I am working on a printable Marble Machine, heavily Wintergatan (link) inspired, for the Marble Run contest.
It will play music like the original, but limited to 10 notes (I think) and a limited loop time, but it will be able to play the original song recognizable. I have extensive Marble Machine design experience and have been featured multiple times on the Wintergatan YT channel. So I believe I am capable of turning out a fully functional design in the coming weeks (I have started conceptualizing this design years ago and the prototypes are promising, the Marble contest is the last push).
The design will also be programmable, you will be able to exchange programming rolls for different musical scores and xylophone bars for different notes
The issue is, I would really like the programming roll to be customizable! And Parametric Model Maker seems like it would be perfect! I will design the primary example roll in SolidWorks but that will be a painfull process where every note has to be sketched individually and non customizable.
So I am looking for somebody that is willing to join ventures and design a compatible programming roll in OpenSCAD so people can compose their own songs. It will be two separate models (one on your profile and one on mine) linked to each other. IF any prize is won, we can split it. And you can choose your own license/exclusivity.
I am not sure if the model will be popular, it needs some Maker Supply parts and a worldwide available (Ali, Amazon, etc) Xylophone. I will upload a video to my modest but dormant YouTube channel with almost exclusively Marble Machine enthousiasts (3.4k).
Please let me know if you are interested in this or have any thoughts or questions below!
Let’s make this thing a reality!
I recommend starting with the non-customizable design in SolidWorks and uploading it to Makerworld. Alongside this, create your YouTube videos and other content to gauge interest in the model. This will help you decide whether it’s worth investing the time to make the design parametric.
Keep in mind that the Parametric Model Maker cannot read STL files, so any parametric aspects will need to be built from scratch in OpenSCAD. That said, completing the initial design in SolidWorks is still valuable, as OpenSCAD designers can use the SolidWorks objects as references when collaborating with you.
I love the marble machines and components from your YouTube videos. I’m not sure if I’ll have time to help, but I have some OpenSCAD experience and love mechanical marble machines.
I’m not even sure that it needs to be modeled in OpenSCAD as it the model could just include instructions on how to space out the peg placement in the drum. It may even be better as a web app where someone can select a song and it converts it into the pattern they need to place the pegs, if that makes sense.
I think having it parametric from the start makes the model more interesting, but yeah, what makes it worth it? I don’t think it will get thousands of downloads, but it will be an awesome project nonetheless. I make all my designs because I like designing, when they become popular it’s a bonus of course
I think I’ll do that as well, but that is even more involved than OpenSCAD I’m afraid. I will make the first drum in SolidWorks but I have to draw and cut almost every cutout separately without having a good way of keeping track of what I’m doing… But yes, I’ll add technical drawings to the listing!
I’m truly impressed with all the models you made using OpenScad! I remember somewhere that you really prefer code to visuals, I have the exact opposite. I can think up and design almost anything but as soon as I see a line of code I feel like the dumbest kid on the block.
Some people have reached out to me and I have a much better understanding of the limitations (but also possibilities) of OpenSCAD for this project. I think for now it is time to get on with the design and hone in on the final dimensions of the roll/sleeve before diving too deep into the OpenSCAD file.
One idea somebody had was to not have a complete roll but only a sleeve on top of a permanent roll. Prints quicker and the gearing system can be on the roll instead of the sleeve:
This is more or less how the sleeve should look in the end, but not with this pattern of course. Right now I’m thinking of 9 tracks and 64 possible positions per track (8 bars with 8 (half)beats).
So I’ll get on with the design and I’ll post some technical drawings here if I reached that stadium. For now the most importand mechanical part seems to work and that is the marble dropper with the escapement mechanism and a dummy programming roll/disc:
It is due to a medical issue called Aphantasia, I can’t form a mental image of anything, loved ones, friends, or objects.
That is possible, there are no limitations that jump out to me.
Detractors are usually prone to denigrate things that they do not understand.
Models that require organic structures are not usually possible or at least very difficult, those that use standard geometry are fine. As you say, many of my models have complex structures.
Latest update on the mechanical side of things, I’m happy with the marble divider and the angle (exactly one bambu magnet box + merci bar). Marbles can get stuck but will be nudged easily by the next one. So I can start to work on the rest of the design and the final drum dimensions:
I didn’t mean impossiblilties, where there is a will there is a way. But I can imagine designing large assemblies would be harder in OpenSCAD compared to Fusion.
I am just stoked that there are so many (free) CAD packages, one for every workflow and every preference. And OpenSCAD clearly enables you in a way no other CAD package ever could have
Thanks! I was quite involved in that project and multiple of the design improvements of the MMX were first designed and prototyped by me (marble divider, escapement dropper). However, I don’t think any of the CAD data would have been helpful with my tabletop version. But without all the knowledge and experience shared by Martin I wouldn’t have been able to design mine either!
Anyway, here is the final result, posted about a week ago:
