tec question

[onewecallgod]

Originally Posted by Ghost742

thats the most helpfull post so far.

Which is ironic, since molexes catching fire are the least of your upcoming worries.

[Teyber]

Originally Posted by Ghost742

ya and its the length of a bible, y would wouldn't i be able to use it on my current air cooler i have see it done with the 120 xtremes.

the titan is also a air cooler with a pelt(even tho it sucks)

and all i was asking is what wattage would sufice

because it depends if you wanna burn your house down or not.

Ricecrispi had a great first post. None, untill you read more about them. Length of a bible, maybe. But clear some time and read it... Im tec noob, but unless you have 850/1kw psu,(high quality), or no sli/high powered card, perhaps. But im not sure what psu you have. For pelt size, depends on how much you plan to o/c it.

You need a good watercooling setup for a tec to be worth cr*p. By good, i mean if its the only thing in loop, a mcr-320, if theres anythign else in loop: thermochill pa 120.3

Please read more before you burn your house down

Additional Comment:

Originally Posted by avgazn91

i have personally never tried and never will try hooking it up to my pc PSU
it makes the PSU more unstable. It's safer to go with a separate one. And i had a 850 watt SLI compatible one. but then again, i never did it, so i don't really know what it would do to your computer

avgazn, i think oru psu's(arent you glad i recomended it to you?) could handle it, but ghost, im not sure what psu you have.

Originally Posted by Ghost742

thats the most helpfull post so far. i suspected a molex wouldnt be able to handle the current.

Look sorry if i seem mean, but a tec is for people who are bored with their watercooling and are looking for a cheap(but not easy) way to go extreme. Its not for someone like you. Youd be better off doing drews ac/chiller mod.

[manicdan]

playing with TECs is one thing you cannot shortcut on. the smallest mistake you can do will end up: frying your psu, melting wires, putting it in backwords, condensation on your chips. and even worse the tecs themself are made cheap and i broke the wires off the unit cause of how brittle they are.

take the time and just read what people have written about them, until you are answering peoples questions you shouldnt be touching this stuff. play with water cooling for a bit and then make the step to this.

[avgazn91]

Originally Posted by Teyber

because it depends if you wanna burn your house down or not.

Ricecrispi had a great first post. None, untill you read more about them. Length of a bible, maybe. But clear some time and read it... Im tec noob, but unless you have 850/1kw psu,(high quality), or no sli/high powered card, perhaps. But im not sure what psu you have. For pelt size, depends on how much you plan to o/c it.

You need a good watercooling setup for a tec to be worth cr*p. By good, i mean if its the only thing in loop, a mcr-320, if theres anythign else in loop: thermochill pa 120.3

Please read more before you burn your house down

Maybe a 120.3 isn't really needed if he does cpu only. a 120.2 would suffice. Unless you are actually doing a SLI system wced too. then u kinda need 2 120.2s or .3s.

Additional Comment:


avgazn, i think oru psu's(arent you glad i recomended it to you?) could handle it, but ghost, im not sure what psu you have.

You recommended me the OCZ proXstream 1kw PSU, and i would recommend it to anyone, especially if you don't have dx10 cards, and what a TEC to be hooked onto it aswell

Look sorry if i seem mean, but a tec is for people who are bored with their watercooling and are looking for a cheap(but not easy) way to go extreme. Its not for someone like you. Youd be better off doing drews ac/chiller mod.

I totally agree with you, Teyber. It would be much much easier and more benificial to just go with a chiller of some sort instead of having to get a watercooling system, TEC, and a new power supply (maybe not needed).

[Teyber]

Originally Posted by avgazn91

I totally agree with you, Teyber. It would be much much easier and more benificial to just go with a chiller of some sort instead of having to get a watercooling system, TEC, and a new power supply (maybe not needed).

thing though avgazn: 120.2 rad for watercooling, but im talking if you are going to throw a pelt in there, you need a really good 120.3 rad.
And avgazn, i think that he should go straight to phase change. Thign with a chiller is its more $$ then a ss,(txv/hx), and you still need a pump and waterblocks. I think a chiller is for extreme upgrade from watercooling...

