45C idle. 27C coolant. Good thermal interface. What gives?
 

UPDATE:
Issue has been updated. See a few posts down...

old post follows...
--------------------

I have a Corsair CWC100 kit I am trying to configure but I'm getting really poor performance. As far as I can see, the whole system is configured properly and everything functions.

Specs:
- EPoX EP-9NPA+SLI
- Athlon64 FX-53 (12x200 / stock)
- Artic Silver 5
- All watercooling hardware provided in Corsair CWC100 kit (http://www.corsair.com/corsair/COOL_water.html)

The CPU idles at 45-50C while the coolant remains 27C. This means poor thermal interface with the block & CPU right? So to try to fix this, I've tried:

• mounting without standoffs / AMD mounting hardware (more direct pressure on IHS)
• AS5 using a small bead. (what I usually do)
• AS5 spread over the CPU
• AS3 spread over the CPU
• White paste spread over the CPU (not recommended! This ripped the CPU out of the socket and bent pins when I tried to remove the block!)
• different mounting hardware & techniques (i.e. Used "mounting-post" method. Used standoffs as recommended by Corsair. Used different mounting hardware (plastic backplate and metal backplate with longer threads & standoff).
• Lapped the base of the block. There is slight concavity, but not enough to cause a 15-20C difference the temps I'm getting and what the expected temps should be.
• Shimmed the retention mechanism to exert a little more pressure on block.

The pump is reporting the expected 3600RPM. Fan rad is fine (120mm Panaflo). No kinks or sharp bends in tubing.

I'm really out of ideas. Everything looks good and everything works... except for the fact there doesn't seem to be any conduction between the CPU and block!

To compare, swapping the block for a Zalman CNPS7700-Cu heatsink/fan gets me 35C idle and 42C load. :shrug:

Any help or suggestions appreciated!

- Jon.

FYI.

Here's Corsair's recommended method of installation:
http://www.pcperspective.com/images/...B-Exploded.gif

Here's what it looks like installed:
http://www.insanetek.com/reviews/coo...s/IMG_3805.jpg

I've tried the above method and variations in mounting (see first post) but still have poor performance.

- Jon.

cpu block piped correctly?
res. piped correctly?
all air bled?

post full pics of system would help....

Yes. Everything is routed and connected properly (block outlet to res, res outlet to pump, pump to rad, rad to block inlet). Checked and double-checked etc.

No air in tubing. Res is trapping air properly at the top of loop.

I'm currently using a table top / open case currently until problems are fixed.

Pump & Block on same level. Rad slightly above, and res at top. I'll take pics of the setup tonight, but there's really nothing special to see. Typical laid out parts, tubing, wires, etc.

- Jon.

Lets see, coolant is cool but the block has high idle temps. That means mostly likely culprit is the mounting between cpu to the waterblock.
That would be the reason why you would get such high idle temps because heat is not being transfer to the block into the coolant. I would remount the waterblock and be very specific in how you mount it. Have your computer run for awhile and see if the water block gets hot by touching it. If it is cool then you have mounting issues.

If the block feels very warm or hot, then the pump isn't corretly working.

Would agree. The idle cpu temp shouldn't be more than 5-10C over water temp and that is generous. Maybe to much thermal compound? Is the block mounted in the right direction?

Geez. Looks like part of the problem was airflow. :rolleyes:

Since I was working on a bench, there was no cooling over the socket area. I pointed a 120mm fan over the socket and the temps are now 41C load.

Load: 41C
Idle: 36C
Coolant: 28C
Ambient: 24C

41C load still seems a bit high since I was getting 42C load using air and less noise (watercooling = 2x120mm fans, aircooling/Zalman = 1x120mm fan).

But watercooling will pay off in overclocking though, so I don't think I'm going to sweat it too much.

Yeah block is mounted correctly (outlet at higher elevation than inlet). Thermal paste applied with only a small bead of AS5. Block feels a little warmer now after running for 4 days straight.

- Jon.

how are you using 2 120 fans?
it looks like the kit only comes with one BIP120
do you have one pushing and one pulling?
fyi: this is a much louder arrangement than a single fan for minimal performance gain

also: what are you measuring temps with? you may be seeing inaccuracy of several probes that amounts to a drastic looing problem

jman,

2 fan WC config: one blowing on rad. one blowing over socket / MOSFETs.
1 fan WC config: one blowing on rad.

The 2 fan configuration above is the only way to go at the moment. Those MOSFETs are way too hot.

Coolant and ambient temps measured using a digital thermometer. CPU temps measured with diode. I know the diode is not the way to measure accurately, but it's useful judging relative performance compared to my old aircooling setup.

hmmmm......

Perhaps your block is clogged? Rare problem, I know, but it is worth considering.

Check list:

1. Proper mounting: Check
2. Proper routing of tubing: Check (I'll take your word on this, no kinks etc.)
3. Waterblock is functional: don't know yet
4. Pump is functional: don't know for sure (RPMs doesn't mean 100% that the pump is working)
5. Rad is functional: probably, as your water temps are good.

Where are you measuring your water temps from, btw? Yeah, those MOSFETs are hot little suckers, and the PCB conducts heat, meaning that if they run hotter than the CPU, they will transfer some of that heat to the CPU (ironic, isn't it?). If you have good airflow through your case, and heatsinks on the MOSFETS (I know the plastic doesn't act as a heatspreader, the MB does, but it still helps) you shouldn't need a fan blowing directly on them.

You might have a crappy diode. On an epox 865 board of mine, my CPU temps on a 2.6c shot up when I installed a swiftech MCX 478V from the stock cooler. The temps went from 35C all the way up to 69C! Sometimes they will plummit to 0! Yet, somehow I was able to push the chip from 2.9ish with stock to 3.2 with the new heatsink. Diodes can be tricky suckers.... don't trust them.

This kit is a swiftech kit rebranded. mcp350, mcw6002, 120mm.
I don't think the blockage would be a problem on these WC blocks.
Diode might be a problem.

I never used the pump before..... Might be pump problems. If the WB is hot and you need fans to cool it, not enough flow. CPU idle temp should be 28-32 C with those coolant temps not 35C. Essentially the waterblock is acting as heatsink and fans are cooling it.

Easy solution is to take everything out and put it back in instead of troubleshooting.
If you still have probems then you might get a better idea which one it is.

I would also remount. Some peple here have more experience than i do mount blocks and they still run into mounting problems on rare occasion. I mount 3-5 times to compare results. Once i find a good mean I try to repeat what i did exactly the same. I mean sometimes directions aren't always correct.....

would greatly help to post pics!