DIY digital piano

aleath, awesome work.

will the project be open source so the community can work together. we might even run a non profit group buy for the hard to make parts like for the printed circuit board, precut keytops.

RIP @luns @n-player @vagfilm @kapelli @HZPiano -- I know some of these are just waiting to jump on the bandwagon when they will be ready

😁 Must resist… must resist… must resist…

😢

Cheers and happy projects,

HZ

@RIP the key tops are made in two layers, 6mm mdf and 3mm white acrylic, which are then glued together. Laser cutting 9mm white acrylic would also work, but thicker material has a more prominent slope from the laser beam taking more material from near the top, and cost increases non-linearly with material thickness, so that it is often cheaper to laser cut two thinner layers and glue them together. I didn't actually find there was much of an issue with black and white keys hitting each other. The displacement of black/white keys towards each other due to rotation around the pivot point proved minimal enough that it could be compensated for with a minimally increased gap between black and white keys. What are hw and sw? The sharps are painted mdf, which I thought would be easier to sand an angle into than plywood. They are laminated, 3 layers of 3mm MDF glued together, but in the next iteration I will just use 6mm MDF and sand out any slope from cutting. I painted and then applied a spray on matte varnish. I would be interested to know the brand of enamel you used.

I outsourced the laser cutting for this prototype, and it wasn't too expensive. I had free laser cutting at my old workplace, and originally planned to laser cut the entire key. The thickest I could do was 12mm mdf, which functioned ok as a key, but had a little too much flex for my liking. But when I moved to a new job and got a quote to laser cut a set of keys I was surprised at how expensive it was. And 12mm was the upper limit of what could be done in MDF or plywood. I tried laminating an 18mm thick key out of three 6mm laser cut layers, but in the end the straight keysticks on a table saw and laser cutting key tops only was much cheaper and not much hassle. Then, using laser cut buttons everywhere reduces labour, because the mortices in the key don't have to be precisely sized.

I was hoping to eliminate as much manual labour as possible with this design, but it still takes a while to assemble… that's the unfortunate trouble with 88 keys! I would like to reduce manual labour further, and in order to do that might try an unweighted controller.

Yes it is such a shame about the lack of successful commercial projects! If narrow key keyboards were firmly established as a part of any child's piano education, then I'm sure that culture would remain firmly established. It is just unfortunate there is such a firmly established standard.

Indeed it isn't really a piano action, and that was something I was nervous about - how would it feel, given how mechanically different it is? Fortunately I quite like the feel, I wouldn't mind playing 'proper' piano repertoire on this kind of action, unlike on an unweighted controller.

I'm familiar with the Cybrid project! I followed it with great interest, and seeing the success of that was one of the things that encouraged me to embark on my own project. That haptic piano looks incredible, I would be fascinated to watch your progress if you tried something like that.

@kawafanboi The repository is a bit messy, but the code is on github: https://github.com/aleathwick/midi_hammer

I will probably make the drawings (for laser cutting etc) available somewhere (maybe as a release on github), once I have fixed them up. They are currently a jumble of many slightly different versions of parts.

aleath

whoa! this i can do!!! a more feasible project ! I love those weights! a very clever use of nails!!!
i just hope the magnets don't go flying on me….

will you be sharing the schematics at instructables as well?

Thanks for getting back to me and once again: awesome project!

aleath the key tops are made in two layers, 6mm mdf and 3mm white acrylic, which are then glued together. Laser cutting 9mm white acrylic would also work, but thicker material has a more prominent slope from the laser beam taking more material from near the top, and cost increases non-linearly with material thickness, so that it is often cheaper to laser cut two thinner layers and glue them together.

Yes that was my point. You find that 6mm is still cost-effective even when done for 50+ pieces (and spares) for a whole piano?

aleath I outsourced the laser cutting for this prototype, and it wasn't too expensive.

May I ask where an how much (ballpark, not exact figures okay)?

aleath The displacement of black/white keys towards each other due to rotation around the pivot point proved minimal enough that it could be compensated for with a minimally increased gap between black and white keys.

