Fixing An Annoying Problem From The Marble Machine X

Support Wintergatan:
Patreon ► / wintergatan
KZbin membership ► bit.ly/4cQVM7C
Marble Machine Engineering Discord Server:
/ discord
Video edited By Martin and Hannes from the Trainerds KZbin Channel:
/ trainerds
-
PATREON ► / wintergatan
KZbin MEMBERSHIP ► bit.ly/4cQVM7C
WINTERGATAN RECORDS ► www.wintergatan...
SPOTIFY ► bit.ly/2oKxXWd
ITUNES ► apple.co/2ntWNsZ
ENGINEERING DISCORD ► / discord
COMMUNITY DISCORD ► / discord
-
Step 3 FTW
Max Sansalone: Funky Drumming:
• Max Sansalone : Funky ...
Edited by Hannes And Martin, check out Hannes Trainerds KZbin channel here:
/ trainerds
-
MUSIC DOWNLOADS ► wintergatan.ba...
WINTERGATAN RECORDS ► www.wintergatan...
SPOTIFY ► bit.ly/2oKxXWd
ITUNES ► apple.co/2ntWNsZ
MERCH ► teespring.com/...
COMMUNITY DISCORD ► / discord
SECOND OFFICIAL CHANNEL:► / wintergatan2021
-

[@aleksandrkinyaev6703]

LONG SHAFTS WARNING! Previous week I wrote post about shafts. In short - 9 meter shaft would twist. I see big red gears on shaft ends. This shaft would twist a few degrees, which would turn into BIG difference in timing. Please measure your shaft twist under load or you'll get into serious problems! Part 1 - timing is not done yet, seriously, please don't ignore!

[@aleksandrkinyaev6703]

Also, interesting thought. Floor wouldn't be perfectly flat, so your music drums would not be placed at exactly the same angle around their axis of rotation. I mean, whole structure of the machine would have some twist in it too. So, it is another source of timing problems.

[@SoranoGuardias]

More shaft matings and stronger profiles?

[@christianvanderstap6257]

He did state that it will not be a single shaft. Not sure if that makes it better or worse. For sure will consume more power at each transition.

[@IchDuForeverExplorering]

connect all programming wheels with each other

[@IchDuForeverExplorering]

@@aleksandrkinyaev6703 i dont see how a not perfectly flat floor would effect timing the frame of the machine when built up in a location will be put up leveled, you just need to adjust every contact point to the ground

[@jaceg810]

The new muting on the profiles sounds like it is adding: - a delay module, - an extra spring, and a lot of precision requirements when it comes to spring tension - a new cable in the interest of simplicity in design Assuming you mute an entire group of instruments, could one not drop the entire programming wheel a peg, or raise the readers, that way they are disconnected, they will jump to neutral (and thus release whatever marble) and you will only require one, maybe 2 raising/dropping modules instead of one for every marble dropper on the instrument group. (this would also remove the need for the linkage and the original disengaging system, thus even further simplifying parts) Hope this is constructive, keep up the good work and good luck.

[@mrboberson7424]

I like the idea of dropping the wheel away from the readers. Simple solution, doesn't need to be super exact, and less parts.

[@FroxyCz]

And if dropping the whole wheel is not an option you can just lift the bank of readers by using some kind of a cam mechanism or something like that. Getting rid of the original disengaging system on each of the readers is a major bonus.

[@Lampe2020]

The problem with that is a) how do you ensure a marble isn't played too early when the group is muted while at least one reader in it is on a profile? b) Is it really less parts, to drop the programming wheel or raise the reader group? The wheel has to be able to get powered and must be precise and the readers have to be precise as well, making them movable makes them less precise.

[@FroxyCz]

@@Lampe2020 Moving the whole reader group by moving the shaft on which they are installed can be more than precise enough. A cam-like mechanism is extremely precise. Or you can just CNC a couple of grooves in which the shaft would be raised and lowered. Depends on the implementation. I really dont think precission would be an issue. Moving the wheen can be more problematic because it is powered that is true.

[@alexandrachernysh7]

Yes, that's exactly what I was thinking as I watched the video, raising the readers makes the whole assembly a lot more simple and reliable

[@RG-At-Large]

I recommend keeping track of torsion in the long shafts. As load increases and the shaft twists, the relative position of one end vs. the other will increase timing error between the programming drums.

[@simonbrooke4065]

Aye, this strikes me too. Transmission wind up in old Land Rovers introduces very nasty harmonics, which in something which is trying to produce tight music would be undesirable. The shafts (and especially the couplings between shaft sections, since these shafts are designed to be separable into shorter lengths for transport) will need to be very torsionally stiff. I also think it's likely that getting the alignment of support bearings for the shafts, in a modular demountable rig, sufficiently tight is going to take some thoughtful engineering. I think the music will be much tighter, and the machine suffer much lower friction losses, if the power transmission shafts are perfectly straight and cannot flex. This suggests to me a large diameter, probably hollow shaft, perhaps in carbon fibre(?), with support bearings on either side of the wheel which drives the shaft, and probably at other suitable stations along the length.

[@simonbrooke4065]

In fact, I definitely think that before you tick off 'plays tight music' as fixed, you need to ensure that you can transmit 250 watts of power down a shaft half the width of your stage set, with joints where you think you'll need them, under varying load, without introducing torsional harmonics which throw the timing out to an extent that's detectable. I mean, I'm fairly confident you *can* do this, but I'm not confident you can do it with small diameter mild steel shafts and any old joints.

[@VinzC]

And also keeping track of loss of energy through friction, deformation of all components, that will in the end affect music precision.

[@Milagoso]

bigger issue: 150 watts is not possible. the energy input that Martin talks about is for biking, where the bike is designed for maximum efficiency and takes our biology into account so maximum force can be inputted per cycle. small example: on a spinbike, you have straps to keep your feet there, so you can pull with one leg pushing with the other, making the peakforce per muscle lower for the same wattage if you can't pull. his energy input calculation is way off...

[@ANoBaka]

​@@Milagosono, the bigger issue certainly is deformation. Power input can be solved by just welding a bike frame onto the stage or having a second person stomp on it. Deformation will introduce serious timing issues that will drive him mad to try to fix.

