Chylld's guide to watercooling

Gooserider · Post by **Gooserider** » Sun May 16, 2004 4:39 pm

I agree on the finer points question to some degree, but I do think the barb material and tightening concerns are somewhat critical, along with the thread sealant. I say this because I see a great many tales of cracking tops or pump bodies by using brass fittings and over tightening them. (Including not using sealant, then trying to stop a leak by tightening the fitting some more :OOPS: ) When I did a 'disaster survey' over on Pro-Cooling, this was one of the more common sorts of failures.

(OTOH, the question of how a res affects performance IMHO does get sort of esoteric

)

Gooserider

chylld · Post by **chylld** » Sun May 16, 2004 5:00 pm

yeh someone on ocau just reported that their eheim pump inlet cracked, possibly due to overtightening. (although cold weather was also a suspected cause)

thing is, most new watercoolers won't have to screw in barbs anyway. the good waterblocks these days come with barbs attached (mcw5002 is an exception, but it's being superceded by the mcw6000) and putting the barbs on the radiator is the radiator man's job. eheim and swiftech pumps already come with barbs on them, and with good clamping there doesn't need to be any modification. (i'm still using the stock short plastic eheim barbs).

jk1 · Post by **jk1** » Tue May 18, 2004 9:18 am

Generally, you want to keep the same tubing diameter throughout your system

Could you explain the advatages of doing this...are there any? I would think that changing tubing size would sometimes result in an improvement of temps. For example, let's say that I had a 1/2" CPU block and 1/4" GPU block. Downconverting from 1/2" to 1/4" in between the CPU and GPU would result in higher pressure, faster flow, and more turbulence in the 1/4" section of tubing -- seems like an advantage to me, but I would like to know for sure...

Post by **Rusty075** » Tue May 18, 2004 9:46 am

Going to a smaller tubing diameter will increase the flow velocity, but it will decrease the more important flow volume by increase resistance in the system. Turbulance in the tubing doesn't help anything. If you have to use a block that has a smaller fitting than your others, you're better off just reducing the tubing size right at the fitting. The longer distance the pump has to push the water through a small tube, the greater the resistence.

The "general" rule: Use as large a tube as is practical.

chylld · Post by **chylld** » Wed May 19, 2004 4:38 am

HammerSandwich wrote:Adding anything to a loop increases the demands on the aircooling side of the system (i.e. radiator and fan). Either you increase airflow or the water temp goes up. When water temp increases, so does CPU temp. With a good setup, this isn't a showstopper, but no one should approach this blindly.

i think one comment i'd like to add here is that yes, the water temp will go up, but when you consider the efficiency of watercooling it makes sense.

i bring this up because i just changed my motherboard (old one was faulty) and as such got a 300mhz increase in cpu speed (see sig) with no added noise. the water is noticeably warmer, the cpu is a few degrees hotter but there's no additional noise.

Sklug · Post by **Sklug** » Wed May 19, 2004 9:01 pm

I have another question.. how do you mount the various pumps out there? I understand some people seem to use some sorthobane from McMaster to dampen vibration, but how does the pump get affixed to the case so that if I move the PC the pump doesn't move around. I haven't seen any good photos of that anywhere.

chylld · Post by **chylld** » Wed May 19, 2004 10:09 pm

a better option than sorbothane is to use vibration dampening mounts like these ones:

here they are in use, successfully decoupling an eheim 1250 from an aluminium case (you can see them just, behind the pump):

however, these aren't absolutely necessary if you're using 1/2" ID tubing. what you'll find is that the pressure exerted by the tubing on the pump keeps it securely in place, and since the pump is pretty heavy it'll stay where you want it to.

for example, in my old system:

the pump isn't affixed to the case at all, yet i was able to turn the case fully upside down and nothing would move. (not even the res!)

luminous · Post by **luminous** » Sun May 23, 2004 1:45 am

Many thanks for the responses to my comments on the guide. I've been having a few problems with my email and internet connection so I have just read your responses.

I'm currently digesting the additional reading that was suggested

Seal · Post by **Seal** » Sun May 23, 2004 2:49 am

SWEet!! that 2nd pic is a dang nice setup! i love the silver pump looks different. I have those sobarthane pump vibration dampers, ive had them for ages but havent installed them yet lol as im *still* trying to find time to re-do my watercooling setup. im planning on stripping it down and writing a little chylld style tutorial on my own personal setup

.