This is very good to know. From the video it seems okay. But I played one piano once with an increased gap between keys and it wasn't fun. So I hope this is really minimal and I plan exploring this myself too. Or alternatively (see idea below) making the lower plies smaller to achieve slope.

aleath I was hoping to eliminate as much manual labour as possible with this design, but it still takes a while to assemble… that's the unfortunate trouble with 88 keys!

That is part of the reason I said it was too complicated and/or expensive. Now that you have made vast majority of the setup in laser-cut parts, I am thinking of a way to reduce labor. It will work best if you could also manually create the laminated part of the keysticks by gluing them together. This sounds like more work, but keep reading and tell me what you think.

Have you considered using matching slots to align all the parts with each other? You could put tiny screws or tiny "blades" to keep all the parts in place, somewhat similar to how a key in a clavichord is kept from going letf-and-right with those slots and blades in the back of the rack (see pictures at http://www.bostonclavichord.org/what-is-a-clavichord/) -- having to glue each keystick would be more labor, but having the slots in place on all parts will make it immediate positioning with two blades per key and a single clamping of the various plies (potentially of different sizes to achieve slope and not having to have enlarged space between keys). All of that will work fine for the bottom parts, but can one make a slot in the white acrylic which is not through, but blind. Is that possible?
And most importantly: what would be the cost of this approach? Are there too many parts to cut to make it cost effective?

aleath I'm familiar with the Cybrid project! I followed it with great interest, and seeing the success of that was one of the things that encouraged me to embark on my own project

@CyberGene has really been the pathfinder for all of us. As you probably have seen from my other message I'm doing @JayKominek 's piano-conversion at the moment (and frustrated by the parts shortage)

aleath The repository is a bit messy, but the code is on github:

This is very cool. And very similar to my own experimentations, which you can check out, if you haven't already. The Pico-related experimentations are below (but do check the rest of the repo too if you have time. https://github.com/davidedelvento/Mybrid/tree/main/RaspberryPiPico

aleath I will probably make the drawings (for laser cutting etc) available somewhere (maybe as a release on github), once I have fixed them up. They are currently a jumble of many slightly different versions of parts.

I'm really looking forward to that. If I can, let me nudge you to share one or two "best" parts you have so people like me (who never used a laser cutter, but I used a 3D printer exactly once) can take a look and start experimenting. Even with the understanding that they are only temporary and not the final design, just to see what it entails.

aleath That haptic piano looks incredible, I would be fascinated to watch your progress if you tried something like that.

If I try something like your project, it will most likely be haptic. The most difficult part so far (besides building the keyboard which you are "solving" for me) is finding a suitable solenoid. Most inexpensive/small ones are too wimpy to produce enough force, and most good/strong one are too big and definitely too expensive (about $50 each, or more!!) to be viable for an 88 key instrument. I think something appropriate ought to exist, just that I haven't found it. Then, most electronics drivers are made to move the actuator in a specific point regardless of force which is typical and natural for any robotics or automation project. For this project, rather, one needs a specific force regardless of position which requires a different driving approach, not the one which you can find in off-the-shelf projects. The good news is that this other driving approach is actually quite simpler to design and implement, nevertheless it's an analog project which requires to be done from scratch.

aleath I would be interested to know the brand of enamel you used.

From your spelling I assume you are not in the USA, so I doubt you will be able to find it, but here it is anyway:

Brand: RustOleum Professional (related, but not the same as the regular RustOleum easily found in all hardware stores)
Model: High Performance Enamel.

As always, carefully manage the application environment (temperature, humidity, absence of wind and dust) is paramount to successful results.

Looking forward to chatting with you about these projects!!!

I always meant to reply - sorry I'm only just getting around to it now. It has been an unusually busy year.

Regarding laser cutting cost: It is still cost effective with 6mm material. At the time I think the ballpark figure for 88 keys was around 300NZD (including supply of material), at a local laser cutting shop in my home city in New Zealand. That was part of the reason I was attracted to this design - by only cutting the front of the keys, rather than the whole length, cutting time is significantly reduced.