[@dragoncoder047]

5:45 I bet you’re going to have a lot of finicky mechanisms if you go with this two-spring solution- it might be worthwhile to think of an alternative plan B.

[@05thepoge]

Use the example from the previous mechanical instruments and slide the programming wheel laterally so the profiles lose contact with the registers when muted. This wouldn't make the drive mechanism any more complicated as the entire drive shaft could shift with the programming wheel. Build in a space of about 3cm between each of the registers that the profiles would shift into when muted. The register would still keep tension if muted while playing and the the profiles could be designed with an angle at the leading edge to pick up the register when unmuting.

[@ChuckSploder]

I was thinking that instead of the new spring design, he should attach the muting cable to the hook handle very close to the hinge, maybe on a pulley, that way it'd pull the hook up and out of the way and pull the reader away from the programming wheel at the same time.

[@-NGC-6302-]

Yeah, I'm getting "sounds cool but also like it might not work well" vibes again. At least verification will show who's right!

[@singulosta]

I also thought about it. My solution would be to instead of moving the reader separately, could we not also use a ratchet or magnet mechanism (whose activation is optimally also controlled via the original muting cable) which ensures that the reader remains in the upper position after it has been lifted by a programming pin. This shouldn't cause any problems as far as I've thought it through with the mentioned edge cases, as it will either drop and engage on unmute or be supported by the current programming pin and engage after the note ends.

[@delanezdelanez5253]

Just start working on plan B, the springs will not work. This will be another proving it could work instead of finding good engineering solution.

[@ixenvire]

I Genuinely liked the mechanical noisiness of the marble machines of the past, it had its own special sound and tempo and was beautifully mechanical, but I've not ever been in the room with it, so I dont know But either way, watching your vision slowly come true as you do these experiments brightens my week every time, i cant wait to see how it turns out!!!!

[@don.3s]

Me too, it was part of the charm just like the mechanical look and inventive ways the marbles were moved around. The new machine is too big, and the hyper focus on effectiveness ("form follows function") has reduced the personality to 0

[@celeron55]

Martin had to do a lot of work to edit the noise out from the videos. What you have heard is not what Martin has heard.

[@ProactiveYellow]

I worry about the spring vs spring combination. That seems a very tight tolerance in order to make it work in this edge case. At the very least, you probably need to be able to individually tune the mute springs based on climate, humidity, and how level the stage is (all factors that will affect the lockup and release). The problem is it would be hard to tune specifically for an edge case that shouldn't be happining that often in the first place. I like the idea of this solution, but i feel like it's too "theoretical" and not enough "physics"

[@jacobfreeman]

The solution for the other edge cases are so clearly good and will be reliable that relying on nebulous spring action for the last edge case feels wrong. There should be a better mechanical solution that will always work for sure.

[@williamflinchbaugh6478]

Definitely agree, it seems like a finicky system that might not be very robust. Honestly, to me it doesn't seem like there's too much wrong with the original design if he just adds a little padding or something.

[@gurt-tkoft]

The only way I can think of around this is to have a two-stage muting action. That requires the operator to make two inputs though.

[@neorogertube]

I totally agree. I think the original design was better and more reliable. The new one I just like the idea of fully separating the programming pin from the lever, so it does not wear, but using springs I think is a very very bad idea. I am wondering if instead of that, it is not possible to make the lever oscilate around a bistable system, that just moves the lever (I am calling it "lever" but I mean the thing that touches the programming pin) center of rotation a bit up, so it does not touches the programming pin? we would need to think on the edge cases, bcs if we do just this we will drop the running ball if you mute in the middle of a lever action, but I think it needs to be a way to do this easier than with springs. Otherwise, if no solution is found, I think we should tackle the fact that the lever does a lot of noise while muted, which at first glance seems wrong to me and maybe could be tackle and keep the old design.

[@joshuapawlak4605]

The issue is that it’s a dynamic motion that you’re relying on, and those are very difficult to calibrate. Fixed states like in the old design will always be more reliable.

[@coco805]

You've turned your machine into a monster stair master! Your legs are gonna get super buff, 200W sustained output is a serious workout.

[@Schlups]

Very sweaty

[@salty.nall42]

maybe replace pedals with bicycle + gearing to make it easier. Totally agree that 200W sustained is not an easy workout (even on a bike).

[@nickfosterxx]

Audience will be queueing up to help out...

[@stuka78]

@@salty.nall42e-bike motor even better

[@lukehunter2833]

Indeed - I also think Martin needs to consider that 200W is possible for a cyclist using equipment that has been tuned over a very long period of time and is hyper efficient for inputting power compared to standing on a lever. I think this should be viewed as a bit of a problem that needs either a mechanical solution, or sufficient headroom.

[@werner-de-jong]

Hi Martin, I think adding the spring + tension, as showed at 6:17 in the video will create many failure spots. Relying on momentum sounds nice but tweaking the exact momentum, tension and the hooks not grapping is a nightmare waiting to happen. (In my humble opinion) Perhaps an easier design is to change the angle of the lower protruding (reader) part , lifting it above the profile all together. OR retract the protruding part into the joint, making it smaller and missing the profile (put tension on that using a spring so that the part wants to stay longer and hit the profile. Making the reader less long using the pulling mechanism sounds mechanically simpler and less chance of the right momentum.

[@omeganik]

Agreed. I'm not sold on the springs either.

[@sebastiancordero27]

You know, I actually liked that mechanical sound mmx did. I think its a big part of a mechanical device to hear sll the sounds it makes.

[@philippthaler702]

Please keep in mind that a cyclist can use 2 legs to produce 200W while with one pedal you just have 1 leg - also you'd be playing music along paddling. I suggest also to verify how much power you can produce comfortably. All the best from Austria!

[@jakobrosenqvist4691]

You only use one leg at a time on a bike to, but 200W sustained while trying to control the machine and play music at the same time probably isn't happening anyway.

[@Bl4astedgnu]

Bike wattage is still calculated using both legs. If you have one sided power meter it literally just doubles the output it see to simulate dual sided. 150 watts for an hour with one leg would be the towards the

[@jakobrosenqvist4691]

@@Bl4astedgnu His design uses both legs too, it's just one leg at a time. It might not be as efficent as a bike but the biomechanics involved are very similar.

