A New Music Programming Wheel

[@C1oudstr1f]

MM1 and MMX are not failures. They were learning tools, and a great source of entertainment and inspiration for millions of people! Their music didn't need to be perfect for us to enjoy it!

[@randomsketchystuff5572]

Exactly!

[@starduck2]

We can learn from failures, but it does not mean that they are not a failure. He had big goals for it that he was not able to achieve, so for his ambitions, yes it was a failure. It was not a "total failure" as there is much to be learnt from it, but we should never bend reality to accommodate our feelings, or else this might lead to a point where we think learning with mistake is equal to actually achieving goals.

[@matthiasengh7935]

They were work of art!! Martin is just tired of building art and finally wants to build a machine

[@falcon1209]

And yet, he is still chasing perfection.... All these "tight music" videos are just digging a deeper and deeper hole.

[@chaichantheshiba5902]

It doesn’t have to be perfect to be great. Don’t fall into the perfection trap.

[@stevenlischer]

Hey Martin, really happy to see you're back and taking first principles approaches. I've designed and worked with drum assemblies and they can be tricky. Here are my recommendations that I hope can save you many headaches: 1. Sheet plastic for precise parts will betray you. Aluminum is your friend. Machinable plastics have young moduli low enough that they become very undesirable for large parts that need to be stiff. Plastics with better stiffness are usually not good for machining. 6065 Aluminum is awesome though and, if you take it slow, might still be machinable on the CNC you're getting. Almost forgot... the sheets of plastic will probably come warped and will need considerable work to true them. Worse, the warp may not be apparent until after cutting the shape out because the weight of the sheet will pull it into flat. 2. Stage your gears. A large helical gear on the exterior does look cool, but it will be a bear to maintain and the gear that drives it will also need to be large which either means you need to cut another custom gear or you will be paying a machinist a lot of money for a large custom gear. Reduce your drive gear considerably and you might be able to find something premade. Then you can use an off the shelf gear box to get the same final rotation speed of the drum. 3. Fix the drum to the shaft and use cross bracing in the interior. Tube steel is great and easy, but you might get away with much lighter/cheaper materials here if done right--maybe even some rigid plastic. I'd also add an interior hub. This will add torsional and axial stiffness the current design is lacking and needs to reduce wear/noise/precision. 4. Separate the drum from the drive gear and use a dog clutch to transfer power from the drive shaft to the drum shaft. If you use a coupling sleeve on the other side, you will be able to very very quickly remove and swamp in another drum. 5. Consider reducing the need for high dimensional accuracy of the programming drum by designing readers first and making "float" along the surface of the drum and only care about the relative height of the "pin" at any given point. This will make the drum much easier to manufacture and still play precisely even if the surface is 1mm off due to manufacturing or warping from temperature changes. Maybe such a reading head would be too hard, but I think worth considering Best of luck Martin! I'm always excited to see the next video of your journey!

[@JonLovestheLord]

These are great points. To actually get them into the group that Martin is listening to, you would want to post these on the Discord for the marble machine. It is quite apparent that Martin is not reading KZbin comments now. Also, be careful of using phrases that don't translate into other languages, idioms like 'it will be a bear to maintain' may not have meaning for Martin, his English is really, really good, but remember he is Swedish and the idioms don't necessarily translate, and worse, the word 'bear' has multiple definitions and meanings. My personal preference for the spoke wheel would be fiberglass composite for weight and stiffness concerns, but, it isn't easy to manufacture.

[@zackbuildit88]

Doesn't aluminum warp in heat though? If it's touring then that could be an issue, in a similar way to how many brass and woodwinds need to be retuned when traveling due to humidity and ambient heat making the materials expand. I'm not an engineer though, so i can't say for sure if it'd be an issue in this case

[@stevenlischer]

@@zackbuildit88 ~ .025mm/M-K. With a 20 degree Celsius operating temperature range and a 1 meter target diameter the total change to the radius would be as much as +/-0.25mm. Not sure if that would be an issue for the readers, but something to consider with any material. Warping probably isn't likely here with a 5 to 10mm plate thickness, but I have seen bad plate stock that comes bent or with poor flatness, but most places one orders from will make guarantees on flatness.

[@mikkokannisto]

I'm worried that number 3. things as well. There is just too little fixed points to hold all stiff and in position. Cross bracing is must!

[@ennocramer4703]

Experience from 2 previous tries unfortunately suggest, that your comment will be unheard, unrecognized or ignored. To prevent the first two issues, it might help to post in Discord channel - no guarantee however.

[@VagabondTE]

I think you should make more than two. Not just for backups, but because you're going to need more time than a single song to replace pins. Your team has to set pins, confirm placement, and need to be ready and alert before they go to make the swap. This is a lifetime on a hectic stage. I would say three wheels at a bare minimum, which gives you a two songs buffer, but honestly I would build six. Three for the performance, a travel back up for immediate problems, and then two more you can leave at home for catastrophic failure backup. Normally I wouldn't worry about making so many redundant parts but the programming wheel is different. This is the heart of the machine and it's also the most handled. It's going to receive the most wear and tear of any part. You need backups for this. Also consider that the wheels are going to change the least in machine redesigns. Because its really the foundation for the rest of the machine and the size is set. If you could replace certain parts of the wheel then I wouldn't worry about this but the wheel needs to be machined as a whole. That means you need whole backups. I would also take a moment to consider adding handles and shipping contact points to reduce that wear and tear. And lastly I would make sure the axle and its contact points are extra sturdy. You don't want the bearing or the axle itself being the interface. That will severely increase your wear and tear. The bearings should only be under stress during operation, not swaps.

[@89lutzy]

At 6:07 he says he going to make two wheels so that a song can be changed while one is on stage being played. That's already the plan Edit: wait you said more than two... my bad you were paying attention, im the one who wasn't

[@VagabondTE]

@@89lutzy no worries

[@kerryabear]

My thought exactly. For touring I’d have a programing wheel for each song plus backups. If not, then exactly what you said.

