Water Block Fun
My overclocking fever started when I was around 12 or so, I remember having a Pentium 150mhz - ooh look - what do all those jumpers do! So my journey into the world of pushing your system to the max took me all the way to 166mhz - a nice 16mhz improvement!
This took me all the way to the present (with a few other things in between) - a custom water cooling system!
So I thought, how do you put water inside the “puter” without those pesky irons shorting the thing? Some research later, I delved into the black art with no money, limited skills, very limited time, some AS-level physics knowledge, armed with some water block designs such as #rotor
’s, my school's materials, and motivation from seeing the pocket-frightening prices of water-cooling equipment of today.The CPU blockTarget: Athlon XP 2400+
First compromise: no copper, as it’s quite expensive; therefore I use the next best thing: aluminum, which offers somewhat inferior performance. As I was sketching designs after seeing commercial blocks, the light bulbs appearing above my head were telling me: use as little metal as possible - for higher heat concentration, therefore faster heat response, have as much inner surface area for water/metal contact as possible, and have as much flow as possible. This led me to a design similar to #rotors, but instead of water flowing over the top of the peaks, the water will be 'passed through' the heat sink-like formation of the aluminum spikes, which were made by drilling lines of holes next to each other then linking them up to larger holes at the barbs. I later realized the need for wider channels, having to saw through the block sideways to connect the drill holes.
After this I had machined off the top and bottom of the block, leaving it in an H shape to make use of the anti-crush pads on the CPU. Now the problems started and many lessons learnt:
1) do NOT use super-glue where there is constant contact with water - found out during preliminary testing when the bugger leaked and leaked from different places, then the aluminum sides “snapped” off as the glue dissolved :-(
2) flow..FLOW!! the restrictive - high surface area meant that the pump was being pushed (1000lph Eheim I purchased off Ebay Germany had a 2M head - therefore much pressure), and I couldn't stop small leaks, coming through the silicone seal between aluminum/plexi, barb fittings, sides of block, therefore having to dismantle the block and widen the channels with my Dremel equivalent...
I completed the CPU block in 3-4 weeks which consisted of me going to the workshop for an hour every 2 day or so. The end result looked promising:
Cost so far - around £3 for 2 1/2" Danger den hi-flow barbs
First tests showed 27°C idle, around 34°C load on stock voltages (1.65Vcore) can’t remember exactly – but overall not bad.