[@brianorca]

Peak bikers clip in to the pedals, so their legs can pull up as well as push down.

[@CornholioTP2]

@@brianorca The pulling action on a bike is very weak and increased power output because of it is a myth. Still its nice to be clipped in, mainly because the feet stay where they are supposed to be

[@BoxOfZoodles]

I've learned this lesson during my years as an engineer: keep sight of your goals. You need to be very clear on making sure timing is not jeopardized in any step. Just because you focused on the gates and got them just right doesn't mean timing is solved for good. Just don't forget your old requirements because every change you make will affect

[@thomasbecker9676]

He has no established goals or requirements. Any goalpost that does get established is move the very next week. All evidence points to Martin wanting to make KZbin content, and therefore revenue; not a music machine.

[@Lavafish2525]

​@@thomasbecker9676 I ain't going to complain cause the content is currently pretty good 👍

[@thomasbecker9676]

@@Lavafish2525 I wonder what your bar for good content is.

[@Lavafish2525]

@@thomasbecker9676 What stake do you have in this? unless you paid money to Martin, I think you should keep this to yourself.

[@Lavafish2525]

@@thomasbecker9676 Why do you have 100 comment's on this channel if you don't enjoy? This clearly means way too much to you. Do you not realize why people ignore your complaining?

[@justinsemple7454]

All the little mechanical clicks, ticks and sounds of marbles rolling are part of why the original Marble Machine song was so good. Good art, even great art, is often defined by its imperfections. The Mona Lisa wouldn't be a masterwork if it was a photograph.

[@Platanov]

I couldn't agree more. Additionally, the sound played for comparison in this video is without padding/damping. I think the soft click-clacking of the machine when it's properly padded will not be a problem against resonating instruments, especially from an audience point of view.

[@lunaeclipse4304]

i will say that part of the appeal of the first two marble machines is the movement of everything, and how it was all packed together. the design processes that you're using are incredibly important and it will absolutely result in an easier and more reliable machine, but im worried that the end product won't be as inticing as the previous two. i look forward to seeing the result!

[@chnebel2740]

i fully agree. the previous ones had the feel of a one man band. something you could encounter on a street corner like the old time barrel organ. all the flaws and mechanical noises just added to its authenticity. at the end of the day he has to do it the way he invisions it but for me it lost all the magic the first ones had.

[@simonbrooke4065]

You should probably look at off the shelf bicycle gearboxes, which are designed for exactly the sort of power input you have (i.e. the power a person can deliver when pedalling). A particular design which might be suitable is the Enviolo, because it is continuously variable so should be able to match tempo very easily; but the Rohloff is also worth considering, because it is very efficient.

[@nickfosterxx]

great idea.

[@DemsW]

Less parts, easily replacable, very smart

[@gedoensful]

Pinion ftw ;)

[@05thepoge]

Using a bicycle gearbox also gives many more tempo options. A car's gearbox typically only has 5 speeds (and reverse) which would limit you to 5 tempo choices.

[@FocusDisorder]

Plus they run on bowden cables, which Martin is already using elsewhere in the build - fewer unique parts for the parts list means easier and cheaper build and maintenance!

[@John_Weiss]

8:55 Martin, I still think that you should have a double-Huygen-drive, and that you should have the weights hanging someplace self-contained and separate, not underneath the power-module. Having a double-Huygen lets you drive both the music and the marble-return without needing to _constantly_ input power, which will be _incredibly exhausting!_ Decoupling where the weights fall eliminates a limitation on your drop-distance, which currently is the height of the power-module. That's not much space for power-storage. Moving the weights to a "tower of power" will add extra drive-belts, it's true, but it lets you put the weights someplace where you and the audience can see them, where they're isolated in the event of a catastrophic failure, and where you can lift them to a very large height, thereby storing a lot of power for both driving the programming wheel and driving the marble-return conveyor belts.

[@CornDogShaun]

You were doing so good. you were removing complications. you were removing points of failure. now, here we are, adding them back in.

[@MaxWithTheSax]

Relying on momentum is usually not a good idea. This design depends a lot on friction and spring force which are not necessarily going to be constant over the lifetime of the machine and under different conditions. I prefer the old design because it is perfectly reliable and probably silent enough with some padding.

[@peterwinter8028]

I really enjoyed the mechanical background noise of the first MM, it was giving it a lot of character and made it (would make each iteration of a MM) unique. I would compare it to the sliding fingers on an acoustic guitar or the breathing of a singer in a recording. Its not intentional and not making the song but its a hard to erase necessity that gives a song character. Love your work 😊

[@1706rasmus]

Hi Martin. Please remember to go back and verify that timing and sound is still on spec when you start fiddling with features and assemble. A classic error in any engineering implementation is that you did your verification and it all worked, but later you add some stuff and it has unforeseen consequences. In software implementation it is called regression testing (verification) All the best, love following along 😊

[@GameHut]

Maybe I’m misunderstanding the problem with muting, but it seems to me that lifting the bar that the reader pivots on away from the wheel would allow you to remove all the breaks in every linkage. So you can hard wire from the reader to the marble drop and just lift the reader vertically away from the drum. The cables give you the movement needed to do this. Right?

[@Jon1010]

Right?

[@CodingSecrets]

Right!

[@skipnyip]

Part 1: The dual spring design seems like a potential failure point as I worry about relying on the momentum of the block isn't going to be consistent enough for what you're looking for.

[@petecomps7260]

As sketched, the "gear box" shows a discrete, selectable quantity of gear ratios. Instead, use a Continuously Variable Transmission. This will not only provide fully variable tempos between the minimum and maximum, but will allow you to change tempo during each song by simply moving a selector lever.

[@Akareyon]

CVT gang unite!

[@specodhec341]

Yeah, a CVT gearbox is a good bet

[@florianzimmermann2730]

this would also make it possible to speed up the music at certain parts or slow it down by moving the ratio

[@chrismofer]

but how stable is a CVT? he cares a lot about tightness

[@randommcranderson5155]

Help him come up with a super tight timing CVT design.

