my Water Cooling 2 cents

[drag]

In reading several articles about water cooling I noticed a trend were people are always shooting for more volume or faster pumps in order to make the cooler more effective and this may not exactly true. I know with automobile radiators people with high performance cars make the mistake of installing high-output water pumps to increase the cooling compacity of the water cooling system, unfortuently this does not always work because the coolent sometimes does not remain long enough in the engine block to absorb the heat or go thru the radiator to quickly to cool the water enough. This has the unintended qualities of acually reducing the amount of cooling compacity.

A effective way to test something like that would be to have a high-output pump and free flowing fittings on a efective cooling block and add a faucet to the pumps output. Vary the flow and decide what flow is best.

I also think that the key to making a effective cooling block is surface area. Lots of fins, rough casting and maybe some bead-blasting for more texture. If you sand-blast (or to a lesser degree bead-blast) a bare peice of sheet metal (steel) it can rust so fast you can see the rust growing, this is because when you blast metal it will increase the effectice surface area by a factor 10 because all the miniture craters and hills that are formed in proccess (more surface area = more area explosed to oxygen and water vapor = more rust faster). The same holds true for cooling ability. You want the bottom of the cooler to be shiny and smooth to more surface contact with cpu, but shiny-ness is bad for the part that comes in contact with the cooling medium (air or water). Also painting something with a extremely thin layer of flat black paint can increase cooling power since darker colors will transfer heat differences more easier that shininess or light colors. (you can use black spay paint primer; spray primer into can top and paint on with fine brush(old soft toothbrush? or toothpick with end frayed)

Also wouldn't make more sense to use a peltier element on a small well insulated water reservior, then directly to the cpu? That way you could have a small cooling system in a computer case with pre-cooled water going from a suberged (ei quiet) pump and into the cooling block with no dependance on a large radiator. You could intergate a large heatsink on the exterior underside of the case with a large slow-moving (ie quiet agian) fan to keep the peltier/reservior chilly. That way if the peltier failed you would still have a water cooled cpu instead of a fried one!

 

[Adul]

a lot of what you say make sense, though i am not sure about the paint. Unless the paint can transfer heat better then copper.

 

[ObiDon]

Also wouldn't make more sense to use a peltier element on a small well insulated water reservior, then directly to the cpu?

I think the reason that this isn't really done is because you would have to insulate the entire liquid-cooling system instead of just the socket/cpu/peltier/block setup. Your liquid would be below room-temperature and you would risk having condensation form on all of your tubing and reservoir.
Besides that, you would still need to cool the peltier that was attached to your reservoir. I suppose you could attach a heatsink/fan combo to it but that would almost negate one of the primary advantages of liquid-cooling

That stuff about too much waterflow makes sense. If I can't find a decent variable-flow pump, I'll pick up a valve to control the flow. I think it would be so much nicer to be able to do that from the pump though...no excess pressure buildup between the pump and the valve

 

[BillDKiller]

You are absolutely right about the water flow. You want the water to stay in the radiator until the water temp. has reached room temp. Moving it too fast will not accomplish this. The water block is not as important, since more water will be coming along with the flow. You also want to use 100% distilled water in your system. Not coolant, rubbing alcohol or mineral oil. The only reason you would want to add anything is to inhibit growth of algae or microbes, or to color it a cool shade. Automotive coolant is not a good conducter of heat energy (not even in an auto system). It is there merely put in to lower the freezing point and coat/inhibit rust and corrosion.

The current peltiers are not really strong enough to cool something of any large size. They also produce quite a lot of heat while doing it, so you would have to keep them outside your system case. They are mainly used to move small amounts of heat from one point to another. I suppose if you used enough of them they could cool a storage tank. But really, how cold do you want to get? Just keeping your system components at room temp is usually good enough. After that and you have to massively insulate everything or condensation will form, and eventually drip on something expensive (Murphy's law). With water cooling you will reach a point where the temp. is no longer a problem, nor the FSB speed, but the voltage you have to apply to keep everyting stable. If you have to turn up the voltage too far you will get leakage across the cpu circuits, and that's the limit you have to stop at.

I am looking into various other fluids/additives for the water cooling system and will post my findings on a thread. But until now I have found nothing better than distilled water, a couple of drops of dishwashing liquid (to break the tension of the water), and a bit of bleach for growth inhibition. A little water-wetter won't hurt, but it is used mainly so that bubbles don't form in pockets on an engine block, which shouldn't occur in these systems. Plus, of course, add a little bit of food coloring dye to make it ultra-cool looking.