As for changing tube sizes, you either have a high pressure setup or a low restriction setup but you can never effectively have both so stick to the same tube sizes!

Sklug · Post by **Sklug** » Sun May 23, 2004 3:44 pm

Anyone know of any online vendors that sell those vibration dampening mounts? I assume it't not something I'm going to find at my local hardware store.

chylld · Post by **chylld** » Sun May 23, 2004 3:50 pm

mcmaster-carr has them, i think they ship worldwide

DynamicalSystem · Post by **DynamicalSystem** » Wed May 26, 2004 4:53 am

WizardDesigns in the UK has them too.

jk1 · Post by **jk1** » Tue Jun 08, 2004 7:40 am

I think this guide is great and is full of good information. That said, I am trying to make a decent watercooling system without making any case mods. Currently, I have an Antec P160. The real problem here is that I can't find a good heatercore or rad that will fit the case. As said above the BIX bulges out and doesn't fit the 120mm fan perfectly (same fan position as the Sonata). Is there a good way around this?...or would it just be easier and better to get something like an Exos?...

chylld · Post by **chylld** » Tue Jun 08, 2004 2:57 pm

jk1: the koolance units aren't particularly designed for high performance, nor are they silent (or even quiet, especially by SPCR standards). it is best to keep in mind that all the commercial watercooling units are just that - commercial watercooling units, and as such are built to a price and have certain compromises (performance, silence, etc) that you do not have to make with a DIY kit. That said, if you still want an exos then feel free to buy it, let us know how it goes; i'm pretty sure it won't be a total waste of your time and money.

if i were you though, i'd get the corsair hydrocool unit as that performs very respectably and is my pick of the commercial watercooling bunch. it, like the exos, doesn't require any case modification.

if the BIX bulges out when mounted to the rear 120mm fan, what you can do is take a 4mm high-quality drill bit and drill 4 new holes in the BIX shroud on the side onto which the fan will be mounted. in doing so you can offset the bix by 4-5mm (or more, if you need to) which is sufficient to keep the BIX inside and the side panel close-able. this is what i did with the BQE, which is even tighter for space (BIX-on-rear-120mm-wise) than the sonata.

Sklug · Post by **Sklug** » Tue Jun 08, 2004 9:06 pm

jk1, my watercooling project is a work in progress (as is my web-log of it). I'm also using a P160 case. I did however make some small mods, nothing too difficult though. You can check out my partially completed log of the project. I'm a bit further than this, but haven't updated the page yet. Also, I'm waiting for a motherboard for my socket 939 athlon to become available, so the system is not in use yet.

Excuse Road Runner's lame webserver that doesn't support the latest FrontPage and therefore messes up my navigation bar:

http://home.san.rr.com/klugs/

jk1 · Post by **jk1** » Wed Jun 09, 2004 9:48 am

Thanks for the suggestions...this is probably the most helpful forum that I have ever been involved in...

zoob · Post by **zoob** » Tue Jun 29, 2004 2:51 pm

chylld wrote:for example, in my old system:

the pump isn't affixed to the case at all, yet i was able to turn the case fully upside down and nothing would move. (not even the res!)

Hey, just curious, how were temperatures with that setup? How noisy was it?

I'm interested in W/C but can't decide on either a front, or top mounted 2x120mm radiator in my Antec SX1030. I'm thinking maybe a dual 80mm mounted at my exhausts would be a nice intro for me to water.

chylld · Post by **chylld** » Wed Jun 30, 2004 6:25 am

the noise level with that setup was acceptable, the loudest thing was the radiator fan (antec bqe stock fan @12v) and more significantly the turbulence noise created by it.

temperature-wise it was quite disappointing actually. cpu peaked in the high 50's.. the radiator (a BIX) was clearly the limiting factor here as it was hot to the touch.

