Using HVAC to cool cpus

Toggle sidebar Toggle sidebar
Gam3r01

Gam3r01

Titan
Jan 12, 2013
25,417
976
115,440
So I have decided to revamp my approach since my older post.

A micro HVAC system should in theory work correct?
A coolant, undecided on what so far, would super cool air, which would then circulate though fins on a cpu cooler such as a stock AMD cooler (covered in a shroud to keep one flow) would then cool the fins, allowing them to take in more heat from the copper heat pipes.
Once heated this air, or rather fresh air would be sent through a compressor with coolant in it, supercooling the air, then put across the cpu again.

Firstly am I correct in assuming that super cooled air would allow for more heat to be dissipated?
Second if not, then would super cooling the fins of the cooler themselves work? (An endothermic reaction inside the cooler)
Third, If it would work, any predictions on how well? It would have to surpass air cooling, and possibly even water to be worth while to research.
If not, then I can fall back on endothermic.

The endothermic approach is reacting inside the cpu to cool the fins themselves.
 
Solution
Drew010
In order for a system like that to work, you would have to have a medium that would transfer the heat away from the CPU IHS/die efficient enough to be cooled, which when using aluminum like the stock heatsinks are mostly made out of, wouldn't work very well. My suggestion would to either build a custom water loop, or buy a closed loop one like the H100i, and run the cold air through the radiator. That in my opinion is the best way to implement your idea.

Unless you are thinking more along the lines of peltier cooling, which is a whole other world.

But to answer your questions:

1) in theory, it will work, yes.
2) effectively, having a system that will super cool the air will allow more heat to be dissipated, thus allowing for higher...
Sort by date Sort by votes
Drew010

Drew010

Honorable
May 11, 2013
1,150
0
11,660
In order for a system like that to work, you would have to have a medium that would transfer the heat away from the CPU IHS/die efficient enough to be cooled, which when using aluminum like the stock heatsinks are mostly made out of, wouldn't work very well. My suggestion would to either build a custom water loop, or buy a closed loop one like the H100i, and run the cold air through the radiator. That in my opinion is the best way to implement your idea.

Unless you are thinking more along the lines of peltier cooling, which is a whole other world.

But to answer your questions:

1) in theory, it will work, yes.
2) effectively, having a system that will super cool the air will allow more heat to be dissipated, thus allowing for higher voltages/overclocks.
3) super cooling the fins of the cooler itself would work, however that would be more like a peltier cooler
4) see first big paragraph
 
Upvote 0 Downvote
Solution
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,445
17,910
184,590
It seems what you're describing is a miniature version of a typical window AC unit. Compressor, some refrigerant, pump, constant power...

In theory, that would work. It would be loud, expensive, fail-prone, and the 'refrigerant' would be a sticking point. But it might work.

I wouldn't.
 
Upvote 0 Downvote
Gam3r01

Gam3r01

Titan
Jan 12, 2013
25,417
976
115,440
The main reason I am considering all of this is for academic purposes. So even if it is improbable now, the idea is to refine it. I am simply looking for ideas at this point. It appears to me and after the input that a endothermic route, like I have been considering earlier, would be a better choice.

I simply need to find a way to get a reaction going inside a waterblock, and a way to get the products of the reaction out of it to be reversed in a different area. If I can get it in the block, reacted, and out the rest is simpler.
 
Upvote 0 Downvote
hunter315

hunter315

Champion
May 12, 2009
10,653
3
51,960
Your first example seems to be just pumping cold air into a computer case which will work just fine, heatsinks work relative to ambient temperature, so if ambient is 30C and the heatsink runs ~15C above ambient at that power load you will be at 45C, if you drop ambient to 10 you will be down to 25C, the problem here is you can't really get air much below about 5-10C with an AC unit, you would have to tap off a freeze to get a bigger drop in air temperature. An easier way to fix this problem is just to get a CPU cooler with a better W/C rating which will give you better cooling without the need for exotic methods that use more power to get rid of the extra heat, look at the frostytech table for reference.

If you want an interesting cooler, there are phase change coolers and peltier coolers that are active coolers, peltier coolers use electricity to syphion heat away from the heatsink but have issues with condensation around them.
 
Upvote 0 Downvote
Gam3r01

Gam3r01

Titan
Jan 12, 2013
25,417
976
115,440


Yeah phase change and peltier seem to be the best base idea for this. The main reason is for this research it needs to be a novel idea, so in other words exotic enough to be unique.
 
Upvote 0 Downvote
Drew010

Drew010

Honorable
May 11, 2013
1,150
0
11,660
Well Peltier cooling is certainly rarely used other than for demonstrational purposes, or just so someone could have bragging rights. What is the scale that your idea can be? This may sound crazy, but I think it would be awesome to have a build where you have the reservoir inside of a mini fridge/freezer where the temperature is kept ~2C to prevent freezing. Then you could use rigid acrylic with some insulation to prevent condensation from shorting out your other components. And keep your beer cold at the same time ;P
 
Upvote 0 Downvote
Gam3r01

Gam3r01

Titan
Jan 12, 2013
25,417
976
115,440
I agree a mini fridge wouldnt work, but something with the same concept should, right? Something designed to cool like that, and to run long without failing.

As for the scale right now its anything goes (within reason, no cargo containers :D)

I think cooling a coolant(redundant much?) then circulating it would be a good idea, finding a coolant with a low, low freezing point would allow for more heat to disperse into it, and I can isolate it to keep condensation from ruing the pc.

Monetary I currently have two dollars to my name (Woo go me!) but I am in the process of applying for a STEM grant, which supports science, tech, etc to fund projects like this one. Best part is it is from 20-75 thousand dollars. So if I can get funding I can really do anything needed. Living in the mountains in CALI I grew up learning to work with my hands. So any fabrication I have the know how to do, and the shop at school to do so.
 
Upvote 0 Downvote
nikoli707

nikoli707

Splendid
Jan 4, 2012
6,008
5
29,965
you would be better off getting a water block for the cpu and modifying it so that it can accept the actual refrigerant flowing through the block. for the hvac system... the refrigerant is at is coldest on the low side after it goes through the expansion valve and is in liquid form before it goes through the evaporator. it is usually ice cold at this point and would be far far better at removing heat from a cpu than a standard air water cooler. i have always wondered by there hasn't been a computer sized hvac system or an entire computer case designed around it. although the compressor is always going to be noisy... much noisier than standard fans... but there is no reason why it shouldn't work.
 
Upvote 0 Downvote
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,445
17,910
184,590
Viable 'solution', or viable 'alternative'?

In the consumer space, air -> water -> active cooling. Increasing complexity and noise.

In the server farm/business space...I want to cool the whole room, rather than each CPU. Above all, it must work 99.999% of the time.

I could see hoe this could work...basically building a mini AC unit and pumping refrigerant past the CPU and then the cooling fins. But who would buy it, outside of a tiny group of extreme OCers?
 
Upvote 0 Downvote
Gam3r01

Gam3r01

Titan
Jan 12, 2013
25,417
976
115,440
At the moment I am not interested marketing it as of yet, I would like to later on, sooner the better, but right now I just need a proof of concept to gain some attention, and some money through a national symposium/academic purposes.

If I can get it to work, I can simplify it, refine it, and improve reliability and then market it.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom