Perhaps your AS2 and tight fitting of the probe are causing your probe to read more accurately then your board is expecting.

I would overclock anyway and see if you lose stability. IF its fine at say 1.55GHz then I'd assume the probe sucks.