This is the story of the Boing Ball case I built for my AmigaOne XE. It started out as a joke, but became a serious project that is now reality.
The sphere is 450 mm diameter and stands on a cylindrical pedestal 250 mm diameter and 100 mm tall. The top half of the sphere lifts off for access to the internals. Several of the "checks" are clear unpainted, so that you can see the AmigaOne inside.
It started on the A1G3Dev Mailing List back in March 2003 - there was a thread going about cases for AmigaOnes, and I jokingly suggested a spherical case with red and white checks on the outside. Well, the more I thought about it, the more I thought it might be fun to build one that way. I was also thinking that if the Amiga Downunder User Group were likely to be doing demonstrations of the AmigaOne and AmigaOS 4, the sight of the machine in a large Boing Ball would be eye-catching, to say the least. So I set about sketching the basic essentials. The design had to support the AmigaOne ATX-sized board (this was before the MicroA1 was released), along with a CD, space for several 3.5" drives, and a floppy drive that had to be accessible, but it was OK if I had to remove the top cover to get to it. Here's my original sketch:
So I started looking around for something I could use. My first thought was one of those polycarbonate spheres that you sometimes see as outside lighting. They are popular in touristy areas like harbour foreshores. In Sydney we have them all around the Sydney Harbour foreshore, and around the Opera House. They look great, but expensive. I finally found the suppliers (Thorn) and their catalogue details. The Thorn sphere is 400 or 500 mm diameter, with a large hole in the bottom. It would have cost nearly A$400 by the time I added tax, and I would have to cut it into two halves to get the computer inside! Too big, too expensive. I grumbled to myself and started asking local suppliers. Eventually I was directed to B & M Plastics in Sydney. A phone call confirmed that yes, they frequently make this sort of thing, and 450 mm is a "standard size" in their repertoire. They would make two 450 mm diameter hemispheres for me, trimmed to fit together and make a sphere.
The way they make them is interesting. How do you mould a large plastic dome? Well, you have two pieces of wood - one is a ring, with the internal diameter that you want. The other is just a flat surface with a short pipe sticking through. You start with a thick sheet of plastic (acrylic plastic for instance), clamp it between the ring and the bottom plate, and heat the whole assembly in an oven. When it's hot, you pull it out, blow air into the pipe, and the soft plastic expands upwards in a nice dome! When it has expanded to the right height, they shut off the air supply and allow the assembly to cool.
My two domes/hemispheres cost me A$300, which I thought was reasonable for something made to order. They were ready for pickup in less than a week from my order, which was even better. At that time, I was still anticipating delivery of my A1 board at any time, but in fact, the A1 was not delivered until the Sydney AOS4 Roadshow in October, so there was no prospect of my having the Boing Case ready for the Roadshow.
Meanwhile, I assembled the hemispheres and a wooden salad bowl as a pedestal for this mockup. The CD shows you how big the sphere is.
I built up a model backplane and webs out of 6 mm compressed fibre board (we call it MDF for Medium Density Fibre). This is cheap and allowed me to make mistakes and changes before I bit the bullet and made it all out of expensive acrylic plastic. The pedestal I made out of thick corrugated cardboard. You can see the pedestal in the picture above.
I had to make a sanity test of the airflow design. Would the path of the airflow through the case be so tortuous that the fan could not cope? Would there be pockets of stagnant air that got too hot? I assembled the mockup with the power supply in the cardboard pedestal and one hemisphere on top. I had not at this stage cut a hole in the "bottom" hemisphere, and I wasn't going to until the last minute! I taped newspaper around the bottom of the assembly to make it airtight and ran the computer as realistically as I could.
The power supply had an automatic temperature sensing circuit to control the fan speed. While I had known about this before, I was not sure what the effect would be. My mockup showed that the fan ran slowly because the hotter air stayed up the top of the sphere, losing heat to the outside through the plastic rather than by circulation. The air flow was greatly improved when I shorted out the control circuit and made the fan run at full speed. I also found out that the slightest restriction would make a large effect on the air flow. These little 80 mm fans are about as gutless as they appear to be.
Building the Internal Assembly
The next stage was to buy the plastic for the internal assembly and cut it up. It sounds simple, but it took weeks of marking out, checking, sawing, filing, drilling and checking again. While doing all the marking out and drilling, I left the protective film on the plastic sheets to reduce scratching, so these shots show the assembly still in its "cocoon".
The electronics assembly is attached to the pedestal by three posts, made of 12x12 mm aluminium angle. These posts are secured to the wall of the pedestal to take the weight of the assembly. Two of the three posts also secure the power supply to the inside of the pedestal.
Building the Pedestal
The original concept was to use the pedestal as support for connectors only. With the removal of the power supply from the sphere to the pedestal, suddenly things became a little more cramped. Also the means of fixing the bottom hemisphere to the pedestal became important. Although the weight of the electronic assembly and pedestal is entirely removed from the hemisphere, anyone picking up the complete computer from a table will put his/her arms around the bottom hemisphere and lift that. Therefore the attachment of the bottom hemisphere to the pedestal has to be able to bear the entire weight of the computer.
I tried to find cylindrical acrylic tubing for the pedestal, but it turned out to be far too expensive in the 2 metre length that I would have to buy. Therefore I bought a metre of 250 mm PVC stormwater pipe. This cost me about A$30 from the plumbers' supply shop. The pipe has a 6 mm wall and is a medium grey colour. The ink writing on the pipe became a problem later.
