The light bulb in my ceiling fan died. It needed a new halogen bulb. The "project" could have ended there. Instead, I decided to take this time to fix a few things I did not like about the lamp. For instance,
Since I did not have to worry about the aesthetics of the light, I decided it was time to replace the fixture with one that gave me much better lighting and utilized LEDs. I could have bought something, but more than likely anything I purchased would have had to be modified to bolt on properly to the existing fan. That's when I decided that I would 3D print my own light fixture.
After dismantling the bottom portion of the ceiling fan, I saw that the halogen bulb was held by the center nut, but there were also 8 holes in a circular pattern around the base. I decided that these holes would be perfect for attaching a new lamp. After taking some measurements, and considering how I would want to hold light bulbs, I began drawing up my design in Autodesk Inventor.
The design I came up with would fit 3 standard E26 (standard US) light bulbs, and would point them at an angle 20 degrees from horizontal. It utilizes the bolt pattern from the ceiling fan to hold it in place, and has some channels cut on the underside to allow wire routing. In order to actually hold the E26 light bulbs, I would need some sort of pre-made socket. I found these lamp sockets at the local hardware store. These ceramic sockets have threads on the outside, and come with plastic nuts, for clamping a lampshade in place. I did not need those plastic nuts, however. I wanted to be able to thread the sockets into my light fixture.
I ran into a slight problem once I came home with these threaded sockets. It turns out that they are not standard, and I could not find threading information anywhere online for them. At this point, I knew I had to create my own custom threads. I took some measurements, and tried my luck on my first threads.
This nut was the first piece I printed with threads, and I was really pleased with the result (note, I stopped the print early, which is why the top is not finished). While it mated, I found that the sharp threads were slightly too deep, and it made it rather difficult to install. After I modified my teeth to be a bit more shallow, and rounded on the ends, the next nut I printed worked perfectly.
Since I knew I had a successful thread, I was able to transfer the coil feature to my fixture, and pattern it to each of the sockets. I also created some threads in the center hole, so that I could screw on a cap later to hide the wiring.
As anybody who has a printer knows, creating your part is only half of the battle. The other half is printing it. Fortunately, because of the fillets, and the angles used, the part could be printed without supports. Unfortunately, it was 175 mm in diameter, and my bed is 200 mm x 200 mm. This meant that my auto bed leveling had to be absolutely perfect, and my print surface had to be perfectly flat (I had issues in the past where my bed was warped, with the center lower than all 4 corners). To top it off, the print would take 14 hours. While this was not the longest print I have ever done, it certainly was the largest.
14 hours later, and I had a finished part on my printer! The problem I had at this point was removing the piece from my bed.
I've become pretty good at printing with ABS on my printer, so it was only natural for me to print this fixture in ABS. To adhere it to the plate, I used ABS slurry. To make the slurry, I just pour a cap full of acetone on the glass, and slide around a failed ABS print. When the acetone evaporates, it leaves behind a film of ABS that is stuck to the glass, and my print sticks to that film. The blue in the photo above is the ABS slurry, from a blue piece that I use to coat my bed.
Usually, when most people have a problem with bed adhesion, it is on the side of not enough. In this case, I had too much adhesion. Usually, as my ABS parts cool, they contract and pop right off the plate. In this case, the part cooled and contracted, but it actually warped the glass with it, pulling the corners off the table!
I had eventually removed the glass plate, but not without breaking it. It was not a total loss, as the part was perfect, and I had plenty of spare sheets of glass. After I threaded up the sockets, it was time to mount the light and wire it up. This is where I made the biggest mistake of this project.
I turned off the breaker for the room with the fan in it, but to be safe, I decided to test the bare wires for the light socket with a digital multimeter. In the process, I found out that they were still live. Unfortunately, I found that out by accidentally shorting the two wires through 1 of the multimeter probes, welding it to the ceiling fan in the process.
The good news was that by shorting this, I definitely killed power to the ceiling fan, and now I knew which breaker had the ceiling fan on it (which was not the one for the rest of the room). The bad news was that before the circuit breaker popped, the remote control receiver for the fan and light popped first.
They say that nobody can ever finish a home improvement project with just one trip to the hardware store. I was determined to not have to go back for a replacement receiver. I took the receiver down, and scrubbed all the soot off with a toothbrush and some acetone. After looking at the damage, it seemed that the FET that controlled the lighting was not damaged, but rather the current was so high that it blew off some copper traces, along with the solder mask that was with it. After some careful scraping and soldering, I repaired the board, and to my surprise, it worked!
All electrical catastrophes aside, I was able to finish mounting the light fixture. I found these E26 75w equivalent LED light bulbs made by CREE, and they were perfect for an extra bright work area.
With 3 of these 1150 lumen bulbs and a 5000 K temperature, my new light would produce about twice as much light, and at a much nicer color, in my opinion, than the old halogen lamp. The light would seem even brighter due to the fact that the majority of the light would be cast downwards, rather than in a 360 degree pattern like the halogen. And at a combined total of 33 watts, I simultaneously cut my energy usage by 80%.
Don't let this photo fool you. The camera was just metered for the brightness of the lights (f/2, 1/969 sec, ISO-50). This fixture ended up turning out great. There are still a few things I would do to improve upon this project, however. First, and foremost, I needed to add the center cap to hid the wires in the last photo. After I find a large enough container, I would also like to acetone vapor finish the piece to get rid of the layer lines. I could have printed with finer layer resolution, but I really did not want to wait close to 40 hours for a print to finish. Lastly, if I were to print this again, I would add some pockets to the bottom of the piece in order to reduce the amount of surface area that would adhere to the print bed, making it much easier to remove.
Oh, and turn off the correct breaker next time...