Safety First! - Marble Powered Clutch

[@Wintergatan]

Competition! Price: Glory Forever. Reply to this comment with your guesses on what the ratchets i show at the end of the video are for. The most specific explanation wins. If several people have the correct answer the earliest comment win. as long as it is precise enough haha. All clear? GO!

[@officialxardeus]

Have no clue, just wanna say hi

[@Frosch1220]

for love ? 🥺🥺

[@MrGonzonator]

Bicycle.

[@thatfrickenweeb]

My guess: You're planning on spring loading the machine and the ratchets are there to assist with winding the mainspring by preventing counter rotation while it's being wound. The teeth on the wheel will interact with pawls that will permit the wheel to only rotate in one direction. That would also explain the extra safety measures you're employing.

[@danieljohnsson8267]

Benig able to turk freeley like a bild or not being able to rotate the wrong direction

[@andyjbauman]

12:04: If you are using that light green disc as a brake, ensure you review the following: 1. The material needs to be an actual friction material, metal-to-metal contact will cause galling. 2. The flat head bolts need to sub-flush of the green disc. 3. The bolts connecting the thicker green disc need to be stiff enough to resist the torque being applied. Bolts are really not designed to bend, which is how you are loading them in this application. 4. The material of the 'brake disc' needs to be thick and stiff enough that it does not turn into a cone when you apply a load axially. The same applies to the pink ring to which the bowden tubes are connected. 5. The two hex nuts between the green disc and the cyan disc would need to be set very accurately to ensure those two plates are perpendicular with the axis of rotation. You many want to replace those two hex nuts with a machined sleeve that can slip over the bolt and accurately set that distance.

[@jansimblank9358]

These are good points. From what I'm understanding, he's not trying to break the flywheel, just the drive wheel which is relatively light, which should brake almost instantly without any significant force needed. He might want to brake the whole machine I don't know

[@quakxy_dukx]

@@jansimblank9358 well that’s what the clutch is for - to disengage from the flywheel so he can brake more easily and safely

[@Benign441]

I suppose you could build into the Bowden system some flexibility for on the fly torque adjustment. Maybe down the road from wear the normal torque gets too high and brakes? Or maybe I don't understand how sensitive that mechanism is.

[@billkeithchannel]

I had no idea that bicycles now use disc brakes for high end models. I also had no idea that the average price is now $5,000 USD. wow!

[@deathofallthingspotato9919]

You seem to have misunderstood the mechanism (or I have). If the brake was engaged, the force would be between the cyan and pink disks, as the green disk is rotating with the cyan one. This makes some of your points largely irrelevant - 2 particularly, but makes the issues with the bending forces so much worse, as the green disk provides the braking force against the alignment system for the springs, which have effectively nothing holding them aligned. Even disengaging the clutch would likely break them.

[@quinnobi42]

My engineering design class talked about scope creep today, and the first thing I thought of was Martin and the MMX.

[@jairo.cabello]

Hello Martin! Take this with a grain of salt, I believe you have over-engineer a bit too much this clutch-brake system. You could very easily add a brake disc from a motorbike/bicycle, with a caliper that is operated by a cable and have an extremely reliable and based on off-the-shelf parts available anywhere for achieving a simple, reliable and lightweight brake system that on top of that is easy to modulate how much strong-ness you need. In general aviation (small ultralight planes), they actually use this approach of reusing something that was already designed and tested as a package. There's a reason why vehicles have moved away from having the brake attached to the clutch :) I bet if you would explore this option, you could reduce considerably the amount of parts required. Cheers!

[@christophercasale7961]

I agree, these are separate problems that could be handled with separate systems. Clutch and brake systems are wear items, going off the shelf for wear surfaces is the best option. I suspect for the torque and energy something from a motorcycle or moped would be perfect.

[@dmor6696]

@@christophercasale7961 agree. simplicity and off the shelf parts are, at least, where you should look for inspiration for their proven success

[@eLBehmo]

I agree, but want to note that I don't understand how the actual break works. If it is just 2 hard-to-replace metal discs grinding on each other, the resulting noise and vibrations won't be compatible with the instrument microphones, right?