TheDarkHacker · Post by **TheDarkHacker** » Tue Jul 13, 2004 9:53 pm

after reading through the guide again i have to disagree with you on a couple points. first the optimal fan directions is not always sucking and is really case dependant. some people may find that blowing provides better performance. it varys from person to person. secondly i have to disagree on saying that pumps put heat into the loop. the heat put in is so minimal that you wont even notice. most of the heat generated by the pump comes from the electro magnet which is another part of the pump. the pumps that people use for water cooling even the most powerful ones are air cooled, not watercooled, they dont add significant amounts of heat to the system, especially when your talking about a 100w heat source such as a CPU. just my opinion

toiletduckuk · Post by **toiletduckuk** » Fri Jul 16, 2004 4:01 am

I'm interested in W/C but can't decide on either a front, or top mounted 2x120mm radiator in my Antec SX1030. I'm thinking maybe a dual 80mm mounted at my exhausts would be a nice intro for me to water.

What about bottom mounting? Is this acceptable? I figure you'd have to blow air down through the rad and out of a hole in the bottom of your case.

In this case, the pushed out air has about 1cm between case and ground.. is this enough clearance for optimal cooling?

cheers

Dave

chylld · Post by **chylld** » Fri Jul 16, 2004 5:59 am

if the radiator's at the bottom of the case it's best to pull air up through it. this way you get fresh air and you let the forces of convection (warm air rises) work with you.

i think you'd need at least an inch clearance for effective cooling, although i haven't experimented with that setup. (yet)

Glitch · Post by **Glitch** » Mon Jul 19, 2004 6:34 pm

for quiet watercooling a heatercore is not what you want. have a look for oil coolers, as these are less restrictive, and will enable you to run alot quieter fans.

secondly a fan sucking, indeed benefits from a shroud, as it forces the fan to pull all its air through the radiator, and also creates an area of lamina flow, which reduces noise.

a fun pushing on the other hand, doesn't benefit from a shroud, as axial fans produce a focused "jet" of air. if you have a fan pushing it would be better to save you money, and mount the fan an inch away, again to reduce turbulence noise.

i believe a fan sucking produces better temperatures because there is little or no "dead spot" and air is drawn at approximatly the same rate across the whole radiator. a fun pushing on the other hand will keep its "dead spot" a factor larger than the diameter, in the case of 120mm fans, i estimate approx 1-2feet. the only way to eliminate a dead spot is to change the air flow direction, which means terbulence, and this cause noise. i believe a fan pushing performs worse than a pull configuration, because a great area of the radiator (center) is subjected to low air flow (dead spot) and the edges (tips of the fan) are subjected to high air flow.

I hope this is informative, and easy to read.

Glitch · Post by **Glitch** » Mon Jul 19, 2004 6:40 pm

chylld wrote:i think you'd need at least an inch clearance for effective cooling, although i haven't experimented with that setup. (yet)

an inch clearence is generally regarded as a good distance to havea fan from the radiator, not because it increases effective cooling, but because it reduces the amount of turbulence noice created by the fan, and the air flowing over the radiator.

chylld wrote:if the radiator's at the bottom of the case it's best to pull air up through it. this way you get fresh air and you let the forces of convection (warm air rises) work with you.

for really extreme cooling, build yourself a 5ft chimley on top of the radiator,. it wont' have any fans, but the air flow because of convection would be pretty significant.

peteamer · Post by **peteamer** » Tue Jul 20, 2004 8:31 am

Whilst on the thought of oil-coolers, certainly on mid 80's Kawasaki's they were designed for low pressure systems (IIRC ~/< 5PSI). I seem to recall Suzuki's were the same, but am open to correction on both.

Pete

HammerSandwich · Post by **HammerSandwich** » Fri Jul 23, 2004 10:32 am

peteamer wrote:Whilst on the thought of oil-coolers, certainly on mid 80's Kawasaki's they were designed for low pressure systems (IIRC ~/< 5PSI). I seem to recall Suzuki's were the same, but am open to correction on both.

Do you mean oil pressure? We're more interested in airflow restriction, I think.

peteamer · Post by **peteamer** » Fri Jul 23, 2004 11:46 am

Indeed I do.....

Surely both are of interest... no point having a cooler that requires a pump the size of an elephant to push/pull the fluid through...

Motorcycle oil coolers are sometimes/often/usually low pressure/high flow designs, just what we're after innit ?

