Pumping acetone through standard heat-sink heat-pipes

[3ldon]

I've been looking at building a new system, so I've been debating how appropriate is it right now to build a system to handle a grossly over-clocked processor. (I have not decided what to go with yet.)

What I've found is there is a lot of data available on 50+ websites detailing the performance of various water blocks, heat sinks, radiators etc.
What i don't like is the price, I'm not about to spend $50 on a water block, $100 on a radiator and $40 on a water pump when i could hack together a system that works just as well for less.

Having said that lets take a look at some heat-sinks, we have generally speaking, 6, 8 and 10 heat pipe designs, all with various flow rates and fin area.
(6 pipes means 6 pipes leave the processor, i'm intentionally ignoring the fact that there may be 3 dual, or 6 separate pipes....)

If you look on frosty tech, the top 5 are all the same design, heat pipes in direct contact with the processor. and they all offer the same 15C temp rise at 150 watts heat or .1C/watt, regardless whether or not they have 8 or 10 heat pipes.

A quick look at what Newegg has available on the cheap, only one currently has exposed heat pipes, the Xigmatic S1283V; frosty places it at 18C/150W, it is a 6 heat pipe design! The rest are 6/8 heat-pipes, soldered into a drilled copper block, or nickel plated aluminum...and they vary between 23 and 28C/150 watt.

What this shows us with is that the heat pipe isn't the limiting factor, its the heat pipe to processor contact, and fin area.
Example:
If we take a 30mm cube of copper and apply 150 watts of heat to one side and an infinite heat sink to the other, it is trivial to see that there will be a 12C difference across it, and the same math for a 2.5 mm thick spreader that the heat-pipes are soldered to is an additional 1C just for the copper alone.

Looking at water blocks, considering that most of them have at least a 2 mm thick base, you can't possibly get better than .01C/watt;
(the APOGEE GT is at .06 to .04C/watt at .5 to 3 gpm)

What I want to discuss here is that it appears that pumping acetone through the existing heat pipes on a Xigmatic S1283V is going to get you better performance than a $60 waterblock, and it offers you a .12C/watt heat-sink for free. (Water is just too viscous to get inside the wick structure, assuming the heat pipes aren't using water to start with, (if they are then acetone will not be that much of an improvement)) This statement is based on the assumption that the thermal resistance is .03C/watt from processor to heat pipe, and .07C/watt from heat pipe to air. The thermal resistance of the heat pipes is almost negligible at .01C/watt.

Reading reviews of 70C core temperatures even with a water block, the thought occurs to me, why not seal off all the way around the processor with silicon and pump acetone underneath it as well?
The thermal conductivity of silicon leaves much to be desired.

[hokiealumnus]

Unfortunately, running anything through heat pipes smacks you up against the wall of restriction. The ID of those pipes is so small that you're hard pressed to get anything through it fast enough to cool effectively.

As far as putting the liquid in direct contact with the CPU, it's not very efficient for removing heat. The problem there is one of surface area, which is very small. I watched a thread at XS with great curiosity. A user lapped his Q6600 so that the cores were exposed (either that or de-lidded it, I'm not sure which) and designed a water "block" that shot a jet of water directly on the dies.

Temperatures were horrible. The reason for that is b/c there is not enough surface area for the water to pull heat from. You probably know water blocks have pins or channels in them carved out of the base. This is to increase the surface area the water hits on its way through. Without that, the water can't do its job properly (and acetone probably can't either).

So, my $.02 - Save yourself a lot of work (and headache when temps aren't what you want) by spending $50 on a decent water block...or even $80 on the best.

EDIT - This isn't to discourage innovation. If it's a project you want to do for fun and you enjoy it, by all means, have at it. We always like a good project. Heck, we could turn it into a frontpage article (whether results are good or bad). But if you're just trying to save money, there is a reason for the hierarchy of Air > Water > Sub-ambient (single stage) > Really cold (cascade/DI/LN2). Any chip with a high enough overclock will reach 70C and then some with water cooling. There is only so much ambient cooling can accomplish.

[Diggrr]

Acetone also would dissolve most plastics (pumps and tubing) as well as the epoxy that seals around the CPU cores, and there's the fact that it's highly flammable.
It would seem to be more expensive bypassing those faults than to simply watercool.

Check out the bong cooling thread to see a cheaper alternative to radiators. You would have to supply a pump still, and either build or purchase a water block, but you wouldn't necessarily have to use one of the expensive "preferred" pumps.
I'd guess though that even with a bong that outperforms most radiators, a heavily overclocked CPU would still see higher temps than what you might be seeking.

:welcome: to the Forums!