If I understand your idea correctly, in the prototype before the one in my original post I think I did something similar - keys were made by laminating multiple layers, with the whole key cut out as one, rather than just the front part of the key cut. I considered using slots to align the layers when gluing, but settled on clamps for alignment. I.e. in addition to the clamps pushing the two glued faces together, first clamp the sides of the keys to align them, even though that is the 'wrong' direction to apply force when gluing. In this design (the one I posted about), the interkey gap was 1.25mm iirc. The kerf from laser cutting would add around 0.25mm to that.

What actually proved the most fiddly part of construction with alignment were the pins in the balance rail and front rails, despite using a drill press. When I 3d printed these in my very first prototype, they were nice and accurate with no effort, but at the time I had access to free 3d printing which I don't anymore.

Regarding the sharps, I've now turned to 6mm black acrylic (thinner sharps to accommodate my fingers between black keys on a reduced width keyboard), which eliminates any work in painting or laminating.

I've attached some example files to give an idea of what is involved, many parts from these files were used in the prototype, although some were tweaked and recut. If you're interested I could attach those as well. Blue lines are cut, green are etch. The 3mm MDF file has black key tops (to be laminated together), templates for guide pin holes, and raisers for the balance rail to lift the keys up where they balance. The 3mm white acrylic file contains white key tops and 'buttons', the bits that screw on to keys and guide them on the pins. The 6mm MDF file contains the bits that glue to keys sticks, providing height and support underneath the keytops. I think the biggest problem with the prototype was that I was overly generous with width between black keys for D, G, and A, at the expense of space for the remaining white key tails.

Congrats on the innovative nail-action!

aleath I always meant to reply - sorry I'm only just getting around to it now. It has been an unusually busy year.

Thanks a lot for the reply and no worries, this is my second project and I am still working at the first one. I would not start this one before finding the right actuator. At the moment I am undecided between custom-designed voice coil or piezo, but both have their drawbacks and would be needed to be squeezed beyond their specifications to achieve what is needed. I'm thinking of some alternative designs there, I'll keep this forum posted.

aleath What actually proved the most fiddly part of construction with alignment were the pins in the balance rail and front rails, despite using a drill press. When I 3d printed these in my very first prototype, they were nice and accurate with no effort, but at the time I had access to free 3d printing which I don't anymore.

Yes, I can see how that could easily become tricky to get right without CNC, especially when scaling to 88 keys!

aleath I've attached some example files to give an idea of what is involved, many parts from these files were used in the prototype, although some were tweaked and recut.

I am not sure I fully understand everything, for example I don't get the asymmetric way in which the laminated part of the black keys the a3-3mm-mdf.pdf -- if you have pictures that would be great, but don't sweat it if you don't.

aleath If you're interested I could attach those as well.

Yes, but again if it doesn't cost you too much since for me this is in the future.

aleath I think the biggest problem with the prototype was that I was overly generous with width between black keys for D, G, and A, at the expense of space for the remaining white key tails.

I think the best would be to go for (or at least start) with one of the DS keyboards, since I assume they tested them a bit before publishing them. I think they have their measures posted, but I can't find them (I love them from a technical point of view, but from a "political" I hate what they do with constant advertisement, surveys, announcement of upcoming products "almost ready" that never materialize and websites that continuously come-and-go). They also have STL to 3D print prototypes, but I can't find them either. Good news is that I do have the STL files. I'd post them here but they are almost 50MB each (15MB if I compress them) and I thing @CyberGene would slap my wrist if I use that much storage, even if I post only one of the 4 available widths that I have, namely 5.5" (138mm), 6.0" (151mm), 6.5" (161mm), and 7.6" (188mm) per octave. I can post them in wetransfer, but they will last only a week, so if anybody wants them and is ready to download them immediately let me know and I'll send the link here.