[@rodedogad]

There's a textbook mechanics/stiffness problem where the student calculates how many rotations at the top of an oil drill rod does it take until the tip of the drill starts to move. Stiffness of your drive shafts given their length is going to matter at some point to avoid rotational deflection that would make the programming wheels out of phase.

[@jeremypearson9019]

This problem is going to come back to bite Martin. Just wait 12 months and he's going to notice that it's a big problem

[@EatMyYeeties]

Meh, the difference is minimal. An oil drill shaft is potentially hundreds of meters long, this is 9m split in half, so 4.5m. I already crunched the numbers, assuming 30Nm of torque and using driveshaft tubing with 80mm outer and 7mm wall thickness, he'd see 0.018 degrees of twist. With 300 Nm he'd 0.18 degrees. He'll be fine with the shafts as long as he selects the correct diameter and thickness tubing.

[@tomboblombo]

Make the mute lever move the read head out of the way of the notes (side to side). It would achieve the desired silence while muted without the use of additional mechanical parts (springs etc). If the read head was flexible, it could avoid the edge cases as well.

[@alexandrachernysh7]

It would break the part (or jam the whole programming wheel) when the part moves back into a passing pin

[@foldionepapyrus3441]

A good idea if there is enough width in the wheels between programming Pin lanes, which I suspect there will not be. I do agree though this method has way too many parts and points of failure for comfort. I think with the expected dimensional constraints making the read head fold its pickup foot up out of the path with his beloved Bowden cable is probably the solution - spring on one side (probably to lift the foot) put tension on the Bowden to make that part effective the single ridged whole that can transfer the programming to the dropper (that way you can just put longer lever and more throw on the 'un-mute' lever until its reliable and easy to operate.

[@EatMyYeeties]

​@@alexandrachernysh7Then he could add a bevel to the profile of the part being moved out of the way to allow it to slip back into place in the event they were going to collide. Similar to how a door latch pops into the frame of a door.

[@KnuckleHunkybuck]

@@alexandrachernysh7 Exactly; it would be like a car changing lanes and sideswiping the car next to it.

[@NickDangerThirdGuy]

You might consider adding additional locations for power input from the other members of the band as a way to spread the physical load.

[@gutschke]

Or maybe just an e-bike motor with torque sensor. You still have to do the pedaling or nothing moves. But the motor dynamically adjusts to reduce the amount of power that the human has to input. 200W of sustained human-powered operation sounds more than optimistic, when the human operator also has to perform music at the same time. An electrically assisted marble machine would take the edge off and give Martin some headroom for his design -- and fortunately, e-bike components are becoming commoditized, so it's increasingly easier to incorporate these type of features into a design.

[@SerNerey]

Martin, try another design of the muting system: instead of lowering the lower hook try to rise the upper hook. The advantage of this change is next: since you altering just the next element to the reader, you can drive it too if you add some stopper. In that way if hook rises slightly, it can rise free, but if it rises higher, it should touch the stopper and rise the reader too. Of course, muting system should rise the hook enought to rise both of the hook and reader

[@userPrehistoricman]

You also need to make sure that raising the upper hook will keep the hooks engaged until the end of the note. That's the edge case where the channel is muted during a note.

[@Kant0sh]

It's great how you can deep dive on specific designs for small parts without having to think about where it will fit! This will make collaboration so much more viable, because not everyone has to know every part and their side effects. But I think you now have to set very clear constraints for what will be designed, otherwise feature creep may become a problem for individual parts, and if a part changes enough for it to influence the parts it is connected to, you get the same interdependent nightmare that the MMX became. The specific task, target size and other parameters for every one of the parts you show at 7:05 should, in my opinion, be documented beforehand, and changes to those documents should be reviewed and scrutinized not to add unnecessary bloat.

[@Culpride]

I love the Idea of a "Power Meter" that shows the hight of the weight. It gives you the option to let the machine run out, like you did in the MM1 video. Add RPM meters for the flywheels and MM3 will become properly Steampunk =D

[@mrping2603]

Love the graphics on the presentation. Your meticulous effort is very satisfying

[@gerbdnas]

I think that belt gearbox is quite nice, probably the most quiet one too! Easy to change ratio, just 3D-print new ones. Look at a solution like on lawnmovers where you activate/ deactivate the mower deck with a tension roll. Could have a loose belt on all gears and just activate the ratio you want with a lever. Another easy solution could be like the "gearbox" on a snowblower that is just a rubber disk and a wheel where you move the wheel in and out on that disk.

[@ah-64apache84]

could there be a problem with repeated fast notes while muting? also maybe drop the requirement of having the axle of the reader fixed. moving it out of the way might be a better solution if you want a silent mechanism. EDIT: also i feel like there are a lot of parts in this assembly for what it does mechanically.

[@markfeeer2149]

I agree this feels like a hack. Similar to the MMX. Extra parts, more point of failiure. Not to mention that weak springs are ware parts. You should think about splitting the reader in two and only lifting the feet not pivoting the whole mech. And with that you migth can get rid of the existing discconect. Your idea to lift the reader is correct but I think that should be the only part that diconnects.

[@ah-64apache84]

right. the sliding connection was great for the MMX, but if you redesign it any way, why not swing the reader out of the way while keeping it connected to the dropper. this saves parts and does not rely on springs that might wear out or produce unpredicted behaviour.

[@izek01]

Idea for your thoughts in regard to a different muting method… Instead of adding and over complicating mechanisms to each ball dropping key… You can add a much more simple and reliable clutch mechanism before each drums program gear that will allow you to stop the rotation of the programming wheel… this will solve many issues. A… you can build up mechanical momentum before your music starts by releasing the clutch B… effectively mute all ball dropping mechanisms in multiple scenarios by adding extra clutches to each programming wheel gears. C… significantly reduces extra moving parts and reducing extra possible failure points. D… save space in your already tightly designed ball dropping mechanism… this will allow to to improve the strength, stability and reliability that you want to maintain. You can also reduce mechanical noises by attaching foam rubber in strategic spots that will not compromise the moving parts… obviously not the main solution but if it can reduce the overall mechanical noises by 5 or 10db, its a win… any way you can reduce sound transference bleeding into all them microphones will ultimately help the front of house sound quality.

