Can you cool a CPU with air to below ambient temp?

[Ike0069]

Ok, there is a guy I know saying that his P4 (I don't know the speed) runs at 57F (yes farenheit), when the room temp is ~70F. Hes is using this HSF.
I immediately called BS, but a friend backed him up. So I do quick search on google and find this article.
Some quotes:
Going Below Ambient - Going below ambient temp ( ambient air around the heat exchanger) is not possible with normal Heat Sink or straight water cooled systems.

100% Efficiency - Its impossible to hit 100% efficient. Its also pretty hard to tell what your system is really running at, but if you get figures that your system is 100% or above... time to re-calibrate some of your temp tools.

Well then he replies with some crap about Bernoulli's Principle of Pressure. Accept it's not crap. It basically states that if you decrease the size of a fluid (gas or liquid) path, the velocity increases while pressure and temp. decrease. He's saying that as the air is forced into the narrower crevices of the heatsink, the velocity of that air is increased, thereby decreasing both pressure and temperature. Now this cooler than ambient air is reducing his CPU temp to below ambient temp. This actaully makes some sense, but I know there is now way it decreases air temp enough to get the CPU down to 57F.

Can someone give me some more insight into this? Any physicists on this board?

 

[Googer]

Originally posted by: Ike0069
Ok, there is a guy I know saying that his P4 (I don't know the speed) runs at 57F (yes farenheit), when the room temp is ~70F. Hes is using this HSF.
I immediately called BS, but a friend backed him up. So I do quick search on google and find this article.
Some quotes:
Going Below Ambient - Going below ambient temp ( ambient air around the heat exchanger) is not possible with normal Heat Sink or straight water cooled systems.

100% Efficiency - Its impossible to hit 100% efficient. Its also pretty hard to tell what your system is really running at, but if you get figures that your system is 100% or above... time to re-calibrate some of your temp tools.

Well then he replies with some crap about Bernoulli's Principle of Pressure. Accept it's not crap. It basically states that if you decrease the size of a fluid (gas or liquid) path, the velocity increases while pressure and temp. decrease. He's saying that as the air is forced into the narrower crevices of the heatsink, the velocity of that air is increased, thereby decreasing both pressure and temperature. Now this cooler than ambient air is reducing his CPU temp to below ambient temp. This actaully makes some sense, but I know there is now way it decreases air temp enough to get the CPU down to 57F.

Can someone give me some more insight into this? Any physicists on this board?

He is probably using an air-cooled pelteir. Thats the only way to make a claim like that.

And I find the farenheit scale to be better for measureing temperatures over celcius. Becase celcius operates on a scale of 1-100 and farenheit is between (for the purpose of this conversation) 1-200, so a 1°f change is like a ½°c change in temp. Most motherboards dont do ½° measurements and it gets rounded off to the next highest value.

 

[MDE]

It's not possible with a traditional air cooled heatsink setup, including the heatsink you linked to. If you're using 50F air to cool a 75F hunk of metal, you're not getting under 50F no matter what.

 

[Ike0069]

Originally posted by: Googer
He is probably using an air-cooled pelteir. Thats the only way to make a claim like that.

And I find the farenheit scale to be better for measureing temperatures over celcius. Becase celcius operates on a scale of 1-100 and farenheit is between (for the purpose of this conversation) 1-200, so a 1°f change is like a ½°c change in temp. Most motherboards dont do ½° measurements and it gets rounded off to the next highest value.

He says he's just that HSF I linked. I'm not sure what exactly a pelteir is though.

 

[sodcha0s]

Impossible. I am aware of that pressure principle you mentioned, and yes it is true, I have seen the effects of it...... but any kind of normal HSF combo isn't capable if actually achieving it. It takes quite a bit of pressure going through a fairly small area to see the effects. If it was in fact getting that cold, there would surely be evidence of condensation on the HS.l

 

[akugami]

I call BS as well. If you can, get one those temperature probes and tape it to the bottom of the heatsink as near as you can to the bottom where the CPU is and I'll bet it gets hotter than 57F. It's probably more likely that it's 57C than 57F. 57F would be a breezy 14C and most people running water coolers with A64's and P4's run in the 30's-40's.

