A Beginner's Guide For WaterCooling Your PC

[HotFoot]

Further to mr_fnord's suggestions, I would say that I am very interested in hearing more about the potential drawbacks of liquid cooling. So far I understand:

Cost
Complexity of installation
Risk of leaks

I'm fairly comfortable with those, and that probably has to do with my familiarity with them. Questions I'm still not sure about, which prevent me from thinking liquid cooling is viable for my own use are:

What's required for maintenance?
What are the failure modes?
- What can be done to prevent damage in case of failure?

The first question you allude to already in your first page, but I still don't know what I'm going to need to do on a regular basis while using a liquid cooled computer. As well, I leave my computers running 24/7 and under constant load. What risks do I take leaving a liquid cooled computer unsupervised? Is pump failure a big concern? What about spontaneous leakage? What can I do to make sure anything that goes wrong with the cooling system doesn't result in damage to expensive electronics?

I think these were the main questions of mine that I would like answered if/when you put out an update to the primer.

Cheers,

HotFoot

 

[sviola]

Further to mr_fnord's suggestions, I would say that I am very interested in hearing more about the potential drawbacks of liquid cooling. So far I understand:

Cost
Complexity of installation
Risk of leaks

I'm fairly comfortable with those, and that probably has to do with my familiarity with them. Questions I'm still not sure about, which prevent me from thinking liquid cooling is viable for my own use are:

What's required for maintenance?
What are the failure modes?
- What can be done to prevent damage in case of failure?

The first question you allude to already in your first page, but I still don't know what I'm going to need to do on a regular basis while using a liquid cooled computer. As well, I leave my computers running 24/7 and under constant load. What risks do I take leaving a liquid cooled computer unsupervised? Is pump failure a big concern? What about spontaneous leakage? What can I do to make sure anything that goes wrong with the cooling system doesn't result in damage to expensive electronics?

I think these were the main questions of mine that I would like answered if/when you put out an update to the primer.

Cheers,

HotFoot

I'm no expert in WC, but there are some important issues that must be taken account, regarding to risks:

- Corrosion:

This is a major issue, cause it can damage your pc, as corrosion will impact on the heat dissipation and can lead to leaks.
The major cause for corrosion is oxidation, due to the appearance of algae in the system (hot water tends to make life appears... :x). This can be prevented using a biocide along with the coolant.
Other thing that can lead to corrosion is the use of Aluminium and Copper water blocks in the same loop. These metals along with water create a chemical reaction that may lead to corrosion.

- Leaks:

One thing that should have been mentioned in the article is to test your setup before mounting on the motherboard. Just set all outside of the case and run in to make sure there are no leaks.
The use of non-conductive coolants diminish the possibility of major issues if a leak happens as well, the con is that these usually have worse heat transference that water and distilled water (which is the best for wc).
If a leak happen and your pc gets wet, turn it off imediately and wait for 24 hours before turning it on, so you can make sure it has dried up you can use a hair blower to dry it also). Even then, your PC may not come back to life.

 

[cleeve]

What's required for maintenance?
What are the failure modes?
- What can be done to prevent damage in case of failure?

Maintainance mostly consists of paying attention, inspecting the coolant regularly for contaminants that would indicate corrosion, and replacing the coolant once in a while (every year or so).

Talking about the danger of mixing aluminum/copper components is something I'd definitely include if I could re-write the piece, the only real regret I have is not mentioning it.

Failure modes... not sure what you mean by that. There would be two main types of failure though, either the coolant would stop pumping or it would begin to leak.

If it stops pumping, there are failsafes. As I mentioned there are options that will shut your PC down in the event of overheating, hell most people have them in the motherboard's BIOS temperature monitor. But there are others, in the case of the EXOS2 you can wire it to the power switch of the mobo to shut down in case of failure.

Leaking... well, that's more of a problem. In the first place an ounce of prevention would equal a ton of cure, you should be very careful when setting up to make sure your fittings are tight & proper.