[@justinnaramor6050]

uh... I'm not sure just stopping the rotation of the programming wheel is a valid option. What if an instrument is muted for say 4 measures, then unmuted on the fifth measure? The wheel would still need to rotate so that the unmuting happens at the fifth measure. The programming wheel is literally where all instruments get their timing from, which includes at what point in the musical measures they're muted and unmuted. So you're solution won't work from a musical perspective. The mute and unmute time needs to follow the rotation of the programming wheel... it's where all musical timing comes from.

[@izek01]

you make a good point but I still think that 1 master clutch the can stop all programming wheels that will mute all instruments while he builds up momentum in the main drive mechanisms will help

[@justinnaramor6050]

@@izek01 You know, that part you mentioned, about stopping all programming wheels while building up momentum, is definitely a good idea. Perhaps Martin's already considered it and just has not yet mentioned it. I literally do not know shit about engineering though, so I've no idea how to implement that. I'm certainly not going to even attempt to offer engineering advice on the project because I'm not an engineer :).

[@kitschypictures2290]

I used to watch for inspiration on my projects, but now I watch like a cautionary tale, as I witness the catastrophic effects of perfectionism at play. The tragedy is Martin's inability to see that the machines imperfections are what make it a work of art.

[@bearpawdgamr2785]

I had a similar feeling. Wondering how big the group of folks is that appreciate the mechanical noises that add to the music itself and make it super unique, while also appreciating the drive to make it perfect. Tight music is beautiful, but so is the mostly wood machine with every creek, squeak, and slightly off beat notes.

[@christianvanderstap6257]

He was not making an art piece however

[@bigbuckey7687]

This is exactly right. I stopped watching after he quit the MMX, this just popped up in my feed and I'm laughing seeing the lack of progress after he completely scrapped an imperfect but working machine. I think he realized he makes a lot more money from youtube and patreon making these videos than he would doing the mythical "world tour". Totally ridiculous that he sold merch with the MMX blueprints on it and then threw it all away because it wasn't perfect.

[@justinnaramor6050]

@@bigbuckey7687 I'm sorry but you have got it all fucking wrong, very wrong. this community needs to scrap that completely false narrative. I'm serious. Some of this community is full of shit because of outrageous claims like this. Jeez! Martin didn't "completely scrap" the MMX because he mentally felt like it failed, he "completely scrapped" it because it actually fucking failed, god damn it. And he didn't just throw it all away in the trash either... he sent the machine away for someone else to fix it up. More specifically, to a mechanical instrument museum somewhere in Germany if I remember correctly. Seriously, this is why "devoted" fans of someone or something are fucking annoying... they get all butthurt whenever their "idol" changes direction or halts a project to start a new one or whatever. Artists have their own hopes and dreams. the least we can do is encourage them to pursue those dreams. Not making false accusations and wining like a baby. Give real, constructive criticism that's actually going to be helpful, but only if you actually have real knowledge on the subject. This will actually help Martin along the way, complaining and acting like a know-it-all when you have zero knowledge on the subject does literally nothing. Also, please remember that the marble machine is not an art piece. No, it's not. It is not some beautiful visual art sculpture where reliability and correctness means nothing, where it's only purpose is to look mind-blowing. It's a musical instrument, and as a musical instrument it needs to be practically perfect, like any other instrument. It needs to exactly follow the musical instructions you give it, as consistently as fucking possible, and if it doesn't, then it has failed entirely.

[@buznee17]

Hey Martin, I was curious if you are also looking into safety. I don’t see that in your pyramid. Usually flywheels with lots of angular momentum can store lots of energy which needs to be contained. I would add lots of safety factor in the design when doing stress analysis and also make it failure tolerant. I’m thinking you will also be doing a static and dynamic balancing of the flywheel, similar to what is done for tires. Maybe a containment system for the flywheel to provide some risk mitigation for a failure? Are you planning on adding a brake?

[@blueneosky]

If you choose to split power in 2 mains axes, I think you should consider an epicyclic gearing (like differenciels for wheels in car). This gear is great for self-dispatching power on axes with different speed.

[@oraziovescovi1922]

everyone calm down with the shaft being torsionally unreliable. With a good choice in material and dimensions (fairly easy to calculate for any mechanical engineer) there's no need to worry at all. The forces here are simply not enough to make it bend, as long as there are no acceleration peaks. If there is an acceleration that powerful, it would probably be a decelaration with something that went VERY wrong (jamming), so hopefully that is not the case!

[@psergiu]

Seeing Title: Oh, no, the MM powertrain exploded ... time for the grinder !

[@Jesse_Carl]

From an engineering and development perspective, making the machine more spaced out and modular is an excellent choice. However, I cannot say I like it aesthetically. There is a reason the curta calculator is the only mechanical calculator I know by name. Desk sized ones do the exact same thing, (and some do more). However, they are not nearly so impressive. When you take the case off of one of them, it is not too hard to follow the gear trains and start to understand how it works just visually, and it is not hard at all to understand how someone designed it. But when you look at the internals of a Curta calculator, it is completely inscrutable, and the design process is a magical mystery. You are left wondering how anyone could have ever come up with it. I think this wonder at the tight, non-modular design is what made the marble machine great, and what the MMX captured perfectly. I watched the MMX design process from close to the start of the video series, so I have heard the thought process and design behind neary every section and part. But even still, I feel the same wonder as when looking at the Curta calculator. I worry that an expanded and modularized design will just not be able to capture that. Again, from the standpoint of an engeer and a musician, you decision seems excellent. But as a mechanical art piece, I think an expanded MM3 will not be able to live up to it's predecessors.

[@Mr_CrowFPV]

It's still winter but our captain is looking at spring already. Grandmaster level thinking, bravo

[@jeremy6732]

Power output is a function of your weight, with the current design. You can not push more that your weight (unless you can push yourself down). So the speed you can put energy into the system is caped. So to increase the rate energy into the machine you need to get heavier or be able to increase the amount of times you are using the foot petal. Increasing the frequency will no longer with with the beat of the music. Love the Project keep up the great work.

[@jakobrosenqvist4691]

Longer levers would increse the max power he can imput without increasing his body weight.