[@PvtHaggard]

I agree that the part count for this design could be reduced but using a brake as a torque limiter/clutch is not a good idea. For a clutch to act as a torque limiter, it would need to slip, which is a recipe for wear and heat. Clutches are not designed to slip for extended periods.

[@Robinlarsson83]

​@@PvtHaggardI think you may have misunderstood the point of the comment. I don't think the intention was to replace the torque limiter with a brake or even off the shelf clutches. The idea is to replace brake part of this mechanism with an off the shelf brake, clearly a better idea. Also, I think it may even be possible to completely do away with the clutch here, as long as the torque limiter stays. The torque limiter most likely have to be custom made and is definitely needed. But if you use a standard motorbike disc brake or similar, you can just brake the machine and let to torque limiter do it's thing if needed. There could even be 2 brakes, one for the flywheel and one for the rest of the machine, if needed. But, maybe there is a real need for a clutch to separate the machine and the flywheel but still let both spin down in their own time?

[@fiskfisk33]

"it's so convoluted, this design" this seems to me like a candidate for simplification

[@stevenmayhew3944]

Hopfully you're not overcomplicating it. I remember how you simplified the marble gates and programming wheels.

[@OrafuDa]

He probably needs to review a couple more times for simplifying the design. That being said, to my untrained eye Martin’s design seems to be in the rough range of complication that a combined clutch / break / flywheel design could get.

[@xXVintersorgXx]

I would argue that the flywheel is the most important part of the machine. He's basically engineering a clock/watch. And this is basically the main spring... The engine so to speak

[@petric334]

He's not manufacturing his own plywood, or forging his own bearings- it might be worth it to consider simply buying in a proven clutch design given the specifications. It would be a shame if literaly reinventing this wheel would sink the project.

[@tommasodemarchi1132]

Hi Martin, great work with what you are doing! I'm a bit worried though you are again complicating what you are designing. I'm also worried about the bearing case and type. In this configuration infact the bearing is carrying not only radial load but also a big axial load, normal ball bearings aren't built for these types of loads, angular ball bearings should be a much better fit. Furthermore the mounting mechanism you came up with will certainly concentrate the load on tiny portions of the external bearing ring which they aren't made for this. I would suggest at least to put some laser cut profile disks in between of the bearing and the bolts to distribute the load on all the bearing housing surface.

[@MostHolyPlace2]

Consider using precision Dowel Pins for the Bearing Housing or at least, Shoulder Screws. Threaded fasteners may cause tolerances to stack up.

[@FlexDRG]

If he measures the bolts he can adjust the hole placement and the difference in bolt&hole size to make them pretty tight. Or leave enough room to use shims to fill it out. That said. A turned housing with a stepdown / shoulder that has a tight tolerance to the hole in the disc should be near perfectly running true.

[@CleverMonkey-jd3du]

I do look forward to seeing how the tests work. We stand on the shoulders of people for hundreds of years of bearing race design. I'm a little dubious of how well the bearing is going to be held, but I'm willing to be surprised.

[@MrTPW26]

I completely agree with stack up issues. Shoulder screws at the very least

[@quentinmangel2265]

He can use shoulder screws but that only solves half of the problem because there are two bearings ! Look at 10:24 we can see that it is the same only one screw holding both bearings in place... i'm not sure how well the threads will hold :/

[@dawson3hill]

Locating pins are the move!

[@SamuelEstenlund]

The last videos have been about improving the old designs by reducing the number parts from 250 to 11, or something like that. This week: Look at awesome 2000-piece design!

[@br52685]

If you're looking for a consistent design-approach or philosophy, you've come to the wrong channel.

[@travelinman70]

@@br52685 the artist is coming out. LOL

[@steveggca]

Hi Martin, May i suggest that you reseach flanged keyless shaft collars and one-way bearing this might make what you are trying to achieve much simpler.