Pete

HammerSandwich · Post by **HammerSandwich** » Sat Jul 31, 2004 7:59 pm

Sorry for the delayed reply.

I'll agree that pressure drop on the liquid side matters, but it plays a far smaller role than the air side's influence on noise. IMHO, the (US & Aus, particularly) WC community has overestimated the need for large water flow rates. Some blocks require much flow to perform well, but many do not.

snutten · Post by **snutten** » Thu Nov 11, 2004 5:42 am

Nice post, chylld!

Some thoughts:
Any heatercore with fins set narrow work best with high cfm fans such as the ones they use on overclocking tests. For silence, use heatercores with less air restriction or it will create a noise even with a 4412 at 5V. (I use 2 x 4412 at about 4V and it still makes a hush from the fins.)

I think you need to address the pump noise issue more thouroughly. Pumps do make noise. It should say so on the "against" list for w/c.
I use a 1048 now, and the type of dampening springs you suggest transfer enough vibrations to make a light alu case vibrate and even if the mounts are really really soft you need soft tubing near the pump to take advantage of it. (My pump is now totally suspended in long elastic cords, touching nothing solid, and I can still hear it if close.)

Also I believe the issue of dumping the heat is somewhat lost on most w/c articles. Focusing on this or that water block with such and such flow isn't going to make the system quiet.
It takes a large water - air heat transfer area. This is the w/c advantage. (I use my whole alu case as a big heatsink.) I believe we should tell our SPCR friends to get several or very big heatercores / radiators. That is the key!

Moreover, the idea that a w/c system is future proof is subjective at best. Not many of us want to throw our old stuff away when buying new. So either way you look at it, the old, or the new components will require new coolers.
Thing is, that by the time you by new there's bound to be something new to cool it with too, making your old cooling set look very un-cool, in a literal sense.

Finally, I think the best thing about going w/c is overlooked: the ability to cool HDDs while sound dampening them. Please add to "pro w/c" list!
Speaking of that, cooling HDDs on the same loop make the whole system play by totally different rules, since water temps then need to be kept below some 35-40 degrees C. I.e. buying a better CPU block is then to absolutely no avail. Obvious enough but just a reminder.

Thanks for taking the time, chylld!

chylld · Post by **chylld** » Thu Nov 11, 2004 2:21 pm

cheers for the response, snutten

i just updated the first post with a couple more paragraphs on choosing a pump, and dealing with noise/vibration etc.

it certainly is key to go with a thick/big heatercore-style radiator, however numerous people just don't have enough confidence in the pulling power of their fans with such a dense monstrosity

low-speed undervolted fans are perfectly capable of pulling air through 2" thick heatercores, and make no more noise than if they were pulling through a thinner, sparser heatercore. and of course you have the performance advantage

watercooling harddrives is something i'm not totally sold on just yet. as you've pointed out already, the temperature to which hard drives will be cooled depends on the temperature of the coolant circulating throughout the loop. with overclocked systems this coolant is likely to be quite hot, and short of deliberately buying underperforming waterblocks there isn't much you can do to stop the propagation of this heat onto the hard drives.

i think a separate watercooling loop is necessary, quite possibly a passive one. i'm using a chinaflo (chinese-made panaflo 80L) at 7v to cool all 3 of my hard drives and their temps never go above 38c.

Bishamon · Post by **Bishamon** » Thu Nov 11, 2004 3:17 pm

As for W/C liquid, destilled water can be mighty expensive...

I used "battery water" instead (£1.5/l at the local petrol-station).

Destilled water is water boiled, steam channeled and cooled, filtered and bottled. "Germ-free" and pretty much "pure" H2O.

Battery water is de-ionized (de-mineralized) water. Basically, same as as destilled water, only not boiled. Not quite as clean as destilled water, but 95% as good at 1/5th the cost.

With proper additives (anti-algae and anti-corrosion) it does the trick perfectly. This is what I use in my Reserator1 om my server...

I need to see how much space my new HDDs will take, but I'm concidering a single-row W/C-system for my workstation (since it already has a 120mm front and back, why not piggy-back a radiator on the outlet).

Chylld's guide to watercooling

Re: Chylld's guide to watercooling