See if you can get a heat temperature gun. Like one of those used in food services to check surface temperatures without actually sticking a probe onto the surface (usually food) you're checking.

If you can, ask him if you can go to his house to check the temperatures because someone else you know said even with Bernoulli's Principle of Pressure it is impossible to get 100% efficiency. Forget getting over 100% which he would need to get below ambient temps. If he refuses you know he's lying out his rear. Bring a digital camera too. I'd really like to know what his setup looks like if he's claiming over 100% efficiency in cooling.

Another reason to call BS is this. If it was that easy to get such an efficient cooler using a HSF, even a fancy one, don't you think AMD and Intel would have them out? At the very least they'd release them for the Opterons, FX's, Xeon's & EE's. And hey, let's not forget about IBM's Power CPU's which in configs like Apple's G5 dual CPU systems run so hot that they require water cooling. Don't you think Apple rather release a solid state cooling method rather than using liquid and the inherent dangers that come with liquid cooling?

 

[Jeff7]

If that were possible, to easily get an object below ambient temperature with just a heatsink, there wouldn't be any problem at all cooling something to absolute zero. But that's not the case, as we need advanced methods like pulsed lasers to cool clouds of atoms to near-absolute zero.

 

[Greenman]

The diode on his motherboard is wonky, as most of them are.

 

[imported_Mitchell]

Not possible.

 

[frank84]

if thta was possible,

water cooling would never sell

 

[L00PY]

Originally posted by: Ike0069
Can someone give me some more insight into this? Any physicists on this board?

Impossible of course. Just going to the manufacturer's webpage shows a non-negative C/W (0.36).

If you really want to prove it's impossible physics wise, you can plug numbers into Hyperphysics Bernoulli Calculator and see that you'd need the air moving at an near infinitely high speed to get a pressure drop high enough to drop temps below ambient.

 

[mooncancook]

that's against the law of physics. He needs his temperature gauges checked.

 

[Ike0069]

Yea, these replys are exactly my thoughts. I did alot of research today on reviews of HSF's. The lowest temp I saw was 5C (9F) ABOVE ambient temp at idle and that was on a XP-120. Most were 10C or more above ambient temp.

I also did some Peltier units and discovered that getting 13F below ambient (assuming 70F ambient for his office) is still almost impossible. It defintely would not be realistic to expect that much cooling under normal conditions.

Now I can't go check this out for myself because he lives 10 hours away, but I do know both of these guys that I'm arguing against and they are not stupid. I'm starting to wonder if they are just not wanting to admit thay are full of it, but actually realize they are. Beacause after doing some quick reasearch, I easily found several sites (5 actually) that state specifically that you can not get below ambient on this type cooling. I linked all the sites for thier review and I'm awaiting their reply.

 

[lobadobadingdong]

my guess is his thermal sensor is f'd up, or that the temp they are seeing is actually Celcius.

 

[ribbon13]

I have an extremely expensive TEC setup and I'm lucky to get 7°C below ambient.

My guess is they're afraid of FIMS (foot in mouth syndrome)

 

[chinkgai]

Originally posted by: akugami
I call BS as well. If you can, get one those temperature probes and tape it to the bottom of the heatsink as near as you can to the bottom where the CPU is and I'll bet it gets hotter than 57F. It's probably more likely that it's 57C than 57F. 57F would be a breezy 14C and most people running water coolers with A64's and P4's run in the 30's-40's.

i bolded the most logical explanation to why your friend is a moron.

 

[harrkev]

My guess is that his temps are 57C, and he just thinks that it is 57F.

 

[Traire]

Originally posted by: harrkev
My guess is that his temps are 57C, and he just thinks that it is 57F.

Which is typical for a Prescott P4 temp.

 

[shira]

Originally posted by: MDE
It's not possible with a traditional air cooled heatsink setup, including the heatsink you linked to. If you're using 50F air to cool a 75F hunk of metal, you're not getting under 50F no matter what.