[@VagabondTE]

@@kerryabear Well, they're pretty big and they have to be manufactured and kept whole. So I do understand why they need to be repined for songs. A typical concert is 20 to 30 songs and even if they only use the marble machine for every other song that's still a truck full of programming wheels. But yea, only two just ain't going to cut it. I'm hopeful that he's going to be able to make large pin groups that have sections of notes, to make swapping out faster and more reliable. They could even have puzzle piece interfaces to make sure a song is put together without errors. But I'll let him figure that out on his own.

[@babybloc]

I was thinking he'd say there would be multiple "tracks" on the same wheel, and you'd shift the wheel horizontally to change songs. All the instruments could be duplicated on each "track" with different information for each song. You'd just be limited by the width of the wheel. And you'd need to have a "neutral" place to shift tracks. Save you having to remove the huge programming wheel.

[@dm45lm]

Instead of attaching the wheels directly to the bearings I would suggest you make a hub and attach it to the shaft with splines. That will lock the wheels to the shaft and ensure that it will not twist over time. Now that the wheels are locked to the shaft and timed with splines on both ends you can just support the shaft with bearings that are mounted to the support frame.

[@liambryant3519]

I was also worried about the two wheels becoming missaligned.... That is a good solution

[@Jowbaka]

Or maybe you could keep the original idea but use a 10-20 cm diameter off the shelf plastic pipe and fittings, that the wheels can be attached to with screews. It would provide more rigidity, while keeping the smoothnes and precision of the metal bearings.

[@tutmondigo]

With this design and using quick release bearing housings on the machine you would only need the two bearings total vs two for every wheel that gets made.

[@blahorgaslisk7763]

With all the problems of the MMX programming wheel it had one big advantage, torsional stability. The large sheets bolted to the rims made them extremely stiff. This new design sacrifices that stiffness which when talking about precision in timing. The idea of using splines on the shaft could help with this, but I'm not sure it's enough. The machining will have to be very precise with absolutely no backlash in the splines. Even a backlash of a hundred of a degree would mean a shift in timing between the two ends of the cylinder. The two rims has to be fixated in relation to each other. I don't claim I have a solution, I can just call attention to a problem that will come back and bite you if ignored for too long.

[@Jako1987]

Or if you want the bearings on the wheel you can put a tube between the wheels. The axle is inside the tube. The tube keeps the wheels from twisting. This looks like a reel from a combine harvester. You can steal parts from those 😅

[@Infernoblade1010]

Martin, to correct your statement at the very beginning, you succeeded once, succeeded again to a greater degree, and are working on much greater success!!

[@5ANDW1CHES]

Amen

[@PandaKnight52]

Hear hear

[@cary3345]

succeeded, then could have succeeded but gave up before finish line, and now is about to succeed :D what I mean by that is that MMX deserved at least one song published here... even the original marbe machine had its moment of glory :')

[@Thegbear]

@@cary3345 He couldn’t. It broke horrendously.

[@boonjabby]

Came to say something similar. They weren't failures. They are artistic iterations / versions of yet another masterpiece

[@gamingscientist7445]

You didn't fail twice! Those machines are still worth their weight in gold thanks to the lessons they taught and the entertainment they've provided. While they aren't world tour worthy, they deserve a place of recognition for the foundation that they've created.

[@ben_1]

"to fail: to be unsuccessful in achieving one's goal" His goal has always been to build a machine that he could go on tour with. He didn't achieve that twice. He _failed_ twice. And there's nothing wrong with that. MM1 and MMX were successful on KZbin and he learned a ton from building them, but that doesn't mean they weren't failures, because he *_failed to achive his goal_* with them. Stop twisting reality just because you don't like the word failure.

[@justintime5021]

@user-sj8ew2ye3v instruments tend to make music. I don't recall the mmx ever doing that. I recall the original marble machine making music though. That's odd... Sure the mmx couldn't tour either... That's fine... It was honestly never about that. But it seems to have utterly failed in being an instrument since the artist was so fixated on achieving perfection he never bothered to play it. I think that's the saddest thing about all this. He even said he was going to make a music video using the mmx before he moved on but since it's now in a museum I don't see that happening. So as someone who backed martin via KZbin membership from early on in this project it's a bit sad that we never got to see it really make music like the original machine did despite it being far more capable. So these days I'm taking a wait and see approach to this channel and to Martin. Hopefully he will wow us in the future with some kind of crazy marble monstrosity... But I'm not going to wait around for it.

[@schucas1899]

i''m so happy to see a highly motivated martin again - keep up the great work!

[@maneyaf]

Martin, you did NOT fail twice. You learned what worked and what did not. And I don't care how many cuts it took to make the entire song video, I absolutely love your first marble machine music video.

[@douwe4254]

It's the same with children. If you have three, the first one you wing it. The second one you understand you can improve. The third one gets the best parents, because they had two to learn from. All three do the same thing. You got your passion from the first. The second was well refined, but needed more improvement. The third one you implement all the lessions you've learned. Not a failure. But it might feel that way. I'm the oldest, and I never feel any different around my blood own father. And what I learned from Martin: The first is to learn the basics. The second you implement the complicated mechanics, but the process teaches you enough to understand you started wrong. The third is the final product. When I program software for my previous boss, I always told him the first version is heavy and slow. The second one works, but will have issues with workarounds. The final version works perfectly. And it often does. Even in games like Oxygen Not Included I just start over if I made too many mistakes that will build towards an ultimate failure of base. To a beginners eye, it is amazingly complex. To me, just a lesson to move forward. And in that game, I can finish it on the hardest map and settings, because I understand everything about the process. Martin is at that point right now. He just can't accept his own perfectionism in the process, only constructive progression.

[@jacobyunderhill3999]

I think he knows this for sure. But perhaps it's ok to still use that word; it's now a matter of redefining it and removing the negative connotation. Reclaiming it, if you will. 'Failure -- listen here you old, scary word. We're not going to take it anymore. You're not going to get us down. We're not scared of you anymore. We're going to learn and we're going to persevere. That's what you mean now and henceforth!' That way, we don't have to come up with a new word. And when kids are called failures, if we do our jobs, they'll know that it means something different, something about experience and opportunity and hope. Not something to run away from.

[@joshw3485]

this is, by far, my favorite series to follow on YT. I like cars and fishing and a whole bunch of stuff but, when these videos drop, that gets me excited. Keep up the great work.

[@daniellima4391]

I see a new Wintergatan video I insta click

[@codemonkey2k5]

Absolutely.

[@DasVERMiT]

If completely rebuilding a historic wooden sailing yacht named Tally Ho sounds interesting to you, you may also enjoy the series by the channel "Sampson Boat Co". I get that same sense of excitement for it!

[@KlausWulfenbach]

1:40 "If we turn them around, we can turn them into design requirements." YES! Martin is really getting the hang of engineering.

[@Lawsome1997]

These videos are inspirational. By going through such hell, you've shown everyone the power of being organised and having clear task lists and requirements.

[@ExiorasMusic]

Right? These videos are a testament to engineering and project management. It's amazing to see just how much the modern processes we use impact a project when done properly.

[@levilukeskytrekker]

+.

[@lilyofluck371]

@@ExiorasMusic It's so disheartening seeing people treat engineering like the devil in a lot of the comments TwT

[@madman2572]

Ikr? These are even better than the build videos imo. So much innovation and learning going on.

[@daverotors]

One of the philosophies I developed during my career in software engineering is that you write every piece of software three times. The first time, you don't know what you're doing and just make stuff up as you go along, and somehow (barely) get it to work as you want. Second time, you try to fix all the mistakes while keeping (or even adding) the features that you really want or need, but you end up in too much complexity, which will overwhelm you. Third time, you're much more aware of the actual use cases/requirements and the pitfalls to weigh them again. This is the time it's actually getting good. Seems like you're a perfect example of just that pattern, and I'm optimistic for your third version :)

[@Dannerrrr]

That wheel is gonna twist like crazy without some reinforcement. The drag on the cams is gonna be strange.

[@deemstyle]

This exactly. But when it does he can create another 58 videos about why we should care. Dude burned that bridge a long time back!

[@maxweber06]

Yeah, both bearing mounts are completely independent of one another while all the force is on only one side. There's almost nothing stopping the torque the programming plates will put on it. I think putting gears on both sides is the most straightforward thing to do. That equalizes the forces and will make it run even smoother.

[@conorstewart2214]

@@maxweber06 adding extra plastic supports in the middle would help with stiffness, also adding extra metal rods or metal strips on the spokes or inside of the wheel would also help. Edit: or lock both wheels to the axle and have the bearing on the support structure and supporting the axle rather than between the axle and wheel.

[@catseye10000]

@@conorstewart2214 I would jsut add a 3rd inner support wheel

[@afslayer]

@@deemstyle hes never claimed to be perfect..that would be boring anyway.

[@joekx9]

I'm currently studying product development, and my professor taught a great principle the other day. If was along the lines of saying that while function took priority over form, you don't have to sacrifice it entirely. It's about the order of operations. Industrial Designers/Engineers will incorporate aesthetics and design to an existing product without compromising the functionality *after* it has been finished. I'd encourage to keep that in mind because I really do love your artistic vibes.

[@voxel9470]

I remember wintergarten mentioning something like that in one of his videos. Maybe my wires are crossed though

[@joekx9]

@@voxel9470 You're probably right honestly, there is a lot of thought and care put into this so i'd assume it's been mentioned at least.

[@ig88bountyhounter]

I guess we are all here to see you succeed. It has made me feel so bad to see you call mm2 a failure, and I know it doesn't help the least bit, but it was extremely good entertainment to follow your progress, and altrough it hasn't panned out, it's an even bigger pleasure to see you back in action. Godspeed Martin

[@jacobshore2851]

Are you worried about the length of the songs that can with this design vs the old designs? With large programming profiles there won’t be nearly as much room around the wheel for longer songs.

[@ericmorrison278]

I just needed this video today. I love this whole process and watching the progress unfold has been truly a sight to behold. Mad respect for sure!!

[@simonrobert3447]

Hey, good design ! However don't you think the two ends of the cylinder (the two wheels) are not enough connected to each other? It feels like if you give torque to the gear, the whole programming cylinder will twist and deform... Maybe you could add a cross piece on the inside at 45° to give it more strength? A bit like the metal cage you added on the back of the MMX :)

[@ninjafruitchilled]

Will that really be a problem? They are connected by the shaft, and power transmission through the gear to the shaft should be pretty solid, and I would have thought the shaft can then accelerate the other wheel pretty cleanly. No need to transmit much power through the structure of the cylinder itself.

[@ExiorasMusic]

Good observation. I personally disagree, thinking that the cage will distribute the force, but we'll just have to see! Either way, adding a cross beam to the center should be easy enough.

[@titaniadioxide6133]

I’m thinking some tension cables on the inside, oriented so that when the near side moves, the cable pulls the far side along with jt

[@simonrobert3447]

@@ninjafruitchilled As I can see on the video it seems that the whole wheel is only connected by bearings, in other words i suppose the programming wheel can freely rotate around the motionless shaft. Now, because the torque is transmitted by a side of the cylinder and the programming pins creating resistance, thus opposite torque, i think it can deform the wheel and create timing accuracy issues 🤔

[@mickmack1213]

Had the same feeling. All the wrapping and connecting parts, are just attached by one screw, such that a twisting motion is only prevented by friction. (Also during machining of the programming holes) I can't really guess how big the forces are, but during machining, and with multiple pins pulling at the wheel simultaneously, they might get quite high. I'd prefer to see if the shaft (or maybe a tube around the shaft) is attached rigidly to the two tube side, such that this central piece can absorb the torsion forces. This has the benefit, that the timing related wrapping pieces are not a structural part. One last point to add to this are the pins connecting the wrapping parts. Martin clearly thought about it. Let's hope he'll just test if the torsion resistance is big enough.

[@kevinbuiied]

Have you considered a Programming Belt instead of a Wheel? Like taking a punch card music box's punch card, and taping the ends together. Benefit are: 1. that the belt would only need to be reinforced near where the marbles are released, and the rest can dangle. 2. variable length songs. 3. can swap out a song simply by splicing a new track onto the current track and then connecting the ends when available. 4. can have a song on Side A and Side B.

[@voxel9470]

This would be good. I hope wintergarten sees this

[@elMuele]

I think he likes the idea of being constrained by the 64 bars the drum offers when he starts composing songs for it. Otherwise he could just use papertape like most other music machines. Or something similar made from a plastic sheet for increased durability. If you scroll far back enough you will see that he has already worked quite a lot with paper based music machines.

