Chylld's guide to watercooling

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Thu Nov 11, 2004 4:51 pm

Bishamon wrote:As for W/C liquid, destilled water can be mighty expensive...
I wonder why distilled water is such a rarity in Europe. In the US, distilled and de-ionized are usually about the same price, and dirt cheap at that.

Just checked the tag on one of the jugs of DI that I have around here: $0.65 a gallon.

That works out to 1.08 Kroner per liter, if I'm doing my math right. (or about 0.13 euro's)
[size=75][b]Senior Contributing Writer, SPCR[/b][/size]

Bishamon
Posts: 7
Joined: Fri Oct 29, 2004 3:02 pm
Location: Norway
Contact:

Post by Bishamon » Thu Nov 11, 2004 4:59 pm

Rusty075 wrote:Just checked the tag on one of the jugs of DI that I have around here: $0.65 a gallon.

That works out to 1.08 Kroner per liter, if I'm doing my math right. (or about 0.13 euro's)
I payed NOK20 per liter for mine.

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Fri Nov 12, 2004 7:23 am

Cheap and effective is to use what cars use: glycol
Nothing grows in something that poisonous, it inhibits any corrosion activities and helps make the water less conductive. Dirt cheap too. I use 1 part glycol to 3 parts distilled water plus a miniscule pinch of detergent to counter the remaining surface tension.
Am I missing something here?

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Fri Nov 12, 2004 2:42 pm

not really. if it works well for you then there really isn't any need to change it :)

the thing about glycol (and most other additives) is that it increases the viscosity of the coolant. what this means is that when the coolant passes over the inside surfaces of the waterblock(s), it'll have a tendency to 'stick' to them and thus not move over these surfaces as quickly. this results in less turbulence in the coolant adjacent to the waterblock surfaces giving lower heat transfer and thus poorer performance.

Straker
Posts: 657
Joined: Fri Jul 23, 2004 11:10 pm
Location: AB, Canada
Contact:

Post by Straker » Sat Nov 13, 2004 1:49 am

same as what chylld said, but also, glycol isn't really optimal; why reduce performance if the coolant will never even be below ambient, let alone below 0C?
and polyethylene glycol is beyond safe... not really sure if it will make a difference to stuff trying to grow in there, but PEG is used in pretty much all cosmetics, lotions etc and it's safe enough to eat.

eidolon
Patron of SPCR
Posts: 37
Joined: Tue Oct 26, 2004 2:08 pm
Location: in the kitchen stirring the pot (Ohio, USA)

Post by eidolon » Tue Nov 16, 2004 1:37 pm

I work within an industry which works within the heat removal realm.
Cooling is cooling.

Has there been any discussion regarding Reynolds numbers? :shock:


Your systems will be better off with no tee's. Manifold everything.
How efficent is your setup?
From what I know basically you want a +3°F delta between your inlet coolant temperature and the outlet temperature. This temperature differential is best achieved by regulating volume and/or pressure while maintaining a turbulent flow.

About using eythl glycol... using it straight is absolutely worthless.
By itself it has worse heat removal properties than plain ole tap water. There is a reason as to why the manufacturers have the consumer mixing the stuff.
[size=84]Fong Kai case (thanks silentpcreview), A8V Deluxe board AMD 3500+ (winchester), Radeon 9700 Pro, Zulman 7000b alcu, 120mm exhaust fan - Nexus, SilenX iXtherma 350w psu, SpeedFan[/size]

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Tue Nov 16, 2004 3:42 pm

there has been some light discussion regarding reynolds and nusselt numbers here and there, but nothing in depth.. simple terms like "flow rate" are quite adequate for our needs :)

i'm interested when you say you want a 3F delta between inlet and outlet temps. i'm not doubting you or anything, i'm genuinely curious. is such a dt indicative of optimum cooling? from my experience (and that of others) the dt has been less than 1c in the best high flow rate systems.

Straker
Posts: 657
Joined: Fri Jul 23, 2004 11:10 pm
Location: AB, Canada
Contact:

Post by Straker » Tue Nov 16, 2004 4:23 pm

eidolon wrote:Your systems will be better off with no tee's. Manifold everything.
Think most people know/would agree with that, it just makes setup more of a hassle, and you're screwed if it's not completely bled, or when you eventually need to replace coolant lost to evaporation. Better a tee than a res...
eidolon wrote:About using eythl glycol... using it straight is absolutely worthless.
By itself it has worse heat removal properties than plain ole tap water. There is a reason as to why the manufacturers have the consumer mixing the stuff.
noone uses it straight, that'd be hilarious :)
chylld wrote:i'm interested when you say you want a 3F delta between inlet and outlet temps. i'm not doubting you or anything, i'm genuinely curious. is such a dt indicative of optimum cooling? from my experience (and that of others) the dt has been less than 1c in the best high flow rate systems.
Just depends on the length of loop, the rads/airflow and the pump... lower is better until you're using a pump that's adding more heat to the water than the extra flow is helping to get rid of. Considering the precision (relative to fans, at least) to which we know the pq curves of pumps, the C/W and pq curves of rads/blocks and the power consumption of different pumps it shouldn't be that hard to simulate and find the optimal pump for any given setup, but I have no idea how I'd go about doing this. :)

edit: what do you know, Cathar just did that.

