sappo
09-08-02, 09:22 AM
I’m pretty sure I’m going to ruffle some feathers here, but I’m wondering what you guys think about a smaller maximum temperature difference on peltiers actually being a GOOD think… (or maybe I’ll just ruffle feathers because this is nothing new, hopefully it’s the former)
We all know the formula: dTload = (1 - (heat load/max cooling power)) * max temp difference
At first glance, it looks like the maximum temp difference is a good thing. If the peltier has a small max temp difference, the peltier must suck, right?
That is certainly one possibility --the peltier just might be inefficient.
However, I think there might be a second reason a peltier could have a small max temp difference --one that is actually GOOD for a Watt hungry CPU.
How can that be? I think that a small maximum temperature difference could be good thing because it shows that the peltier is more thermally conductive. This would especially be a desirable thing with some of today’s Tbred’s that are putting out more heat than my desk lamp!! Plus, it’s not a happy thought knowing that if your peliter stops working for some reason, you essentially are stuck with an insulated CPU.
As my understanding goes, the dTload equation just takes into account how much cooler it will make the CPU if you ignore the amount of heat that is actually CONDUCTED through the peliter.
I’m thinking that with today’s CPU’s it’s best to get the best of both worlds: to transfer heat with conduction AND thermoelectrically. Realistically, to cool a power hungry Tbred you need a peltier in the 200+ watt range. I’m not sure if that’s worth it where live. Power is expensive here. Running that big of a peltier here for a month straight, I’d be paying about $15/month. That’s about the difference between dialup and DSL.
So here’s the million dollar question: If you have an extremely conductive (say close to aluminum) and extremely thin (to make it even more conductive) peltier, that’s running at a given wattage, would it make sense to subtract a percentage of the peltier wattage from the CPU (because that heat is being taken out thermoelectronically) and assume that the rest of the wattage from the CPU be carried out via conduction? If the answer is yes, you could have low wattage peltiers on an 80 watt Tbred and it would bring down the temp quite a bit.
So for example: Your ultra-conductive peltier is 50 watts. You have an 80 watt CPU. Thermoelectricly, lets say it will transfer 50% of those watts directly to the heatsink, convection will carry the rest. So essentially what you end up with is a 55 watt CPU.
Think about it. If the power was off, a heatsink/peltier combo would still be very conductive, and if you turned it on, it would be even better. Right? Or would the temperature difference on different sides of the peltier create energy that would prevent the peltier from functioning??
Gone are the days that you can realistically bring a CPU below ambient with a decent sized peltier, but if my ultra-high conductive peltier theory is plausible/possible the difference between the CPU and ambient temperature can be very efficiently reduced..
I just have a funny feeling that peltiers would work a whole lot better if they were much more conductive.
So I’m asking, does this make any sense, or am I talking out of bodily orifices (orifici?) that ought not be talked out of?
We all know the formula: dTload = (1 - (heat load/max cooling power)) * max temp difference
At first glance, it looks like the maximum temp difference is a good thing. If the peltier has a small max temp difference, the peltier must suck, right?
That is certainly one possibility --the peltier just might be inefficient.
However, I think there might be a second reason a peltier could have a small max temp difference --one that is actually GOOD for a Watt hungry CPU.
How can that be? I think that a small maximum temperature difference could be good thing because it shows that the peltier is more thermally conductive. This would especially be a desirable thing with some of today’s Tbred’s that are putting out more heat than my desk lamp!! Plus, it’s not a happy thought knowing that if your peliter stops working for some reason, you essentially are stuck with an insulated CPU.
As my understanding goes, the dTload equation just takes into account how much cooler it will make the CPU if you ignore the amount of heat that is actually CONDUCTED through the peliter.
I’m thinking that with today’s CPU’s it’s best to get the best of both worlds: to transfer heat with conduction AND thermoelectrically. Realistically, to cool a power hungry Tbred you need a peltier in the 200+ watt range. I’m not sure if that’s worth it where live. Power is expensive here. Running that big of a peltier here for a month straight, I’d be paying about $15/month. That’s about the difference between dialup and DSL.
So here’s the million dollar question: If you have an extremely conductive (say close to aluminum) and extremely thin (to make it even more conductive) peltier, that’s running at a given wattage, would it make sense to subtract a percentage of the peltier wattage from the CPU (because that heat is being taken out thermoelectronically) and assume that the rest of the wattage from the CPU be carried out via conduction? If the answer is yes, you could have low wattage peltiers on an 80 watt Tbred and it would bring down the temp quite a bit.
So for example: Your ultra-conductive peltier is 50 watts. You have an 80 watt CPU. Thermoelectricly, lets say it will transfer 50% of those watts directly to the heatsink, convection will carry the rest. So essentially what you end up with is a 55 watt CPU.
Think about it. If the power was off, a heatsink/peltier combo would still be very conductive, and if you turned it on, it would be even better. Right? Or would the temperature difference on different sides of the peltier create energy that would prevent the peltier from functioning??
Gone are the days that you can realistically bring a CPU below ambient with a decent sized peltier, but if my ultra-high conductive peltier theory is plausible/possible the difference between the CPU and ambient temperature can be very efficiently reduced..
I just have a funny feeling that peltiers would work a whole lot better if they were much more conductive.
So I’m asking, does this make any sense, or am I talking out of bodily orifices (orifici?) that ought not be talked out of?