As processors became faster, and doing so at a phenomenal pace, the drive to overclock began to fade and my focus was now turning to dual processor setups. At the time Windows 2000 did not exist, so it was Windows NT 4.0 as the operating system of choice. Over time however, I went thought several different configurations (upgrades) to where I finally am now. It was a long road, but I have to say that dual CPU setups have certainly come into their own.
During that upgrade period, the modding scene began to unfold. At first it seemed to be a natural evolution to overclocking as many were installing additional fans and cleaning up the wiring (me included) as a necessity. As it continued to evolve though, ascetics seemed to come into play as a lot of people began doing their own home grown mods.
In the search for more and more mods, I began discovering other web sites dedicated to the craft of modding. Virtual Hideout was probably the one that truly pushed me over to the modding scene completely. I remember seeing a mod with skateboard wheels instead of the typical casters and thought that it was a really cool idea. I began the search for skateboard trucks and wheels, but soon realized that skateboard wheels were not going to work, or more specifically, skateboard trucks. They were too small for the case I was working with, the Enermax FS2200 (the big brother to the FS1500 that I used for Heavy Metal). Then one evening, as I was over at my cousins for a birthday party, he took me in the basement to show me his new racing go-kart. It hit me right there that this was the solution to my dilemma; go-kart wheels!
Above, you can see the custom made chassis and the kart parts attached. The chassis was designed in AutoCAD and sent over to a local machine shop. The axles used are both rear 40mm kart axles that were cut down to fit the new chassis. Almost everything else is off the shelf go-kart parts. The last two shots are with the belly pan on to check for fit. It was at this point that I realized additional support was going to be needed to handle the weight of the computer. The reason for the additional support is due to the mounting point now being in the center of the pan as opposed to the edges. A shot of the support plate will be shown later in the final assembly. For those interested, almost all the go-kart parts were purchased from TS Racing.
These are the intake and exhaust stacks back from the machine shop and before polishing. The three you don't see are the ones that didn't make it out of the machine shop in tact. The rear 120mm exhaust stacks proved to be quite the challenge to machine.
One picture I seem to forget to take in the process of the build was that of the front door headers. The original doors had plastic headers that really didn't seem to fit the theme. When I designed the new headers to be made out of aluminum, one of the biggest goals was to give the case an overall cleaner look. The new headers were made out of aluminum and polished to accent the front doors.
Trial assembly: Here you can see the new aluminum headers before they were polished. Also visible are the pressure cut flames, tachometer, lighted switches, window, pressure cut power supply mount, wing, and one chrome face plate. My first thought after the trial assembly was that the tires looked too small and would have to be replaced. I eventually decided on 6" Burris wheels with an 8" width for the rear and 7" for the front. As you will see in the final build, it made a huge difference.
I have gotten several questions on the flames and the window on how I got everything to line up. First, all the panels, including the Lexan for the window, were all pressure cut at the same time the flames were. I gave Cutting Edge Water Jet Service CAD drawings for all the parts, which included under-sizing for holes that were going to be tapped. I knew that if the parts and panels were all pressure cut that the end result would be a perfect fit.
The wing on the back from Taylor Wings, was originally designed as a fender wing for a semi-tractor. It is stainless steel and the only one I found (outside of designing one from scratch) that was roughly the right size.
The power supply mount was redone because the old one had a spot for a 92mm fan. Since the redesign had the exhaust fans now exiting the top, it was no longer required. When looking at the back after paint, you will also notice that several of the serial and SCSI ports are covered as well. I had them welded up at the body shop before paint.
The tachometer is something I will discuss a little bit later in the build. I get an enormous amount of questions regarding how it's hooked up and controlled, and it is a fairly lengthy explanation.
Final build: In the first image you can see the support that was made for the belly pan to accommodate the mount to the chassis. Almost all of the internal chassis parts, drive supports, and face plates were chromed. I was mostly happy with the chrome job, but I will go to a plating shop that specializes in doing show chrome as opposed to one that does it on the side next time.
You can see in these pictures what a huge difference the new wheels and tires make. You can also see part of the custom wire harness that I soldered up from scratch. In fact, the wiring harness for everything was made from scratch with all of the connections soldered. You can also get a glimpse of the custom interface that was built for the tachometer.
Some items that I never did get a shot of are the control board for monitoring temperature, fans, and PSU, and the PC Rider board for the drive lights. Since this is a server case it included thermal, fan, and PSU monitoring. If a fan stops, one of the PSUs goes off-line, or the temp climes out of control, the system triggers an alarm. The PC Rider board (version II as I mistakenly connected the 12v rail to what was suppose to be 5v) does a Night Rider pattern on the 8 drive lights. It has several features that are controlled via a button on a lower face plate, but I thought it was perfect fit for the theme.