Ghost: sorry, extreme cooling isn't easy, nor is it something to tinker around with...

[Ghost742]

well i guess il save for some water cooling, il make the water block myself tho.
il read up on the pelt sticky

[manicdan]

how do u plan on making your own block? i could think of a few ideas but they usually need a CNC milling machine

[Ghost742]

i will find a way, i have a ton of tools in my garage, or i can go to a place that has a mill.

[carbinefreak]

the copper and the fittings will cost almost as much as a decent water block and will probably be a lot better at disapating heat. i mean its up to you but i draw the line at making my own blocks, for leakage reasons.

[Ghost742]

small rubber seal will keep leaking away, teflon tape around the barbs will do that to

[Teyber]

Originally Posted by Ghost742

small rubber seal will keep leaking away, teflon tape around the barbs will do that to

For leaking, just mill a small ring around the outside and get a o-ring set on ebay or harborfreight.

But he has a point; copper prices are ridiculous... You will end up spending 40$ on copper anyways. Im not trying to discourage you though! whawt tools do you have?

[Ghost742]

well my garage is a mess i dont know what all the things we have are at the moment and getting through my garage is like climbing k2.

il get the smallest piece of copper i can i dont need a huge block, maybe 5-7mm thick and 6-7cm wide and long, some on on here might even have some left over. u never know

[manicdan]

i was thinking about making one out of an old heatsink that had decent not so thin fins and cover it with an acrylic "box." i know it sounds overly simplified but im sure i dont have to go over the small obvious details.

[Ghost742]

they should make a block with a o-ring that when seated lets the water touche the cpus heat spreader directly

[teebob21]

Originally Posted by Ghost742

they should make a block with a o-ring that when seated lets the water touche the cpus heat spreader directly

WARNING: SCIENCE AND MATH LESSON AHEAD. READ UP AND SPARE YOUR CPU.

Actually, you really don't want that. Water has a thermal conductivity (k) of just .58 W/m-K while copper's k is just over 400 W/m-K/. Air chimes in at about .024 W/m-K. In layman's terms, copper is going to absorb heat from your CPU 689 times faster than than water, and over 16,000 times faster than air alone. Water is 24 times better than air at conducting heat. Air is a great insulator; that's why parkas and fiberglass insulation are so puffy: they trap air to insulate. That's also why when used to cool a processor alone (no HS) air is the best CPU killer, hands down. If you didn't strap any copper, aluminum, or any metal at all on the top of your CPU, water would kill it just as sufficiently, just 24 times slower. (Assuming no built-in thermal protection routines kick in) Water's higher thermal conductivity also explains why you die of hypothermia many times faster when wet in the cold compared to keeping dry.

Water's claim to fame stems from its high specific heat (4.181 J/gK). Water can absorb massive amounts of energy, and experience a very small temperature change. Copper, on the other hand, has a quite low specific heat: 0.385 J/gK. J/gk means Joules per grams times degrees Kelvin. Simply put, to raise a material's temperature by one degree Kelvin (one degree Celcius), simply add x number of Joules.

If your CPU is cranking out 200 W of heat energy, we need to remove 200 W from it to maintain a stable temperature. 1W is 1 joule per second, so 200 W = 200 J/s. Those 200 joules of energy per second would raise the temperature of a single gram of copper by 519 degrees Kelvin every single second if we applied it in a vacuum where the air could not passively cool it! (200 J/s / 0.385 (J/gram x Kelvin) x 1 gram= 519 K/s) Good thing heatsinks and copper blocks weigh more than a gram! I'm going to estimate a decent water block weighs 250 g, about a half pound. Your water block (and CPU!) is going to increase in temperature by 2.59 K/s if the heat has no where to go. (Same math as before, just put 200 in for the 1).