 

[drag]

I ment that the peltier would be on the resevior and the peltier would be against the side of the case. You would then cut a hole or something in the side and attach a large heatsink to backside of the peltier the heatsink would radiate the heat from the peltier/resevior on the outside of the case and keep things cool in the case. With a nice instalation and large fan you could have a very cool cpu without having to put up with all the noise that comes from a normal overclocking machine.

Oh and you would just use a much larger pelteir to cool the resevior, that way you could use a much cheaper one then the little high-power ones used for cpu's. I've seen beer coolers you plug into a cigeratte lighter that use peltiers to function so I know they manufacture ones that would work.

Another thing I wouldn't use the peltier to cool the water down so far that it condenses It would be just used to create a more compact efficicant cooler for CPU cases than a big clunky transmission cooler with gigantic fans attached to it . and If I were to cool it down farther I don't suppose it would be to hard to insualate the lines with some sort of foam/rubber insulation spray or use large diameter line with a interior coating.

you could create a super overclock machine in something as small a desktop case instead of usng a tower. It would be a sleeper like the guy who stuck a V8 in a volkswagen!

 

[Insane3D]

Cases and Cooling Forum

 

[drag]

sorry my bad I be a newb here I learn for next time

 

[Insane3D]

Originally posted by: drag
sorry my bad I be a newb here I learn for next time

Oh, no problem. I just thought you would get more input over there. There are a bunch of watercooling nuts over there that might have some ideas for ya.

 

[woolmilk]

I know with automobile radiators people with high performance cars

I m not sure about that statement. If you increase the flow of water, the temperature at the end of the radiator might rise, but on the other hand, the water temperature would rise less at the prozessor.
From the point of energy, the system is always stationary. The thermal power is taken from the processor and dissipated by the radiator.
Since the water is used for transport, the cooling should get better, the more m^3/s of water you put through.

 

[BillDKiller]

Your thinking is correct. The more water and the faster the flow the better the cooling. But only to a certain extent There comes a point where the water is moving too fast through the radiator to effectively transfer its heat. Dont forget that the surface area of the radiator is fairly small. You really need to look at where the velocity/temp function slope levels out and eventually rises at a given volume of water. Now this may be much higher than a regular pump could get, that I don't know yet. But it will be easy enough to check once I get my system up and running. I plan to either put a regulator on the flow line, or vary the voltage to the pump to slow it down.

Of course we could always try a car radiator and fan. I guarantee you couldn't overpower that with a pump. Huge surface area.

btw, has anyone seen those cheap plastic refridgerator/heaters that you can use in your car? The ones that hold only about 8 or 9 cans of soda. They're about $40.00 bucks at Target and Home Depot. The inside of one is almost the exact size of the comp cooling radiator. I wonder if you could use one of those, or if the constant cooling would be too much for it. I am sure they use some kind of a peltier system in them. Hopefully there would be enough room left to keep a few beers cold.

 

[Big Lar]

Heres my $.02 as to Flow and Pelts on reservoirs; Flow in most cases is in fact extremely overdone IMHO, as in I have heard of peeps using upwards of 400gph pumps to merely run thru a single water block, this isn't needed, nor is it efficient. A 250gph or slightly less will in fact push plenty of fluid, and absorb the heat from the block. Now then, the Peltier; I tried something to this effect quite awhile ago, and thought back then that it was a good idea, simply put, 1 would need a minimum of 2 pelts to cool off even a small amount of fluid, and then the problem arises, which is Cooling the Hot side of the pelts, if air cooled, then it goes 1 step furhter= cooling off the room they are in. Peltiers give off a gawd awfull amount of heat. 1 could water cool the hot side of the pelts, but that entails an entirely new loop of cooling, which again, is not efficient, as more electric is used, more heat rendered into the atmosphere, and more noise from another pump. Sorry for babbling, but I have tried em all, and have found that cooling the water in a small fridge, has been My best option, your mileage may vary.

 

[DHL]

Drag, do u have some references about the impact of flow rate on cooling capacity, I tend to agree with woolmilk but I have no info on this

Here is my $0.02 cents. A black surface can enhance the radiative heat transfer, which is proportional to (temperature diff)^4. In an automobile radiator it is useful as the temperature difference is larger. In a PC the effect is not that significant.

