Homebuilt water cooler for CPU

[AndrewDV]

If you have not read this off of Tom's site, you need to:

http://www6.tomshardware.com/cpu/01q2/010528/index.html

Personally, I'd go for a nice blue coloring for the water.

Now, where did I put that all of that tubing?

[IM_Riktar]

Not bad if you want to build yer own but by the time you add up the case, PS and your time I would prefer going with the following:

http://www.koolance.com/main.html

Read the reviews. Quite the product for the money.

[mikiem]

You guys sound like my son... Every winter I have to hide all the extension cords & bolt the windows, lest he take advantage of nature's cooling system! If he could contain the cold from spreading thru the place no problem, but that's not on his list of priorities.

That said, sciplus.com usually runs a decent deal on take out peltiers you can gang in the dry world.

[IM_Riktar]

Hey, every little bit helps. Ya know I was thinking that since I am forever looking for ways to increase cooling, and the fact that I am forever getting up for another beer from the fridge, maybe, just maybe I could find a new way to save myself a few steps while computing.

[elfredo]

Hey 155 $ does not sound very low-priced.

There used to be a site http://www.overclockers.com which contains a lot of tips for real home made watercooling, but somehow it´s down right now.
You can also find some nice articles here:

http://www.overclockers.com.au/techstuff/index.shtml

I personally built my own watercooling system a year ago for less than 30 $, which is equal to a good air cooler.

Friedrich

[Brian Ellis]

Under normal conditions, this kind of thing is totally unnecessary. I've never had any problems with PIIIs running intensively at ambient temps up to 40°C at their rated speed. There is absolutely no advantage in dropping the CPU temp by 10°C or 20°C, provided that the junctions are under about 120°C. If extra cooling should be required, a Peltier cooler is more efficient. An alternative is to use a heat pipe, which has an effective thermal conductivity of about 20 x that of solid copper with the same dimensions. Neither of these have the severe disadvantages of water cooling. For your info, the traditional cooling medium (using just plain convection) for supercomputers is a perfluorocarbon liquid, which is inert (no damage in case of leaks), insulating, very expensive (typically $500/kg), long-lived, maintenance-free, does not support algae growth etc. The real problem with a pumped system, such as described, is that a) the plastic pipes harden with the plasticiser leaching into the water and b) there is no expansion vase. This means that, if the system is purged of air, the pressure will rise as the water warms up and the pressure may crack an old pipe with the results that may be imagined. Also, the efficiency is less than would be imagined as the kinetic energy of the water being moved by the pump is dissipated as heat.

IMHO, if you need to get the last ounce of speed out of a computer, buy a faster CPU, rather than try to go beyond what the manufacturer intended with a slower one. If you do the latter, you probably have no safety margin. That having been said, ask yourself the question, what do I gain by overclocking? Very little and, for most practical purposes, absolutely nothing at all other than the psychological kick of "getting something for nothing" (and that nothing may result in a dead CPU or worse). Pl-e-e-e-ase, give the CPU manufacturers enough credit for knowing what their devices are designed to run safely at.

I've little doubt that this will raise the ire of the overclockers but, as an engineer, I cannot recommend exceeding manufacturers' specifications. If 40 tonne trucks consistently use a bridge rated at 20 tonnes, sooner or later something will give and a driver will get himself drowned when it happens.

------------------
Brian (the terrible)

[Flying dutchman]

Brian, I must correct you here. There's a fundamental equation (Arrhenius equation) that states that chemical reactions run twice as fast if the temperatures increases by 10 Kelvin. (See http://www.shodor.org/UNChem/advance...arrhenius.html for more info)

The alteration/aging of silicon is a chemical process too. You can double the lifespan of your CPU if you cool it down by 10 degrees Celsius. Quadruple it if you cool it 20 degrees etc. Not that it matters much - a Pentium 3 will still be scrap iron in a few years...

[Greebe]

If you haven't already noticed, the MURC forums has a section call General Hardware/Software and that's where this thread belongs.

Chris, not to be picky, but I have noticed that many whom post here don't even know that there is more to our forums than just Desktop Video.

[Brian Ellis]

Fliegender Hollander,

I agree with what Arrhenius said (he also was the first to numerically calculate the effect of man-made CO2 from fossil fuels on global warming, in 1896: nothing new under the sun!).

However, the chemical degradation of passivated silicon, sealed in a hermetic housing, is such that the estimated lifetime of semiconductors at full rating (i.e., with a junction temperature of 100-120°C) is probably about 100 years (short of an electrical fault). The very first transistorised device I made (a 0 dbm 600 ohm 1 kHz oscillator for calibrating telecomms equipment) was in 1955, using germanium point contact transistors (notoriously delicate and made in the infancy of the semiconductor industry) still works, although I grant it has not run at full power for all of the nearly 50 years, but it has been subjected to ambient temps up to nearly 50°C, which is nearly at the limit for a Ge semiconductor. I strongly suspect that 90% of Pentiums would last for many decades running at full rated power. The major degradation mechanisms of correctly-specified semiconductors are neither chemical nor directly thermal (although it can happen in many cases, especially with glob-top COB devices or even some plastic-cased ICs).

As you say, Pentiums generally hit the trash can long before they are defunct.



------------------
Brian (the terrible)

[KRSESQ]

Something a little sad about that...

Kevin

[DentyCracker]

Damn it I'll overclock if I bloody feel like
I'm an overclocking wuss though, I prefer to just get the best Air Cooling out there, water ain't going anywhere near my computer. In any case I NEED good cooling as I live somewhere where it is constantly hot (ambient temperature usually 30 degrees or above) and I have a Duron 700 that I paid $110 for last August so I feel justified running it at 950 (a 950 Duron is selling for $82 at the same place now) and I have genome@home running on it almost constantly.
I have a Globalwin WBK38 currently and am going to get a Millennium Glaciator (cool name) to replace it.

[Michel Carleer]

Two rapid remarks:
- I was under the impression that semiconductor aging had something to do with the mobility of the dopants and impurities, more than with a chemical reaction.
- More importantly, the Arrhenius equation applies to the gas phase, not the solid phase.
Sorry, FD.
Michka


[This message has been edited by Michel Carleer (edited 12 June 2001).]

[This_Idiot]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by Greebe:
If you haven't already noticed, the MURC forums has a section call General Hardware/Software and that's where this thread belongs.

Chris, not to be picky, but I have noticed that many whom post here don't even know that there is more to our forums than just Desktop Video. </font>

No worries Greebe, be as picky as you like. By all means contact them all and let them know

I think you're wrong tho, I've noticed many familiar names posting to other forums. Besides, it's obvious innit, there's a bl00dy great link to the other forums at the top of the page.

You're right about this belonging to HW tho. I was probably asleep when I read it.

ATB,

Chris
(T_I)