If it actually decides to leak, well, that's a pickle. you could get non-conductive coolant if it was a real worry for you - if that leaks it won't harm the system. Or you could mount the motherboard in such a way that leaking coolant wouldn't touch the mobo, but that's pretty extreme. But if leaking conductive coolant actually happens there's not too much you can do, there's probably going to be some dead hardware.

 

[Labrat636]

Wikipedia has a good article about heatpipes, and explains how they can work over a broader range than the working fluid at atmospheric pressure.

http://en.wikipedia.org/wiki/Heatpipe

An interesting comparison would be heatpipes v watercooling. Heatpipes have many orders of magnitude greater thermal capacity in their operating range, but water coolers have much more effective dissipation with their external radiators.

Thanks for the link. I had never heard of heat pipes before - I like the idea of the thermosiphon http://en.wikipedia.org/wiki/Thermosiphon

I wonder how big it would have to be to be effective?
Anyone know if this has been tried on a PC processor?

 

[HotFoot]

Cool, thanks!

 

[sviola]

Maintainance mostly consists of paying attention, inspecting the coolant regularly for contaminants that would indicate corrosion, and replacing the coolant once in a while (every year or so).

Talking about the danger of mixing aluminum/copper components is something I'd definitely include if I could re-write the piece, the only real regret I have is not mentioning it.

Well, you can add a part two and write about it.
I really enjoyed your article (text was objective, nice written and easy to understand) and would like to see a series on it.

You could write next a "Dangers and Pitfalls of watercooling".
Then a "How to choose a setup: going with a kit or a DYI?", where you could review some kits and add some comparisons with some member's DIY setup. Also could discuss the diferences of a setup with a single loop and Y conectors or a 2 pump/loops setup.
You could also write an article on coolants and tubing (diferences in coolants, water and distilled water, impact of tubing on flow).

One article that would really be great would be something on pumps...how does it work, how to choose for your setup, etc.

 

[HotFoot]

I like that idea. Oh, and once you're done, package it all together and call it a book! I'm being serious, not sarcastic.

 

[HotFoot]

Thanks for the link. I had never heard of heat pipes before - I like the idea of the thermosiphon http://en.wikipedia.org/wiki/Thermosiphon

I wonder how big it would have to be to be effective?
Anyone know if this has been tried on a PC processor?

The phase-change thermosyphon is very similar to the heatpipe, but it requires gravity for liquid feedback. There is such a cooler on the market: the Asetek VapoChill Micro. The big difference is the lack of a wick inside the tubes, and the diameter of the tubes. I suppose that, technically, a thermosiphon would have a separate return path, to form a loop instead of a pipe, but that might just fall under design variation.

 

[cleeve]

Actually, I'm thinking of writing an article specifically on the VGA card cooling - stock cooling VS premium air cooling (Thermalright HR-03), vs. water cooling - and really hone in on that.

To be honest, I prefer to split up complex subjects into more than one article...

It's difficult to make an article accessible to the masses if you make it too cumbersome, or delve too deeply into the minutia. That's why articles targeted at an uninformed audience rarely work with the knowledgable audience, and vice versa.

I've written articles that focus on a small part of PC tech, and they get great feedback from the hardcore, but newbs find them intimidating and don't even read them. So there's always a balance you have to strike, you have to really know what you're trying to accomplish and focus on it. But you can't please everybody, someone's always going to want more - or less - than you give. Part of the writing biz I suppose.

 

[sviola]

Actually, I'm thinking of writing an article specifically on the VGA card cooling - stock cooling VS premium air cooling (Thermalright HR-03), vs. water cooling - and really hone in on that.

To be honest, I prefer to split up complex subjects into more than one article...

It's difficult to make an article accessible to the masses if you make it too cumbersome, or delve too deeply into the minutia. That's why articles targeted at an uninformed audience rarely work with the knowledgable audience, and vice versa.