[@Joost.]

Its wednesday my dudes! :D

[@BWITHYURI]

AAAAAAAAAAAAAAAAAAAAAAAAAAAA-😩

[@Joost.]

@@BWITHYURI yes

[@BWITHYURI]

@@Joost. uh-huh

[@Giacomo_Lee_Oakenstand]

I'm so impressed by the ingenuity of your solution in Part 1! I think the key would be to test different strengths and types of springs. If the Weak spring (Spr. B) is not weak enough, it may cause a delay in timing due to pulling against the Strong spring (Spr. A). We would need a (Spr. B) that doesn't deform when stretched, although the idea of two springs may help negate that. And (Spr. A) can't be too strong because it might lockup the whole system.

[@SL-mk9jr]

This is an enormous undertaking ! 😵‍💫

[@MrPixelTech]

“I’m going to stay on point by doing form from function because that’s the way I kept running into issues… now does everyone remember our old friend feature creep??? HE’S BACK, here today to talk to us about running before you can crawl and drive!” From one feature creeper to another, Learn the Lesson before you’re 5 years in without even knowing if it’ll work.

[@thomasbecker9676]

But feature creep allows for more content to be created for KZbin.

[@voarnes]

Not sure if adding complexity for muting would be a good value add. I assume there will almost always be unmuted instruments running. So you are just reducing the frequency of the too loud clicks. I would look more in the direction of making them quieter all the time. Either by padding the paddles, changing the shapes, or shielding. Keep up the amazing process.

[@MaxSansalone-y9r]

So Cool! Thx for your interest in my video :) Can't wait to hear the finished product.

[@vladimirpain3942]

Honestly, I am not even that much into music myself, but what I love on your channel is the way you are making consistent and effective progress. It is valuable for me to see the way you manages this project. Helpfull.

[@thomasbecker9676]

I'd like to refer you to MMX.

[@tomasn3]

Reducing the overall noise it top notch! I found the overall noise of the previous machines too much. So really happy to to see it been taken care of 🤩

[@KilianKlein]

The way the power train looks makes me think that you should get yourself an Elliptical trainer (or even several) and modify it to be the power input, it would allow you to use arms and legs in a synchronised matter which would probably make it easier on you (so long as you remove the original resistance of course). You could even have people from the audience come up and be the power source for each song :)

[@Jesse_Carl]

Very practical, but I think it would look way too goofy

[@ivanangeli]

I have to say, my job is to make stuff from LEGO bricks, and I have been doing it over a decade now - sometimes I loose my enthusiasm. But every time I come to this chanel, I get boost of energy and desire to start designing and building again :)

[@nealcarter]

On the muting function, listen to your own voice, the deleted part is the best part, (less is more) , go back to first principles. You are adapting a design that was made against a different set of requirements.

[@DHoberer]

I love your work and your music. I actually quite like the mechanical noises, so hearing them in this video without marbles was really cool to me. I enjoy clicks and clanks, but realize that’s not what you’re going for in an end product. Way to go and I can’t wait to see the finished machine!

[@ardemus]

Thanks for the update, I love to see the progress, and the more mature development approach. That said, please be careful assuming that you can match the power generation of a typical fit cyclist. That cyclist may be able to maintain 200w for an hour, their effort tends to fluctuate in intensity (periods of high, low, and no effort), and a bike is a highly refined machine. On that last point, cyclists get a very efficient transfer of work to power. They generate power on the full circular stroke of the pedal, up and down but also (to some degree) forward and back, Your design looks like it only generates power on the down stroke? Bicycles are optimized from generations of iterative improvements for low mechanical losses. Bikes are also a minimalistic machine. Have you factored in mechanic losses on your much larger and more complex machine?

[@Milagoso]

👏👏👏, I am a bit concerned about this as well...

[@ferran1818]

I would recommend to use just one programming wheel. No need to have 3, they are connected by cables to the readers so you don't need to have them behind every module. I feel like having threee is just unnecessary and ads complexity

[@Kazutoification]

Have you considered the possibility of a pre- or mid-performance injury? For instance, if you were to injure one of your legs or feet, would you still be able to operate the exploded powertrain?

[@earthplusplastics]

he is slowly inching towards the inevitable MIDI marble synth

[@grachiki]

Всем привет. Спасибо за труд ❤🎉

[@zachstover1500]

i'm super happy to see martin getting back into the passion of his project, instead of feeling lost with the uphill battle of the mmx, this process is so great!

[@thomasbecker9676]

How is this progress?

[@kevinschafer2713]

I feel like you need to re-examine whether you need to break the link between the reader and the marble dropping. The purpose of breaking the link to mute was to not drop a marble when the reader was activated. If the reader will not be activated, is there a need to break the link?

[@teblack2]

MM4 Would be a machine in humanoid form to play like us, and you will have dynamic, timing, sound like never before

[@Santrial]

Wintergatan Wednesdays always boost my mood

[@Dannerrrr]

It stresses me out as he adds more fragile features instead of making things robust.

[@plig88]

Two random ideas for dynamic music to throw in the hopper: - Additional second channel with smaller marbles on the same instrument. - Using the rubber strips that you used in the vibraphone sound quality video. You could imagine the height of these from the instrument being controlled by cam profiles on the programming wheel and/or with a second lever from the main control area.

[@nickfosterxx]

Neat. I too was thinking immediately of smaller marbles (ex drummer) but the rubber strip height is interesting idea too. Edit: perhaps the rubber strip solution would also work instead of the possibly more complex smaller marbles on drums too.

[@rossaylen7441]

Would be interesting to see the kind of power output you can produce through the pedal design you are proposing.

[@Milagoso]

I think this might be indeed much more of an issue than we realize

[@jeffs1571]

Regarding drums: When learning to play, our drumline was taught "You don't whack the drum harder, you just raise the stick higher." Generally a loud hit would be as close to 90 degrees as you could reasonably get, a standard hit would be around 45 degrees, and then ghost notes would be 20 or so. You also get quieter the closer to the rim of the drum you get, so you may be able to optimize it to two heights and two locations on the drum to help control dynamics.

