W17: Mechanical & Machine Design

The idea

As I mentioned in more detail in the Principles and Practices, I've done some long-exposure light painting photography in the past, like the shot below:

To take these kinds of shots I had to completely shut down a room to prevent the light from coming in. In a run-of-the-mill family apartment, this is quite a feat.

Therefore, the idea of having a small, self-contained and portable studio of sorts to create light paintings seemed like both useful and interesting.

In the following drawing, we can see a closed up box —no light comes in— with a small hole to take shots with a mobile phone's camera. Inside, there are a number of RGB LEDs that would move, creating the painting with their motion.

This assignment ended up becoming my final project.

Designing and building the parts

Case/Box

I though I'd use 10mm plywood, which was easy to source locally, light and manageable, all while being sturdy enough. The pieces are designed to fit tightly together, although I'll be applying glue close to the end of the project.

I though I would add a bit of decoration in relation to the theme of the machine, so I thought something like these circled patterns and letters were easy enough to machine yet added a nice touch.

Inside, the box was sprayed in matte black:

Assembling was fairly easy, as the plywood has enough tolerance yet the whole structure was tight.

Here's the case for the machine:

There's quite bit more detail in the Computer Controlled Machining assignment.

Frame

As seen in the sketch above, my iniital design for the frame was a circle. Here we can see a couple different takes on the base:

However, early prototypes showed this approach didn't work well with the focal length of a mobile camera working in the constraints of the case, so I tried other designs:

Finally I ended up settling on the following one becasue it provided a constant an ideal distance to the top hole in the case, and that would become key when shooting.

To finish the frame, there needs to be a hole through the base to attach it through its axis of rotation.

I broke a couple at this point, as it turned out to be too thin to withstand the hole at the base.

I ended up glueing two frames together to make a thicker, 20mm frame

The frame, along with the base, was also painted black:

I then installed the neopixel strip. Here's a first prototype:

I used black tape to hide all the wires and Neopixel substrate:

I made a mistake connecting the parts of the stripe and had to desolder, turn around and then resolder. My bad for not paying attention to the labels. Finally, I got the full strip working.

Fixture

At the beginning, my idea was to pass on information and power to the frame through a set of concentric tracks, and fix everything together with just a screw and a nut, with a set of washer in the middle:

I made a hole in the base of the case to host a bearing:

Inside the bearing I made some extra "cushion" with a couple of plexiglass washers. It took a few tries to get them just tight enough and a good fit for the screw.

I made the tracks out of PCB stock:

Giving it a spin manually it seemd like everything was going to go smoothly. I was wrong.

I devised a number of ways to get the cables to connect to the tracks:

However, there was too much slack and contact it didn't make for a consistent connection. There would be enough drops in continuity to make it unusable.

I tried different designs and approaches, and ended up going on a different path: using a slip ring with a 3D printed fixture to hold everything together:

The first design was a both bigger and simpler than other iterations. It features a longer attachment base, according to the first iterations of the frame, and also featured a simple hexagonal ending meant to be inserted in the gears below.

After a few iterations, the frame had evolved into a shorter base, and I had also added a small bearing beneath the case's bed, to help with the tilting. This fixture still features the haxgonal fit for the gear, but also an axis that would go into that second bearing.

More info about the fixture can be found in the 3D Printing assignment

Gears

At first I planned on using the step motor I had tried in the Output Assignment, but it was pretty obvious the speed was too slow for the use case and I would need to multiply it with a system of gears.

I used the online tool Gear Generator to create a couple of gears that would speed up the spinning of the frame. I went witrh the biggest and smallest size I could fit, which resulted in a 1:3 ratio approximately.

Gear Generator exports SVGs that can be later imported into AutoCAD, where I cleaned them (there's a lot of text within the gear with its details). I also added a shape used to fix the gears to the motor:

These are the first iteration

Eventually, after a number of tries and iterations, I ran out of 5mm white plexiglass and had to move to 3mm black plexiglass. I would then "glue" both pieces with chloroform:

As we can see, I tried to use the biggest and smallest gears I could fit beneath the bed.

More info about the fixture can be found in the 3D Printing assignment

Motor

As mentioned previously, the step motor I had in mind was too slow and didn't seem to have enough speed:

It also didn't seem to have enough torque to move the gears:

I could've tried to make the gears lighter by removing material with a different design, but I still needed more speed. Since speed control was not a requirement for my project, I thought I'd try a servo motor. A quick proof of concept showed promise:

A quick set up with the gears engaged confirmed this was a good solution.

Controls

I had a pretty good idea from the get go of the interace to control the machine, which was for better or worse constrained by the switches and buttons I had at my disposal at the time, since sourcing materials was complicated with covid slowdown.

It was fairly straighforward moving from a drawing to a digital design that I could use with the laser cutter. I measured all components and made a digital template.

Black 3mm plexiglass was what I had at my disposal, so black it was.

Wiring was pretty straightforward too:

I don't think it gets any more 80s than this:

Operation

As seen in the sketch above, very early on I had the idea of two modes of operation: manual and automatic. This would allow for exploration both in interaction and programming.

Modes are selected with the main switch in the middle: Manual mode is on the left side and auto mode on the right side.

Manual Operation

The color for the LEDs in the frame is chosen randomly by clicking this button:

The color is preserved when switching between manual and auto modes

At first I thought I'd use the rotary knob for brightness, but after a number of tries I concluded that it was not an interesting feature, as long-exposure shots couldn't handle more than around 5% brightness anyway.

Therefore I chose to use this control both to switch between the full stripe of LEDs and moving one of them. This proved to be superfun in practice :)

The left-most switch controls whether the frame rotates clockwise or counter-clockwise. The green button is used to start the motor, which will spin as long as it is pressed.

With manual we can get videos like this one:

And shots like this:

Automatic Operation

In automatic mode, there are three programs to choose from:

The first program will use the current color and move a single pixel while rotating the frame. It produces shots like this:

The second program is like repeating the first program 10 times with random colors each time. It produces shots likes this:

The third program is similar to program two, but it will randomize the number of "passes" as well as the delay between passes. It produces shots like this: