"sp3" is a term that chemists use to describe the hybridization of orbitals of a particular atom in a molecule. It is, generally, a fancy way of saying that an atom's substituents are arranged in a tetrahedron around it. It seemed an appropriate name for this lamp, which is both tetrahedral in form and inspired by the tetrahedral, or sp3, carbon atom, which is the atom of which life is primarily composed.
This is one of those it-all-started-five-years-ago-with-a-toothbrush-in-Papua-New-Guinea type deals. 'Round about the end of 2003, I think it was, during the summer, there was scuttlebutt around the University of Texas, where I was going to school, that the on-campus parking situation was going to be revised to include a number of special "loading zone" spaces, in which anybody could park at any time so long as they left their hazards blinking. Anybody without blinking hazards got a ticket, otherwise it was okey-dokey. This, in fact, was NOT the policy that went into place, because a car with a well-charged battery can run its hazards for way longer than the average university class, but nonetheless I got it in my head, at the time, that I wanted to install LED hazard bulbs in anticipation of this policy so that I could park my car and leave it all day without worry. Parking at UT is mare rare, and after you've been there a while you start to think like that, especially in the late summer when its so hot in Austin, TX, that people have been known to go crazy and just start shooting.
Anyway. So I procured a set of four LED bulbs to replace my old incandescent hazards, got them installed, and--as it figures--they don't blink. Turns out the flasher in my '98 Jeep Cherokee is an old-style electromechanical unit with a bimetallic strip and everything, and it needs a certain minimum current to heat the strip and break the circuit during the "on" cycle, and the super-efficient LED bulbs just don't draw enough. The lights just come on and stay on. The only way to make them flash is to wire big honkin' resistors into the circuit, which of course defeats the purpose. Of course, it should be possible for some bright soul to design an aftermarket flasher unit for the older vehicles that would use digital logic and an integrated circuit rather than a bimetallic strip to flash the hazards even if they are LEDs, but as far as I can tell nobody's done such a thing as of yet and it's way more work than I wanted to put in for what was really only a half-baked wild-hair-up-the-ass kinda notion to begin with. So I took the LED bulbs out and put my old incandescent flashers back in and they work just fine.
The upshot, however, is that I was left with these four super-efficient amber-colored LED DC bulbs. They were moderately expensive at the time, and I didn't feel right returning them, so I had it in mind to dream up a use for them. About the same time I happened to pay a visit to PetSmart with Melody, and browsing there I discovered the glory that is fluorescent plastic hamster balls, like these:
I've never had nor particularly wanted a hamster, but looking at them it occurred to me that there must be all sorts of interesting off-label uses for inexpensive, hollow, festively-colored translucent acrylic spheres in a range of sizes from 5 to 12 inches with built-in resealable access ports. As a chemist, when I see a sphere the first thing I think of is an atom, and when I see a bunch of spheres of different sizes I start thinking of molecular models. What's more, I realized that, by removing the circular access cover on the hamster ball, another ball could be "nested" in the circular opening, eliminating the need for "sticks" to represent bonds and giving the impression of a space-filling, or CPK (Corey-Pauling-Koltun), model. The simplest molecule to start with would be a simple tetrahedron, like methane.
Which was an interesting idea in and of itself. But when it dawned on me that I could put my four surplus LED bulbs into the four geminal "hydrogen" atoms and have a lamp, I was committed. I bought four 5" water clear balls for "hydrogens" and a single 8" yellow-green clear ball for the central "carbon." The access ports of the four smaller balls were removed to allow the central ball to nest into them, and the access port of the central ball was modified to hold the battery bank that would give power, allowing it to be removed and the batteries changed if necessary. I opted for battery power because it would allow me to leave the exterior of the lamp unmarred by a power cord.
Holes were drilled in a tetrahedral arrangement around the center ball, and the LED bulbs mounted facing outwards. When I started thinking about where to put the ON/OFF switch, it occurred to me that I could avoid using an external switch altogether if I included a tilt switch that would turn the thing on when it was positioned on one face and turn it off when it was positioned on another. I bought a mercury switch at Radio Shack and wired it into the battery block, then wired up the lamps in some replacement tail-light sockets I bought at the auto parts store.