So now we've transferred the heat from the CPU die to the copper, but now what? It's going to melt if the heat energy keeps getting pumped in. In comes water to save the day! Water is capable of removing those 200 joules per second from the copper and only increasing in temperature by 0.19 deg C (assuming 250 mL of water). Most WC setups use WAY more than 250 mL, so the total heat capacity of the coolant loop is quite high. Let's pretend we have 1 L total coolant in the system. Our warmed water passes to the radiator, and back through the pump, and on and on and on.

Eventually, the heat must be expelled from the radiator, where the limiting factor is the air. It takes a HUGE volume of air (enough to push a mass of air equal to ~24 times the mass of the heat-expelling water, see thermal conductivity comparison above) to remove the heat contained. Let's get rid of the 1 second's worth of heat from that water in our cooling loop. Imagine an ideal rad, with perfect thermal transfer.

At STP (25 C, 1 atm pressure) 1 mole of air weighs 29 grams and takes up 22.4 liters of space. We need 24x29 grams or 696 grams of air. That's also equal to 24x22.4 liters of air, or about 538 L. 1 L is equal to 0.0353 cubic feet, so we need ~19 cubic feet of air to remove the heat in a 250 mL loop. That's PER second! Thankfully, we went with the 1 L coolant loop, so we only need 1/4 the wind. 19 CF x60sec gives us an ideal fan of 1140 CFM for 250 ml, and 285 CFM for a 1 L loop to keep our 200 W CPU running at ambient temps. 285 CFM is going to be crazy loud, but for the criminally insane, by chance I found a 1140 CFM fan. The highest CFM I found for a fan available on Newegg was under 135 CFM, less than half of what an ideal fan would need to push (in the thermodynamic sense of the word, not the aesthetic or technical sense).

The CPU heats the copper, and the copper heats up faster than the water. The water in the radiator heats up faster than the air can remove it. Heat flows from the area where it is more concentrated to where it is less concentrated, and the bigger the difference, the higher the flux. Heat flux is also dependent on surface area, which is why rads and heatsinks have fins, and water blocks have protrusions on the inside. (That portion of the math lesson is beyond the scope of this forum) Eventually, we'll reach a point where the CPU can no longer heat the system at a greater rate than it is removed, but instead at the same rate it is expelled from the rad into the air. This is because as your rad temperature increases, ambient air becomes more effective at removing heat. This equilibrium point is your final load temperature. The only way to decrease it is to remove more heat from the rad per second, or add less heat to the loop per second. Maybe for you that means a higher CPM fan, or a chiller loop (which is actually the same process repeated in a secondary closed system). I'm certain adding less heat via a slower OC is out of the question for all of us.

In the end, it's just thermodynamics, baby.

EDIT: Since I had all the numbers I needed, I figured the equilibrium point for this setup:

200 W CPU
1 L pure water coolant
250g pure copper water block
250mL/min flow rate (HIGH flow rate: ~3.95 Gallons/Hour)
Perfect copper radiator measuring 10cm x 10 cm x 1cm, with fins spaced 1 mm apart
Ambient temp: 25C
Ambient air pressure: 1.00 atm
120 mm Rad Fan: 125 CFM
Perfect Insulator (TM) Tubing (ZERO heat loss through the tubing)
Delta T between CPU and Water Block (contact surface area: 16 cm^2, 5mm thickness): 0.71 C
Delta T between Water Block and Water: (contact surface area: 36 cm^2, 5 mm depth of water): 4.55 C
Delta T between Water in Rad and Forced Ambient Air: (contact surface area: 1000 cm^2, 0.1 mm fin thickness): 14.73 C
Total Delta T CPU to Air: +19.99 C
Load temperature: 44.99 C

Granted, those temps are possibly off a bit, but this is a theoretical setup under imaginary, ideal circumstances on an incredibly hot CPU. A more reasonable estimate of a 100W TDP CPU gives a load temp of around 33C in this "perfect" WC setup. (Plus, some of my calculus may have been off. It's been a LONG time) It gives you an idea of how many factors come into play in your cooling setup, and where each bit of heat is handed off in the system from the CPU until it is vented into the environment.

[manicdan]

i read very little of all that, its just to early for me, but your right, the copper block is used to create a larger surface area for the water to steal the heat away from the cpu/gpu