[@voxel9470]

@@elMuele I saw someone on the discord server say that a tape system wouldn't work out because it wouldn't be able to apply the mechanical force required to actuate the marble droppers

[@RobFarley74]

Funnily enough I suggested multiple flat plates, they'd pass through the machine flat and join without any gap. Probably pulled through the machine like an old tractor feed printer

[@abrodeur]

I am still not convinced that he needs physical programming system for his minimum viable product. Sure doing some kind digital system like what he using in testing could play extremely tight music. Then once every other part is working simply switch out the digital system to physical system.

[@dexterm2003]

Plastic is very flexible and rarely fat. It also can warp and sag over time. I think you would be much better off with a nice stiff aluminum like 7075. While it is nice to be able to machine it in house you would be much better served by a better material. Also you should really consider having a stiffening ring or two on the interior of the drum. The plastic peg holes could also be a problem in that they are on the same order of strength as your pegs. You really don't want the drum peg holes wearing out you would rather have extra pegs for when they get loose. Please consider doing the plastic design in aluminum.

[@youareliedtobythemedia]

Ah for a prototype plastic is good enough. Aluminium is pretty expensive, and he can make one if the plastik one works

[@foldionepapyrus3441]

@@youareliedtobythemedia The materials properties of the two different family of material are so wildly different the plastic one is relatively likely not to work, where the Aluminium one should for the same design - this is one of those parts where its so damn large, will take ages to machine and with so many different locations forces will be applied creating so many varied and inconsistent twisting forces trying to economise and build in cheaper materials is foolish. This is the part you build bomb proof solid the first time you build it and save money by doing so as even in the cheapest materials its going to cost a huge amount of time.

[@SkigBiggler]

Engineering plastics tend to be thermoset, think Bakelite, more brittle than soft. Same thing with acrylic. All plastics can deform when heated, which is something to be aware of, particularly when touring, but they’re more than capable of being used for the programming wheel, plastics aren’t prone to warping or deforming under normal loads. In addition, I’m not sure whether his CNC is capable of cutting metal. Plasma cutting isn’t terribly precise, and would require more finishing steps, where as plastic can be easily cut on any CNC machine that is capable of handling hardwoods. Agree on stiffening up the drum though

[@SpinThwomp]

So great that you’re doing another one! Excited to be along for the ride. Good luck

[@gizanked]

It IS Wednesday my dudes.

[@Incognito-rb4tz]

it's thursday my dudes

[@Ribo138]

Some might say a Wintergatan Wednesday

[@daniellima4391]

Back to the old times

[@billkeithchannel]

Somewhere @DMadHacks is smiling.

[@jksanrio]

The level of inspirations and joy MM1 brought to people shows it had so much value during its time on this world! No failure at all!

[@SAM_FaZe]

love how u put engineering and music together. keep up the great content 💯💯💯

[@flyn2274]

Martin, you are truly inspirational. One piece of feedback that I've been dying to give is that I think you can solve the problem of the precision of the diameter of the programming wheel with a simple design change. I believe what you care most about is the linear speed of the programming pins as they pass by the arm reading them. When you're gearing the edge of the wheel, like you are, the linear speed of the pin can be independent of the diameter. Essentially, you have a tube of programming tape like the music boxes. That axle can even float somewhat, and as long as you're gearing the drum in line with where you're reading it, you'd remove the diameter from the equation. Hope that helps.

[@PadreGamer]

Piece of Advice: You should manufacture a base where to place the programming wheel to hold it while in the world tour and concerts. That way it should be easily stored while waiting to switch the wheels during concerts and also it should provide an easy way to attach the programing profiles, that means the programming wheel should be able to rotate easily while it is placed in the base. I hope this comment helps.

[@psy0rz]

i would say: have a dedicated programming wheel for each song and make them props on stage! light them with stage lamps for extra effect

[@TheQuicksilver115]

I freaking love this series, from how you talk through your design from issues faced previously to new design specifications to how the new design satisfies the constraints to the end goal of making awesome, expressive music! I am so excited for the world tour and the rest of the journey along the way!!

[@d0n0x]

Damn it, I was just turning off the phone to go to sleep and this video dropped. I guess I ain't sleepin

[@transendinghuman]

One concern i have about concentricity is the central hole where you put the bearing and axel. If minimal errors is the x- and y-axis make the outer circumference wobbly, that same problem should apply to that hole. Putting it on the rotary axel might make the oputer ring concentric, but if the inner hole isnt round enough the wheel will be unbalanced and intoduce vibrations into the whole system

[@skadoo63]

The “perfect” engineering solution would be a programmed digital instrument. The beauty of the first example was how good it worked even with the lack of design discipline. I miss that!

[@seedmole]

This is awesome. I've spent the last few months making my own MIDI sequencer in pure data, and it's amazing how all the fundamental issues are the same. Like how to solve the quantization vs resolution issue, and how to maintain the potential for variety while keeping things simple enough that they can actually be done properly. I like how the note profiles are able to separate the issue of defining the timing of notes from the resolution of the grid. Like a DASDR envelope rather than just an ADSR envelope.

[@avasam06]

I loved the handcrafted, wooden aesthetics of the original machines. But now that we had them for looks, it's time to get a working one ;)

[@LinearNetworking]

Make some music, I need some fresh Wintergatan toons, Martin! Keep heart, every step is a step forward!

[@Sonderax]

I like the idea of multiple wheels. What stops you (other then cost) making 5-6 wheels before a show to ensure 0 downtime

[@DeanMcKenzie]

Loving the series... just a small bit of feedback, you said that "using the technic pins was a really really bad idea..." Was a fantastic idea! It allowed your idea to come to life. Without which you wouldn't have even been able to iterate on it.

[@_JohnHardy]

Nice! I would love to see the programming pins attached to some fabric that can wrap around the wheel and click into place; reducing the labour/mistakes for programming songs and making them easier to store.

[@sliceofsparta8985]

Love the new design, Martin! Should handle the rigors of the road way better! Cheers, and never give up!

[@itsleotv8487]

This next music machine will be amazing! keep up the work.