Doing this as a prototype, I'm still undecided if I should go with custom-made parts such as these, or instead try to find existing things that would work with minimal labor, e.g. with existing parts such as the following the labor is certainly minimized (but then keyboard size is "standard", not "reduced" -- perhaps okay to experiment with things when not having access to laser cutter and 3D printers which I don't have and would not have where to place even if I decided to buy them). These could perhaps be attached to rectangular stock dowels as keysticks...

https://www.howardpianoindustries.com/piano-keytops-without-fronts-set-of-52/

https://www.howardpianoindustries.com/german-piano-keytops-with-attached-fronts-white/

https://www.howardpianoindustries.com/key-punchings/

https://www.howardpianoindustries.com/piano-key-buttons-set-of-15/

https://www.howardpianoindustries.com/piano-center-pins-1-oz/

RIP I'd post them here but they are almost 50MB each (15MB if I compress them) and I thing @CyberGene would slap my wrist if I use that much storage

No problem, AWS S3 is ridiculously cheap, I pay $0.02 a month for our entire forum storage.

I’ve currently limited the max file size upload to be 30 MB though. That’s per file.

CyberGene I’ve currently limited the max file size upload to be 30 MB though. That’s per file.

Awesome thanks. If anybody expresses interest I will upload them then. As I said it's about 15MB compressed, are zip or tgz files okay? The content will be stl as I said.

RIP are zip or tgz files okay?

Zips are OK. Not sure if tgz will work, although I can configure it.

RIP I am not sure I fully understand everything, for example I don't get the asymmetric way in which the laminated part of the black keys the a3-3mm-mdf.pdf -- if you have pictures that would be great, but don't sweat it if you don't.

I did take some photos during assembly but I'm not sure where they've got to. I'm happy to clarify anything if you have any questions. There is nothing particular about the black key arrangement in the drawing - it just me trying to arrange them in such a way that saves space by sharing edges and rotating/reflecting. Then 3 x identical 3mm pieces are glued together to laminate to 9mm thick sharps, which in retrospect was a bit too thick for a reduced width keyboard.

Here are the other files, which were done a bit after those originals I posted. Lots of buttons, this time with side entry notches for screws so that screws don't need to be undone all the way to remove them, and many different slot widths to test different clearances. Blue is cut, green is etch, so I can mark the different width slots:

And a two octave version of the templates for marking holes for the balance and front rails:

RIP I think the best would be to go for (or at least start) with one of the DS keyboards, since I assume they tested them a bit before publishing them.

I am satisfied with my resolution to this issue, which is to adjust the spacing somewhat - now the widths of the tails of C, D, and E equal, as are F, G, A, and B

RIP Doing this as a prototype, I'm still undecided if I should go with custom-made parts such as these, or instead try to find existing things that would work with minimal labor, e.g. with existing parts such as the following the labor is certainly minimized

Speaking of minimizing labour, I have just recently assembled a new 5 key prototype (with no electronics) that, in exchange for higher laser cutting cost (perhaps double), greatly reduces labour over my last prototype. More or less everything is laser cut, with no need to manually drill holes for pins. I will post some photos sometime soon, maybe in a new thread.

aleath What actually proved the most fiddly part of construction with alignment were the pins in the balance rail and front rails, despite using a drill press. When I 3d printed these in my very first prototype, they were nice and accurate with no effort, but at the time I had access to free 3d printing which I don't anymore.

drill press yes, but you want to make sure to precisely pre-tap the hole with the shortest possible centering bit.

the drill press run-out and bit length matters alot. if you do it on those $200 presses from harbor freight, you can not get the bit to go down stright.

you need the heavy industrial press, and you need a test dial to make sure the bottom tray is at the right angle.

but with wood, it's never perfect. they bend those pins quite ALOT to get wood keys to fit right, and they're making it on giant perfectly milled out beds.

aleath Here are the other files, which were done a bit after those originals I posted.

Thanks a lot!

aleath Speaking of minimizing labour, I have just recently assembled a new 5 key prototype (with no electronics) that, in exchange for higher laser cutting cost (perhaps double), greatly reduces labour over my last prototype.

Great to hear!

aleath I will post some photos sometime soon, maybe in a new thread.

Looking forward to that