Building Simple, Accurate, Precise, and Cheap Player Pianos for Entertainment in Public and Private Spaces
Advisor: Mr. Daniel Ong
Internship Location: East Bay Music
Onsite Mentor: Vanessa Chen
We often use speakers as a way to listen to music, radio, podcasts and more. Modern day speakers are a technological marvel in their ability to reproduce sound, however when it comes to music, they will never surpass live music in terms of audio quality. You may ask how one would have live music nearby at their beck and call in a way that could compete with a speaker. The answer is simple: player pianos aka self-playing pianos. Player pianos have been around since the mid 1800s, however they have always been an expensive commodity. Some early mechanical player pianos could cost up to 30,000$ in today's currency, and were prone to malfunctioning under the wrong circumstances. Modern electric player pianos are better in that regard, but haven’t improved much in price. This is a problem that I seek to solve by first building a player piano from a regular piano with off the shelf electronics, and then making an open source kit and instruction set in order to allow others to modify their pianos in a similar way, allowing people to enjoy a plate piano for only a few hundred dollars. Player pianos were once a common sight in the homes of everyone and I think that something of value was lost when they went out of style in favor of other ways of listening to music. I believe that by building my own player piano, and making it easy for others to do the same, I can allow people to experience the joy that those who had player pianos back in the day experienced.
Project Blog
Week 0: Blog: What's a Player Piano?
If you’ve seen a piano playing itself before, you may think that self playing pianos are a technological marvel of the digital age, however self-playing pianos a.k.a “Player Pianos” have been around for a long time. Player pianos first started showing up in America in the 1880’s, and were powered by a system of pneumatics powered by foot pedals. They were very common in households across the nation by the turn of the century, and people generally enjoyed using these pianos as a way to enjoy music without having to put in effort to learn how to play the instrument. However, they had a big problem in that they were expensive, difficult to move and difficult to repair. Because of these problems, most people stopped using them after the 1920’s and 30’s when better methods of playing music at home became more widespread.
Modern electromechanical player pianos are available and solve some reliability issues that the old pneumatic ones had, however they are mostly sold by big name brands like Steinway and Yamaha for a very high price. (High enough that you have to call them to find out the price!) I plan to fix this problem by building a precise and well functioning player piano for relatively cheap (< 500$) using off the shelf components and by posting designs and instructions online to allow other people to replicate the player mechanism on their own pianos.
Personally, I have been fascinated by mechanical music machines including player pianos, coin pianos, fotoplayers, and more all my life. I believe that the sound quality produced by these machines is far superior to any sound produced by speakers or other means. I’ve already found a piano that I will be using for this project and I’ve begun fixing it up to get it to sound better. Hopefully by next week I’ll have some more information on the mechanism and circuits that I will use to bring this piano to life. For now, here are some pictures of the piano in its current state and as well as a video of me playing it so you can get a feel for the piano’s current sound: Image 1 Image 2 Image 3 Video
Week 1: Blog: Progress
The first step of making an electric player piano work is getting the first key to play. This is easier said than done however, since a solenoid needs to have enough power to fully push the key down while being small enough to fit inside the piano. Last week I ordered 3 different solenoids so I could test each of them and see if any of them would work for this project. From my testing, I was able to get one of the solenoids to play a note reliably. However, I had to overvolt the solenoid in order to get enough power, which caused the solenoid to heat up significantly more than I would like. To alleviate these issues, I have ordered some more powerful solenoids which I will be testing next week and should hopefully work.
Another challenge which I have been working on this week is getting the logic circuit for decoding MIDI signals to work. This is essential for the piano to work since it will allow for a computer to communicate with the piano and tell it when to turn notes on or off. For the most part I have finished building a prototype version of the decoder, and I will be ready to test it next week.
The last part that needs to come together for this project to work is a circuit which takes the output of the decoder and turns the solenoids on or off. I finished one design of this circuit and I am currently ordering parts for it.
Overall, It has been very motivating thus far since I have made a lot more progress than expected this week and since I am well ahead of schedule.
Week 2: Blog: Delays
So far, this week has been slow. I started out the week with working on designs for a volume control circuit since I had placed several orders over the previous weekend, and they hadn't arrived yet. Luckily, two of them arrived in the middle of the week and they had enough parts to allow me to continue testing.
One of these packages contained 4 solenoids that someone salvaged from an old first generation Pianodisc Player piano and sold to me on eBay. I got to testing these solenoids right away and found that these solenoids performed significantly better than the other ones I tested, outperforming them in every metric. These new solenoids were significantly more powerful, and had almost twice the push force as the other solenoids, while not drawing ridiculous amounts of current. They also run on 12 volts compared to the 24 volts that some of the other solenoids required, which makes sourcing a power supply for this project significantly easier. These new solenoids are also much more robust, and have stronger mounting hardware, allowing for more stable operation and less noise. The old solenoids also had a problem where they would stick in the extended position for around a second after the power supply was disconnected. These new solenoids do not have this problem which will be important when playing music that has lots of short (staccato) notes. The new solenoids do have one downside though, in that they are significantly bigger, to big to be fit side-by-side in the piano. However, they fix this problem by having long and short solenoids, allowing them to be staggered so that way they can fit within the required space constraints.
Exhibit 1: Fitting short and long solenoids into the piano.