[@gerrardhickson9471]

Just a word of caution regarding machining the indexing features on the outer edge using the rotary axis - Your application demands that each hole is exactly the same distance apart, but you're controlling the distance apart using the stepper motor and there are a lot of components between the holes and your stepper motors which means there is a high probability of error - consider a small angular discrepancy on the stepper motor multiplied by 40 times (the gear ratio) multiplied by the radius of the timing wheel. The rotary axis will solve the concentricity problem, but introduce a precision problem. A more accurate way to control the precision is to drive the stepper motor directly to the back of your MDF back plate (assuming you can make the drive mechanism concentric too). In most applications, this wouldn't be a big problem, but I see how much effort you go to, so you want to get it right! Secondly - machining in the timing pin holes will have the same issue - I suggest doing a test piece (full diameter, part width) to test the angular accuracy of the indexing mechanism. All the best - keep up the good work.

[@valentin_boes]

feels so good to have these weekly episodes again!

[@j.v.9936]

Awesome dude! I love youtube because of people like you. I love Collins series about digging a tunnel between his house and bunker, and now i love your series about everything that goes into designing and building this marble machine 3. looking forward to your next video.

[@TreyVaswal]

Dimensional stability! He said it! He said it! My soul cried out for Martin as he desperately wrestled with those poly sheets. But now he understands.

[@Lethgar_Smith]

Ive been following this ever since that first video went viral. I remember when you where working out of a storage trailer on an island.

[@aviphysics]

I think you would be happier with a row of locator pins or roll pins vs. just screws. Screws are great for clamping, but terrible for holding precise registration. Every time I have to deal with a precision device at work I ask for them to use locator pins and we always regret it when they don't.

[@JonathanKayne]

Another thing to consider: how precise do you actually need to be? A valid test or analysis would be to see how much play there can be before it affects musical timing in a noticeable way (I stress the noticeable part because you have already shown a tendency to shoot for perfection when you had already gone well past what any human being can detect) Remember, you are making a musical instrument, not a jet engine

[@Jonathan-ex3sl]

All the round parts need to be PERFECT! +/- 0.0000 mm

[@blacquejacqueshellaque6373]

The helical gear should also be quieter than a regular gear. I think you may need some stiffeners on the bars with the slots for the programming pieces. The way you have it drawn, there may be some flex in the centre of the bars which would throw the timing off between the middle notes and the outer ones. I hope you get what I am trying to explain. One final note, I would not try to make the timing too perfect, the charm of live music is that it isn't perfect, it makes it more human. Don't get me going on autotune.

[@charleshettrick2408]

Stiffeners for both tangential and rotational. The end wheels can twist from handling, maching, etc, skewing the bars. Aging, mishandling, cam forces, etc, can cause the bars to bend. Possibly 1 or 2 more inner wheels lightly cross braced to at least 1 outer wheel should give you reliability and durability for global performances.

[@bow-tiedengineer4453]

Honestly, I'm kinda worried that he will actually succeed in chasing down all of the stray sounds and inconsistencies of the marble machine, and in doing so destroy the magical feel of it.

[@80211Denver]

helical gear on both sides to reduce twist from side to side

[@dannyjepp985]

Just going to leave this comment here before the video about why the helical gear should and should not be centered on the programming wheel.

[@17thstellation]

@@bow-tiedengineer4453 A lot of people would already say that there's no "magic" or what-have-you in the marble machine's music because it's made by a machine and not by a human with their bare hands. Everyone has different ideas of what the essence of music is, and I personally don't understand the obsession with those accidental little Imperfections™ the artist has no control over, instead of the actual ideas that they spent so long deliberately perfecting in order to express what they wanted to express. To me at least, it feels very dismissive to the whole point of creativity and art.

[@perishark234]

i like how your video format has changed martin! these new streamlined ones are great!

[@vincev57]

I really hope you get somewhere with the new design. Looking forward to see you live!

[@egpaul]

I wish i could press the like button more than once. The maturity of your design philosophy is really coming through. Kudos :)

[@Toastmaster_5000]

My only concern with this new approach is that the pins are pretty large, and I would think they'd greatly limit how many notes can be played. This seems a little too obvious though, so I'm sure Martin has that figured out.

[@Yay295]

I think the idea is that each pin could actually play more than one note.

[@yvan2563]

That was addressed in a previous video, it basically gives him nearly limitless possibilities as far as notes and timing goes.

[@billkeithchannel]

4 notes played in a row would have 4 notches in one clip. This reduces song note programming by 4.

[@terranosuchus]

If the goal is precision with timing, I agree with others in that there are potential issues with this programming wheel. There's no support between the two spokes, so all that torque on one side will cause the programming wheel to twist and bend. A pretty straightforward solution would be to have supports running between the two spokes. You could also have an inner spoke in the center (or both for extra stability) Having inner supports will help with this, but another thing to consider is that the plastic programming parts will deform over time, especially on contact points. Would it be feasible for some of the programmable parts to be made of metal to minimize deformation? Aluminum 6061 maybe?

[@Soularchitector]

I believe you need gear on other side as well, otherwize programming wheel will be twisted along rotational axis and will broke eventually. But I'm not sure :)

[@ajbh2o]

I don't think it will break, it will twist. My concern is the music could be out of time. The instruments close to the gear will be on time the instruments far away will always be late. Martin, you should do FEA on the wheel to see how much twist there is given the force from the motor on the gear. A sure fire way to cut the twist in half is to put the gear in the center. An additional possibility would be to put a tube about 1/3 the diameter into the center to act as the bottom of the torsion box.

[@CScottAnanian]

It's suggested elsewhere on this thread that the axle be considered part of the wheel and fixed to the wheel on both sides, with the bearing living on the housing. That would also help eliminate twist, since the axle itself would then be part of the stiffening mechanism, instead of transmitting the torque *solely* through the programming plates.

[@ajbh2o]

@@CScottAnanian If the design were changed such that the shaft was fixed it would help a little. But not much, the shaft diameter is too small and the lever arm created by the diameter of the outer cylinder will see the steel shaft like a piece of string. A better idea would be to remove the shaft all together, use a piece of large diameter plastic drain pipe. Then for the side support you would use cam rollers and the pipe or shaft could smoothly ride on them while being driven by the gear. Minimum would be 2 cam rollers per side to create a v shaped nest for the tube to rest on. Of course, turn 0.2 mm off the tube to make it concentric with the rest of the cylinder.

