Understanding & Mastering Water Cooling

Discussion in 'Reviews & Articles' started by Dashken, Jan 2, 2007.

  1. strawroot

    strawroot I Lurrrve Panda Biscuit!

    We have updated the article! :wave: :wave:

    Ever wondered about water cooling your PC? Well, wonder no more. AlmostThere will show you everything you need to know about water cooling.

    [​IMG]

    Link : Understanding & Mastering Water Cooling
     
  2. neverthar

    neverthar Newbie

    You therefore do not seem to understand that water blocks are also very dependant on head pressure as more restrictive block design would require greater head pressure to overcome resistance in flow. Reduced water flow in the block would mean there would be saturation in heat in the water which would not be beneficial at all. Which is why Delta charts exist. Otherwise explain this chart borrowed from Swiftech. This is the same chart people use to do comparative analysis of pressure drops versus thermal resistance

    [​IMG]
    Disclaimer:this is an example chart, not verifiable independantly
     
  3. Chai

    Chai Administrator Staff Member

    I think you don't get what he meant.

    The flow rate at the intake and the output of the pump will always be the same, as long as the tubes are in serial, and the total length of the tube in the loop is the same. The pressure drop in the loop is equalised.

    This is the same myth as same people claimed, slower flow rate in the radiator improves cooling, which is not true. The water spend the same amount of time in the radiator regardless of the flow rate, and in fact, higher flow rate = better performance.
     
  4. Olle P

    Olle P Newbie

    Exactly! It doesn't matter where the restriction is added as long as it's somewhere in the entire loop.
    That's a new one (myth, that is) to me...

    The fact is that what happens with higher flow is that the water spend less time in the blocks as well as in the radiator.
    That results in a lesser temperature difference between the "hot" and "cool" water.
    Heat transfer is always improved exponentially with the temperature gradient, which means that in the radiator (as well as the cooling block) it's the water that just enter that provide the most heat transfer, and the longer it "hang around", the less heat is transferred (because the water temperature is closer to that of the surroundings).
    By increasing the water flow you make the water about to leave the component almost as good in heat transfer as the one just entering (because it has virtually the same temperature), thereby improving the cooling efficiency.

    Cheers
    Olle
     
  5. Olle P

    Olle P Newbie

    Comments on part 3 of the article:

    1. Page 13 is missing! (Replaced by page 11.)

    2. Page 11, cleaning radiator and blocks.
    - You tell us to use vinegar for cleaning, but don't say why. Why use vinegar?
    - Given that vinegar is used, then there's no reason to not use (filtered) tap water for the initial rinses, since the acid will remove any remaining mineral stains.
    - Why have the block's base soak in vinegar before polishing and cleaning with alcohol? The alcohol should suffice for cleaning.

    3. Page 11, vinegar in the coolant.
    What should be prevented by a proper selection of coolant is:
    - Chemically altered tubes. (They might be dissolved or swollen.)
    - Growth of algae, fungus or bacteria.
    - Galvanic corrosion.
    - Short circuits in the event of a leakage.

    Vinegar will;
    - have no to little impact on the tubing, if the material is properly chosen.
    - help prevent the growth of life in the system.
    - work as an electrolyte that promote galvanic corrosion. Can be countered by making sure that all metal parts in contact with the liquid are of the same material, preferably exactly the same alloy composition!
    - work as an electrolyte that promote short circuits in the event of a leakage.

    4. Page 12. Checking for leaks.
    One common supplemental test is to simply assemble and run the cooling system outside the computer before letting it close to any sensitive components.
    That will disclose any faulty components with a minimum of fuzz.
    It does not get rid of the need for testing after installation though, as the installation itself may have produced leaks in the tube fittings.

    5. Page 12. Temperatures.
    Not quite true.
    The measurements may (and most often do) have a lousy accuracy, but they also have a fairly good precision!
    This means that as long as you use the same BIOS and software for measuring the temperature the different results are fully comparable to each other when it comes to show differences in temperature.

    /Olle
     
  6. Chai

    Chai Administrator Staff Member

    Those myth are constantly brought up by noobs. hehe. "Myth: Water must slow down to fully absorb heat"

    http://www.overclockers.com/articles1088/

    Well, not completely related to the current discussion. But the thing is, no matter how long or short the tube from the pump to the waterblock, as long as the total length of the loop remains the same, there shouldn't be any major temperature difference (same amount of flow rate and pressure across the loop, and same amount of heat generated by the pump will reach the waterblock), unless there're other factors like altering the waterblock mount (which is highly possible even though you didn't physically remount, because changing the tubing alone will definitely move the waterblock in some ways). Even the most professional water cooling gurus cannot guarantee the same optimum mount everytime they mount the waterblock, which is why they have to remount and retest several times to ensure consistency when testing waterblocks, which I'm sure you are fully aware of.

