Prototype #1
I've always thought it would be cool to have a device with a few knobs and switches that can sit on your desktop and allow you to control the volume of each program you have running on your computer, without having to interact with the computer itself. There are a variety of devices on the market that basically achieve this, but all of them seem to either have screens, RGB LEDs, plastic construction, or various rather gaudy design choices.
I've also had an interest in getting more into electronics as a hobby and building something that has a physical product design component to it. While my experience with electronics was pretty slim, and I had no experience with physical product design, the idea described above seemed like a reasonable project to attempt as it's based so heavily around fairly basic electronic components (buttons and knobs/potentiometers) which could give me headroom to figure out the physical product design.
Throughout the project I documented various milestones which I list in this article along with pictures and a brief summary. While this project has a software component, it is fairly simple/uninteresting so I don't discuss any of that here. I find it interesting to see how this project went from a concept to a fully-realized device at the hands of someone with very little experience. I hope you feel the same, enjoy.
Code for microcontroller and parent computer found here.
When I decided I wanted to commit to this project, I came up with a few design goals I would stick to:
I did a small bit of prototyping to ensure I could connect a switch to a microcontroller and send that value to my computer, and everything worked as expected. Unfortunately I don't have any pictures of this.
Elegant right? It did work though, you can see in the top portion of the enclosure (the box my mouse came in) the potentiometers and switch are soldered to some jump cables and those cables are plugged into the microcontroller. Funny enough, in this state it could basically do exactly what I want (telling my computer the position of the knob and whether a switch is on or off).
Designing a PCB (printed circuit board) seemed like a good next step since there were so many wires with just 2 potentiometers and a switch, plus it didn't seem too out of reach because of the modern services that let you order personal-sized batches of custom PCBs for a pretty reasonable price. The board above allows a switch, LED, and potentiometer to be soldered in and then controlled using 5 pins at the bottom. This ended up being pretty straightforward to design and order, and got me familiar with the PCB process.
While the PCB design above allowed some consolidation of wires, a bit more was preferable since I wanted to have 5 potentiometer-switch-LED columns, and I wanted to have all of them controlled through one board.
I took a digital logic class in college so I vaguely remembered the concept of multiplexers and shift registers. The multiplexer allows me to interface with 16 outputs using only 5 of the pins on my microcontroller, and the shift register allows me to set the on/off state of all LEDs using only 3 of the pins so I implemented both of those in the prototype and switched to using an ESP32 because of its size and price.
You know how printed circuit board are supposed to make it so your circuit is fully printed on a board? Well, I was not aware that some multiplexers don't cooperate when you leave some of their pins unconnected (hence the wires on the left chip of the board). The shift register on the other hand... I just sold on, straight up did not design the board with all the connections it needed (hence the wires on the right chip of the board).
Those issues aside, it works and the wire situation is much-improved.
The previous PCB was definitely along the lines of what I wanted, but it had some major design issues and was a lot larger than it needed to be, so I redesigned it with that in mind. To celebrate the design breakthrough I upgraded to the box my light bulbs came in!
Now that I was onto designing an enclosure I was 100% out of my element.
The enclosure needed to be fabricated at a reasonable cost which removed the idea of CNC milling from consideration. When looking at other techniques I learned of sheet metal fabrication which consists of cutting metal from large sheets and bending it into the desired shape. I saw good reviews of the site SendCutSend, which allows you to upload CAD files and they cut/bend them into the desired shape and ship them to you.
Next, I pretty arbitrarily chose Fusion360 as the CAD software I'd learn, which was fairly straightforward for something this basic. However, designing around a sheet metal fabrication process presents a few constraints, such as not being able to bend metal into a box shape, and less precision on cuts/bends.
You can see from the picture above I sold again and somehow indicated I wanted a bend in the middle of the sheet that was going to be the façade of my potentiometers and switches.
Even though the previous attempt at an enclosure was rough, it made me realize all the knobs and switches were too spread out and the enclosure would end up too large (I decided I really did not want the device to be taller than it was deep as it would get pushed over when someone tried using the controls). So, after much deliberation I decided on a new layout for the faceplate that I then designed the rest of the enclosure around it.
I remember this being particularly difficult, the sheet metal has a lot of rules around how it can be bent, and I wanted to minimize the amount of screws that I had to use since they would be visible on the outside. I decided on having a faceplate, backplate, and baseplate, then having an elongated "U" shaped top-plate to cover everything up. I also recessed the face and back plates under the top-plate since it looks cool.
So, the basic enclosure design and electronics were done, but I was stuck in limbo trying to think of how to actually connect all the pieces of sheet metal and how I would mount the PCB and microcontroller inside of it.
I stopped working on the project much after this, work and life got busy and I didn't have any clear next steps. However, I thought about it often and had much support from my girlfriend who even bankrolled some forward progress <3. So during this time the thought of finishing just sat in the back of my mind.
I had always harbored a slight anti-3D-printer sentiment because I personally don't think their output is all that pretty. But, I've come to accept that it's totally fine for some applications, and in the case of my project, it can be used to print pieces for inside the enclosure.
Having this made it much easier to reason about and iterate on the design, eventually I was able to print brackets that hold the front and back plate in place, and a mount to hold the microcontroller and PCB in place.
Having finished the design, I ordered the last iteration of sheet metal parts. The fabricator offered powder coating (if you can stomach the price) which I went for in the name of aesthetics and the end result is exactly what I was hoping for: a black metal exterior, small silver screws, low-brightness red LEDs, and black knurled metal knobs with one white arrow to indicate position. For those curious, it was $125 for the sheet metal, and with the cost of components and the board this final version cost me about $185 total.
For connecting the top-plate to the rest of the enclosure, there are magnets embedded under the thumb-shaped cutouts and magnets superglued to the top-plate (which fit into the thumb-shaped cutouts) seen in the image. I ended up really liking this mechanism because it's easy to remove the top-plate and it has a really nice tactile feel when interacting with it in person.
When I first started this project, I thought "a box with holes for switches and knobs will be easy enough to design", and sure enough it is, until I wanted it to actually materialize in the real world.
I've found it much easier to consider any constraints around bringing your project into the real world upfront, rather than trying to disfigure an impractical plan/design into cooperating (which tends to be futile).
Don't let your ego derail you. Every project will have issues, but some should be looked past at times for the sake of actually achieving the end goal or you'll get caught up infinitely reworking some specific aspect that blocks you from any practical progress and burns your finite time and money.
Make informed decisions on what is worth fixing/improving based on your goals, but remember that common sentiment: "perfect is the enemy of done "— and something that is done tends to be wildly more practical, interesting, and useful than something that is not.
Don't get too caught up in what you do or don't have experience in, just start at a reasonable spot and you can figure it out along the way. With enough iteration and failure (learning) it's highly likely you end up completing whatever you want.