The second package I received contained a MIDI host device as well as a MIDI cable. One of the problems I faced when designing the decoder was the problem of easily generating MIDI signals to test it with. These two devices would allow me to generate MIDI signals using a digital piano that I already have by connecting it to the MIDI Host device using USB and then connecting a MIDI cable between the host device and the input to my decoder circuit. When I press a key on my digital piano, It will send a signal to the decoder circuit with the information about the note press. Although in the future the digital piano will be replaced by a computer reading from a file, the digital piano will allow me to test the circuitry before I have all that set up by simply pressing a key. I connected the circuit after triple checking all of the connections with my multimeter and probed it with my oscilloscope, expecting to see it work, and I got nothing. I tried many different things; playing with settings on my oscilloscope, checking connections, turning it off and on again, all to no avail until after 3 hours of debugging, I realized that one of the microchips was plugged in backwards. Once I discovered this, it was an easy fix and it mostly worked first try after that. (I forgot to ground the "Enable Output" pins on some of the latches I was using, but this issue didn't take long to find).
Exhibit 2: Solenoid actuating a piano key as I'm holding it.
Exhibit 3 above: One of the new long solenoids extended on top of a power supply.
Exhibit 4: Oscilloscope measuring decoder input signal (yellow) and an output bit signal (blue).
Overall, even with the delays, I am proud of the progress that I made this week and I hope to have a note or two playing automatically by early next week. (Assuming FedEx doesn't drop-kick my packages...)
Week 3: Blog: Quick Progress Update
I don't have a whole lot that I can show for this week, hence why this blog post is short. I started off this week by building a circuit that can take the output from the decoder (Which I got working last week) and manipulate it to control individual notes. While I have this circuit mostly working, I cannot show it off properly due to some problems with one of the microchips I used. I am working on a workaround, however I was not able to get it done in time for this blog post, so I will show this circuit working next week. I also hope to start putting some of these circuits on printed circuit boards to start moving towards a final product. I talked to one of my father's colleagues who works with circuit boards and circuit board design and he gave me some tips on what software I should use and how I should go about getting these boards manufactured. Hopefully I will have this mostly done by the end of next week so that way my wiring can stop looking like some very colorful spaghetti. (See last weeks images)
Week 4: Blog: Light at the end of the tunnel
This week, I think I'm finally getting close to a working prototype. I have managed to get one solenoid to extend and retract when a the correct input is given (shown in the video below) and I have begun mocking up a mounting mechanism for holding the solenoids inside the piano. I have had a few issues with both of these things, one of them being that the transistors that I bought to control the solenoid tend to overheat and fry themselves fairly quickly when in use. I plan to solve this by buying some more rugged transistors that can handle more current, but in the meantime i've gotten around this by not leaving the circuit plugged in for too long. The second main issue that i've encountered is that the frame of aluminum extrusion that I am using to hold the solenoids in the piano is not super sturdy and has some problems with solenoids moving out of alignment. So far, I'm still looking for a better solution for this and hopefuly I will have it by next weeks blog.
Showcasing the frame made out of aluminum extrusion
Week 5: Blog Post: Production
Since I got the decoder circuit working last week, my main efforts this week have gone towards finalizing the design and getting circuit boards manufactured. So far I have been testing all of the circuits on solderless breadboards, which are amazing for prototyping and allow me to find problems and fix them quickly, circuits on breadboards are a bad final product due to their flimsy connections, size and messiness. This is precisely the reason printed circuit boards (PCBs) exist, and why I am designing one for the decoder circuit part of my project.
Screenshot of the EAGLE Editor I am using to make the blueprints for the PCB
As for the solenoid control and volume control circuit, I am still waiting for parts to allow me to test the design I have in mind, and if everything goes well, I can start on a circuit board design for that part next week.
Week 6: Blog Post: Production Delays
This week I mostly worked on finalizing the circuit board design. (Pictured below) It was a challenge to route all of the traces and I ended up needing to use 6 layers in the PC board. I also added circuits for power delivery and output control to allow this board to interface with the other circuits that will actuate the solenoids inside the piano. I have ordered the circuit boards from JLCPCB, however they are taking some time to manufacture and have to be shipped from China, so I will hopefully have them by next week!
Showing the circuit board in the EAGLE editor with layer 1 in red, layer 2 in yellow, layer 3 in green, layer 4 in purple, layer 5 in orange, and layer 6 in blue.
Week 7: Blog Post: Production
So far this week I have worked on assembling the printed circuit boards that I designed last week. I had one issue related to the fact that the footprint for the transistors I was using on the circuit board was incorrectly configured causing two of the pins' locations to be swapped on the board. This is an unfortunate issue, but I think I can solve it by finding a suitable replacement that will work with the different configuration. Aside from that, I tested some of the other parts of the circuit board which all seem to function properly, and I have started testing the volume control circuitry, which seems promising so far. By next week I hope to have the solenoid control boards in production and have the transistor issue fixed.
Post 9: Senior Project Blog Post #8: Volume Control
This week I mostly worked on the circuit board for volume control since that is the last main component I need to get this project working. The circuit works by providing 24 volts to the solenoid when a note is turned on initially, and then stepping the voltage down to 12 volts after a certain period of time. This period of time is varied by the volume data supplied from the MIDI Decoder. this works using a latch and a binary counter such that when a note's state is changed, the latch is set (1 = note on, 0 = note off) and the velocity information is stored in the counter. The counter is then set to count down, and keeps the voltage at 24 volts, setting it to 12 volts when the counter reaches zero. This way higher velocity notes will have the solenoid have more power (24 volts) for longer, in theory producing a louder sound. I wasn't able to get as much progress done on the circuit board design as I would have liked to this week, but I hope to have a finished board (and possibly notes playing with varied volume!) next week.