[@fletchro789]

One way bearings are also known as Sprag clutch. 👍

[@frollard]

original machine already had one way bearings (in the this-old-tony elevator)

[@doro5751]

Reminder: If anyone has some great suggestions for the marble machine, I recommend you to bring them up on the discord server, since they are more likely to be seen there. Thank you!

[@nelsonion3529]

oh hi!! fancy seeing you here

[@doro5751]

@@nelsonion3529 Hiho! I am everywhere :D No, not really, but I try to keep an eye on the comments and to keep them nice and remove spam...

[@Stephen_The_Waxing_Lyricist]

I'm not on Discord, so could you please pass on my question about preload on the bearings? It's not enough to clamp the outer bearing race in position; you need a positive axial load to be applied from the inner race through to the outer race. A common method is to use a pair of belville washers next to the inner race, and apply a clamping force through that into the race. This axial load (and you need it on both bearings on the shaft) will apply the preload needed to reduce bearing wear.

[@smit5983]

Love the single take with no edits! Much prefer this raw style knowing it helps you get more done. The quality was still plenty good

[@NaisanSama]

Not great for most people (cuz attention span) but it is quite chill unedited

[@valovanonym]

@@NaisanSama Martin has been developping mmx and mm3 for more than 5 years now, attention span is probably not a problem for it's fidel audience lol

[@GethinColes]

Yeah nah. I get why, and Ive been following this project since the beginning but I'm sorry, I hate to say it, but to be honest, I was bored. There are very very few people in the world that can extemporise without benefitting from editing.

[@Nooticus]

@@GethinColes you do have a point tbh...

[@aaronpaul5990]

I would suggest to use some actual break pads either from bikes/car or create something similar ... and find one that will be relatively silent ... and can be adjusted/exchanged without having to dis sample the whole clutch mechanism. Because at the moment you have to take out the whole disks if/when it is worn down .. in addition the heat will create tension in the whole thing ... meanwhile with pads it would be concentrated in those first.

[@DJShadesUK]

I don't think Martin is intending to "brake" the flywheel but to "break" the pedals connection with the large pulley using the torque limiter for this secondary purpose. Grinding between the stationary disc attached to the bowden cables and the rotating disc tensioning the ball bearing springs is going to be the problem. EDIT: Oh, wait... do you mean use brake pads on the stationary disc so its the pads that wear and can be swapped out easily when worn instead of having to replace the discs themselves? So not necessarily for braking but to eliminate metal-on-metal grinding?

[@aaronpaul5990]

@@DJShadesUK yep that's what I meant

[@mihalydozsa2254]