I've written articles that focus on a small part of PC tech, and they get great feedback from the hardcore, but newbs find them intimidating and don't even read them. So there's always a balance you have to strike, you have to really know what you're trying to accomplish and focus on it. But you can't please everybody, someone's always going to want more - or less - than you give. Part of the writing biz I suppose.

Yeah...I know, you can never please everyone. But you can always increase the complexity with a "chapter" format. Start easier for the uninformed ones and increase gradually, until you can reach the knowledgeable crowd. This way you not only post info that appeals to all sometime, but you create a bigger knowledgeable crowd. (mmm...I think that this is what they call a course ... lol).

 

[bkiserx7]

Besides the fact that the graphs are not there, lol. Is that not an overkill of a PSU lol.

 

[cleeve]

Is that not an overkill of a PSU lol.

Yeah, that 1200w beast is crazy. I'm running it against two other 1200 watters - thermaltake's and E-power's - in an upcoming PSU review.

 

[bkiserx7]

Is that not an overkill of a PSU lol.

Yeah, that 1200w beast is crazy. I'm running it against two other 1200 watters - thermaltake's and E-power's - in an upcoming PSU review.

I'll handle the 2000w microwave review once you complete that.

 

[Kurz]

Non Conductive Fluids like FluidXP suck.
They leave residue and eventually become conductive later on as they pick up ions from the blocks.
Distilled water is great as long as you clean out the parts before running water through them. Because of residue from the machining process isnt cleaned out. Alcohol 91% and distilled water is great to clean stuff up.

Mixing metal is a no no... either stick with copper and brass or get all aluninum.

Another thing... Its better to pick your own parts. Koolance being a good company, doesnt have the best blocks or best radiator for the money.
They also use weak pumps and small ass tubing that slows down flow.
Low flow = low performance...
And check out all those 90 degree metal elbows around the blocks... those kill flow as well.

Frankly that setup is pretty crappy... The tubes could've been cut shorter, less tubes = better performance. Even if it takes cutting a few holes in the case it'll increase performance.

I put together these parts,
CoolerMaster Stacker - case
yate loon fans cheap fans that push alot of air and are silent.
Fuzion Dtek - CPU water block
MCW60 - GPU water block
Some Memory heatsinks
PA120.3 radiator - Thermochill
EK resviour - which feeds the pump
DDC+ Pump
some Tygon 1/2in 2075 tubing www.mcmaster.com
Distilled water of course,
and some algecide

All that cost me only... $350
And it gets better performance than Koolance's setup.

You can read more at the forum.
http://www.xtremesystems.org/forums/forumdisplay.php?f=70

 

[HotFoot]

I keep coming across statements that 90 degree bends and smaller tubing significantly restrict flow and reduce performance, but I've never seen data to back it up. There have been threads where I've shown, at least analytically, there to be negligible impact of using smaller tubing and 90 degree bends. The restriction to the flow in good water blocks is great enough that the losses in the tubing should be negligible.

A 90 degree bend in 3/8" tubing causes about 1/50 the head loss as a Swiftech Strom waterblock. Say you have a system with 10 90 degree bends, which is a lot: the flow degredation due to the bends is still small compared to only one of the functional loop components.

If there is some emperical data on the subject, I would be very interested to read more on the subject.

 

[joex444]

Once again, this is a primer for the uninitiated, not a how-to overclocking article. It's also not a premium air cooling vs. water cooling article....
If you guys would like to see a premium air cooling vs. water article, let me know and I'll consider it for the future.

It's just that if you are considering water cooling you aren't new to the PC world, and the only reason you would want to try water is if the aftermarket air coolers aren't cutting it for you. Quietness wouldn't get involved for the price.

So to compare Intel's stock cooler against water isn't really useful. Kinda like comparing a Veyron to a Camry. It would be better to see how more effective water is than what the best air coolers can do. Like how the Tesla roadster should be compared against gasoline sports cars, not the EV1.