Once I was sure that the circuit worked properly, I mounted the smaller spheres in position on the surface of the central sphere using a special epoxy for use on plastics, which turned out to be a mistake because outgassing fumes from the glue have over the months crazed the once-flawless interior of the outer spheres, a problem that's familiar to anyone who's ever built a model airplane and glued the canopy in place with regular plastic model cement. The bond is strong, at least, and those outer spheres have stayed securely mounted for a couple years now despite supporting the weight of all those D-cells in the center sphere.
Ultimately, however, I was more excited about the sp3 prototype while I was building it than afterwards. Although everything worked more-or-less as I intended, there were a number of problems.
Despite my own dissatisfaction, however, the prototype garnered rave reviews from friends. Christie Clark, in particular, was very enthusiastic and several times offered to pay for materials if I would build another one for her. Eventually I was persuaded that I was on to something and should attempt an improved version. While browsing the local hardware store, I came upon a rack of replacement globe shades and realized they'd be perfect for the job, if only they weren't made of glass. Then I went up and touched one and realized--eureka!--that it was, in fact, made of some kind of plastic.
I decided at that point that I'd attempt a more sophisticated version of my hamster-ball contraption using four light fixtures with globe shades arranged tetrahedrally around some sort of central frame. Then the hunt was on for the most suitable fixture. Much web browsing eventually led me to the Progress Lighting P5797-16 Hard-Nox outdoor model, and everything else fell into place.
|Part #||Part||Vendor||Price||TX Sales Tax||Shipping||Total|
|1||4 Progress Lighting Model P5797-16 Hard-Nox Outdoor Light Fixtures||www.ebay.com||$87.96||$0.00||$11.10||$99.06|
|2||9 sheets aluminum 6961 T6 0.065" thick, cut||www.onlinemetals.com||$25.90||$0.00||$11.00||$36.90|
|3||1 DuraKool A1-002 Tilt Switch||www.newarkinone.com||$15.25||$0.00||$7.80||$23.05|
|4||4 pack 12W spiral fluorescent bulbs||Breed & Co., Austin||$8.49||$0.70||$0.00||$9.19|
|5||1 6' Computer Equipment Cord, 18AWG, 3-conductor black jacket||Discount Electronics, Austin||$3.95||$0.33||$0.00||$4.28|
|6||24 0.375" 4/40 SS machine screws||Breed & Co., Austin||$2.64||$0.22||$0.00||$2.86|
|7||9 1.5" 8/32 SS machine screws||Breed & Co., Austin||$1.53||$0.13||$0.00||$1.66|
|8||24 SS #6 split washers||Breed & Co., Austin||$1.20||$0.10||$0.00||$1.30|
|9||1 Serv-a-lite G723 Rubber Grommet 1/2" ID 3/4" OD||Breed & Co., Austin||$0.63||$0.05||$0.00||$0.68|
|10||9 8/32 SS nuts||Breed & Co., Austin||$0.54||$0.04||$0.00||$0.58|
|11||9 SS #8 split washers||Breed & Co., Austin||$0.45||$0.04||$0.00||$0.49|
|12||9 SS #8 flat washers||Breed & Co., Austin||$0.45||$0.04||$0.00||$0.49|
|13||1 Serv-a-lite G403 Rubber Grommet 1/4" ID 3/8" OD||Breed & Co., Austin||$0.33||$0.03||$0.00||$0.36|
|14||2 Yellow Wire Nuts||Breed & Co., Austin||$0.16||$0.01||$0.00||$0.17|
|15||1 Orange Wire Nut||Breed & Co., Austin||$0.06||$0.00||$0.00||$0.06|
|Items in yellow were puchased in surplus.|
Obviously, selection of the light fixtures was important. The Model P5797-16 Hard-Nox outdoor fixture, by Progress Lighting, proved ideal: The bundled globe is polycarbonate, rather than glass, which is important to prevent breakage in the sp3 because the globes themselves are load-bearing. Also, the globes are threaded and screw into the fixture-bases, rather than attaching, like many globe shades, by tensioning machine screws in the fixture against a lip in the shade. This latter type is undesirable because the machine screws break the infinite radial symmetry of the fixture, the fit of the globe into the base is more "wobbly," and the screws themselves are small and easily lost. As for the fixture itself, it is cast aluminum with a brushed finish, which I knew would be easy to match against the aluminum sheet metal used to form the central frame. Also, the fixture has infinite radial symmetry, which enhances the geometric appeal of the completed piece. Finally, as a higher-end outdoor fixture, the P5797-16 is generally tougher and more robust than most other light fixtures.
