peltier/TEC

T

Teknokid

[H]ard|Gawd
Joined
Feb 17, 2007
Messages
1,292
ive heard a lot of people around the forums mention these, but what are they?
im looking at ways of getting my cpu temps down, for a big ass overclock, and these seem the way to go.
Can someone give me some examples, or tell me more PLEASE :)
cheers
 
TEC's are electrical devices that move heat from 1 of their sides to the other. During this movement, they consume energy. They're little white pads of ceramic and peltier junctions that generally require their own 12v PSU to run.

TEC's are very inefficient, and usually add anywhere from 30-50% more heat than they move. For example, moving 65w of heat (stock conroe) across the device creates almost 90w of heat on the hot side.

Generally, to use a TEC you need to calculate the heat output of your chip (P = C*V^2*F) at it's overclock, multiply that by 1.4, then make sure your watercooling loop can dissipate all that heat. My CPU @ 3400mhz puts out (0.02 * 1.40^2 * 3400) 133w of heat, factoring in the innefficiency (40% innefficiency) means I'd need a peltier rated for about 190w (and a 12v PSU that can supply the required amperage... 190 / 12v = 15amps). Additionally, I'd need to be able to remove these 190w of heat with some sort of cooling... So, at least a double radiator'ed watercooling setup.

Large wattage peltiers that get your temperatures down below ambient (250+) require insulation and motherboard waterproofing to prevent shorts or corrosion due to condensed water.
 
Forget TEC unless you aren't paying for the electricity. Otherwise in order of cooling capabilty:

  • Alcohol chillers
  • Phase

And this is where you want to go to discuss them.
 
many thanks, so a tec isnt the way to go, but i dont pay the electricity bill :)
what about a peltier? what do they do?
sorry, im a noob, i want something in my water loop, ive already looked at phase but then im limited to cpu only :(
cheers
 
sorry, found out, there the same thing :(
can you run a tec without power and use it as a normal block? then power it up for heavy overclock time :)?
cheers
 
Arcygenical said:
T
TEC's are very inefficient, and usually add anywhere from 30-50% more heat than they move. For example, moving 65w of heat (stock conroe) across the device creates almost 90w of heat on the hot side.

Generally, to use a TEC you need to calculate the heat output of your chip (P = C*V^2*F) at it's overclock, multiply that by 1.4, then make sure your watercooling loop can dissipate all that heat. My CPU @ 3400mhz puts out (0.02 * 1.40^2 * 3400) 133w of heat, factoring in the innefficiency (40% innefficiency) means I'd need a peltier rated for about 190w (and a 12v PSU that can supply the required amperage... 190 / 12v = 15amps). Additionally, I'd need to be able to remove these 190w of heat with some sort of cooling... So, at least a double radiator'ed watercooling setup.
.
Click to expand...

You raise some good points but there are certainly some additional considerations. First, it does not need to be the goal of a tec-assisted water cooling loop to remove all or even most of the cpu generated heat. Providing chilled water an an otherwise serviceable cooling loop can give us a great advantage. the success of systems such as FreeZone and Amanda seem to demonstrate this.

Secondly, the inefficiency and current draw (implied also by the memeber who refers to the cost of using tec's in the comment following yours) are problems that rise with the voltage input and decrease as driving voltage decreases. All we have to do to start making the pelt look pretty good is to use say, 7 volts. Efficiency improves, cost decreases and hot side cooling is more manageable. Granted, that 320 watt pelt we just bought is only going to give us a 70 watt heatsucker when we use that approach. But, hey, lets just get a couple.

Parallel tec systems used in chilled fluid reservoirs or tec-assisted rads and driven by voltages closer to the bottom of the range than the top may have a lot to offer.
 
ickywu said:
You raise some good points but there are certainly some additional considerations. First, it does not need to be the goal of a tec-assisted water cooling loop to remove all or even most of the cpu generated heat. Providing chilled water an an otherwise serviceable cooling loop can give us a great advantage. the success of systems such as FreeZone and Amanda seem to demonstrate this.

Secondly, the inefficiency and current draw (implied also by the memeber who refers to the cost of using tec's in the comment following yours) are problems that rise with the voltage input and decrease as driving voltage decreases. All we have to do to start making the pelt look pretty good is to use say, 7 volts. Efficiency improves, cost decreases and hot side cooling is more manageable. Granted, that 320 watt pelt we just bought is only going to give us a 70 watt heatsucker when we use that approach. But, hey, lets just get a couple.

Parallel tec systems used in chilled fluid reservoirs or tec-assisted rads and driven by voltages closer to the bottom of the range than the top may have a lot to offer.
Click to expand...


ickywu said:
You raise some good points but there are certainly some additional considerations. First, it does not need to be the goal of a tec-assisted water cooling loop to remove all or even most of the cpu generated heat. Providing chilled water an an otherwise serviceable cooling loop can give us a great advantage. the success of systems such as FreeZone and Amanda seem to demonstrate this.

Secondly, the inefficiency and current draw (implied also by the memeber who refers to the cost of using tec's in the comment following yours) are problems that rise with the voltage input and decrease as driving voltage decreases. All we have to do to start making the pelt look pretty good is to use say, 7 volts. Efficiency improves, cost decreases and hot side cooling is more manageable. Granted, that 320 watt pelt we just bought is only going to give us a 70 watt heatsucker when we use that approach. But, hey, lets just get a couple.

Parallel tec systems used in chilled fluid reservoirs or tec-assisted rads and driven by voltages closer to the bottom of the range than the top may have a lot to offer.
Click to expand...


All your points are too true, except for the Freezone, which is no better than a Thermalright 120U (at overclocking heat loads that is).

But TEC aided loops have their problems too... Unless you're using a huge TEC (or a bank in parallel) to actually chill the water (which would need it's own secondary cooling loop too...) you're just not going to remove as much heat as a triple radiator. If you use a radiator plus a TEC, you'll just be venting negative heat (cold air) from the radiator, and your water temperatures will stabilize around room temperature anyways.

About running that tec at 7v... Now, most TEC's peak efficiency at around 8-9v... There are many measures of efficiency, so isn't that inefficient in it's-self? Running a 3xxw tec at <100w?
 
Arcygenical said:
But TEC aided loops have their problems too... Unless you're using a huge TEC (or a bank in parallel) to actually chill the water (which would need it's own secondary cooling loop too...)
Click to expand...
Well, if we pick the right tec's we should be able to handle hot side cooling with virtually any decent air cooler. You're right, parallel tecs will be required and some attention to insulation. But, it actually looks like we can end up with a stable system that runs chilled water at pretty reasonable cost.

Arcygenical said:
About running that tec at 7v... Now, most TEC's peak efficiency at around 8-9v... There are many measures of efficiency, so isn't that inefficient in it's-self? Running a 3xxw tec at <100w?
?
Click to expand...

<g> Only financially for whoever is buying the tec. I betcha Porches' get pretty good gas mileage at 60mph...
 
thanksfor the replies guys, after reading this, i think ill stick to a 'standard' water loop.
 
You must log in or register to reply here.
Back
Top