[@zerstorer1ss]

If the mechanical noise of the reader riding the profile will be present when music is played, shouldn't that root cause be addressed? That would remove the need for more fiddly linkages and reliance on spring tensions to be "just right" when muted.

[@benmauro1022]

Hey Martin, I'm glad you are feeling inspired to continue, I have always loved your music and work! I hope you build a prototype of the double spring mute system, as I would rather you be able to guarantee it's functionality before it was installed into the marble machine than find it not up to your standard and have to rework. As always I hope you remember we all wish you well, and a lot of success on this journey. Even if we don't always write it in our comments.

[@Steve-lu1nc]

I might be completely wrong since I don't know much physics but for a while I've been thinking about this whole whatt thing. Martin keeps comparing the energy required to a person on a bike. But the energy will be input with a floor pedal (one foot at a time) rather than constant two pedals like on a bike. From what I understand a bike is a very efficient way for humans to produce energy though movement, and the same amount of effort on a different device is not equal to that output. Would really appreciate others thoughts on this

[@ProactiveYellow]

There is a dead spot in a bike power train where the pedals are vertical and you can't put serious energy into it (before you can push down on the top pedal but after you've reached the bottom of the low pedal). Periodic power output should work pretty well with the flywheels to smooth out the losses, it's like how the wheels of the bike and your mass hold momentum during that dead spot in the pedal cycle. Yes the mechanics will be different because of where the power's going and how it's being applied, but it's fairly similar in a broad sense.

[@Steve-lu1nc]

@@ProactiveYellow makes sense, thanks!

[@Milagoso]

I was searching for this comment! I am agreeing that the energy input per cycle might be the same, but the force that you can input into the system will fluctuate. I think pedaling has larger force peaks than a good bike.@@ProactiveYellow

[@gavdownes100]

Dr Frankenstein would be so proud of the monster you are creating. The after affects of this is going to be interesting as well. The tours and interviews and your fame you are going to have.... I'm glad I've been on your journey from almost the beginning

[@iancovill8854]

Why don't you put the mute and the locking link behind the reader so that the you only need one action to get the reader and the dropper muted? Obviously you'd still need to solve the dropper problem, but I think adding more action for the same reason will cause headaches down the road.

[@alexanderpouwels5042]

Wintergatan, i just heared your song on the national dutch radio. It was awsome. Keep up the good work.

[@peterpansplayground]

I wonder if you’ll make a song that integrates the machine’s ticking.. I am no engineer, but it’s always such a delight watching you..

[@thomasbecker9676]

I *am* an engineer, and I wouldn't use the word "delight."

[@GregorDuckman]

the original marble machine song sounds kinda hollow without the mechanical noise and chattering.

[@tobythepotato99]

after almost giving up im glad you didn't because to most situations of creating something there are always a proper solution you haven't yet tried which is the golden piece you needed ! 😊

[@dataandroid2036]

Hallo, ich würde die mechanischen Geräusche haben wollen. Viele kaufen sich wieder LP's weil sie das Rauschen und Knacken vermissen. Ohne Nebengeräusche wird es steril. Grüße

[@Atomic314]

This project is getting me so excited about mechanics

[@chadelliott7629]

All these springs and cables are opportunities tolerance errors. You're drifting away from the "best part is no part" philosophy that saved the project. Please take a step back and evaluate again. Choose a simple, rigid, reliable design.

[@thomasbecker9676]

That would be less content to post on KZbin.

[@peanut_buddah]

Hey there Mate(s). It's very inspiring that you keep on developing the marble machine. I have to admit that its nice to watch and hear, although not really being my cup of tea. But its the sheer and brutal willpower to do this, to achieve this and to perfect this that leaves me in awe and wanting to create stuff on my own, so thanks for motivating and inspiring! 😃

[@KevinRedmondWA]

Overly complicated.

[@lhovo]

If you haven’t already considered it, I would recommend designing in a shear pin to break under unusual high torque for your flywheel. It will save you and the machine if something would to break down and seize unexpectedly

[@QuestionMan]

Nothing wrong with dreaming big. It's good to find the edges early in the process. Can hardly wait to see what happens!

[@theherk]

So glad you used Max as an example. He plays some of the funkiest beats ever.

[@howtoappearincompletely9739]

I'm so glad you're still working on this project. It's absolutely marvellous.

[@Sharlenwar]

Always look forward to your videos. Glad to see you continuing to work on the dream, and man, each day you are getting closer!

[@zactron1997]

For the muting problem, I think the simpler solution would be to use the programming wheel itself to unmute a channel. Spring-spring interactions (especially when relying on momentum) are very hard to tune, and increase the amount of power you need to route into the programming wheel (stronger springs mean more force to overcome them). The way you want your mute lever to work is: 1. When activated, the channel cannot release a marble 2. When deactivated, the channel can release a marble on the next programming pin. Hard to describe the geometry in a youtube comment, but you want something that interacts with the wheel. As for the power shafts, they are going to be a timing nightmare at that length. Nine meters long with ~200w of power and a large gear at the end. Even the slightest torque will add a few degrees of twist to the shaft, which could translate to an entire note of error across that machine. You can keep the shafts as is, but you will need a mechanism to keep the different wheels synced. Finally, for the gearbox you probably want a CVT (Continuously Variable Transmission). Normally they're disliked in cars because they're not able to handle really high power levels. But for the Marble Machine, you're basically dealing with a bicycle at the size of a truck. A CVT will let you choose an arbitrary tempo for your music, and they can be setup to even give you a lever at your control platform, so you could change tempo during play (you don't need a clutch to disengage play with a CVT).

[@deefadale]

Hey drummer here: - to get that 'pop' sound on the snare you hit the rim at the same time as the snare with your drum stick, which causes the 'whip' - to get the dynamics you're chasing, with the subtle nuances, I think you're going to have to drop your marbles onto different drumsticks (on levers, springs etc) and not directly onto the snare - to get softer notes, either longer/shorter levers, stiffer/lighter springs, different padding on each drum stick - most of the nuances could probably be solved by dropping marbles directly onto the snare at different places with different padding/foam - then only need one 'pop' mechanism eg via drum stick or other crazy invention you come up with 😆 Love your work and so excited to see what solution you come up with here! 🥁

[@nickd3873]

Quick though on dynamic of your drums. Most of the dynamic in drumming is caused by how hard or softly you hit the head. I’ve thought of two ways to add the dynamics, one would be to change the hight of the drops, higher for harder hits, lower for softer hits, main problem would be that you are adding additional motion problems. 2nd idea would be to use different sized marbles on the same drum, larger for harder hits, smaller for softer hits. This adds a problem of marble collection, but maybe this is the easier problem to figure out. Love your channel!

