Martinm210

Seems like I'm not the only one that has installed a brand new loop only to find my temps were less than I had hoped for. I spent a fair amount of time tinkering with different things and have learned alot doing it. I found that there are several relatively inexpensive tools and tests that you can do to improve areas which can either accumulate to a decent improvement or pinpoint an obvious problem. This is my newbie attempt at sharing the tests I've done to improve my system.

TOOLS NEEDED
First the thermometers, this will allow you to figure out what is going on from the Air to the coolant, and allow you to rule out if you have a radiator/fan deficiency. See the pictures under test two, you need some sort of digital probe temperature tool. For example I currently have two probes running, you can do alot with just two. Bought these at my local PETCO for $10 ea. One I've run down my CPU loop T line until it's right at the bottom of the T fitting but not quiet in the main flow. The second one is at the front of my radiator.

The last tool you need is a 5 gallon bucket and about 10' of extra tubing to do a flow rate test.

TEST 1 Room Temp Vs. Rad Intake Temp
The first test is to determing how channelized your room air is, and to find out how close the air entering your radiator is to the ambient air. So right now in the below photo I'm loosing about 1C from hot air blowing around my radiators and getting sucked back in. This was about 4-5C before I realized just how much hot air I was sucking in. With as much air as we move with our radiators, channelizing the cold air to meet the intake side of your radiator alone can make a huge difference. Espcially if your case is under your desk or your rads exhaust against a wall, chances are you're recycling some of that hot air back through the radiator. Now you're cooling with hot air...not good. You want this as close to ambient as possible.

TEST 2 Rad Intake Temp Vs. Coolant T-line or Res Temp
This second test will tell you how well your radiator and fan combination is doing. It will rule out if you're under capacity or if you need a better fan or shroud combination. In my below photo example, I can see that my water temperature at idle is 1C above the radiator intake. Even under TAT 100% load (or orthos small FTT) my temperature only rises to be 2C above the rad intake temp. This is a great tool to show me that my radiator and fan performance is doing very well. If after a long benchmark you were seeing 10-15C or more, that would be an indication that you need to focus on your radiators/fans/shrouds. Either get a bigger or better radiator, get a second one, use some higher speed (better pressure) fans, and or build some shrouds to minimize the dead spot from the fan hub.


TEST 3 Flow Rate
This is not a quick test and requires draining of the loop, but it can rule out if you have a pump/flow rate problem. All it takes is a little extra test tubing and a 5 gallon bucket/stop watch. Take the 5 gallon bucket and fill it up a gallon at a time with some type of accurate 1 gallon measuring jug and make marks at 1 gallon through 4 gallons. Now fill up the bucket to above the 4 gallon mark and attach one piece of test tubing to the intake of your pump into this bucket. Take the second piece of tubing at the outlet of your loop into a sink to drain. Now first prime the hole system so you get all the air bubles out of the pump and inlet line. Then fill the bucket back up again above the 4 gallon mark, now get the watch ready and start up your pump again, when the water draws to the 4 gallon mark start the time and when it reaches the 1 gallon mark stop. Now divide 3 by the time in decimal minutes. 2m 5 seconds would be 3/(2+5/60) = 1.44 gal/minute. Average is about 1.5 gal/min if you have alot less you might have a pump/restriction problem, if you have about 1.5 or better then it's not your pump or your restriction.

This is how Petra from Petra's Tech Shop (http://www.petrastechshop.com) does his testing. I'm showing this example because mine is much more redneck style. In this case he was pumping from a container that was constatly supplied by a garden hose, and he measured the outlet but either way works, just make sure you don't suck any air bubbles into the pump, that's why I prefer measuring at the inlet side.



TEST 4 (Thermal Paste Pattern)
If the above 3 tests come back with favorable results, the only thing remaining is the water block itself, the contact, or the processor is just another hot one. The third test is a visual check on your thermal paste spread. This is done by simply mounting the block with TIM and then removing it. The paste pattern left over is very revealing about your contact.

Use this AS link for different applications.
http://www.articsilver.com/arctic_si...structions.htm
Thin line for C2D, rice dot in the middle for single core, etc:

This "Art of the Bow" link is very helpful in showing examples, preferrably you want almost a transparent thickness of TIM when you remove the block where your core is at the center. Remember TIM that too thick will insulate, you want a close to metal on metal as possible so you should see metal, not paste in the area of your core. You might also considering bowing the block if you're brave enough to do so, note you can't go back:
http://www.xtremesystems.org/forums/...+fuzion+apogee

This is an earlier example of what my processor looked like before lapping. This is an bad example of using TOO MUCH paste, but note the shaded red area is the approximate die location of a C2D, this is the most important area, and you want very little TIM here which this is a good example of. Note how the paste is thin in the middle and thicker on the outer edge, this indicates either my water block or my processor has some slight bow which is good. If it was the other way around, thin on the outside and thick in the middle that would be bad.