The tachometer: Sorry for the repeated image, but I wanted to illustrate what was attached to the bottom of the tachometer mount. First off, no, this is not a tachometer kit from Xoxide. The Xoxide tachometer kit comes with a Sun 8000 RPM tachometer and is software driven via the serial port. The problem with being software driven is that the tachometer requires perpetual updating to work which results in unnecessary CPU cycles being chewed up. In addition, software alone cannot drive the tachometer with the precision that most would want, nor drive a tachometer much beyond the 8000 RPM limit. That in conjunction with what I consider to be clunky software brought me to the conclusion that I wanted something better.
My first step was to find a 10000 RPM tachometer that I liked, and then determine what would be necessary to drive it properly. I decided on an Autometer 5" Pro-Comp tachometer with memory recall. I knew that an interface was going to be necessary to drive the tachometer properly, but it was first necessary to understand how the tachometer worked. Like most automotive tachometers, the Autometer was designed for 4 cycle engines which fire on every other revolution. So for a 1 cylinder engine the tachometer would see a signal .5 times per minute for 1 RPM, 50 times per minute for 100 RPM, etc. For a 2 cylinder engine it would see a signal for each revolution, 4 cylinder would see 2 per revolution, 6 cylinder would see 3 per revolution, and finally 8 cylinders would see 4 per revolution. I chose the smallest number of cylinders the tachometer would support (4 cylinders), which translated to 2 signals per revolution. Since the tachometer is indexed in hundreds and 10000 divided by 100 equals 100 which corresponds perfectly to 1% - 100%, that was the resolution chosen to emulate. Therefore, 1% CPU usage would translate to 100 RPM, 2% to 200 RPM, etc.
It was determined that 5v would drive the tachometer fine, but shooting a fixed duration signal was not. It was necessary to have the signal length decay linearly as the RPM increased to simulate an actual engine. By not doing this the signal became too short and infrequent at lower RPMs and would ultimately fail to register.
I had already written the software to control the tachometer, so I just need someone to build the interface. After a fairly lengthy search for an electronics firm willing to do a 1 or 2 off prototype without breaking the bank, I found Custom Electronics located in the UK. They, or should I say Mike, was great to work with, knew his stuff, and delivered a great interface.
The last picture above shows the tachometer in action. It will take CPU usage and translate it to 100-10000 RPM based on CPU usage's between 1% - 100% in 1% increments. Also in the last picture are the illuminated switches to control the PC. From left to right: alarm reset, PC reset, lights, and power.
Parts, parts, and more parts. After receiving most of the parts for the new lower chassis, it was time to begin the CAD drawings. Over a month of drawing, plotting, and fitting before finally sending them off to the machine shop.
The result from the first round is what is shown above. Initially the machine shop suggested that it would make more sense to weld the chassis and charger assemblies as opposed to bolting them together. The welding turned out to be a disaster due to the heat deforming the assemblies to the point to where they could no longer be assembled! I actually blame the welder more than the machine shop on this one, but nonetheless, this version was scrapped and a new bolt together version was designed and machined. If there was a silver lining to the welding disaster (and additional cost), it was the opportunity to refine the design.
With the new parts back from the machine shop, it was time to go through its first assemble and test. It took several weeks to get all of the polishing done before assembly, during which time I also soldered together the wiring harness. The first test went really well as can be seen by some "crop circles" left in the carpeting. I knew the chassis was overpowered, but wow, I didn't realize it would be by so much!
The last two pictures above are of the charger assembly unpolished and then polished. When I was designing the chassis, the one goal I was striving for was ease of use. I didn't want to pull Ni-Cads out and charge them constantly. So I opted for some very powerful dry cells that conform to military specifications and can be charged very rapidly without being encumbered by memory. The charger is a smart charger designed specifically for these batteries, and can be left charging for extended periods, as it will go into maintenance mode after it has reached a full charge. Since there needed to be a charger behind the computer when it was not moving, it seemed a good idea to incorporate a power strip.
I knew that I need something on the sides to give the appearance of a wider stance. Going back to my original go-kart reference, I contemplated modifying go-kart fiberglass pods but ultimately decided to make them out of aluminum to be consistent with the rest of the theme. I also decided to use the old wing, but with a new mount that set it back and down from its old position. These two small changes had a tremendous impact on the appearance and achieved a look that I was striving for from the beginning.
One thing you may notice looking at the top three images is that the rear tires are wider than the front. All along in the design process, convenience was a big factor, so the ability to make it through doors easily was a big deal. The rear tires made it just a hair too big to get through the door, so while having the new pods fabricated, I ordered new rear tires and wheels that matched the front.