Actually, with evaporative cooling, it IS possible to get temperatures below ambient. But in order to accomplish this, you'd need to place a wet pad (presumably wicking water from a nearby bucket) in close proximity to the CPU.

The principle is: When a liquid changes state to a gas, it must get energy from the surrounding environment, thus cooling the surroundings. This is why we sweat: evaporation of sweat cools us off.

 

[Nnyan]

Your friend is for lack of a better term, Full of Hot Air. What we are talking about here is Forced Convection Heat Transfer (To be more specific you get Conduction along the HS/fin and convection from it surface to the air):

"The heat transfer coefficient is related to the heat flux and the temperature difference between the cooled surface and the coolant (air/water) by Newton's Law of Cooling, given by Q = hA (Ts - Tf) where h is defined by h = q"/(Ts - Tf)".

You heatsink (HS) is always cooler then your CPU see if your friend knows why? Its impossible to get 100% thermal transfer from the CPU to the HS. You can increase transfer by using some really good TIC (AS5) and lapping your block to increase contact but you will still get "air pockets" which will decrease your transfer. But in the end nothing is a perfect black body with perfect emissivity.

Material Thermal Conductivity
silver 422 W/mK
copper 402 W/mK
gold 298 W/mK
aluminium 226 W/mK
steel 73.3 W/mK
lead 34.8 W/mK

With a HS the heat transferred is proportional to the difference between the HS temp and the air temp. All things being constant doubling the differential doubles the heat transfer.

Using Bernoulli's Principle of Pressure to scam you is pretty lame. You will not get enough of a pressure drop to enable you to go below ambient. Before you even got close the pressure drop would actually block airflow reducing your cooling.

Thats not even bringing up variables like boundry layer, fin joint connections and type of materials used in the HS and materials specific heat.

 

[Ike0069]

Originally posted by: Nnyan
Your friend is for lack of a better term, Full of Hot Air. What we are talking about here is Forced Convection Heat Transfer (To be more specific you get Conduction along the HS/fin and convection from it surface to the air):

"The heat transfer coefficient is related to the heat flux and the temperature difference between the cooled surface and the coolant (air/water) by Newton's Law of Cooling, given by Q = hA (Ts - Tf) where h is defined by h = q"/(Ts - Tf)".

You heatsink (HS) is always cooler then your CPU see if your friend knows why? Its impossible to get 100% thermal transfer from the CPU to the HS. You can increase transfer by using some really good TIC (AS5) and lapping your block to increase contact but you will still get "air pockets" which will decrease your transfer. But in the end nothing is a perfect black body with perfect emissivity.

Material Thermal Conductivity
silver 422 W/mK
copper 402 W/mK
gold 298 W/mK
aluminium 226 W/mK
steel 73.3 W/mK
lead 34.8 W/mK

With a HS the heat transferred is proportional to the difference between the HS temp and the air temp. All things being constant doubling the differential doubles the heat transfer.

Using Bernoulli's Principle of Pressure to scam you is pretty lame. You will not get enough of a pressure drop to enable you to go below ambient. Before you even got close the pressure drop would actually block airflow reducing your cooling.

Thats not even bringing up variables like boundry layer, fin joint connections and type of materials used in the HS and materials specific heat.

Very nice Nnyan. Study much? LOL :laugh:

 

[The Boston Dangler]

Not to mention, the higher velocity of the air means less time to absorb the heat from the HSF. Formula 1 cars fine tune air chambers for "perfect" heat transfer and flow, not Coolermaster. That heat sink DOES look pretty badass tho.

 

[Nnyan]

Don't get me talking about the condensation he would see around the CPU. = )

 

[ribbon13]

Nothing to date competes with it except for cascade using the same VP method or LN2.

 

[Ike0069]

Originally posted by: ribbon13
I have an extremely expensive TEC setup and I'm lucky to get 7°C below ambient.

My guess is they're afraid of FIMS (foot in mouth syndrome)

It looks as though you are correct ribbon. I haven't heard any more comments since my last attempt to educate them. Maybe it finally worked.