peltier?

GJS1

GJS1

GJS1

Associate
Joined
13 Aug 2009
Posts
316
Hey, i was reading around that people oncidered peltier to not be good enough for modern CPUs what do you think?
thanks
 
That is an exceptionally vague question, I can't work out what you mean. I'm learning about them at present, started a thread here which may be of interest.

Are you getting at the cpu wattage is too high to deal with using a single peltier? If so you're pretty much correct, but you can just use multiple peltiers. ?
 
Not "not good enough" just pointless. When they were fashionable, there wasn't such a thing as phase and water-cooling was nowhere near as good as it is now. On top of that, a peltier system needs exceptional cooling making it nigh on as expensive as doing it properly with phase-change.
 
JonJ678 said:
That is an exceptionally vague question, I can't work out what you mean. I'm learning about them at present, started a thread here which may be of interest.

Are you getting at the cpu wattage is too high to deal with using a single peltier? If so you're pretty much correct, but you can just use multiple peltiers. ?
Click to expand...

would that be with a copper plate inbetween the CPU and peltiers?
 
MikeTimbers said:
Not "not good enough" just pointless. When they were fashionable, there wasn't such a thing as phase and water-cooling was nowhere near as good as it is now. On top of that, a peltier system needs exceptional cooling making it nigh on as expensive as doing it properly with phase-change.
Click to expand...

I was just asking, why have you gone mad?
 
I presume you haven't noticed that the higher wattage peltiers are 50x50mm or 62x62mm then. You can't fit lots of them between a cpu and a waterblock for really obvious geometrical reasons.

The approach which makes the most sense to me is a loop with the cpu in going to one side of many peltiers, and another loop with radiators in going to the other side. So all they're used for is an active heat exchanger, at which point you can move however much heat you have space and electricity for.

I'm not sure Mike, I think there's a use for them. In particular I think it's possible to cool a greater wattage using a single 240 radiator by having an array of peltiers cooling the components, and the radiator cooling the peltier, as radiator efficiency improves with temperature. I'm struggling to model this at present so it's just a hunch that it'll work out well. This would be good news for matx builds.

The other obvious use is maintaining the water temperature at ambient, or just below but above dew point, instead of accepting a water temperature 5 or 10 degrees over ambient.
 
Do some reading, there's enough threads on the forums and articles on google. Peltiers move heat from one side to the other, the more heat they're forced to remove the harder they have to work and more heat is generated. So you have to keep them under a certain load to see their optimum performance. As processors are producing more and more heat peltiers are becoming less affective, so to get the most out of them now you have to increase the cooling on the hot side with watercooling.

MW
 
JonJ678 said:
I presume you haven't noticed that the higher wattage peltiers are 50x50mm or 62x62mm then. You can't fit lots of them between a cpu and a waterblock for really obvious geometrical reasons.

The approach which makes the most sense to me is a loop with the cpu in going to one side of many peltiers, and another loop with radiators in going to the other side. So all they're used for is an active heat exchanger, at which point you can move however much heat you have space and electricity for.

I'm not sure Mike, I think there's a use for them. In particular I think it's possible to cool a greater wattage using a single 240 radiator by having an array of peltiers cooling the components, and the radiator cooling the peltier, as radiator efficiency improves with temperature. I'm struggling to model this at present so it's just a hunch that it'll work out well. This would be good news for matx builds.

The other obvious use is maintaining the water temperature at ambient, or just below but above dew point, instead of accepting a water temperature 5 or 10 degrees over ambient.
Click to expand...

Remember you need to cool the component, plus each tec. TECs are also hideously inefficient.
 
jellybeard999 said:
Remember you need to cool the component, plus each tec. TECs are also hideously inefficient.
Click to expand...

Indeed. Coefficients of performance of 1.5 or so seem to be the most readily achieved, so 200W of tec will move 300W of component heat, leaving me with 500W of heat to get rid of just to break even. After a few emails it looks like the hot loop is limited to 60 degrees by the pump. The tubing I'm using is good up to 74, the ddc until 60 and feser haven't bothered to reply.

I fear I'm going to have to conclude that I'll always need more radiators, not less, unless using many, many peltiers. Not done trying to get some hope out of the equations yet though. I think theres some charts about for the wattage thermochill radiators can remove under various conditions, perhaps efficiency improves more sharply with temperature than I thought. [currently working to double temperature, double efficiency, which is crude]
 
GJS1 said:
i thought the tec would be placed on the component and you would only have to cool the hot side of the tec?
no?
:S
Click to expand...
Yeah, you only need to cool the hot side of the TEC, but you need to remove what ever heat the CPU is having sucked out of it by the TEC, plus the heat generated by the TEC itself, as JonJ678 not only do you need to remove the 200W from an overclocked CPU, but the 300 or so W that the TEC generates in maintaining the delta T across it's faces. 500W is a lot of power to dispose of in a watercooling setup.

Edit: If you fail to remove the heat from the hotside, the TEC will break down leading to the heat flowing back the way to the cold side/CPU.
 
I put it at 300W cpu/etc, 200W waste from the tecs at a coefficient of 1.5, about what you hit around 6V with enough peltiers. The point remains though that 500W is a lot of heat to get rid off, even if the radiators are at 60 centigrade. Struggling to find a way to make this workable without just using enormous radiators.
 
Jokester said:
Yeah, you only need to cool the hot side of the TEC, but you need to remove what ever heat the CPU is having sucked out of it by the TEC, plus the heat generated by the TEC itself, as JonJ678 not only do you need to remove the 200W from an overclocked CPU, but the 300 or so W that the TEC generates in maintaining the delta T across it's faces. 500W is a lot of power to dispose of in a watercooling setup.
Click to expand...

ok, just slightly confused - if the tec is on the components hwo do you attach a waterblock to the component aswell?
 
TECs draw heat from one side to the other, you stack TECs ontop of each other to cool the hot side of each down to put it simply, but as said above you need watercooling/phase to remove the heat from the last TEC, your waterblock will be attached to this by some kind of custom made bracket.

what kind of OC are you aiming for? if its not majorly serious then stick with a good water setup.
 
breadturbo said:
TECs draw heat from one side to the other, you stack TECs ontop of each other to cool the hot side of each down to put it simply, but as said above you need watercooling/phase to remove the heat from the last TEC, your waterblock will be attached to this by some kind of custom made bracket.

what kind of OC are you aiming for? if its not majorly serious then stick with a good water setup.
Click to expand...

oh i was thinking of this for n/b s/b mosfets

also, if you had phase wouldn't you just use that?
 
You must log in or register to reply here.
Back
Top Bottom