The surface roughness seems like a good idea

 

[ST4RCUTTER]

Overclockers.com did some testing that proved fairly concisely that flow through the waterblock should be maximized for heat removal from the die. Water flow through the radiator should be slow enough to allow the maximum amount of heat dissipation to the radiator element.

 

[Mingon]

So perhaps different sized tubing would help, smaller before the die for increased flow moving to larger before the radiator to reduce it again. (think that would work - flame if wrong)

 

[drag]

Very interesting. If I was to use a peltier to cool the resevior i would use a much larger on then the ones people use on the cpu's. Also the heatsink for the peltior would be massive, very large I would use something like a block of copper with fins soldiered to it. maybe and the peltier's heat sink would be located on the OUTside of the case thru a hole. Since the resevior can be located anywhere I would put it toward the back so the peltier could protrude a bit outside thru a hole.

My next idea would be to remotely mount the resevior on the outside of the case. This resevior would be much larger. Then I would gut a hotel-size refrigirator.
They use a alchohol-based refrigerant, and I would put the cooling coils in to the water.

 

[DeadHead]

Everything seems to make sense except the paint. That will only apply if its radiant heat which it is not.

 

[SocrPlyr]

one thing that must be said is that the paint thing doesn't really work out in the real world... it is true that the darker something is the higher it tendency to absorb and trap heat... if you paint your block this hurts you because the heat is now trapped in the block in the normal air cooling doesn't happen as well and the block itself might start to heat up too much not allowing the quick transfer of heat from the cpu to the block... (remember this is because heat dissapation is directly proportional to the temperature difference so trapping the heat is not what you want to do...) there is a reason why heatsinks are made of metal... metals generally absorb heat very quick, which in turn means they radiate heat very quickly as well... this principle is shown by if you go into a room with a wooden chair and a metal chair and sit down in each of them the metal one will feel cooler because it absorbs the heat quicker and then dissapates it quick... so in escence i hope somewhere in here i have proven that you would actually like a light colored paint versus a dark one, but in reality the metal itself does a better job than any paint...
just as a little more on the subject, my point is easily proven by the thermal paste/gel/glue makers... if you notice all of them are light in color because they transfer the heat away faster (what they are intended to do) and really if you look at artic silver, it has silver in it, showing that they think a metal transfers better than just paste
another thing you mention is the speed at which the water travels through the block, there are some merits to your arguement, but not the ones that you mentioned... the faster the water moves through the block the better, because the heat dissapation is directly proportional to the temperature difference the water will be cooler since it is given less time to absorb the heat and warm up so the temperature differential is higher. now the downfall to the faster and faster theory is friction, the faster the water moves the more heat is caused by the friction of the water with the rest of the apparatus. where does the friction overtake the temperature differential, i don't know, i didn't say i was an expert.
i agree with your comments about the smoothness of the heatsink, that is why many people here sand down their heatsinks and processors with very fine grit paper...
i am not so sure what you are trying to say at the bottom about the slow stuff, but precooled water would be a plus, but rather espensive, also remember that if you cool the water you have to chill it pretty well because just a few degrees won't do you any good because it will absorb the heat from the regular room, but once you get it so cold you start to run into problems w/ condensation... so in effect i don't know what to tell you here...

oh well feel free to correct me as you please, but this is what my knowledge has taught me

Josh

 

[drag]

Na this is cool.

I just have some weird ideas about stuff.

The paint stuff maybe is not a good idea with the heat sinks on cpus, however I still think my reasoning is sound it does have some good hings about it. Black absorbs heat because of the thermal transfer stuff, right? But I think it goes both ways. I think dark colors will absord cool temps more easily. The reason white will not absorb heat as fast a black is that the white reflects more radiation, so in reverse I think that white will also retain more heat, because the light colors serve to inhibit heat transfer, from hot to cold and cold to hot.