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Wed Nov 17, 2004 10:25 am

Cathars article is excellent and well worth reading.

His conclusions are a bit off for SPCR though, because Procooling serves the o/c crowd mostly. I believe the numbers serves as final proof that all this endless talk about flow rate is just a nerdy waste of time. Unless you're at the overclocking world championship finals, who the **** cares about one or two degrees CPU temp difference???
Focus when building a silent system should be on how to dump the heat silently and how to isolate the pump.

Note also that flow rate has a very small impact on the radiator's efficiency even with that 35 dBA fan he used in his calculations. With much lower cfm fans, e.g. 4412s @4V, then of course higher flow is completely uninteresting.

Great article link. I shall use it to bash overclocking must-have-biggest-pump-coolest-block penis-measuring schmucks in their heads.

BillA
-- Vendor --
Posts: 10
Joined: Sun Aug 24, 2003 10:04 am

Post by BillA » Sat Feb 19, 2005 2:50 pm

snutten
if you would preface your posts with this quote (from you);
"who the **** cares about one or two degrees CPU temp difference???"

this will enable others to understand your perspective
you are on noise, that is fine - for you
others are on performance, and that is fine - for them

try and remember, different strokes for different folks
all your mods decrease noise and performance, others may choose differently

another quote of yours:
"Great article link. I shall use it to bash overclocking must-have-biggest-pump-coolest-block penis-measuring schmucks in their heads."

so far you have demonstrated several things about yourself

BTW, manifolding in WCing systems is hydraulic 'overhead' only decreasing performance

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Sat Feb 19, 2005 3:05 pm

Actually Bill, considering what site this is on, all posts are pretty much prefaced with that sentiment. :wink:
[size=75][b]Senior Contributing Writer, SPCR[/b][/size]

BillA
-- Vendor --
Posts: 10
Joined: Sun Aug 24, 2003 10:04 am

Post by BillA » Sat Feb 19, 2005 3:16 pm

Rusty
I do appreciate that this is SPCR (and a nice place it is),
and I DO like the specific focus

snutten however is promoting a nonsensical discourse by means of 'loose' terms and mythical values
I'm just suggesting that he concentrate on the quality of his numbers, his conclusions might be different
- and in any case more credible

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

best tubing clamps

Post by frankgehry » Sat Aug 20, 2005 10:12 pm

I'm planning to use tygon 1/2 id x 3/4 od tubing. Maybe Clearflex. From searching through some of the post at procooling it seems that most people prefer the metal spring type clamps. The plastic clamps that edward ng uses here look pretty good too. What would be best? Is using two of the plastic clamps a good idea? Is it a good idea to protect the tubing by using a piece of tape between the clamp and tubing? TIA - FG
[size=75]
[color=#3366ff]e8400, Corsair XMS2-5400, Asus 8400GS, Gigabyte EP45-UD3R, Chenbro SR209/Nexus 120, T-Balancer, Enhance ENP-5150GH 500W, Seagate 7200.7 160MB [/color][/size][size=75][color=#cc0033](Retired: XP-120, FSP530-60GNA, Antec SP2.0 500W, Antec SLK3000B)[/color][/size]

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sat Aug 20, 2005 10:31 pm

tygon and clearflex are both good, the price of each will probably be the determinant factor. if you're rich you can run clearflex/tygon throughout your whole system, but you can also get some cheap pvc tubing from the hardware shop to span long distances if you're running an external cooling box (that method saved me a good 20 bucks)

i'm not quite sure what you mean by metal spring type clamps or edward ng's plastic clamps - if you could post some pics or paste some links that'd be awesome :) but with the metal ones, if you're referring to the wormdrive clamps that you screw to make tight, i'd highly recommend using those with 1/2" ID tubing because plastic ones (i've only tried a few different types) have never worked for me with my 1/2" ID clearflex (not enough strength, the tubing is too stiff.)

protecting the tubing from the metal clamps is a sensible idea, personally i don't do it but i don't see any harm in it. with 1/2" ID / 3/4" OD tubing, the tube wall is a good 1/8" (3mm) thick and i prefer tighting my clamps until they actually start to eat away some of the tubing, that way i know they've got a nice tight grip :) if you don't want to mar your tubing though you can either put something in between the clamp and the tubing, or just refrain from tightening the clamp to the point of damage.