I bought 0.065" (about 1/16") thick 6961 T6 sheet aluminum from www.onlinemetals.com. I bought 9 pieces, one for each face of the truncated tetrahedral frame and one extra, and paid to have them cut roughly to size. Four of them were 4x4" and five of them were 5x6". The extra turned out to be unnecessary.
The tilt switch is a DuraKool A1-002 model, which is a non-mercury type for those of you who care. It's rated for 1.0A at 120 VAC, and for a 100W tungsten lamp, which is well in excess of the approximately 40 watts drawn by the four fluorescent bulbs shown here. It is secured inside the lamp by push-fitting into the large grommet in the "special" hexagonal face, with the wires extending into the volume of the lamp.
Because the finished lamp would host four 120VAC lamp fixtures, because it would be made mostly of thermally-conductive metal, and because it would be positioned with the bulbs very close to the display surface, heat build-up was a concern from the very beginning. Low-wattage spiral fluorescent bulbs were part of the original design, and the four-pack by General Lighting pictured above proved ideal. Each of these high-efficiency bulbs generates the light of a 60W incandescent for only 12W, thus the total power consumption of the sp3 using these spiral fluorescent bulbs is only 48W, as compared to 240W with conventional incandescent bulbs. Even after extended use, the globes are still cool enough to be comfortably touched with bare hands.
The cord I used was cannibalized from a dead computer power supply. It is black, about 5' long, and is 1/4" in diameter, which is a dimension that must be matched to the inner diameter of the grommet (Part 9) which will pass it through the case. "Grounded," for those who do not know, means that the plug has three prongs and the cord carries three wires, all of which are necessary. In order to be safe, the lamp's metal case must be grounded and the ground lead in the power cord (usually green) is essential for doing so.
These are 4/40 pan head, philips drive, 3/8" stainless steel machine screws. 24 are needed. They were bought at the local hardware store. They are about twice as long as they need to be, but I couldn't find any that were shorter. The 3/8" length was cut down to give approximately 1/4" of threaded screw using electricians pliers. One is installed at each corner of each hexagonal face, passing through an unthreaded hole in the hexagonal face and tightened against the threads of a tapped hole in one of the triangular face's bent flanges behind. They should be tightened gently, only so far as is necessary to compress the split washers.
These are 8/32 pan head, philips drive, 1.5" stainless steel machine screws. 8 are needed. 1.5" proved a bit long for this application; 1.25" would be plenty. They run through the mounting holes in the light fixture housings through the 1/4" holes in the hexagonal faces and are secured with flat washers, split washers, and nuts. I insisted on stainless steel whenever possible in choosing hardware for the sp3 lamp. The small extra cost is worth the reassurance against corrosion.
These are #6 stainless steel split washers. 24 are needed. Lock washers are essential in securing threaded fasteners in metal. They provide tension against the threads which prevents the screws from loosening over time due to vibration. There are multiple types of lock washers, but split washers are the best-looking, in my opinion. They go between the screw head and the outer plate at each corner of each hexagonal face.
This is a Serv-a-lite G723 rubber grommet. Only 1 is needed. It has an inner hole of 1/2" in diameter, and fits into a hole 3/4" inch in diameter in 1/16" material. The absolute outer diameter of the grommet, including the lip, is about 1". This grommet is mounted in the 3/4" hole in the "special" hexagonal face, and the tilt-switch installed by push-fitting into the inner hole. Although it was necessary to modify this grommet when I constructed my sp3, if you follow the template file I provide above no modification should be necessary.
These are 8/32 hexagonal stainless steel retaining nuts. 8 are needed. They are used together with #8 split and flat washers to secure the light fixture housings against the hexagonal faces of the sp3 lamp. These should be tightened down until the split washers are fully compressed.
These are #8 stainless steel split washers. 8 are needed. Together with the #8 flat washers and 8/32 hex nuts, they secure the light fixture housings against the hexagonal faces of the sp3 lamp. These go between the flat washers and the nuts.
