Use a peltier to chill air ??

[Soul99]

600 CFM is loud. unless it's loads of 120mm fans in 7v or something.

[Daniel ~]

Last I spoke to Gabe (Swiftech) He Did Not recommend the use of teck for higher wattage CPU's.

The need for a second power supply and the negligible results as well as their limited capacity made them a poor choice.

Unless we see lower wattage CPU's in the future the "age of CPU pelter's" may well be over.

[whatever]

Originally Posted by Daniel ~

Last I spoke to Gabe (Swiftech) He Did Not recommend the use of teck for higher wattage CPU's.

The need for a second power supply and the negligible results as well as their limited capacity made them a poor choice.

Unless we see lower wattage CPU's in the future the "age of CPU pelter's" may well be over.

well said.

yes, that's why heat output level from source such as CPU, GPU, or any chipset being low is so critical to the performance of the pelt cooling. The lower the heat output, the better/easiler for the pelt to handle. Not a great overclock result will be attained by applying a pelt to the highly clocked & high voltage CPU due to its heat and additonal amount of heat that needs to be dumped out from the peltier. Most often the problem is the source that needs to cool the peltier isn't strong/hefty enough to get a nice steady result of the cpu overclock. The Temperature is almost everything. Getting rid of that heat is what makes it so challenging. Yes, I agree with you. Using the pelt isn't always encouraged. However, if the person really wishes to use pelt to cool the cpu and expect a high overclock, the person is gonna have to supercool the supercooling device, which is peltier, in this case. Supercooling the pelt, you will get fantastic result. This is why just an ordinary watercooling to cool the peltier doesn't give the greatest result. Another problem is that the water temperature rises as you cool these hot pelt and cpu which can also affect overclocking performance as you run it for long time.


I don't use peltier for my CPU though. The pelt is still not very efficient. I could do a superchilling the pelt and that pelt cools my CPU which will get higher overclock result, but I i think it's good enough having just chilled water to cool my cpu which gets me to 2794mhz on a xp1700. My video card is 80watt pelt cooled + waterchilled however.

[Aedan]

Originally Posted by Daniel ~

Unless we see lower wattage CPU's in the future the "age of CPU pelter's" may well be over.

Don't write them off just yet. There's plenty of research going into making them more efficient. Until their efficiency goes up, they're not going to cut it however.

The biggest problem with peltier devices is that they're rather leaky, thermally that is. In the ideal world, none of the heat that's pumped to the hot side would find it's way back to the cold side. Unfortunately, a real world peltier effect device is far more leaky than that, leaking vast amounts of heat back across the junctions, which makes it perform poorly.

This is the area where research is going at the moment. Effects such as thermal tunnelling of electrons is being investigated and new devices show promise, but they're not here yet. Fix the leakiness of peltier effect devices, and they'll be up there with phase change devices. That'd mean that a 230W "super peltier" might well be able to cool your 90W CPU down to -40C more effectively than a phase change system!

[Daniel ~]

We await the coming of that blessed day! ":O}

[Soul99]

Air cooling a pelt is useless, but watercooling it is fine, even with hot chips my 226watt pelt holds out, i dont need -40 temps (thos my rig is capable of that temp ) I havent done any math but i figure with my current results, i could even chill a presscott!!

As long as you build it right and well pelts are still good, tho with the total i've spent on the watercooling adn pelt setup i could got a vapo or prommie, but wheres the fun in that?

[Aedan]

Originally Posted by Soul99

Air cooling a pelt is useless, but watercooling it is fine, even with hot chips my 226watt pelt holds out, i dont need -40 temps (thos my rig is capable of that temp

Remember, that the bigger the temp difference between the hot side and cold side, the more power a TEC has to consume just to overcome the thermal leakage.

Also, remember that the bigger temperature difference between ambient air and the heatsink, the more power the heatsink can dissipate. This means if the TEC leakage is vastly reduced, then the hotside can be hotter, and hence able to remove more heat. That might make it more practical to run aircooling on a TEC without a problem.

Aircooling, done right, is quite capable of handling KW worth of power with forced cooling. The only reason you don't see it in a PC is that it's really too expensive, heavy and large to do.

Incidently the same goes for watercooling. The higher the delta between your coolant and the ambient air, the more power you can extract with the same amount of air. It would be good to see radiator temperatures that are 20-60C above ambient, as the radiators would be far more effective. This requires heat pumps however.

[Kaitain]

Originally Posted by Áedán

Aircooling, done right, is quite capable of handling KW worth of power with forced cooling. The only reason you don't see it in a PC is that it's really too expensive, heavy and large to do.

Air cooling is used quite a lot in industry to dissipate terrawatts worth of heat. Power stations in temperate climates frequently make use of air-cooled pipework, which is one of the many reasons why they can be a little, er, unstable during an unusually hot summer. *cough*

Difference is that the temperature is usually reduced from several hundred degrees above ambient to several tens.

The problem with computers is you're trying to dissipate heat from a very small area (contact area between HS and core is the size of a postage stamp) and you're trying to maintain cooling of this very small system to as close to the minimum possible temperature that thermodynamics allows.

Ditto for watercooling: where watercooling really scores big is that the rate of heat transfer through core -> metal -> water is very much faster than core -> metal -> air, and you move the really slow stage (transfer to air) to somewhere else. Because this "somewhere else" is not necessarily constrained by the same size and weight limits, you can increase the surface area for heat transfer as much as you like. I.E. add a huge radiator!

What increasing the water temperature does, as Áedán mentioned, is allow you to transfer more heat from less area. Sounds paradoxical but it's true. Given a normal water temperature of 60C you could transfer your 100W CPU load in about a quarter the radiator surface area of a system running with a normal water temperature of 30C. Or more realistically, transfer the 300W your CPU + TEC put out using the same size radiator, but keep your CPU significantly cooler.

[no1]

no wait y dotn u jsut build a mini refrigerator in a seperate mini box case beneth the main case and circulate air thrue the case and in to the frigerator for cooled air( both cases would be sealed)....

[Aedan]

If said mini fridge uses a peltier effect device, it won't have the cooling capacity to handle a PC, and hence temperatures in the PC will go up, rather than down!

If the mini fridge uses a carnot engine, then you will probably hit problems with the evaporator icing up. Ice is actually a resonable insulator, so it would end up insulating the evaporator from the air!