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

Post by frankgehry » Sat Aug 20, 2005 10:48 pm

Image
I first noticed these from the article - Project: Double header 3 at procooling in the article section.



Image
These are similar to those that ed used. He said he used a pair of pliers to make them tight enough not to leak. I would really like to avoid leaks, even as a beginner if possible. Thanks - FG
[size=75]
[color=#3366ff]e8400, Corsair XMS2-5400, Asus 8400GS, Gigabyte EP45-UD3R, Chenbro SR209/Nexus 120, T-Balancer, Enhance ENP-5150GH 500W, Seagate 7200.7 160MB [/color][/size][size=75][color=#cc0033](Retired: XP-120, FSP530-60GNA, Antec SP2.0 500W, Antec SLK3000B)[/color][/size]

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sat Aug 20, 2005 10:59 pm

yep both of those will probably be fine, and neither of them would really require protection for the tubing. these are the worm drive clamps i talked about in my last post:

Image
(page 212 of the mcmaster-carr catalogue)

the only problem i foresee with the spring metal clamp you show is that it'll have a set tightness, beyond which it can't provide any further strength. the plastic clamps alleviate that problem somewhat since you can use pliers as ed did to tighten them. i would have used the plastic ones but i couldn't find a suitable size for 3/4" OD tubing, hopefully you'll fare better!

also, i think with both of these they have to be inserted over the tubing before you actually connect it to the waterblock/radiator/whathaveyou. worm drive clamps can be fitted onto existing tubing easily, but they cost a fair bit more. if you can find the right size of any of these 3 types of clamps, however, i doubt you'll have any problems. good luck!

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Sun Aug 21, 2005 7:26 am

Just been plumbing/filling my first watercooled system, and I had a big leak on the video card that was driving me nuts until I removed the clamp I had (worm drive clamp like above - known as a Jubilee Clip in the UK) on it. Turns out one of those fitted to the Maze 4 block I had (1/2" ID, 3/4" OD Tygon tubing) was causing a leak. Once it's removed it doesn't leak any more.
Just running the pump/fans for 24h to check for any more leaks and give the video card a chance to dry out before I boot it up again and see if I've killed anything in the process (video card did get thumped once or twice... nothing fallen off, although there are a few loose capacitors :shock: ).

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

nylon tubing clamps

Post by frankgehry » Sun Aug 21, 2005 11:06 am

Apparently there are two variations of the plastic tubing clamp.
Image
The inside surface that contacts the tubing dovetails at the overlap to make a smooth clamping surface.

Image
In this one the, the inside diameter is made smooth by tapering one of the overlaping ends.

I'm not sure if it really matters but I like the first one best. The first type is sold at www.voyeurmods.com. - FG
[size=75]
[color=#3366ff]e8400, Corsair XMS2-5400, Asus 8400GS, Gigabyte EP45-UD3R, Chenbro SR209/Nexus 120, T-Balancer, Enhance ENP-5150GH 500W, Seagate 7200.7 160MB [/color][/size][size=75][color=#cc0033](Retired: XP-120, FSP530-60GNA, Antec SP2.0 500W, Antec SLK3000B)[/color][/size]

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sun Aug 21, 2005 4:04 pm

keep in mind that you may not be using all of the teeth when these clamps are tightened, so with the first one there will probably be a small section of hose that isn't physically touching the clamp. but that's a small detail and i'm sure either one will work fine :)

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

Post by frankgehry » Sun Aug 21, 2005 5:38 pm

Good point - Thx
[size=75]
[color=#3366ff]e8400, Corsair XMS2-5400, Asus 8400GS, Gigabyte EP45-UD3R, Chenbro SR209/Nexus 120, T-Balancer, Enhance ENP-5150GH 500W, Seagate 7200.7 160MB [/color][/size][size=75][color=#cc0033](Retired: XP-120, FSP530-60GNA, Antec SP2.0 500W, Antec SLK3000B)[/color][/size]

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Tue Nov 29, 2005 1:18 pm

We also aughta warn folks about the possible future problems with tubes going old and cracking and stuff like that.

Bill A: Sorry, no insult intended. But you have to admit that if you actually DO care about two degrees temp-rise in your CPU, then you have a phletora of informative sites to chose from that serve your needs. Please let's not turn SPCR into yet another one!