These are #8 stainless steel fkat washers. 8 are needed. Together with the #8 split washers and 8/32 hex nuts, they secure the light fixture housings against the hexagonal faces of the sp3 lamp. In temporal order, the flat washers, then the split washers, then finally the nuts go over the exposed screw threads.
This is a Serv-a-lite G403 rubber grommet. It has an inner hole 1/4" in diameter and fits into a 3/8" hole in 1/16" thick material. The overall outer diameter of the grommet, including its lip, is about 1/2". This grommet is installed in the "special" triangular face, which should be positioned opposite the "special" hexagonal face in the finished lamp, to allow for passage of the power cord into the assembled case.
These are 10-18 gauge yellow wire nuts. 2 are needed. They are used to secure the bundled light fixture wires to power and switch leads.
This is a 12-18 gauge orange wire nut. Only 1 is needed. It is used to connect one AC power lead to one side of the tilt switch.
Construction of the sp3 lamp amounts to building the central frame and mounting the four light fixtures to it. The central frame is a truncated tetrahedron, which is the solid having four perfect hexagonal faces and four equilateral triangular faces, with each hexagonal face parallel and opposite to a triangular face. In my construction, the triangular faces bear auxiliary "tabs" along their sides which, when bent to meet at their edges, form the angles of a truncated tetrahedron. The hexagonal faces bear holes at the corners through which screws are fastened into threaded holes in the "tabs" on the triangular pieces. Examination of the accompanying photographs should make this clearer.
The first step in construction of the sp3 was to cut the parts which comprise the faces of the truncated tetrahedron out of sheet aluminum. I contemplated outsourcing this job at eMachineShop; they provide a complimentary full-featured CAD program which I used to layout the parts templates and which incorporates a spiffy built-in on-line pricing option. Although the process didn't work for me as smoothly as it's supposed to, and the price they finally quoted seemed extreme, I still feel like those folks deserve some thanks for providing quality free CAD software to the maker community. Their package produces .DXF files, which are the industry standard for CAD drawings. The parts blank files I used are available here. The file contains one of each type of face. I printed four copies of the file onto Avery full-sheet mailing labels, giving four hexagonal and four triangular patterns. These were cut out carefully with scissors and affixed to eight separate aluminum sheet blanks.
NOTE: One triangular and one hexagonal face are "special," meaning that they have additional holes to allow passage of the power cord, in the first case, and to mount the tilt-switch, in the second. Although all the parts templates bear markings showing these holes, they should actually be cut in only one of each type. These "special" holes are indicated on the parts template. I recommed marking the two "special" parts before any cutting is done.
With the patterns securely affixed, the parts were cut from the blanks using a Delta scroll saw. About one fresh blade was needed for every two parts cut. Once the crude profiles were cut, all holes were drilled on a benchtop drill-press using a step-drilling technique for maximum accuracy. (Step-drilling, if you don't know, means starting the hole with a very small drill bit and working slowly up to full diameter; this is done because it is much easier to accurately position a small bit than a large one.)
Once the holes were drilled, the tabs of the triangular pieces were clamped in a vise between two steel-bar vise inserts (as shown) and bent down manually until their edges met as closely as possible. The ends of the tabs have been designed at such an angle that, if they are simply bent to meet one another, the tabs will be at the correct angle for assembly of the frame. The vise-inserts are necessary to allow continuous clamping of the tab through the full range of the bending motion.
Once the triangular pieces had been bent to shape, the holes in the tabs were threaded to accept the 4/40 case assembly screws and the entire case assembled to test the fit of all parts. It was necessary to cut the case assembly screws down from their original length of 3/8" to a length having about 1/4" inch of threaded screw so that they would not interfere with each other in the close interior confines of the frame's vertices. This was easily done with electrician's pliers, which cut the screws readily and created no problems with the threads.
The frame is assembled by passing the cut screw through a split washer, then through the hole in a corner of a hexagonal piece, and finally threading it into a tab-hole in a triangular piece. The screws are tightened gently with a screwdriver against the threads in the tab-holes.
When I was sure the frame was going to fit together the way I'd intended, I took it apart again and dressed the edges of the hexagonal faces by manual sanding against a piece of 220-grit wet-or-dry sandpaper on a cutting board. It was impossible, unfortunately, to dress the edges of the triangular pieces in the same fashion because they'd already been bent. For this reason, I advise anyone following these instructions to dress the edges of all parts before any bending is undertaken. The finished construction does expose the short edges of the triangular pieces and its appearance would have been slightly improved if they had been dressed.
