full test
 

Hello boys (and girls) !

I am disappointed to see partial results in most of tests I can find on the net, even on overclockers.com

What I mean is that results should be sufficient to simulate any situation to estimate reality.

In this way, what we need is some curves like :

C/W vs flow for rad and waterblocks
pressure drop vs flow for all stuffs

with this 2 curves, we can determine flow in a system, C/W for the given flow, and finally Delta T between CPU and water if you are able to determinate the power of heating from your CPU. Do the same for radiators and you can estimate Delta T between CPU and air, which isn't very accurate, but gives an estimation which is useful to compare with other simulations to determinate the best assembly or component.

pressure drop vs flow is given by manufacturers with a good precision for most pumps

C/W vs flow is given on very high level websites only
Some other good websites give Delta T between CPU and water vs flow, which depends on the power of the CPU, so not really useful. All others gives some useless data like temp from their CPU...

These very high level websites give only a pressure drop for one flow, often 1 GPM, which isn't sufficient as real pressure drop vs flow curve isn't of h=K*Q² type, but h=K1*Q²+K2*Q which need at least 2 points (plus origin) to identify K1 and K2, and more to be more accurate.

Maybe it seems useless for you if you think only high flow systems are powerful and high or moderate power pumps are necessary, but in Europe we are interested by another way (not all, we are divided into 2 tendencies) : small pumps and low flow. In this perspective, our optimisation is focus on taking just the necessary.

On this way, simulating things is useful to determine which pump is suffcient, which diameter of pipe is sufficient (since bigger are always better for performances, but difference between 1/2" and 1/4" ID is not signifiant with low flow) etc.

So I wonder if some websites are already doing this, and why others aren't doing this to give full results ?

Sorry for my poor english (like most of frenchies, shame on us), hope it's understandable.

I believe that pH is in the process presently of getting/using something for PD figures.

all here
www.thermal-management-testing.com

Swiftech does such testing, check some of the product descriptions
http://www.swiftnets.com/products/mcw6000.asp (scroll halfway down)

"Performance: Close to 2°C better than the previous 5000 & 5002 series"
Perhaps someone should add a heatload there, bill ;)? 60w compared to 110 makes quite a difference, id imagine.

/OT

all at 100W
this is the 'shorthand' that seems to have evolved,
a difference in the C/W can be stated as a temp if at 100W (i.e. 0.020°C/W = 2.0° @ 100W)

Bienvenue!

pHaestus does plan to run the flow test, but no, very few people run it.

The main issue there is that even if you have the info, you won't find it for the radiator, save those tested by Bill, here:
http://overclockers.com/articles481/

Tubing pressure drop can easily be calculated, twists and turns aside.

I know this one but it isn't updated (I know why BillA)

I didn't know swifttech have this data on his product's web page. It's a good point for it, and in fact all serious manufacturer should do the same !

But, I didn't see any other tester doing same thing, for now you're the only one BillA (to my knowledge) and I don't know why when I see some curve on flow or pressure drop which indicate that tester has all he needs to do it.

Some people are testing radiators... but appart from BillA, I never saw something serious.

Rosco on www.cooling-masters.com build the pressure drop vs flow curve for radiators (in fact he did it for the only one he tested for now), but he doesn't have the tools needed for C/W mesure to mu knwoledge...

For tubing pressure, I have a good formula to calculate it, so it's the less useful, but in a perfect world, some good tubes should be tested since rugosity is different for all products, even if materials are the same, so it introduces a little incertitude. But it's not really significant since you choose a tube for his price, disponibility or "radius of curvature" (don't know if they are the good words) but not for their pressure drop, so we can expect you use always same tube and we are doing the same error on each calcul and so, it doesn't matter.

Yeah, that's what I meant ;)

Do you use Darcy or Hazen-Williams?

the difficulty is the equipment having the requisite capability, and the time
even using an environmental chamber as we do now, the time required is prodigious
I'm testing rads now and its 2 hours for a 'hot' change (i.e. a simple fan change) to 5 hours from a cold start
- who has that kind of time ? (why I'm here on Saturday posting on procooling, lol)

What a surprise Uncle'Buzz http://forum.hardware.fr/images/perso/ddr555.gif

dare I hope that procooling is becoming French-friendly ??

Yep, revolution's coming ;)

Colebrook

:p Roscal (le chacal huhu)

Private Joke : va publier ton dossier :mad: ! et fissa ! (not really a joke in fact lol)

Vive le Quebec libre, LOL! (no, I'm a federalist, really!)

where is the Bastille ? :D

Serious manufacturers maybe ?

I know it's difficult, but it could be amazing to have all this data !

why I know more is because I test more, quite simple really
if you notice the difference between my statements and 'commonly accepted knowledge', the basis for it is testing
I cannot discuss our R&D activities, nor does Swiftech publish data on products we do not sell;
- but I will eventually do a comparative report similar to that done on pumps

the 'problem' is that many discount such testing by a mfgr alleging bias, etc.
(OCAU being a prime example, but other sites as well)
why should I bother when that is the general reaction ?

while there is a ton of crap on some sites, and no performance data at all on others;
I am bummed at being put into the same group

is it possible to envisage you could reveal some of your data in private ?

Any idea on how to get around the bias? Maybe a collaborative effort with a third party? (i.e. pHaestus)

Otherwise I think you suffer from bias because of where you chose to put up your articles: Overclockers. You have your own site, why didn't you just stick with it?

You'll find a way, I'm sure. In the mean time, here's a new emoticon I picked up in a french Forum, for your usage:

if 'published', the spreadsheets - sure (for example our wbs)
I have provided data to many
unpublished R&D stuff, no
this is a road map to product design (via deductive reasoning), cannot take the corporate risk
nothing personal, eh ?

that really is singularly appropriate Ben, better as an avatar even
but I don't use such
I put up that site after the fact, as nothing can be found on OCers

so pH can be tarred with the same brush (running dog of Swiftech ?)
no, pH is fine as an independent

LOL Why do they pick on Bill for? He's a nice guy. ;)

The flow rate data is something that is needed for this site's testing. It would be a nice touch. I got one question from you guru's though. Let's say if PH tested with one pump and then switched pumps. What effect would that have? Now what about different head ratings? How do you regulate the flow? When you regulate the flow does the head rating change with the pump? The pressure the pump will produce will still be the same no matter how you regulate flow? Right? hmm...It takes a certain amount of pressure to produce any given flow rate, so the head rating of the pump doesn't matter much with testing. As long as you have a decent pump to encompass a full range of pumps?

I think im confusing myself which wouldn't be the first time.

how does that go ? turn about's fair play ?

is pressure related to "h" ?
Roscal can pound us on this one
j/k