I have always been the sort of person to try something different and when It comes to watercooling, I look at the cost and I think it's just not worth it! For that money I could have bought a better processor. That is why my quest from now on is to design and build my own watercooling from things around the house (and tech department at school). My first idea which has been in my head for a while now was instead of using pumps and radiators and such like - why not just put a water container straight on top of a CPU? This is where the jam jar came in handy.

This is the theory behind my jam jar idea. Use metal on either side of the lid as a medium of transferring heat from the CPU to the water. Fill the jar only 3/4 of the way, this will leave an air gap at the top. As the water heats up and evaporates it condenses on the cool glass (thanks to cool case air) and drops back into the water (therefore reducing saturation). I tried this using my Celeron 400@500 using 2.0v.

In order to build this contraption I needed a jam jar that was small enough to fit in-between the capacitors on my BP6 motherboard and unfortunately that's quite small. Luckily I found a jam jar that was small at the bottom and largerat the top.

Second thing to find was an aluminum plate. Being a mountain bike trials rider, I have plenty of old broken parts lying about - I found an old worn out "rock ring" which is basically a 3/4cm thick aluminum disc. It costs about £20. I superglued this to the top of the jam jar and began searching for a heatsink.

As the jar lid was quite small I didn't have much choice. My first find was a heatsink from an old 386. It's about 1cm high with 3 small fins. Much like 3 "W" shapes on top of each other. I could see it wasn't really up to the job but decided to try it. I superglued it to the lid and filled the jar 3/4 with water. Hey presto!

It sat snuggly on top of the processor and had good contact with the CPU.

At idle it was 29c - not bad at all (but then little heat is produced at idle). After 4 hours of stability checker it went up to 46c. To test for saturation, I stopped the tester and within 3 seconds it went down to 35c. After a short while it went back to 29c.

As you can see above, my theory worked and the condensation occurred - stopping saturation. I was not expecting the 386 sink to work that well at all, infact I didn't have that much faith in the system full stop, but from the reasonable results I decided to try a bigger sink. The problem was the heatsink I wanted to use was too big, so I had to chop the corners off. I lapped it for good contact.

I find the next results slightly surprising. The bigger heatsink made no difference to temperature. The only difference I did find was that temperatures changed slower as the heatsink acted as a store for heat. The 386 sink tended to go from extremes in a short space of time while the bigger sink took a while.

I'm not recommending this at all - it wasn't that much better than a standard H/S fan combo and just gives more worries about water leakage's. Not to mention the fact retaining it there would be difficult. Hopefully all my experiments will lead me to designing a new and compact form of water cooling? Watch this space.

Spode