[@afkerex7113]

This sort of engineering walkthrough for something very specific in a passion project will always scratch an itch I have a hard time reaching. Good job!

[@Sqwince23]

HEADS UP: When you put your drum on the AVID CNC it looks like it's going to interfere with your motor and gearbox. Also - the operation to machine the holes in the programming plates may not work as well as you think because there isn't a whole lot of radial support across that plate. Meaning, there will be a lot of chatter and movement in the programming plate when the cutter tries to cut into those plates.. make sure the plates are stiff across the span to avoid this issue.

[@xuko6792]

I personally do not see the need to machine plates on the wheel. Since concentricity will be provided by machining the wheels itself on their axis, and outer shape of the wheel is a... hexacontatetragon 😅 you need only to align flat face to a flat face. So the requirement for the programming plate - to be perfectly flat (and all the plates have to be the exact same thickness). Moreover, holes in the plates provide failure point - they will be exposed to lots of fastening and unfastening, wearing down the perfect fit. Having to reinstall the whole programming drum into cnc router to machine just one plate because of just one loose hole is a bit too much.

[@aussiegarbo752]

Yeseeessss my excitement is uncontrollable! Go Martin!

[@sambarrett4590]

Don't call me a hater - I love this project and have been following for years. Love the new direction with first principal design. However, this video was about concepts which have not (yet) been tested. The marble dropper test series was proving the design, before designing the final implementation. This video swaps those steps, and I sense it will be a great risk to the project to have not proven this direction with testing first.

[@ThePhilbox]

These videos are so great for showing how exciting and fulfilling the engineering process can be. Its very enjoyable to watch.

[@Paulvarrgas]

All of this is really promissing, I can't wait to see the result! I was so affraid to see you stop and surrender after all the trouble of the MMX, but you seems on the good road now and litlle by little, with far better aproche, you reach your goal. I'm so happy, for you, and for us too Good luck, I hope you will be able to tick all these boxes

[@lostnthsauze1440]

I wish you wouldn't call your last 2 marble machines failures. I think I watched the first one when I was 13? I immediately was in love with what you did and have since watched it multiple times. You made something beautiful, and I'm excited to see more of your work

[@AlexxSymmm]

The programming pins/holes around the centre of the wheel may deflect inwards/outwards. I would consider if another support in the middle in addition to the ones at the ends is necessary. Great stuff Martin! :)

[@alltheeasynamesweregone]

This channel has blue balled me for years. I just want to see Martin on stage playing the Marble Machine song on a Marble Machine

[@SVSQZ]

The first song in the album should be called “blue marbles” 🔵🔵

[@constantinosschinas4503]

​@@SVSQZ lol lol lol

[@sararebecapalacios]

Honestly same.

[@NateTmi]

the thing I like about plastic is, that your system seems easy to replace parts, if they break or need updates.

[@Chris-ru9yy]

have a serious think about how the plastic for the notes will fatigue when the wheels are being set up(and also how much force will be on them when reading the notes). it could introduce timing issues

[@yvan2563]

There's so many types of plastics, we can't assume anything for now. He needs to tells us exactly which plastic he's planning on using so that people with the technical knowledge can inform him and the rest of us about his choice and any alternative he should be using instead - if that initial choice is the wrong one, of course.

[@foldionepapyrus3441]

@@yvan2563 Afraid I'm with Chris on this - the clip design as it stands is for me flawed in its very nature - doesn't really matter what materials are used there, its not going to be enough - Martin has put a completely insane amount of effort chasing gates with timing tighter than he can really measure so any programming pin registered only by a deliberately flexible click action that doesn't need bearing puller to remove and hammers to install will not be able to meet that spec unless the read head of the machine imparts damn nearly zero load - the click action he is demonstrating is looking like a travel of well over 1mm for the click action and in the read direction - which is really poor design - the clips should be designed to flex perpendicular to the expected read head load so they can be as stiff as possible in the important direction. They also look to be really really easy to do by hand - so just to make it possible to click in and out so easily it can be expected to have that sort of ballpark for slop under load - though without knowing the expected rotation speed how much that will throw the timing is impossible to calculate. If he adds a fixed registration pin and the click is just to hold the pin to wheel while the timing is handled by the precise and inflexible pin he is in with a shot of reaching timing he can accept. But even that doesn't solve the fatigue issue on its own, and picking the right type of plastic to have good fatigue life that you are then battling to get sufficient stiffness for timing (even with a fixed pin its got to be relatively stiff), low friction and hard wearing enough to last at the same time along with many other considerations for a world tour machine - how does it handle baking under stage lights? Is this a plastic that will creep on its own more readily? Its not a problem that can't be solved, but getting the one plastic that does all the things at once well enough may not even be possible (will depend on factors we have not seen yet), and moving into double shot multi-material type stuff would be interesting but expensive.

[@SkigBiggler]

The new pegs are simple and easy to replace, I wouldn’t worry about them wearing out. Even when on tour, he can make a bucket of each type, and have plenty of spares. Plastic also isn’t prone to fatigue in quite the same way as metal, it tends to stay in one piece for longer than metal under fatiguing conditions, although it is certainly still vulnerable to repeated loading cycles. The main issue with regards to wear on the new notes is during insertion and removal, and I think that’s addressable without a design change.

[@VagabondTE]

Certain Plastics like Delrin or Polyethylene can be more durable than wood or metal. I recycle household plastics by hand and I have an example piece that I've been beating with a hammer for years. It's never shown a dent. You could lift an entire person with a single grocery bag. Imagine how tough they would be if they were half an inch thick.

[@CScottAnanian]

When he says "engineering plastic" martin almost certainly means delrin, which is great stuff (though pricey). I think wear over time can be handled by treating the drum as a consumable and making multiples, since they are designed to be swapped in and out.

[@halfmanhalfnerd4986]

Have to agree with the crowd. Worried about twisting, use inner rings to stiffen and aid with keeping bars in place, at distance. You are bound to have someone drop the thing, or something apply enough pressure on the cylinder during transport to distort the cylinder (Break tabs maybe as well).

