Well.. Not really, but it's the most powerful computer the BrainLube lineup has ever had.
I've always been in love with the machines not quite attainable. I like the style, fonts, and shapes/sizes of the 70's, 80's, and early 90's high dollar machines. Why not have some fun with it! The " - McAlpine - Universal Workstation and Control" fictious company takes design queues from the SUN Workstations, The NEXTstaions, and the DEC Mini-Computers. Along with a few SGI design aspects. I'm no designer, just me having fun with Inkscape and CAD.
The logo is a diamond or square paying homage to the Next and SUN machines, where The 'U' (for Universal), with the following 'U' (for Workstation), preceeds the 'C' (for Control). Just like the SUN logo did so many years ago. If one really tries, one can also see a "mirrored in the design: M C A L P I N E.
The computer is your basic x86 machine except for a few additions. I have created an arduino device that (amungst the other things it does) there is a silicon based scentillation counter (gieger counter) that I strapped a gamma/beta emitter to (urainium test source.) I use it as a True Random Number Generator. The idea is, there is a bit that is flipped at the speed of the processor. When a scentillation is registered, the bit "registers". Meaning the bit quits flipping and is set (at either a 1, or a 0.) Next, the process is repeated, and if the result is the same, the bit is recorded as the first bit in a 32 bit integer, if they "vote" and the results are different, the process is repeated until both are a 1 or both are a zero. The arduino does this 31 more times, until a 32 bit integer is made. I then grab this number from a serial terminal interface and have my random number. Each number takes about 5 minutes to produce. Now ask me why. I don't know, but it's great for coming up with REALLY random passwords! ... So that's in there as well.
The cooling system is also somewhat different. There is a peltier junction array in the machine as well. Not to directly cool the CPU, but just to lower the temperature of the thermal cooling mass of the water circuit. The idea is not to cool the system (the hard and ineffecient way), but to keep the CPU from going into the effecientcy cores as quickly. People do these long speed tests, but in my workday I only have spurts of time where I need massive processing power.
BIG. Really Big. I want a gaudy mega computer in my house.
Mission accomplished! The frame is make from what is called 80/20 (8020.net). It is aluminum extrusions to a standard that allow standard fittings to combine extrusions, and add features that are already made. The panels of the computer are .25" Linen Micarta. These panels are from a company called sendcutsend.com, we live in an exciting time, where I can get the material cut/machined for cheaper than I can purchase a single sheet of the material if buying in small quantities.
The front terminal is a "Clockwork Pi" my friend Rob was good enough to give me for Christmas. The idea here is to use it to SSH into the main computer through the "Cool-Retro-Term" shell emulator and run something like top, btop, htop, bpytop, etc. to see the current processes of the main machine.
The bottom panel is made from 7075 aluminum (only a few dollars more shipped than 6061 after machining), and is designed to mount 15 120mm Radiators/Fans. Two of the slots are taken up by casters, an afterthought but I can't imagine needing them!
The motherboard is mounted to an ATX "Open Test Chassis" and that frame is then mounted to the micarta plate via munting blocks I 3d printed out of PC-CF, that is, poly-carbonate, with a carbon fiber (short chopped) matrix. This is a material from Prusa and it works great on even open 3D-printers. The carbon fiber, aside from being SUPER tough, also helps prevent warping on big parts.
The Panels were sized around what needed to be hung on them in the inside. With the 80/20 rail extrusions this comes out to pieces (1" extrusions) that are 28.4125" and 21.35" long. This is for 4 panels all the same size (though three of them are .25" micarta, and one is aluminum) and the remaining 2 front panels are larger. It can be a bear fitting the micarta panels into the 80/20 because of tolerance stacking, so some of the panels must be sanded (wet sanded, don't breath micarta dust), and some of the rails were file-ed on the inside. That, and a rubber mallet was all that was needed to fit the panels.
Because some of the panels were such a tight fit, taking the 80/20 back apart when more part's arrived was sometimes a chore. One of the reasons was that you must send the "T-Nuts" down the 80/20 from one end or the other. So for non critical fasteners, I sanded the sides. This allowed me to place them in from the broad sides of the 80/20. When the screw turned the nuts during assembly they held just fine. (Sort of like "M-Lock".)