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Wed Jan 04, 2006 6:26 pm

Oh, another thing always coming up in the forums is the pump vibration problem. Rubber studs, cellfoam pads and similar isn't good enough if you aim for silence. I keep recommending folks to completetely de-couple the pump by hanging it in long silicon cords and using long soft water tubing at inlet and outlet.

res0r9lm
Posts: 22
Joined: Tue Mar 27, 2007 12:11 pm

Post by res0r9lm » Thu Apr 19, 2007 8:21 pm

heater cores are designed for high pressure fans not the best choice if you are looking for silence

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Fri Apr 20, 2007 1:12 am

res0r9lm wrote:heater cores are designed for high pressure fans not the best choice if you are looking for silence
they are designed for high pressure fans, yes, but they work equally well with low pressure fans. in my tests with 2 papst FGLs @ 5v, i got MUCH better performance with a heater core than with a black ice radiator (which has a much more sparse fin arrangement.)

res0r9lm
Posts: 22
Joined: Tue Mar 27, 2007 12:11 pm

Post by res0r9lm » Wed Apr 25, 2007 7:35 am

what model blackice rad do you use? some are for high output fan over 80cfm.

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Wed Apr 25, 2007 1:31 pm

res0r9lm wrote:what model blackice rad do you use? some are for high output fan over 80cfm.
i've tried both the pro and the extreme variants. neither of them held a candle to the heatercore in performance at both max fan speed or undervolted.

add to that the fact that a heatercore cost me almost half what the black ice extreme costs, and i have a very hard time recommending the black ice products.

res0r9lm
Posts: 22
Joined: Tue Mar 27, 2007 12:11 pm

Post by res0r9lm » Sat Apr 28, 2007 10:34 pm

I have the 240 extreme and that likes either a powerful fans or quieter fans in a push pull setup but I got a 360 gt stealth and their suppose to be better with low cfm. they are thinner with more fins per inch. I wanted to have my radiator mounted in such a way that it expelled heat out of case so a heatercore wouldn't would with my situation

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sat Apr 28, 2007 10:57 pm

res0r9lm wrote:I have the 240 extreme and that likes either a powerful fans or quieter fans in a push pull setup but I got a 360 gt stealth and their suppose to be better with low cfm. they are thinner with more fins per inch. I wanted to have my radiator mounted in such a way that it expelled heat out of case so a heatercore wouldn't would with my situation
the good thing about going with an automotive heatercore is that you have a much much much wider range of shapes and sizes to choose from. you have to keep in mind that every blackice radiator is essentially a lower-density heatercore, and as such, you are paying a 50-100% premium for the brand name and some fan mounting holes.

the fan mounting holes themselves aren't actually very good; mounting the fan to the radiator directly (be it an automotive heatercore or a blackice radiator) will result in a dead spot directly underneath the fan hub, where very little airflow will occur. optimally the fans should be separated from the radiator by at least 1 fan hub's diameter to remove this dead spot, however this is often only practical in external watercooling setups.

the other marketing ploy you have to be careful of is when watercooling radiator companies advertise certain models as performing better or worse with high/low airflow. certainly, the only data i have seen to date suggests that, within reasonable limits, no matter whether you are running a high-speed fan or a low-speed fan, a heatercore/radiator with a denser fin arrangement and thus a higher surface area will yield better cooling results.

porky133
Posts: 11
Joined: Mon May 21, 2007 7:19 pm

fin spacing

Post by porky133 » Mon Sep 03, 2007 8:09 pm

I know this is old but there is so much goodness here.

If more fins on the rad equals better cooling then how would fewer fins with more spacing (to let air through more easily) work out?

Wouldn't it be quieter with the air not being forced into smaller spaces? Or possibly even run a lower rpm to maintain same airflow through the rad?

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Re: fin spacing

Post by chylld » Tue Sep 04, 2007 1:26 am

porky133 wrote:I know this is old but there is so much goodness here.

If more fins on the rad equals better cooling then how would fewer fins with more spacing (to let air through more easily) work out?

Wouldn't it be quieter with the air not being forced into smaller spaces? Or possibly even run a lower rpm to maintain same airflow through the rad?
i have to say I was quite surprised to get an email notification from this thread :)

airflow is a very complex subject and it is extremely hard to predict how air will behave when forced through either thinly or thickly spaced fins. from experience though, i can tell you that:

1) air flows through dense radiators quite easily, even in low airflow conditions (I'm running 2 undervolted low-speed papst 120mm fans)
2) i can't hear the air at all :)
3) i tried a radiator with much more space between the fins, and the cooling performance was so much worse that i had to turn the fans up anyway, resulting in a higher noise level and higher temps.

there are a lot of preconceptions floating around wrt airflow and thin/thickly spaced fins that are probably best left at the door :)

Post Reply