    Although I have some differences with Olle P, but I fully agree with him on this.
     
  7. neverthar

    neverthar Newbie

    I get pretty damn well what he means but he's deviating very far from original point of argument which is about pump to block head pressure. This is despite the fact it is generally understood that pump pressure at head is vital for many impingement based designs

    Vinegar is used as it is a light acid which is enough to dissolve the residue materials/chemicals which were there during fabrication.

    Rinsing with filtered tap water is to get rid of the vinegar itself unless you think leaving the residue of vinegar inside is very good for you.

    The base as in the inner base of the waterblock to remove any residue of milling coolants whatsoever left as residue, Unless you have a better idea of getting rid of residue in the channels.

    while I agree with Olle on the pros of using vinegar but it was never said to be for long term use. Only for the initial cycle to make sure everything comes out. And it is always assumed never to be mixed in terms of materials as regardless with or without any promoter of ion carriers, such a setup would promote oxidisation (May need to consider editing the article for further fine tuning)

    For testing, why the added hassle of testing outside the CPU? If the CPU is fully disconnected, shorting would not occur. Only a damned fool would plug in the AC plug while testing the watercooling components. Besides that, it is better due to the reason you stated, faulty tubing clamping detection.

    As for the temp measurement, for daily usage maybe it would be ok to use the motherboard monitors but for ndetermination as to exact reduction in temps, it's always better to use proper thermal probes to measure. Older systems tend to have thermal diodes which would be affected by many factors like airflow around CPU socket and such.

    And as for waterblock, I've explained what needs to be explained as I stress again it is a tip for weaker water pumps (Very weak I might add). If you cannot accept such a principle, then so be it. I will not argue or elaborate on any further armchair discussion anymore as I feel this useless and unhelpful for others as it has gone beyond the point of a newbie's understanding which defeats the purpose of the article, which is to help newbies. If anyone of you can propose or write a much better article which would be understandable by 90% of the readers out there, I sincerely welcome and support the idea. I have done the best that I can with whatever limitations that I faced. Whether my efforts are appereciated or not, I do not care. I am only just trying to help and I do not ask for anything in return.

    Any request for assistance or advice in PM I will gladly entertain but I take my leave from this thread.
     
  8. PsYkHoTiK

    PsYkHoTiK Admin nerd

    o_0

    Apart from vinegar, I use ketchup too... :lol:
     
  9. wiLL*lianG

    wiLL*lianG Newbie

    Page 14 "Up Next" missing... is it part 4 continue page?

    BTW, any other can replace vinegar? :D
     
  10. Olle P

    Olle P Newbie

    That type of blocks provide a huge restriction on the flow, thus creating a relatively large pressure drop that increase with the flow rate.
    In order to work properly they still need a sufficient flow, which require a sufficiently powerful pump to overcome the pressure drop.
    But there's no difference in efficiency between a set-up with a block inlet pressure of 50 psi and outlet pressure 45 psi, and one with an inlet pressure of 10 psi and outlet 5 psi.
    That's not what I (nor you) suggested.
    In the article you write:
    1. Rinse with deionised water.
    2. Rinse with vinegar.
    3. Rinse with deionised water again.

    My suggestion is to not use deionised, but regular, water before the vinegar.
    That is a bit unclear by reading the article, especially since you stress that it shouldn't touch the bowl. The inside just won't touch the bowl anyway.
    Assuming that the reader use a full metal water block, why not just drench the whole thing in vinegar? No reason to leave machining residue in the upper half of the block, is it?
    It's most definitely not said to be for short term use. Therefore long term use is assumed.
    Mixed materials is the norm, don't you know?
    - Block bases of "pure" copper.
    - Barbs made of brass, stainless steel or some plated material.
    - Radiators made of copper but with soldering material exposed.
    Mix-up of terminology here? The CPU is a pretty small component, I can't see how anybody could fit a water cooling system in there...
    With push-lock fittings on the barbs testing outside the computer case is a quick and simple way to make an initial check for leakages that save you lots of work if there is a major leak.
    As you point out yourself in the article; the probes in use are "proper", it's just the firmware that isn't properly calibrated.
    If you measure the "before" temperature a couple of times with all settings (CPU work load, room temperature, relative humidity, air pressure, etc.) equal, you'll end up with an average that's somewhere within 10 degrees of the correct temperature because of the crappy accuracy due to poor calibration.
    That's useless if you are to compare that to a measurement from any other computer!
    Now to the interesting part: We're not going to compare it to another computer, but the very same!
    Enter observation #2: Thanks to the proper hardware used for the measurements, the temperature readings done are all within a very narrow interval, like +-0.1 degrees of the average, a trademark of good precision.
    Using the same tools for a series of "after" temperature readings will provide reliable numbers for the difference, since these new readings will deviate from the true value exactly as much as the previous readings did.

