On the left, the very first 3D model of Project Nova; on the right, the model that would comprise the design of the first physical prototype.

It all started - Project Nova, Kimera Industries, everything - because I'm really bad about having slews of memory-hogging programs open, all at once. There are times where I'll have 20-30 tabs in Chrome and Firefox, running alongside Photoshop, SketchUp, a rendering suite, a virtual machine... and often a few other programs, simultaneously. And although my computer at the time, enclosed in NCASE's M1, had the power to run everything, I found myself limited to 16GB of memory by the Mini-ITX form factor. Which, as you can imagine, was a ceiling I frequently hit, and hard.

So, I started considering how I could switch back to a microATX rig (my previous computer was a LGA 1156 mATX build, inside an Antec Mini P180) since those boards support upwards of 32GB of RAM. But after getting used to the small size and minimalist aesthetics of the M1, I was disappointed to find that none of the mATX cases on the market appealed to me, at all. Anything that looked half-decent was far too big, and was usually made cheaply with plastic and thin paneling. And the only real exception to this, SilverStone's SG10, was just plain ugly. It seemed like nobody had really tried all that hard to create a solid, quality SFF case for mATX motherboards that was actually small.

Upon realizing this, I recalled how there had been considerable discussion, during the M1's development, of adding to the height and depth of the case, to support the larger mATX standard. In the end, NCASE stuck to Mini-ITX, but out of curiosity I started playing around with that original idea in SketchUp. Soon enough, I saw that I could, in fact, create an adapted layout that had mATX support, with only a few additional liters in volume. And intrigued by this, and inspired from the dearth of decent cases on the market, I thought: why not design and create my own case? The NCASE guys had shown that it could be done, after all, so it couldn't be that hard, right?

...Well, that was all the way back in April. Little did I know that it wouldn't be until December, a full eight months later, before I had a physical prototype ready to test and use.



I spent my very first weekend on the project modeling my concept, and eventually took some screenshots so as to share my work on HardForum's SFF section and get feedback from the community. Over the next few weeks, as forum members began to critique the idea and offer suggestions, additional features (such as the side bracket and handle) were added to the design.

Early on, though, as more and more users began to voice their opinions, it became obvious that we could debate endlessly about what the best internal layout would be. What was badly needed, on the other hand, was some actual data, to verify our postulations. So I took a spare SG02 I had laying around at work, and promptly tore down and rebuilt it to serve as a mockup frame, complete with cardboard panels that could be easily modified and adjusted. Using a socket 775 board, a power-hungry Pentium D 840 (to simulate a LGA 2011 CPU), and an R9 290X, I then tested several different fan and component configurations, using a wattmeter, thermocouple thermometer, and thermal imager, to collect measurements. Pouring over the results, the best-performing layout of the test set was then identified and ultimately selected.

By now, it was May, and most of that month was spent further refining the design. It was at that time, for example, that the case incorporated a side-vented front panel, as well as the fifth expansion slot. But as the weeks went by, the changes became more minor, and in early June the design was more or less finished, with the model already closely resembling the eventual prototype. 

However, a flurry of roadblocks quickly began to pile up - the initial freelance engineer tapped to create the necessary CAD files had stopped responding, and some unexpected costs hampered my efforts to continue research and development. With the next generation of graphics and processors on the horizon, and with the pile of issues leaving me burnt out, it was clear that I would have to wait.



Come September, Project Nova quickly returned to the front burner, in no small part due to a perfect storm of product releases: SilverStone's 600W SFX PSU had just entered retail, Intel’s Haswell-E was now out, and, to top it all off, rumors of incredibly power-efficient graphics cards from NVIDIA were at a fever pitch. Eventually, leaked specifications for the GTX 980 were uncovered - exposing flagship performance at just 165 watts - and I made an immediate realization: Haswell-E and two GTX 980s could be used in Project Nova, with power to spare. In a case that’s smaller than many of the popular Mini-ITX cases in the market, let alone mATX! Suddenly, Project Nova had an unbeatable configuration, on top of everything else that already made it so great.

I simply had to get my hands on a pair of 980s to test the configuration – and knowing how hard it would be to get them, I snatched the cards at 2 AM, the minute they were available. When they arrived, I quickly built the second test system in a temporary test bench case, and spent the weekend verifying that everything would run well using Silverstone’s 600W SFX PSU. The verdict: at stock speeds, even this beast of a system could run full tilt, with plenty of power to spare.

Further encouraged by these findings, I set out again to get the engineering work done, so that a prototype could be made. However, there was a catch: although my SketchUp models were a great starting point, they were hard to have taken seriously by most manufacturers, since they’re used to working with CAD files from expensive and sophisticated programs like SolidWorks. Thus, I ended up spending most of October tweaking the design in SketchUp, so it’d be easier for an engineer to work with come time to manufacture.

Luckily, in the end I contacted Protocase of Nova Scotia, Canada, and they were very friendly and more than willing to work with the models I had made. After a misplaced decimal caused a misquote of $8000 for the initial estimate (an error that inspired a week of design optimizations), they eventually quoted me at $2600 to create the very first unit of Project Nova. And by mid-November, I gave them the go-ahead to start the engineering work.

Of course, as with everything else in this project, it wasn't that simple - I ended up spending the next several weeks working with the engineers to fine-tune the design for manufacturing. But, by the end of December, the prototype would finally arrive at my doorstep, ready to be tested.