Not a test, but lapping your processor can help shave a couple of C off the temps, it takes the roughness off the surface and helps ensure a flat surface.
Here is an example of my E6600 after lapping (NOTE THIS VOIDS YOUR WARRANTY!):


Another thing you can do to ensure better block/CPU contact is getting a retaining backplate if your waterblock didn't come with one. I bought this one for my C2D, helps prevent the motherboard from flexing:
http://www.petrastechshop.com/scsunrekit.html

TEST 5 (Block Dissassembly Obstruction Check)
The other thing I learned is water blocks are becoming very "Filter Like" and most components in a water cooling system are not necessarily "Fluid Ready" in that they could and often do have a bunch of junk in them. Here is my Fuzion after taking it apart. Note the gunk in the pins and the poor leftover stringy gasket crud on the left:

Dryadsoul first pointed this out, if you buy a fuzion, better check! The lower gasket ring has a bunch of casting leftovers that spill out and hang off the gasket you can see on the left above. But check regardless other obstructions could exist, I know Ira-K mentioned one guy was having trouble with temps and found a piece of a packaging peanut inside

Summary
That's it, hopefully one of the above five test/checks will help you with your high temperature woes, still working on mine, but I'm happy to say I'm on my way to better temps. It wasn't any one thing for me, but a little here and little there and by the time they all add up...it's much better.

The last thing to recognize in all this is two main points. First ambient temperature controls the system, if you have really high ambients, you're temperature will be higher because you're cooling with the ambient temp. Second is the fact that there are significant variations in processors. I've heard many stories where someone has two or more of the exact same processor, yet they get 5C or more difference in temperatures. Luck of the draw, if you have a hot processor...you'll have hot temps.

Last thing....don't worry too much about temps as long as you get the overclock you want and your load temps are within a reasonable maximum.

Hope this helps..
Martin

AMD Dave

Great guide!!
You've done a lot of work and compiled it all into one nice and clean guide!!
I smell another sticky

__________________

Apollo4g

Nice guide!!!!

dpawl31

Holy lapped goodness

__________________

Martinm210

Thanks, just figured I'd summarize my journey over the last couple of months.

ira-k

Nice guide..I'd just add to it that its not only the Fuzion blocks that should be checked..Some have hanging shavings and bent and broken pins...I think I already told you about the guy who came through getting crap temp's and when he took his block apart there was a piece of packing peanut clogging it up...

__________________
...***Beware all ye Men born of Woman, beware the Hemogoblins***...
**E8600 @ 5751Mhz ***OCN Water Cooling Club And Picture Gallery ***150 Opty @ 3242Mhz **

Dryadsoul

Well done! And the original Apogee WB's were notorious for mill tailings and bent pins. But to be fair most retail WB's could employ far better quality checks.

__________________

ira-k

Needs a bump......This is in the stickies guys, might be a good link for the new guys having temp troubles.

__________________
...***Beware all ye Men born of Woman, beware the Hemogoblins***...
**E8600 @ 5751Mhz ***OCN Water Cooling Club And Picture Gallery ***150 Opty @ 3242Mhz **

GriZzlEnLS

Haha Thanks Ira, I was just searching for this exact topic

__________________

We Are Gr33n Machine

ira-k

Hey NP, Glad it helped you a little! Glad Martin took the time to make it to!..

Always really work your box around to work air out, slap your rad or shake it if you can to work out all the small bubbles and do at least 4 or 5 re-mounts over a couple of weeks time then work it back down to the lowest temp that you had.

Re-mounts can make a big difference, the more you do the better you'll get at it. Play around with mounting pressure to, some mobo's like a hard mount, some a light, thats just something you have to play around with to get tweaked right.

Adding a ounce of glycerin or a drop or two of blue "Dawn" dish-soap to your loop will help you to bleed faster to, they are both a surfactant or water wetter and break the waters surface tension, that lets all those little micro bubbles come of the tubing/block walls and lets the coolant get down in all the nooks and crannies. Don't over use either one or it can hurt your temp's, a little goes a long way.

I make my own back-plates out of 3/8" HDPE, you can buy it on eBay cheap or just cut up a white cutting board. Use 6-32 all thread rod for your studs, if you make one a inch or two longer then the others it's easier to stab your block on it. A thin piece of neoprene or a old rubber mouse pad is a good thing to put between it and the mobo to cushion it a little. I really like them because it makes it easier ot get a real good block mount with one.

Once you get the hang of it theres nothing to it, just a little practice is all it takes!

__________________
...***Beware all ye Men born of Woman, beware the Hemogoblins***...
**E8600 @ 5751Mhz ***OCN Water Cooling Club And Picture Gallery ***150 Opty @ 3242Mhz **