The 250 W power supply is, like all computer power supplies, a standard size and shape. It sucks air from the inside of the computer case and blows it out the back. I wanted to reverse this flow, so that it sucked in outside air and blew it up over the motherboard, over the top of the backplane and down over the disk drives, and finally out through vents in the pedestal. The power dissipated in a 250 W supply running an AmigaOne is negligible - the motherboard only takes about 25 W at 800 MHz - so the air coming out of the power supply is quite cool enough to blow over the motherboard (only about 1 deg C more than the inlet air). To make the air go where I wanted it to, I made the backplane come right down to the full width of the pedestal, with the power supply up hard against it. An aluminium cowl ducts the air from the inlet port to the fan. I had planned to include filter material inside this cowl, but I found it interfered too much with the air flow.
A partition blocks the other side of the power supply so that the two halves of the pedestal are sealed from each other. The "exhaust" side has an 80 mm fan-sized hole and wire guard for the second fan on the exhaust side. The holes in the power supply cover are all blocked with duct tape except at the top where the air has to exhaust over the motherboard. Note also that I have had to split up the cables into two looms, since the motherboard is on one side of the backplane and the drives are on the other. The power to the drives is terminated in two 4-way sockets, so that the loom can be connected to the drives without the power supply. The mains input has been reterminated, a relay added internally to provide a switched output, and a 4-way connector is used for mains in and out. There is a screw terminal for the Earth connection.
The two hemispheres are identical (save for manufacturing tolerances). After studying pictures of the "approved" Boing Ball, I settled on a 12x12 chequered pattern. That is, 12 checks around the "equator", with six bands of "latitude". If you think of the latitude as going from the South pole up through six bands to the North pole, then down the other side again, you can see why there are 12 checks around the equator and 12 around the poles. The numbers have to be the same to make the equatorial rectangles into squares.
I marked out the lines by tracing the circle of each hemisphere onto a large sheet of plywood and dividing it up into 12. I then placed the hemisphere over the circle and measured equal arcs from each side to meet at the "pole". My latitude bands are equal angles of latitude, not the equal vertical projection that you commonly see when you wrap a chequered pattern around a sphere in a paint program.
To make the proper tilt (and to make it easier to mount the CD-ROM drive), the axis of the sphere is tilted 15 degrees from the vertical. There is a 250 mm hole in the base of the sphere so that it is open to the pedestal. I marked out a 240 mm hole and cut it very carefully with a hand-held hacksaw blade.
As mentioned above, I used 12 screws to secure the hemisphere to the pedestal. To spread the load of each screw head, I glued small plastic wedges to the inside of the hemisphere, to provide a flat, level surface for each screw head. The top edge of the pedestal is tapered to match the slope of the hemisphere where they meet.
Having built up the electronics assembly and constructed all the fastening devices (brackets, screws, etc), I had to put it all together to make sure it went together. I also wanted to see how the power supply fan coped with all the internal cables in the way, a hot day, etc, etc. So I assembled the whole machine, complete except that the hot air was exhausting out the base of the pedestal, rather than the side. I had to make a fresh air duct to prevent the air from being sucked back into the inlet. I was also suspicious that the CPU fan was recirculating some air, although I had never suspected it before. In previous tests, the motherboard had always been lying horizontally on the table, now it was sitting vertically with a definite airflow over the board. Certainly the CPU seemed to be running hotter, but the ambient temperature was 30 degrees instead of the cooler temperatures of earlier tests. The temperature rise between the inlet air and the exhaust air was about 7 degrees, the CPU indicated (by its "uncalibrated" sensor) temperatures of 50-54 degrees.
The sphere has to be painted in red and white checks. You also have to be able to see through the top hemisphere to see the (Ooh! Ahh!) AmigaOne XE inside. So some checks have to remain unpainted, both red and white ones. I bought some spray paint and masked off the hemispheres.
I was hoping to have some of the squares with just an outline of the square's colour. In other words, it would look like a dirty window with a clear spot in the middle where someone has cleaned it. I would have preferred to spray the outside of the square with a "vignetting" effect. However, I would only get one attempt at it, and it had to work first time, so I used a rectangular "frame" around each window. Only windows that abut each other have this frame.
So, I sanded all the red paint off the pedestal (that took a couple of hours) and started again. The second attempt was much better, the red was good and bright.
Next I tore all the masking tape off the sphere (except where the checks were intended to be clear transparent) and sprayed it all a thin white coat. I left that for two days to harden, then taped it over again and sprayed the red checks. That took a day of several minor coats, leaving it to dry for an hour or two after each.
Next I had to carefully remove the tape from the white parts, tape over the red parts, and repeat. Each time it took about three hours to tape up the whole sphere. I also made a small window that slides out with the tray of the CD-ROM drive. The whole painting job took about a week!
At last the time came for the final assembly (well, when I say final, maybe I mean the "first" final). I'm glad I didn't try to squash it all into a 400 mm sphere. It takes some squeezing to get all the cables into place, without getting in the way of the airflow.
The final assembly can be seen in the pictures. I expected it to occupy a large footprint on the desk, but because of the height of the pedestal (100 mm), things like keyboard, mouse/trackball and cables can slide underneath the sphere so that it takes up less room than the tower of my A4000.