After removal of the pattern-labels using Goo-Gone and acetone, the faces of the hexagonal pieces were also polished briefly against 220-grit paper, long enough to remove any noticeable scratches from the areas near the edges and corners of the hexagons which would be exposed in the finished piece. The finish, again, might have benefited from rigorous polishing of the triangular faces, as well, but they weren't in bad shape to begin with and I was getting impatient with all the sanding. At this point the rubber grommets were inserted into the holes in the "special" parts.
With the parts finished and dressed, the first step in final assembly was the attachment of a light fixture to each of the hexagonal pieces. This was accomplished by use of long screws, washers, retaining washers, and nuts. Screws were inserted through the fixture side and secured on the hexagon side, i.e. what would be the interior volume of the lamp when it is assembled. The outside diameter of the fixture bases should be perfectly circumscribed by the hexagonal pieces, as shown. The black and white wires running from the lamp socket were threaded through the center opening in the hexagonal piece. A turn of electrical tape was wrapped around the tilt-switch housing to provide a snug fit, and the switch itself push-fit into the grommet.
The frame was then assembled again, as before, stopping with only one hexagonal face removed in order to access the frame's interior for wiring. I found it easiest to leave off the hexagonal face opposite the "special" triangle, i.e. opposite the triangle bearing the hole for power-cord entry. This will leave off the hexagonal face with the tilt-switch mounting grommet.
The first step in wiring the interior was to install the power cord. The female end of the cord was cut off and the cable covering stripped to expose three wires: black and white to carry alternating current and green for grounding. The ends of the wires were stripped, and the entire cord threaded through the 1/4" ID rubber grommet in the triangular face. A strain-relief knot was tied in the cord on the inside of the lamp.
A ring-terminal was crimped to the end of the green grounding wire and this was attached underneath one of the flat washers retaining the fixture mounting screws, in order to ground the lamp's case.
The black wires from the fixtures and power cord were ganged together and secured with a yellow wire nut.
The white wire from the power cord was connected to one side of the tilt switch using the orange wire nut.
Finally, the white wires from the lamp fixtures were ganged with the wire from the other side of the tilt switch and secured with the other yellow wire nut.
With the internal wiring complete, the action of the tilt switch was tested. Bulbs were installed and the unattached face bearing the switch rotated to open and close the circuit. When correct operation was verified, I attempted to install the final, unattached face bearing the tilt switch. At this point a slight problem was discovered: As positioned, the edge of the rubber grommet used to install the tilt-switch interfered with the internal alignment of the adjacent assembly tab, and it proved necessary to remove part of the edge of the grommet on the inside face to alleviate this interference. This operation was performed with a razor blade. The position of the switch-mounting hole has been corrected in the template file I supply on this page, so this modification should not be necessary if you use my template file.
With the grommet modified to alleviate its interference with the assembly tab, the final hexagonal face was installed smoothly as before. All that remained was to screw in the plastic globe shades and to put the completed lamp in place.
The grommets should be installed during test assembly of the frame.
The center holes in the hexagonal parts are larger than strictly necessary to allow wire passage. Their diameter may safely be reduced to allow the switch-mounting hole to be moved further from the edge of the "special" hexagon, alleviating the grommet's interference with the adjacent assembly tab. THESE CORRECTIONS HAVE BEEN MADE IN THE SUPPLIED TEMPLATE FILE.
At least the short edges of the triangular pieces should be dressed smooth with sandpaper prior to bending the tabs.
The tilt switch is a neat idea but not strictly necessary and not entirely practical in operation. The lamp has a large "footprint" as it is, and allowing extra room to roll the entire assembly from one face to another for on/off action requires more space than is convenient in most locations. I tend to leave the lamp in the upright "on" position and control it using a socket in my living room which is controlled by the wall-mounted light switch. The lamp could be manufactured more cheaply by omitting the switch altogether and wiring it "always on," provided power control is available at the socket. Alternatively, an in-line switch could be installed in the cord or, perhaps most attractively, a capacitance "touch" switch installed in the frame.