What methods are there?

There are essentially two ways of cooling a PC.  The first is "passive" cooling.  This is using coolant (be it air, water, or whatever) that is at room temperature.  Heatsinks, with or without fans, and basic water cooled systems are this kind of cooler.  They cannot cool to below room temperature, but should still be able to cool things to within their specified temperature range.  The other way is active cooling.  This uses coolant that is colder than room temperature.  Not only will these remove the excess (above room temperature) heat, but they will also extract heat from the processor, taking it to below the ambient temperature.  Obviously, this provides better cooling, but it also has a number of problems associated with it.

There are a number of ways of achieving this cooling.  I have detailed all the ones I can think of (mostly variations on the same theme) and given a list of the pros and cons that I can think of.

 

Method		Description		Pros		Cons
Passive methods
		Any method that uses room temperature coolant.		No condensation.  No expensive coolers needed.  Simpler design.  Water cooled methods can use normal water.		Not as good as active methods.
A heatsink	A large metal lump is attached to the processor, which is cooled by the air which circulates round the PC.	It's very cheap and simple.  For low heat outputs (like I presently have) it works fine.  It's silent.  It's reliable.  It's cheap.	It needs good airflow round the case.  It's not suitable for hot cases, or overclocked processors.  It's not very interesting, either.
A heatsink with one or more fans	A heatsink (as above) that has one or more fans on it, which are normally connected to the motherboard.	The fans make the heatsink far more effective.  With large fans it can be good for some overclocking.  There are commercially available systems available, with up to three fans.	It still needs quite good airflow.  It doesn't actually get the heat outside of the case, so the rest of the machine will still get warmed by the processor.  It's quite expensive to buy the set-ups available, because the good ones seem only to be available in the USA, and so I would have to pay trans-Atlantic shipping.  It's still not very interesting.
Water cooled processor	Pass room temperature water over the processor (not directly over it; over a heatsink attached to it) which goes outside the case, to a radiator, which can then lose heat to the environment.	It works better than air cooling.  It gets the heat out of the case.  It doesn't rely on airflow.  It won't cost too much to build.	The design needs careful thought.  I have to be very careful not to get any leaks.
Water cooled processor, with fan	As the water cooled processor, but a turbocharged -- fan cooled -- radiator.	Even better cooling than the plain water cooler.  Probably the most effective warm-water cooler that I could make.	Same as the water cooled processor.
Active methods
		Any method that cools to below room temperature.		More effective than passive cooling, allowing for greater speeds.  It is independent (more or less) of the ambient temperature, giving good year-round cooling.  The fact that silicon chips do actually work faster the colder they are.  The street-cred associated with having a processor that is -40°C.		The big one is condensation.  Because air inside the case will get cooled, it will start to lose its moisture, and deposit it on my motherboard.  This would not be good.  With careful insulation and a bit of dessiccant in the case it can mostly be eliminated, but it could be a big problem.  Plain water will be unusable, as it will begin to freeze.  The introduction of anti-freeze will stop this, but is problematic because it reduces the thermal capacity of the water, and increases the conductivity (a lot).
On-chip peltier, with heatsink and fan	An on-chip peltier cooler, which is then cooled by heatsink/fan combination	It's very cold.	The peltier will make the case very hot.  The power consumption of peltiers is large -- I would probably need to uprate the PSU.  The heat still needs removing from the case.  There are questions about the reliability of peltiers.
On-chip peltier, with water cooling	Instead of water cooling the processor, water cool the peltier	It's very cold.  It gets most of the heat out of the case.  All the water cooling and peltier benefits.	All the water cooling and peltier drawbacks.  The biggest problem is still the power consumption, and the fact that peltiers spit out a lot of heat, so the case may still get heated too much.
Water cooled processor with peltier cooled radiator	The standard water cooling arrangement, but using a peltier, not a fan, to cool the radiator.	All the benefits of water and peltier cooling, but probably not quite so cold as water cooling the peltier.  The peltier won't heat the case any more.  The peltier no longer needs to draw power from the PSU.	Not quite so efficient as cooling the processor directly with the peltier.
Air conditioning	This is not my idea.  Use a peltier to cool the incoming air, so keeping the whole case cold.	It keeps everything cool.  It would mean that I wouldn't have to dismantle the processor.	It probably wouldn't get things that cool, and would need to be used in conjunction with a fan and heatsink on the processor.  It would need to make some sort of an airtight box to fit over the case fans.  It would create a nasty draught in my room.
Vapour-phase cooling	Like what you have in your fridge.  When liquefied gases vapourize, they take in large amounts of thermal energy, that is, they make their surrounds cold.  It's what Kryotech use.	It's pretty cold, and pretty effective.  A number of projects use this method of cooling in conjunction with peltiers.  This can work pretty well.	It's also a complete bastard to engineer.  You're working with gas under pressure, so it makes things that bit more difficult.  All the really good refrigerants are illegal/toxic/very hard to get hold of, and you need special equipment to handle them.  It's pretty expensive.  Compressors can be very noisy.

Back to Top