Remember in science class: heat is only a measure of the speed (or excitemnet) at which molecules move (or the amount of themal energy) . There is no natural diference between 'hot' and 'cool' these are completely dependant on your point of view per say. It just the difference of from one state of thermal energy over another state of thermal energy based from their distances from 0 degrees kelvin (the theoredical point were all thermal energy is exhausted (-273 degrees C))

So if I am right: anything thing that can speed up the rate at which something absorbes heat will then speed up the rate at which something releases heat.
(I know this is probably simplistic, but thats about all I know about thermal transfer stuff) ie a themos will keep coffie hot and it will also keep your pop cool.
If a black car heats up quiker than a white car in a hot summer it will also tend to cool off faster in a cold winter,

Metal is a good way to transfer heat and it is also a bad way to transfer heat. There are some weird chemical and physical properities to objects that don't make sense a first. Metal does a very good job at dissapating heat if a object is touching it, but has horrid thermal transfer to air. Carbon does a bad job at transfering heat thru contact but does a teriffic job at transfering heat in to air. Hold you hand over a hot pan, You can feel the heat but it won't burn you, however if you were to touch it you hold get burned immidiately. Hold your hand over hot coal and you will get burned, however you can grasp the coal quickle and throw it and it won't hurt you. (you have to be very quick though. I have in the past picked up a dropped cigeratte cherry and by rolling it from hand to hand I was able to move it to a ash tray with out burning anything. (nice white carpet you see))

Thats why the color of the Artic White is immeterial, because the thermal properties of transfering heat from solids contacting is very different from thermal transfer into air. That is also why a cpu is able to transfer more heat into the heatsink from such a small contact point than the heatsink is able to radiate into the air over such a broad surface area as fins or pins. That is also why overclockers are moving to water, because metal is such poor transferer of heat into air that you have to use the water to transfer heat away from the cooling block and into such a much more massive heatsink as the radiator and thats why rads are usually black. --- whew said it one breath too.

 

[OrganizedChaos]

lots of info here

 

[ceZium]

I am just scared to attempt it and fail and oops, there goes my equipment. I have had friends that have had good results come out of it but I just do not think I could do it and not have a leak or something.

 

[BillDKiller]

Don't forget that distilled water is not really conductive. So even if you had a small leak there would be time to detect it and redo that water connection. I dropped my cell phone into a hot tub one time, and it was ok after I let it sit out in the sun for a few days. Luckily it was off at the time it got drowned. Your fine cleaning up water spills from almost all electrical equipment as long as you let them dry completely and don't turn on the power. I hear you should run the new water system for several hours before you turn on the computer, to work the bubbles out of the system. So you have plenty of time to find leaks and problems. I'm a little concerned to, but plenty of people have built these type of systems and they work great.

I'm going to build my new P4 system with the stock heatsink/fan, and then switch over to the water system after I have all of the bugs worked out.

 

[drag]

Ok heres the scoop on the paint stuff i did some research.

there are 3 ways that thermal energy is transfered:

conduction -- direct contact solid to solid.

convection -- from a solid to a fluid (water or air)

radiation -- energy transfer due to electro-magnetic radiation.

Air is a very very poor way of transfering heat IF it is still. By convection's nature a fluid that is hot will rise and cold air will move into replace it so always cool air will be moving against the heated surface. Foams are used as insulation because they are mostly air, the air is held in place in plastic bubbles. Air is such a good insulator when it is still that 90 perfect of the heat that goes thru the foam goes thru the plastic bits holding the air into bubbles. But as long as air is made to move it is a average conductor of heat. the same is for water, so you want to move the air fast thru the cooling block and slow thru the radiator, so it would seem to be a good thing that the rads tubes should be much larger than the rest of the system.

Conduction is already very good for heat sink with the termal paste stuff.

Radiation is very interesting, it is not dependant on any known material and can pass thru a vaccum were convection and conduction are worthless . The color of a object doesn't matter to convection or conduction, however it is very important in radiation. Light is a form of electromagnetic radiation. Dark colors will heat faster in the sun because they soak up the light radiation, therefore they apear black. Red colors soak up all light radiation except red, the red is reflected back so they appear to be red in color. White reflects all light so it appears to be white, since white is what we perceive when we experiance all the visible wavelengths. Black paints tend also to be black in more then just the visible light spectrum, they tend to be be black in other spectrums, too (this is just a unintended, and just the nature of the pigments used). This is why of course stealth fighters are black, their paints are specificly designed to soak most forms of radiation including the visible spectrum. All heated bodies change various amounts of heat into electomagnitic radiation, and this radiation is naturaly occurs all over the spectrum but is peaks in the infrared spectrum and visible to sensitive types of camera equipement.

Like radio antennas (which a good receiver design is also a good sender design, example: satalite dishes), Colors that are good suckers are also good blowers of electrico-magnetic radiator . So generally darker colors will aid in the transfer of heat energy thru increasing the ease at which radiation is transfered. However there is just on caviat: Just because it is black in the visible spectrum it doesn't mean it is black in the infrared one, and visa versa.