[@ryanvoots9827]

When you make this I hope you stream all the cnc runs. It'd be cool to watch it go for a few hours

[@ChrisB...]

I love that you're using the wheel as it's own jig, the whole build just seems right. On a side note, CAD is getting super impressive these days!

[@ForestGramps]

My main question is will there be enough space to play enough notes on a single loop? I am probably not understanding the profiles correctly, but with the current design how many notes can be played before repetition? Super excited to see progress!

[@antoine_nedelec]

I was thinking exactly the same ! There must be something we don't see ;)

[@ForestGramps]

@@antoine_nedelec I hope so, otherwise I think we'll be hearing a lot of 4-chord progressions lol

[@ninjafruitchilled]

Pretty sure the profiles can be cut to play multiple notes per profile. Or single notes at any musical position within their reach.

[@level12lobster8]

The wheel is now the length of time while the pins represent ta section of that time. This is why the amount of machined slots along the wheel was reduced by x10. It doesn't need a channel for every type of note (1/4, 1/8, 1/16, 1/4 off timing, 1/8 off timing, 1/16 off timing, etc). Just think of each programmable pin as a single bar in a measure. The dips cut in to the pins will be the actual notes and contain the timing. If you need further explanation, each pin represents the time between each tick on a metronome.

[@ForestGramps]

@@level12lobster8 ahh okay! Very well explained. Tysm. I should probably re-watch the video about them.

[@onlooker251]

Thanks

[@NSGolova]

Martin: The First Marble Machine was sooo bad! Me: rewatching that video now and then for 7 years ❤

[@verebellus]

This is such a cool project, i loved the v1, loved the mmx (it looked incredible) and i love this mm3

[@dak1st]

Martin: *wants to reduce the manual processes to nothing* Also Martin: *designs a whole CNC lathe to be built manually, to avoid another manual process*

[@Dero88]

Happy to say that i'm starting to find interest in this again. I loved your work on mmx. But I'm not gonna lie, it felt kind of a chore in the end to stay updated with live streams, dual channels and all. This is feeling really good and it's short and to the point. Interesting and well explained. Thanks for all your hard work! Och kul att du är hemma igen!

[@sionn1]

Am I the only one who doesn't like the new design at all and thinks it will never be finished? I just wish he finished the last machine which was like 90% done and looked way cooler.

[@highseassailor]

Don't forget safety, considering the torque in your design. To further fine tune the accuracy, balance EVERYTHING that spins, like a tire on a rim. Thanks for sharing your progress, the design looks awesome! It's especially great to see your perseverance on display.

[@johanbertelsen8412]

This comment is for everyone who waved at their screen when Martin waved goodbye to plywood

[@ritamoore8346]

You are awesome. Making a new instrument is amazing. I also like listening to the music it produces. AMAZING!!!

[@callmeniles6060]

Gotta be honest im very skeptical of this design

[@Jonathan-ex3sl]

How come?

[@justmasses8751]

I'm really glad to be watching your videos somewhat regularly again, its always a pleasure to watch you delve into the madness as you create history. Honestly missed this after you labelled the the MM1 and MMX as failures, despite them being important steps towards your pursuit of perfection. Since the first attempts its been a pleasure to see the growth in design, so long as there are more vids to come I will forever be watching this dream become reality.

[@krelsen7]

i'm still a little worried about the new programming pins, it seems like you will need a lot of different types of pins to allow flexibility

[@CScottAnanian]

I think that's considered a feature? Ie, any particular musical rhythmic challenge can be handled by "just" making a custom pin for the job. OTOH i agree that getting the programming correct when songs are swapped mid-show could be quite tricky if you've got a dozen different pin variants in boxes you have to pull from. My personal feeling is that it would be worth it to manufacture multiple drums so that you never have to reprogram during a show, just swap out one drum for another. Reprogramming could be done only when not under time pressure, ie when breaking the machine down to move it, or during the creative composition process.

[@billkeithchannel]

@@CScottAnanian It is that swapping that has me concerned. We saw how solid the MMX had to be in the frame so it didn't shake. It will be interesting to see how he designs it to be solid but still easily accessed for removal.

[@auronoxe]

The new pins only help for notes of the same pitch, right? Can be complicated accross different notes to use these pins… Maybe this should be simulated up front for a song, before building the wheel like this. The number of holes went down by a factor of 10, as far as I understood. If you would have to cover every pattern (of ten holes that are now just 1 single pin) from 1111111111, 1111111110, …. To 0000000000 - this would need 2^10 different pins = 1024 different pins 😬

[@krelsen7]

@@auronoxe i am by no means an engineer, but that's exactly my concern. the song writing process would also be more tedious if he wants to use the machine, because if a "pattern" does not exist on one of the pins, he would have to manufacture it to even try out how it sounds. if he intends to make the songs digitally first it may not be as big of an issue, but overall the system is less dynamic, and while he has removed a lot of parts, this will certainly add parts and complexity to the programming of the machine. with the magnets, while they had issues, everything was dynamic, with this system he would be in trouble if a specific pin broke on tour and he didn't have a spare. maybe he has considered all of that, it just didn't seem like it in the video.

[@FredrikOstrozanszky]

If you had some kind of durable but flexible paper sheet with holes in on a spool you would have even more flexibility with programming notes. You cut as big holes as you need. Then you use the programmingwheel part to bring the sheet taught and wind it up on another spool. Probably less complex to manufacture and easier to change songs. Paper sheets might not be as durable but are easy to make new ones.

[@timangelico6813]

I may have missed something, but how are you going to make sure the wheel isn't going to collapse across the middle? Even the stiffest plastics will sag over a metre, and if you're chasing the milliseconds, I think the wheel might let you down.

[@Jonathan-ex3sl]

I would bet money Martin ads lots of stiffness once he starts testing

[@ejnordberg]

I'm going to be that guy:. Go with electronic marble drop actuaters driven by MIDI. This would provide a better control, timing adjustment, and creativity. It would also make song transitions in a live environment faster and seamless.

[@robertdickinson4253]

Since your new note posts cover a much larger arc of your when than the previous version, will you find that you cannot repeat a given note as rapidly as yiou could previously?

