Originally Posted by Shrimpykins
$20 says Ira had a probe hooked up to the CPU and wasn't using some type of software monitoring.
My understanding of physics is also limited but if I am understanding everything, the bottom line is, there is a difference and it is beneficial, no matter how small the benefit is. Just because it can't be measured in units of 1C or larger doesn't mean anything. .01C could be an extra 10mhz out of my CPU.
IIRC there is something somewhere that describes what you are supposed to do to accurately measure your CPU temp. It involves carving a channel in the CPU IHS and placing a probe there.
Of course back in the slot A days a lot of us would stick a K probe or a compu-nurse tail to the side of the die. Then with any tops that you can pop off you can do the same, or you can also just cut a slot in the side of the IHS while it's off and and silicone the IHS back on if you want to.
Some of them would groove the die and try putting a probe at dead center, with mixed results. I always just stuck it on the side.
Now i just use software mostly. I'm to lazy now...
I do need to get a Fluke for my DICE pot though.
Originally Posted by Nordar
I agree with you as well that the calculations based on the heat transfer equation is simplified. However, the other factors you mention only have little effect and the effect they have will all work towards reducing the temperature drop (in equilbrium). For instance the tubing and resevoir will work as passive radiators
Lets hope the do work as passive rad's, if not your flow has gone laminar on you.
Small things to be sure, but the can add up when working with miniature systems like we use.
Now materials and thickness can play a big part if your not using standard commercial parts, which a lot of us don't. But even that's really pretty well null since we use C110 copper anyway and get the base as thin as humanly possible.
I admit that I am borderlining my understanding of physics (but that just makes it more interesting). So do take my posts as an attempt to learn.
I was under the assumption that for most radiator tests the water to air delta shown are based on average temperature of the water in the loop (and not rad out water temp). If this is the case is it then possible to compare the 1C temperature drop presented in my calculations above with a 5C delta (for instance)? The average water temperature in the loop is pretty much independent on the radiator placement in the loop.
I'm not into physics at all, other then a very basic idea of basic functions for WC'ing. After a while you hear so many arguments some of it has to stick.
Yeah the rad out to ambient is the very basic building block of your loops effectiveness in dissipating heat. All rad testing and loop health is based on rad out temp, ambient or sometimes seen as rad air in temp, same thing, flow and fan cfm.
The reason the rad out coolant temp is used is because that's the lowest temp in a loop.
Take a look in our stickies for a lot of great info. Here are a few-
Radiator Thermals and Radiator Calculators by Martin
Huge Radiators: Roundup
Great stuff from one of the pioneers Bill Adams-
An Evaluation of the ThermoChill Radiators | Overclockers
Anything you can find to read by Cathar, Bill Adams and the RoboTec (Lee Garbut) is time well spent.
This is very interesting (especially for the serious benchmarker) and is a different experience than posted by others in this thread (that state they do not experience any difference in CPU temp using different loop orders). Maybe it's only measurable at extreme overclocks? or so small an effect that it is hard to meassure? Anyways it encourages one to experiment. .
Sure That's what I generally tell the guys is just try it and see, that's half the fun. Use what works for you and keep the other in mind for possible future use. And always keep an open mind.
I understand your points but it is I believe the only test available that looks into this matter and even though not conducted using an ideal test setup it still points in the direction that the temperature drop over a radiator is small which the theory backs up.
If you'll look at those links I left you almost any rad testing worth anything will show your rad out to ambient DeltaT, that's far from the only testing. You have to keep in mind that your rad out/air in DeltaT above ambient is going to be a multiple.
A real rough way to get an idea at least would go like this. If your loading at 25C above the room ambient and your rad out to ambient DeltaT was 5C that would equal 25C divided by 5, or 5C for every 1C of the DeltaT rad out/air in. I would think that would be a rough approximation anyway, maybe..
If you were loading at 30C above room ambient and your rad out was 3C each C would be equal to 10C, it would be a higher DeltaT ratio because of the lower rad out DeltaT because your rad/fan and general loop set up was more efficient at dissipating heat.
Just about the only modern tester that I have had any confidence in is Martin, to many of the others are just in it trying to get free parts and money, their just hacks.
No dis-respect to the others, but to me reading test results is just a way to kill time. Even the best are going to have a +-20% accuracy rate and the testing isn't done in a climate controlled environment, so your chances of having the same conditions and results are pretty small, they are at best a very broad out-line of the results that you "may" have with that component.
I did learn a lot from Martin, he was good.
Really WC'ing is simple. Just over-pump (with an acceptable heat dump) over rad, over fan and get a decent block and you'll have about as good of temp's as anyone. Just follow the basics setting up, keep tubing lengths and fittings as few as practical and do 4 or 5 re-mounts and work back down to your lowest, simple things like that.The rest is just playing around and personal preference mostly.Edited by ira-k - 8/23/10 at 11:44pm