Ironicly some clear plastics are transperent in the visible spectrum but completely opaque in to infrared one. So if someone is trying to spot with IR goggles you can be pretty much invisble to them by hiding behind a clear peace or plastic! (that is of couse until your heat warms up something on the other side of the plastic thru convection or conduction).

Electricions have recorded that transformers installed indoors painted matt black were only 78% percent as hot as those painted with shiny aluminum paint. (outdoors though the suns radiation heats up the transformer more than they release it if they are painted black.)

I am going to guess that you may see a 3-7% drop in temps by painting a heatsink black, compaired to one that is shiny. The best technic for thin black coats of paint is to spray some flat black paint into a container (lid) and take a water color brush (or a toothpick with a smooshed end) and paint the spray paint on. It remains to be seen though it it will help any and I don't have the resources to try this, my heatsink is one of those copper ones with the folded fins and I can't get in their to paint. I would expect that if you were to do this with a shiny aluminum you would see a much bigger difference than a copper heatsink and the key would be to create a very thin layer of paint.

 

[h2sammo]

I have answered this question before, and i will answer it again.

Based on the formula, Q=mCdT, where Q is the work done by the radiator, m is flow rate of coolant, C is the heat capacity of the coolant, and dT is the difference in temps at the inelt and oulet of the radiator, the radiator will cool better with a higher flow rate of the coolant.

Be advised, Bill's article on radiators, has been revised. You can see revised graphs on the stickied (up on top) thread on flowrate in the overclocker's cooling section.
In theory, Q will increase linearly with flow rate, but in practice, due to differences in convection rates between water and Cu and air and Cu, the line will form a horizontal asymptote and flatten out eventually. This means that as you increase flow rate, the work done will benefit less and less (but still benefit nonetheless). Bill's new graphs show this phenomenon.

Please folks, stop debating this particular item. Its physics, it works every time. Having the coolant longer in the radiator will never be better than running the liquid faster through it.

Now about the black paint. Very ingenious, but pigments that appear black to our eyes only trap photons more efficiently that pigments that appear to us in other colors. The notion of colors itself is very mischievous. Color exists only in our minds. Different pigments absorb different wavelentghs of light and refract the rest. Our eyes/brain intereprets this phenomenon as colors.
In fact, the color red means that the red pigment traps light waves at any visible frequency but the frequency for red light. Red colored pigments refract red light. This means that a red car is everything but red...confusing?
Consequently, black pigments trap every visible light wave frequencies, nothing gets refracted, so our brains makes up a void in our minds eyes - -black--.

Again, black will only trap light waves or photons. Black paint will not have any effect whatsoever on heat absorption (like any other color pigment for that matter). Light waves (photons) and heat are different concepts. Photons can be measured, they are mass praticles that travel in wave patterns. Heat is the transfer of kinetic energy from one medium or object to another, or from an energy source to a medium or object.


cheers

 

[drag]

Na na.

All heated bodies release radiation. All a light bulb is is a device that allows you to heat up a wire so much that ALL eccess heat is released as elctro-magnetic radiation. Visible light is but a small fraction of the possible electrono-magnetic radiation wavelegths. Microwaves are the same things as light, but they are just a different wave length and frecency (or something like that).

Move radiation absorbed = more heat (its conservation of energy; it's got to go somewhere). Releasing radiation takes energy, and the source of this energy is usually thermal.

Of course in a computer the VAST majority of heat is simply transfered thru matter (air/water (convection) or solid object (conduction)), however if we facilitate the release of radiation of the heatsink, temps will drop with only a slight increase of heat in the air (air is not good at absorbing radiation). Black paint can help with this in a small way, and it may be enough to make a difference in temp (it works on air-cooled volkswagen engines).

 

[h2sammo]

Drag, if what youre writing is correct, thean the hot parts in the computer are a source of visible light.

Try turning all the lights in your room off,so its pitch dark, and look inside your computer (given that you have no leds on). Do you see anyhing?

If you do, i can think of two things that are happening:

1. your eyes detect light at frequencies way under the visible spectrum. you could make money off of this, just think the possibilities...

2. the hot parts of the computer do give off visible light, and humankind has been fooled for centuries about the properties of hot metals. maybe thats why new york is so bright at night...