After the "success" of the first water cooling with a jam jar project I decided to try a few different things. Firstly was to try and make the worlds smallest water cooling device using a very very small jam jar. The second thing I wanted to try was not just water as a solution - by suggestion of one of our readers, I wanted to experiment with alcohol(!).

The worlds smallest water cooler.


I think I can officially say that this creation is the worlds smallest water cooler. No one could ever be silly enough apart from me to try this (and that says a lot). I am using a 25ml jam jar and a 386 heatsink in combination to create the worlds smallest water cooler. I ran out of superglue in the last experiment but I got some more so the experiment went smoother. There were two experiment tried with the small jar.

Experiment 1

On the bottom of the lid, for better contact I super glued a lapped piece of copper (actually the remains of my dead celeron!) - on top of the lid I super glued the heat sink. After leaving it to dry I had good thermal transfer between the CPU and into the jar. In the jar I placed water.

When a CPU is at idle it gives out absolutely hardly any heat. Even this very poor cooler was capable of 28c at idle. I loaded up quake and played for about 2 minutes and the temperatures shot up to 53c. I decided to stop their and loaded up Macromedia Dreamweaver to start writing up this report - I thought it would not rise much but in a matter of 2 minutes, it hit 65c!! This shows two things.

1) The celeron processor is very stable when overclocked.

2) Damn that water was saturated!

Experiment 2

This was the biggest waste of time I have ever done. I lapped the glass bottom of the glass from a very arched curve to being completely flat. Unfortunately like this, I could not attack a heatsink as the bottom of glass inside was still an arch and there was no way of lapping it from the inside. But either way it was still pants.

Although I have made the smallest water cooler in the world, it truly is terrible. Although my evaporation technique to stop saturation works with the bigger jar, it obviously wasn't fast enough to cope with little water, it needs a certain volume to work efficiently. All together this was a rather fruitless experiment which only made my wrist ache.

Experiment with alcohol

Everyone who knows anything about water cooling know that water is the best solution for a waterblock/pump method water cooling - it has the best specific heat capacity. But in the case of my experiment what might help is faster evaporation as it would keep the water cooler if it evaporated more. Alcohol has a low boiling point, but also has not got a very good specific heat capacity. There is also the problem with vapours being ignited.

To conquer both these problems I have solutions. Firstly it is in an enclosed container and temperatures will probably be only just enough to evaporate it, not ignite it - if it did there would not be enough oxygen to burn much. Secondly I will be using a mixture of water and methalated spirits. That way the water will hold the heat from the heatsink because of it's good specific heat capacity and then transfer it to the alcohol which will quickly evaporate off and cool down the solution. Using the same setup as my original jam jar experiment with the super glued heatsink on and aluminium 3/4cm thick plate, I tried it. I used about 50/50 solution.

Bascially it didn't perform very well. After 10 minutes of 100% processor load I got 53c. I think these series of experiments have been pretty useless as a cooling device, but it has cer4tainly opened my mind. I do not at all suggest trying this. I do have a reasonable suggestion based on this which involves, A waterblock with an almost minuture pump to slowly move the water around and a coil of thin copper tube with a fan attatched to it. The pump would move the water round the copper tubing and be cooled down by the fan. Because of the size of it, it could also be properly attatched to the cpu and with the case vertical. This would also aid the convection currents. Maybe for another day...

Spode