[@queeg6473]

For the ghost notes on the snare drum have a 2nd gate open closer to the drum with a smaller marble that drops at the same time as the main marble. This can then bounce a few times as it travels across the drum head.

[@starblaiz1986]

So glad to see that you calculated the power input and output in detail and it looks like it's in the realms of feasable, because that was worrying me. 200w is still a serious workout, but I think it's more than feasable that a few smart design decisions to reduce the load in some key places would fix that, and at least it's not something crazy like "10x what top athletes can achieve" or anything wild like that. And if it comes to it, you could always get a second drive train involved and get one of your band mates to pump some extra juice in to double the energy input. You are doing amazing though and this project is developing really nicely, keep it up! 😊❤

[@Milagoso]

It would work if the input was a bike. pedals are not a bike since they have a totally different load profile than a bike.

[@stefanmadsen5605]

i recommend getting a cvt as ‘gearbox’ as this could be setup to work perfectly with your hyigen drive also, you need to seriously consider the torque on the 9m shaft, it WILL twist/warp under load - and will completely mess up your quest for ‘tightness’

With a "marble machine physics" spreadsheet you not only can see what are the results of any changes to the design, but you can employ *solvers* to find out what you need to change (and how much) to achieve desired goal.

[@RASpitzer23]

Some notes: 5:27 (spring on the rear of the profile reader) - Works in theory, but the weaker springs are going to make a lot more noise on the disconnect if it's muted mid-note, and if you have the forward springs on the dropper strong enough to force a disconnect when the lever is set to mute, then you're looking at possibly bending the rear springs as well. To break the noise component of that statement down: the rear springs are going to vibrate quite a lot during the recoil phase when the latch disconnects and it snaps back, plus you might also have noise from the reader arm assembly hitting the end-of-travel on whatever it's connected to when it's lifted under tension. To break the spring tension component of that statement down: You're going to need enough forward spring tension so that when the note is finished, the latch assembly can move to the right, but move to the right with enough force and inertia for the latch to come apart for long enough to drop down. Depending on the tension of the forward spring, this could very easily bend the rear springs if you don't select a heavy enough spring. It may be a better idea to look at only raising part of the reader arm, however I can't give you any pointers there - I'm a technician, not an engineer. --- 7:05 (Powertrain) - I don't really have any input here that everyone else hasn't already covered - just make sure the twisting of the shafts is properly accounted for. Might be better to make multiple points of torque exchange between the programming wheels and the Huygen drive so that all the torque is being more evenly distributed, instead of at the centre of the shaft. Alternatively - and this one I DO have a solution for - it may be wise to look into a proper Cardan shaft. While I don't understand HOW they work (specifically, how having a shell with a U-joint prevents torsion) to know WHY they work, I do know that they work for high-torque applications like heavy machinery and automobiles. You can also look down the bolts on either end of the driveshaft and see if it's being affected by torsion. It won't be elegant, but it might work better for your use case.

[@MacroAggressor]

You may also need a weak spring on the muting cam to pull the release mechanism's hook down.

[@icarvs_vivit]

MARTIN think escapement for the muting mechanism. You can disengage the read head from the gate and move the read head away from the wheel WITH ONE PIECE.

[@horlogeur]

The double spring solution could be a good idea, but I think it can create a lot of problems and need tons of adjustment, especially with the strength ratio between them. If it's too weak, the release never happens. If it's too strong, you can have a worst noise because your piece hit the programming wheel and keep producing the mechanical noise you try to avoid. With all that, the springs herself can react differently if you play in hot or cold weather. Maybe you can add something to adjust the tension of both springs for more precise results.

[@Cyber-Riot]

Look up Continuous Variable Transmission (CVT). They're used in motor scooters and some other lightweight vehicles. It's basically a belt drive with pulleys that dynamically change size. The variable pulleys are made by placing two chamfered disks facing each other. Then by varying the distance between the disks, it causes the belt to move inward or outward along the radius.

[@ror3D]

Two things: - on the mute function, you have a cam already next to the reading foot, you could attach a string/something that just lifts up the foot when it's muted, I believe it would solve the edge cases easily and it would not require almost anything extra todo. - I didn't look very closely into the entire spreadsheet, but I didn't see any accounting for losses due to friction, noise, etc. I don't know what those will be but I don't think they can be disregarded, if you design with very low tolerance it might come to haunt you later

[@Eagle3302PL]

For a muting system I have an even simpler idea than other commenters. Add a bypass channel, basically move a cushioned bypass half pipe in front of the marble dropper to intercept. It'd be completely mechanically decoupled from the rest of the machine and would not add any new failure points or other issues. You could simplify the entire mechanism this way too. Also it would not introduce any continuous noise.

[@dragonboss7995]

Amazing video! I do have a suggestion. On the little arm that is supposed to fall when you mute the device. You should add another spring(I know right) but this spring will make the mute position it's default thus also allowing for the small gap of time where it's disconnected, to have another force acting upon it to persuade it to move down. Cheers!

[@utilka5415]

when presented that recording of readers clicking... i kinda liked that mechanical noize. while i understand why one would not want stray noises in a musical instrument, but it did add to that mechanicm-gears-clock astetic

[@chellybub]

Seeing you work through this takes me back to university. Love from Australia ❤

[@Tex777_]

To accomplish the "dynamics" with the drums, you'd likely need to design an intermediate striker system (think of a drumstick like a seesaw) and have the ability to change the fulcrum point so that the output velocity changes without needing to adjust the marble system, only the fulcrum point. A good example is BMW's Valvetronic which they use to adjust valve lift. The "input" camshaft profile is the same but they can control the leverage it has with a secondary camshaft that adjusts the fulcrum.