Hello Martin, I'm a quality technician so just to be pesky and annoying I'll :D I hope you make this videos to get honest feedback and improve on it so I'll come up with the worst cases I can for this. Pedal: I think the pedal should be higher for ease of use. I believe from horizontal you will have to push to much down It will be tiring, you may need a pivot on the arm to not be so high and risk tripping over it, it seems too long, but you may be able to animate and see how much you need to push it down like this. It is more easy to lift just your leg than crouching to push it, and lift your self up every time. Bearing: You can make an adjustable bearing housing, this you can't adjust, also the bolts can bend. And it will because you forget about gravity pushing down on them. Also just look up Laser cutting accuracy and bearing accuracy, there is a reason for nobody doing what you planned for your self here. Fast rotating heavy things are not a joke. You don't want it to shake it self apart. Also Think about transportation, It needs to be able to withstand that too without much disassembly. With two way forces there is a different type bearing I don't know about the frictions in your system, but If you plan to pull on axle direction you may need "thrust spherical roller bearing" it comes with different angles to your needs, but it does not have marbles. :D Clutch: Where is the part limit on this part? :D I think this is the case when you fell in love with designing and took way to much time designing and now you are in love with your design. When you need more than a few part with another technology you should not make sacrifices on your design for sake of using less technology. Supplier count: If you want to cut down the supplier count, find a supplier who can laser cut and have a CNC machine too. Those plates in your design just to not use CNC lathe looks dangerous too, probably not a big risk of shearing those bolts, but you need as less of a point of failure as you can have. And if you still want to cut down of the manufacturing techniques try to see if you can do everything that you can now with a different machine. Or just try to not overcompensate your last mistake and use what you need to create the most practical and useful machine you can. Don't be afraid of making the same mistake, you learned a lot now you know that everything have a place and a use, when you know something should be done one way, don't create an another obstacle in your way. Brake: Your brake I thing will brake the machine a different way that you hope for. With those long bolts from either side the torque will twist them and in the best case will pull away the toque limiter and jam the wheel, and the worst case snap and if the ratchets are not for emergency stop I don't know how else would you stop the wheel. Also the friction will weaken the plate you use as brake pad, and lose brake force, maybe even bend out, and ruin your bolt heads so if you want to repair the machine for whatever reason you will have a really bad time. The worst case the heating and the shear forces from the rotation will just pop those screw heads off. (not at the first few brake tests when you can catch it, in the middle of the tour the worst possible time) Design: I see you don't want to give up design, but as you have a functioning form, you can design it while keeping it's function. You don't have to think in mass production, you can make it pretty too. 90deg is nice and easy but does it matter under 10 machines, when you have precision tools in your arsenal? You are not just your little shop alone anymore, you can dream a little. There was something in marble machine 1 where you could hear the rolling marbles on the wood and some mechanical noises. It felt like the machine is part of the music not just sterile music like from anything else. Ps.: You don't want a marble machine 4, you want to finish this machine for the meetup. Make it not controversial. People here love you, and won't go against your will even if they know It may not work, they believe in you. But faith won't work against physics, you have to control your self, you know what is right.

[@constantinosschinas4503]

He is psychotic and autistic, why do we even try to give advice. He will never understand that the quality of MMX1 was it's flaws. Wintergatan produced romantic mediocre music and questionable performances, but is was at least happening. Now it is just a criep lashing out his psychotic creativity, trying to become an engineer. I know the style. I suffer from the same, now and then. But learned to hold a bit on creativity, to live a more human life.

[@travelinman70]

I totally agree with this statement, the magic of it being a marble machine is not just the movement and the music, it's the contraption sound.

[@Sembazuru]

Using the D profile instead of a key is an interesting design. I don't know how well it will work, but that is what prototypes are for. 😉 Might I make one suggestion. Use a double D design (two flats on the shaft 180° from each other) and then the split disks won't be two designs, rather two of the same design with rotational symmetry. More unique part reduction. Right now with this single D design each disk is split into two unique parts. But with a double D each disk would be split into two identical parts. I hope I explained that well. (I'm borrowing the "single D" and "double D" terms from panel mounted round items like toggle switches, connectors, etc. Hopefully those terms translate well to this design.)

[@DJShadesUK]

Thats a very good idea to reduce the unique part count.

[@moparmadness0180]

It really depends on the workload on that pulley of how well it will work. The design is flawed in that the the load between the shaft and pulley will be trying to seperate the two halves of the pulley itself. This also will cause problems as the two halves are also being held together by the same bolts that hold the entire rest of the assembly together. If the force of the shaft trying to split the pulley pieces exceeds the strength of that hardware (even by just slightly bending it) it can cause issues with all the other things those bolts are holding together, mainly the bearing.... A really simple way to add reassurance is to add 2 more split halves to each side of that pulley that attach to the pulley itself to help tie the pieces together. However, this adds more parts and overall is just overcomplicating a solution to a problem that has already been solved previously with fewer parts and proven reliability. To simplify, should've just used a tried and true keyway. He's overthinking this a bit.

[@ojsh_]

@@moparmadness0180 Nailed it. Shaft drive torque is converted into separation force in the driven member, kind of bananas to assume Keyways are just for fun.... they exist in industry on driveshafts for very good reason. Going to be very difficult to achieve acceptably small rotational play/slop with that single milled flat spot too.

[@radomane]

Things were so beautifully simple, the design was so basic yet incredibly efficient and precise. This clutch scares me.

[@Forr0n]

I can't wait to see your simplified redesign next week!

[@Joseu124]

Hey Martin, As an Engineer working in mechanical construction, i want to give you my input for your design. To me it sounds like an overcomplicated system for a system that could be done by a simple bycicle disc break operated by a bowden cable clamping on the same disc you are using now. You will also have problems pulling all 4 cables with the same force. About your bearing housing think back to where you said at the beginning of mmx that threades rods are not good for usage as axels. Same goes for centering a bearing. Threads are no good for centering anything. If you really have to use rods for centering, use smooth rods with threads on only on the ends, but id recommend you use a centering flange that is clamped between the bearing and the wheel, hooking on the wheelbore with a lip and the outside of the bearing. You could also use a bearing with a flanged outer race or a bearing with a snapring for securing. For the flat spot, I can tell you that its a possible weakpoint. I suggest you use taper bushings, with that you only need a smooth shaft on both sides, thats the easiest thing to use when connecting two parts. If you got qestions about anything ive said, feel free to ask. Greetings Luke

[@DeDeNoM]

Why not go with of the shelf flange bearings? No custon parts, no strange bolt-circle-bearing-housing

[@killerbreadsticks]

This dude has me checking his channel every Wed and Thurs-.- he said he was gonna be consistent but missed like every Wednesday but one this year

[@fresk9139]

I feel the same, I check every week and I'm always frustrated when there isn't a new video out. But I believe he also said that the youtube videos are taking him a lot of time, it's probably for the better for him to reduce a bit the amount of videos... (Sadly for us)

[@MeriaDuck]

I really like the series, for the last 6 years. And form follows function saves us marble release finger discussions 😍 I do miss the 'oops I got distracted by musical instruments ' moments 🤣. Please be a musician too once in a while 🎶

[@Nooticus]

agreed. i miss that.

[@Duffman-gj4fs]

Ratchets are for the playing card mechanism. Just a nostalgic nod to the old bicycle days with cards pegged to the frame hitting the spokes

[@shahriarrs5213]

This man is so dedicated. He inspires me to straighten up my life

[@NaisanSama]

Same, and even as far as making something

[@CWO3-uscg]

Let us know how that works out for you

[@bryanc1975]

Same.

[@RayFromTheHayclan]

Yes. Having hobbies and interests in life makes all that other crap; in life, job, bills Etc more tolerable.

[@TheUnrealPownament]

I love how martin is putting the same madness we all know, but into an industrial design instead of a garage-project design.

[@user-mu2ze3xi1y]

have you considered an automotive/bicycle style disc brake on the same shaft as the main flywheel? edit: I worry about heat from braking within your clutch stack not having an easy way to cool off, and possibly throwing things out of tolerance and inducing undue stress. an automotive style brake allows you to basically drop in known good things such as silent brake pads and rotors without redesigning the wheel.

[@1itim]

And a clutch assembly from a small motorcycle instead of the homemade one

[@turnerkilgore6818]

Your fusion skills have truly gotten next level. You're killing it. From your early days exploring in the MMX. Been with you the whole way and you are a serious inspiration. Thank you

[@99Mrjulius]

22:30 "I skipped a wohle part there..." (by using a million bolts lol) love it! can't wait to see what the ratchets are for

[@tannermcmurtray2948]

This clutch design is amazing. Looking at it reminds me of a dragster clutch. So excited to see this matching come to life. It'll be the greatest music box ever built!

[@XZapper919X]

Great video! Just one thought though: aren't you creating more parts by skipping the keyway? Yes, you skip cutting and making the keyway, but in turn you need to split all of those plates you have? Doesn't this increase part count?

[@DominusFeles]

Math of reality strikes again!