Also, how did you measure these temps? 62C on a Core2Duo measured in the core through TAT isn't a problem, it would be if that were through the BIOS or a program that reads the BIOS temp, like MBM.

 

[gm0n3y]

Nice article. I don't want to get involved in the drama that seems to be unfolding here (at least no "fanboi" comments yet), but I have a few comments.

I've never tried water cooling before, I have looked into it, but the cost, work, and dangers associated with it have always been a deterent. This article makes it sound pretty easy, and I feel confident that with this, and numerous other (often more technical and in depth) articles, I could do this myself.

The only thing missing from this article is a comparison of the sound output of the 2 setups (air / water). More overclocking benches would have been nice, but the thermal charts pretty much speak for themselves. A list of common problems or some links to other more technically detailed articles would have been a nice addition also.

To me this article is pretty much "A Beginner's Guide For WaterCooling Your PC", so it gave me pretty much exactly what I expected. I didn't expect any crazy in depth benches or elaborate water cooling comparisons, just a simple explaination and an example of how to do it.

 

[HotFoot]

The point isn't at all the few benchmarks. Those are just there to demonstrate the aforementioned pro side of watercooling. If you want to know how well watercooling stacks up against good air coolers there's plenty of those types of reviews around.

 

[gm0n3y]

It's just that if you are considering water cooling you aren't new to the PC world, and the only reason you would want to try water is if the aftermarket air coolers aren't cutting it for you. Quietness wouldn't get involved for the price.

I am considering water cooling my system pretty much for the sole purpose of having a quiet PC. As long as it cools better than the stock HSF, that's good for me. Overclocking potential is good too, but for me, not the main purpose of water cooling.

 

[HotFoot]

For quietness I know of two water cooling systems that have no fans at all. There's the Zalman Reserator (v2 is best) and one by Silverstone that looks pretty sharp. Either of these should do as well as a high-end air cooler, but at several times the price.

For silence, you could also go with a fanless CPU cooler. I used a TT SonicTower on my Pentium D for months in fanless mode. You do need a case with good airflow and room enough to fit the cooler: it's 150mm tall.

 

[SuperFly03]

I basically had the same setup as cleeve about 2 years ago, except my Exos was integrated into the case. I didn't have the WC'd PSU either, but overall on the lowest setting it was damn quiet, barely audible. At high it got moderately loud but no where near what I have heard form some air coolers.

--Kurz

I would love to know how less tubing means better performance. Less tubing would lead to less total water which would lead to a quicker change in overall water temp probably, but I don't see it changing the overall cooling performance.

 

[gm0n3y]

All of the reading that I have done on watercooling, especially in forums, has always led me to believe that it is better and usually cheaper to just build your own setup. Is this true?

 

[skyguy]

Great article for the masses, newbies, and the curious. Well written, informative, thorough, and accessible. Good job Cleeve!


I do think that a "series" of articles would be fantastic. Not only would it provide more info, but also greater depth of info that would appeal to the variety of newbies/enthusiasts. The series could progressively cover more info, and more in depth of specialty areas as well.

This would solve a number of comments/complaints at the same time, provide more info that can't be done in simply one article, and hook readers to come back to Tom's for more and more. Can't argue with increased readership, right?

 

[SuperFly03]

All of the reading that I have done on watercooling, especially in forums, has always led me to believe that it is better and usually cheaper to just build your own setup. Is this true?

If you do your research it will yield a better result, yes.

There are numerous threads (as you have no doubt read) and people always willing to help/comment right before you purchase, if you decide you want a second opinion.

 

[gwolfman]

Non Conductive Fluids like FluidXP suck.
They leave residue and eventually become conductive later on as they pick up ions from the blocks.

Any details/proof to back that up? If FluidXP were to eventually pick up ions, then so would distilled water.