I chose mechanical movement 420 for obvious reasons–Dr. Wettergreen said our plasma cut piece has to be functional and move, and what is better than vibrating to produce music? Thus, I decided to make a music box that would play a glockenspiel.

I assumed keys would be the most difficult part of my concept and was not sure if it would be feasible, so I experimented with it first. I decided to use aluminum because it was light and short lengths could resonate easily. I began by experimenting with width and length. I also experimented with the sounds before and after each step of post processing -filing, wire brushing, sand blasting, polishing and painting. I wanted to make sure I did not accidentally distort the sound later on, but it turns out the only thing that truly matters is length.

Tuning was much harder than I thought. The first two keys I made happened to create a tritone, AKA “the devil’s interval.” Additionally, I found out that the tuning apps that I downloaded on my phone did not really work due to the overtones.  I used the belt sander to raise the pitch of the keys, but after hours of frustration decided to go all out and cut a bunch of keys. I made a major scale and observed that each interval translated to approximately 5mm in length. The math was very interesting and goes back to Helmholtz and Pythagoras! However, I realized that I was getting carried away and settled for the 3 keys that made up a major chord (11cm., 10 cm., and 9 cm.).

 

The other major challenge was bar resonance. I experimented with different ways of fixing the keys to get the maximum resonance. Four holes per key ended up being the best. Each bar has two nodes with the least vibration, and if your holes are off of these nodes by a few millimeters, the sound will be a thud rather than a ring. I located the nodes by pinching the keys with my fingernails and letting the keys hang vertically while striking them with a mallet. At first, I tried laying them flat and pouring salt on top… like you would with wooden bars, but that just made a mess in the kitchen.

…Now onto the actual mechanical part!

To anyone reading this, do not waste prototyping time on cardboard! I was very confused with spacing and am not mechanically inclined, so for some reason I concluded that I should make 3 cardboard prototypes before moving to wood. Do not do that, do the opposite!

 

It was quite a struggle to figure out how to get the right amount  of torque. How much contact should the teeth have with the mallet? What angle should the mallet be at, how should the spring be placed? How big should the mallet swing be, at what height should the keys be positioned? These were the questions swirling in my head much too late at night. I ended up in a rut for more hours than I care to admit until Dr. Wettergreen told me that I should immediately switch to wood and cut slots for testing and increase the surface area of the gear teeth.

I know it’s very obvious, but honestly that suggestion was life changing. I really did not understand the beauty of rapid prototyping and having access to a laser cutter until  then. I definitely did not think enough about the importance of structural support and rigidity, even with my final prototype.After this prototype, I was finally able to convince myself that this idea could actually work. I found a tradeoff between how easy it is to strike the keys and the quality of the sound generated. I’m sure there is a solution for this, but in the next prototype I decided to orient the gears and mallets such a way that it was hard to strike the keys, but if you were successful then you would generate the best sound possible.

Putting together the final prototype was actually relaxing. Unfortunately, the last thing I integrated was the crank. I assumed I didn’t have to worry about it cause it was simpler than everything else. However, it’s the user interface. The crank does not provide the torque needed for the best mallet strike. It’s a slow continuous motion, whereas a mallet strike is quick and violent. I tried gears with 4, 2, and 1 teeth and settled on 1 so that there was more time for the teeth to accelerate before hitting a mallet.  If I had to do this again, I would design a system of gears to make it easier to get the right amount of torque.  The string suspending the keys needs to be very taut, so I added a dowel that could be turned to tighten the string. Here is the final product:

 

It could definitely use a few improvements, but I am happy to say that with the right technique this box can be heard upstairs from the basement of the OEDK!

Cost Breakdown:

This took about 40+ hours due to some fun and very educational failures. Labor is $300 ($7.50 x 40), wood costs are $30, string is 1 cent, and the cost of the aluminum is $8. I’m going to assume laser cutter, plasma cutter, sand blaster, and wirebrush/polish time will be $200 ($20/hr for 10 hours). This brings the total to  $538.01.

 

UPDATE:

I would like to thank everyone who showed me how to use tools and gave me prototyping advice! If it weren’t for 5+ classmates and lab assistants I would have wasted even more time than I did and struggled a lot. One of the most fun parts of this experience was learning from everyone around me, late at night.

For instance, I did not know how to use wood glue properly and three pieces fell off during my midterm presentation. But after, I learned that I needed more pressure, surface area, and time (lol). Cu

rrently there is a knob instead of a crank so that it’s easier for people to walk up and try.