Introduction (The Idea)
I can't remember exactly when we first came up with the idea for the x86 still, but it's been discussed on a number of occasions over the last few weeks by Richard (the Co-webmaster at Exaflop) and I.
The basics of distillation are simple, you heat alcoholic fluid until the alcohol starts to evaporate, then you cool it so it condenses elsewhere into pure alcohol. To put it simply, the key components are a heating device and a cooling device. Now the average PC, has both a heating device (The processor) and a cooling device (The fan). The question we asked was "By separating the processor from its fan and adding a few small components, could we convince an ordinary PC to distil alcohol?"
A lazy Saturday afternoon, a quick trip to the model shop and I discovered the answer.
The Construction
Not wanting to risk any of my regularly used hardware I built the still onto a previously discarded AMD K6 (233mhz at 3.3v). I had originally wanted to build this onto my main system (The 'Pentium still' sounds better) but I figured liquids and deliberate overheating were a little too risky for the machine I depend on.
The Heating Chamber
To heat the alcoholic fluid I had to manufacture an airtight box with a single exit for the vapors. The box I decided should be 5x5x1cm, large enough to hold a full short and just the right size to fit snugly on top of the processor
I made the box from a single sheet of very thin brass bought from my local model shop, the first image is the cut metal folded into shape. The second image shows the completed heating chamber with the exit (and filling) tube. The tube was a short length of fine brass tubing bought from the model shop. The parts were soldered together and after a few dunks in the kitchen sink I managed to convince myself it was airtight.
The Cooling Assembly
I had originally planned to use a peltier heat pump in addition to the fan but I didn't have one laying about. The flexible tubing I had (that fitted over the brass tube) was to wide to weave into the heat sink so instead I cut the remaining brass pipe into three short lengths. The pipes were tube pushed into the fan and joined up with short pieces of the flexible tube.
The second image shows the completed assembly attached to the chosen motherboard. I was still a little unsure of the cooling arrangement so I put the cooling assembly into the freezer while I got everything else setup.
The Final Setup
These two images show the final assembly up and running. I booted to Linux and ran a small piece of code that stressed the processor. One piece of apparatus not shown here was a small bowl of chilled water the collection bottle was stood in (to further help with cooling). The heating chamber was half filled with rum, the nice dark brown variety.
About 10 minutes after startup I started to notice some condensation on the inside of the flexible tubing, at about the same time I noticed the heating chamber was very hot. Another couple of minutes and I witnessed the first drip of clear fluid from into the collection bottle. I ran the experiment for about 1hour before finally shutting down, I had needed to reboot the machine twice in that time (due to over heating) and decided to call it a day after Linux failed to restart (at the third reboot attempt).
The Results
After an hour of running the collection bottle contained about 2ml of clear fluid, a quick taste test revealed it was pretty strong. The first image shows the collection bottle (with a ruler for scale).
The ultimate test was of course burning it, the burn test provided a strong blue flame that was much brighter and larger than an earlier test using just the original rum. When the flame burned out there was a slight residue left behind showing that some water had been present (but only a small fraction). In a real still water is removed by repeating the process for purification, careful temperature control also produces better results.
All things considered the still worked and the experiment was a success.
If you want to link to this document, please link to
http://www.exaflop.org/docs/x86still/.