[@e.fauser2574]

love it to see an update. nice format now btw. can't wait to see it in action ( even tho the engineering part will be finished by then.)

[@pseudo_nym]

I do understand all your improvements of precision, but what about the _resolution?_ *How many notes (marbles) can be played per programming row and 360° wheel turn?*

[@timothybates8501]

That is exactly what I kept thinking looking at the new design. I fail to see how this can play tight music (tight as in rapidly repeated notes)

[@zediax]

Keep up the good work ! You're doing very good at explaining things and it's really fun to learn with you :D

[@jeremypearson9019]

There were tons of people that criticized Martin about abandoning MMX, but I knew that it was a good idea to start over and be more methodical and modular in the design process. He needs a machine that is stable, reliable, and portable so he can reach his dream of doing a world tour. And, he could even build a few of them so that they can be shipped ahead of time so that they can arrive all unpacked and tested before the show.

[@Pixelarter]

Wouldn't be better to ger rid of the shaft completely, and make the outer wheels roll over a spaced pair of bearings on each side? It would make the assembly simpler, lighter to transport, and easier to change programming drums since it is kind of self centering. It just needs some movable mechanisms with additional bearings at the top to lock the drum in place. Also, since Martin will precision machine the wholes in place anyway, wouldn't be easier to use 3 or 4 sheet sections bent on the outer skin instead of hundreds of screwed ribs? It could also add to torsional stiffness, simplify a lot the manufacturing process, and save a ton of bolts and screws. By turning the outer skin on the CNC, perfect concentricity could be achieved.

[@j0nthegreat]

"i don't trust CNC to make a 2D circle, but i trust it to make 1920 slots and a giant circular double helical gear on 3 axes"

[@deemstyle]

Yup!! The irony... Ha!

[@nathangeorge4017]

I think the difference isn't the amount of trust, but rather that that large 2D circle MUST be perfectly circular. The other parts don't have as strict of requirements on their shape. (Meoiswa has correct answer below)

[@fifty-plus]

It's not like there is a way to test and tune that either.

[@nixpix19]

Think you missed the statement at 12:35

[@gredangeo]

It doesn't help that the CNC machine he has is very cheap. It's only meant for light duty work, and ornamental stuff. The precision that Wintergatan is asking for apparently, is something meant for aircraft parts, and technically should be using a $250,000+ gantry machine. But uh..good luck with affording that.

[@WeslomPo]

Super glad that you are back with your videos. I so missed them T_T. Mmx1 and mmx2 not a failure. They are success on their own.

[@thenashus4]

As much as I enjoy the journey and progression of these videos, I've been thinking about older videos where the charm of the design was a major goal for the MMX. I remember Marting specifically gunning for the plywood and steel aesthetic and I was fully on board with that, but now I'm worried that charm is being disregarded and function is now the sole focus; which from an engineering standpoint makes sense, but will people resonate with something industrial and machined as much as they would with something home-made and hand-built? I believe that's what people loved about the first machine, it was something one person built by hand in their garage. Not to say this strays too far from that, but without the wood and all these precisely machined and bought parts, it's starting to lose that personable feel. Still a fan, love the videos, these are just my current concerns.

[@jp-ny2pd]

Might be worth getting a servo for the rotary axis in order to get the precision. Treat it like a fifth axis on the CNC. I don't think I would trust a regular stepper and counting steps on the rotary axis, specially when you get to the locating pins and pin grid. Might also be worth looking at putting a torsion tube between the two spokes. Something like a 4" or 10cm pipe that you can glue or chemically weld to the spokes would help a lot with any potential twisting. It also wouldn't need to be a precision piece other then the very ends where you are interfacing to the spokes.

[@Jonathan-ex3sl]

That's a great point, especially with how precise martin wants everything. Seems like it would be beneficial to have some kind of encoder feedback as close to the drum as possible.

[@thomasbecker9676]

I'm still not sure Martin understands what precision actually means in an engineering sense.

[@voxel9470]

What do you mean? I'm curious

[@thomasbecker9676]

@@voxel9470 In engineering, precision and accuracy are similar but different. Precision is more about repeatability, but a lack of accuracy can skew all efforts. Also, increasing precision is usually diminishing return.

[@voxel9470]

@@thomasbecker9676 thanks.

[@theviewerofart]

As someone who was never going to watch this on tour and just wanted to see a really cool machine that can play music, The MMX is the pinnacle to me. It may not have been tour-worthy, but it was both beautiful and functional in a way that captured the charm of the original marble machine in style. This new machine looks promising, and you should be proud of what you're doing with it! But for me, it's not what I came to this channel for. Good luck Martin!

[@snabbott]

I'm loving these videos! ❤ My one suggestion is that you make sure you are testing the integration of different components as early as possible / practical. That way, if you come across an issue that requires you to redesign a component, you don't end up redesigning everything that interacts with it.

[@dirkv.9013]

👏👏👏 LOVE. THIS. Designing the wheel using a very stable material, reducing complexity and part count, and machining it using the rotary attachment to get good concentricity - you've upped your engineering game so much. Even though the past machines didn't work as well as hoped, so much learning has happened from them, that they have been very valuable experiences - for you and for all of us who are following along. Really impressive, Martin!

[@arnauds2222]

Lots of great suggestions here to improve rigidity. As an aerospace engineering student I can give you this analogy: you made a wing with no spars and ribs. You could very easily add some lightweight reinforcements.

[@WarWolfX1]

I think you need to add cross supports inside the wheel, it looks like there might be some twisting flex across the wheel. I have a feeling that the tightness of the notes will be affected from the far right note to the far left note. maybe put the gear in the center of the wheel? imagine the gear putting load on the gear side of the wheel and then there are 8 notes on the left side of the wheel twisting the wheel. the whole wheel will want to twist like a helical.

[@stevezozuk9622]

Wow you have managed to tighten everything with wonderful new decisions on design from function. It's beautiful, simple is often the hardest thing to achieve. Congratulations on sticking with your goals for this wonderful machine. I always figured that you would. Cheers my friend

[@drakebletl7754]

Martin you will always be an inspiration to me. Although I was very disappointed when you gave up on the mmx. Because of you I became a precision machinist. Nothing's impossible with hard work