    /Olle
     
  11. Chai

    Chai Administrator Staff Member

    For accuracy, I think Core Temp is as close as you can get. Intel is relying on that temp to trigger the thermal throttling when the temperature threshold is reached.
     
  12. empire23

    empire23 BRB. Attacking Russia

    Vinegar seems to be the cheapest acid you can get off the shelf. Using tapwater first has an effect of leaving trace amounts of water in block, adding more to work, plus creating what is called dillution points, but my point is, before and after it's the same, but putting water in seems like more work. Don't need to add to the vinegar's work do we? It's just logical, because not everything is removed, thus better not introduce it in the first place. And mind you, not all minerals dissolve just as easily under acid ;)

    You'll have to soak the base in vinegar due to certain head coolants they use in the cutting and lapping process, and mirconized copper left over can be removed with a good acid scoring rather than just an alcohol. Alcohol can remove most residues, granted, but when faced with metal, you better use an acid.

    On the vinegar in the water part, first of all, when water bursts, it doesn't really matter of what you used, the electrical conductivity from the water mixing with a minimal amount of case dust is more than enough. Unless you're using Florinert from 3M, expect your stuff to blow up when water is added into the equation, regardless of what you put into the water. It's not the water, water is a terrible conductor electricity (how many free electrons does it have anyways eh?) it's the stuff that the water comes into contact into when it goes kapow! :D

    And it's kind of common knowledge that most living things especially small ones don't live in acid too well, i guess that's just common sense dude :)

    On the part of the installation itself, well AT has described that all powered components are to be removed while testing, If you ask me, reinstalling the HDDs, GCs or even heatsink shouldn't provide any extra stress on the sealings, or infact be much of a concern to warrant extra testing. It's like putting humans in a car and crash testing it and taking them out and crash testing it, once is enough, and besides weight, the humans doesn't really affect the crash test that much, except the screaming that is :p

    and on the temp part, not everyone can afford the luxury of calibrated thermometers and multiple thermocouples, and heck the variance shouldn't be that far off from the real world anyways, which should be generally acceptable. Flukes aren't cheap, and so it their calibration;)
     
  13. Chai

    Chai Administrator Staff Member

    A normal Fluke thermometer costs RM1200 when I called up, it's meant for industrial use... :(
     
  14. empire23

    empire23 BRB. Attacking Russia

    The prices are real killers, planning to sell my 179 and get a Fluke 189 once i'm done with my oscilloscope. But damn flukes are well made, drop them, dunk them, whack them, slap them and they still keep running.
     
  15. Olle P

    Olle P Newbie

    I seriously doubt that, given the low voltages in use.
    As you point out; pure water is an insulator. It needs free ions to become a conductor. Therefore adding a string/stream of (pure) water between to conductors won't cause a short circuit unless it causes a chemical reaction at the surface. ... And I don't think that a few pieces of insulating dust will become a good conductor just by moistening.
    Here I can't see what you're getting at. You propose one test to be done, but do you suggest it to be done before or after installation?
    Many cooling blocks require access to the back of the motherboard for mounting, so getting everything but the PSU and disk drives out of the case to start with is pretty much mandatory...

    /Olle
     
  16. empire23

    empire23 BRB. Attacking Russia

    Ahh, you forget that capacitance can charge a material with a suitably low dielectric constant and release it via static charge. Of course when taking stuff into a pure voltage context based on perfect conditions, this is fine, but when you put other factors in, it becomes far more complex.

    Depends on the content of dust and how it reacts with water, we shouldn't just assume perfect conditions here, and usually dust doesn't come in pieces, it usually comes in a more powdery coat more like it. Anyways, anything with a high enough capacitive effect = enough potential to go past a dielectric.

    Done before, since in my view, remounting doesn't induce enough stress, not by a longshot, to produce leaks. What AT is trying to get at i assume, is that you put WC system on, minus sensitive components, test it, and put the components back in and walla, you're ready to go. If it passed the test without the sensitive stuff, why would it fail with stuff?
     
  17. wiLL*lianG

    wiLL*lianG Newbie

    Wanna know is there any BIG different on the temp between 'Push' and 'Pull' on da radiator? Something like 5c and above?
     
  18. Skip Da Shu

    Skip Da Shu Official BOG Supporter!

    No personal knowledge but I read someplace... maybe here, might have been in Never's article... the theory and it favored push (has to do with air density of the cold vs the hot air). Also read someplace on a particular setup it was 2~4c.
     
  19. Chai

    Chai Administrator Staff Member

    I don't think the difference is as much as 4C. I think you should arrange the fans according to casing layout than the optimum performance of the fan arrangement for your radiator.
     
  20. wiLL*lianG

    wiLL*lianG Newbie

    when the part 4 will release? :lol:
     

Share This Page