[@Robinlarsson83]

Yes, you are of course correct, this is sadly just one of the "bad" ideas in this design. The brake, the bearing mounts, loading screws in the wrong way, and increased part count/complexity to adapt to specific machining capabilities are some of the others... Of course you need to design for manufacturing and try to adapt to in-house capabilities or to those of your suppliers, but you can't go to far. I say this as someone who's done the same type of mistake or suboptimal designs more than a few times, and it almost always comes back to bite you in the end :) It seems that Martin gets a bit fixated on a specific process, a few years ago it was CNC-router to cut plywood, now it's lasercut sheet metal and profiles :) I wonder how much he knows about CNC lathes for example? If you need 20-40 bearing housings for a one-off project, you design your machine around the same bearings as far as possible (including using a bit to big bearings on parts with lower loads if possible). Then you design a housing that can be used for most or all of the bearings, and have it made by a shop with a good CNC lathe or even a 4-5 axis Mill if needed. You don't need a machinist to manually turn 40 identical parts on a manual lathe these days :) I wouldn't be surprised if Martin could design a bearing housing that he could machine to good enough tolerances in aluminium on his own Avid CNC-router :)

[@M4nusky]

IMO As soon as a good amount of force or shock is applied the shape of the shaft will try to split the 2 halves and all the bolts will simply bend away leaving the disc loose over the shaft...

[@mattcollier5150]

​@@M4nusky Agree!!! this design will slip in no time! at least use a shaft collar with a set screw or better yet a keyway.

[@kotiachiy]

That brake assembly will have alot of flex in it. Also, synchronizing 4 bowden cables... i hope Martin will once again step back and simplify this design.

[@isaacrobbins9526]

This gives MM2 vibes. Something tells me we could do exponentially better and simpler with a couple live streams or iterations.

[@alvaxo7]

It´s amazing watching over the years how martin has transform from a charming woodworker to a pretty competetent engineer who even thinks about precausions for his machines, i couldn´t imagine the martin from 6 years ago doing this amazing contractions.

[@billstrozberg3932]

For the paddle, I have an idea that might work. Split the pedal shaft in 2 and connect them with a one way hinge. That wai it stays still when you press down on it, but if there is any force (from the tip of it) pushing it the other way (like a foot underneath it) is justs bends upwards with the hinge. Systems like these are used on these types of mechanisms where you want to have force applied to once side but basically disingage if force is applied from the other.

[@scaredyfish]

Sure, it would bend upwards, but wouldn't it just keep whacking you in the foot? I think it should be a treadle, anchored to a pivot point low enough that you can't get your foot under it and/or a ratchet mechanism that means the pedal only moves when you're actively pedaling it.

[@billstrozberg3932]

@@scaredyfish it would whack you but only with the weight of the pedal after the hinge. So very mild whack. Its not something you would constantly do, its a safeguard against chopping your foot off. There should be also a guard, but these are never 100% failproof. So double protection is the way to go

[@D3nn1s]

​@@scaredyfishadd a light sprint to it and youre good. Ive posted basically the same comment under the last vid, even if you add vertical wall all around the pedal, one small screw is enough to *screw* up your day at best, or scar you for life at worst

[@NaisanSama]

He should see this

[@Hoggaforfan]

That system was on Marbel X and yea, don't see the advantage with the new system over the old

[@jcKobeh]

I love these videos not just because of the music and the machine and Martin, but because I get to learn new engineering concepts on each one, AND THEN I get to read up on those same concepts and practical applications down in the comments. Learning never stops.

[@KruddMan]

What if the marble machine 3 became the most dangerous and deadly instrument to play, but it played such beautiful and complex music that people were compelled to sacrifice their lives to it.

[@constantinosschinas4503]

lol. MMZ level of sarcasm.

[@valovanonym]

So happy to see you back and healthy! Great video, I like the calm explanations without a lot of edits, it feels like a good friend is explaning me a project he's working on. I simply hope this design is not overengineered as many other more capable people already said. I also want to tell you Martin you've been a huge inspiration for me over the past 6/7 years. I love your music, which has the power to strongly move me, and I love your MMX and MM3 projects which really show your resilience, intelligence and humility. Thank you for everything you have done and everything you are doing!

[@Lolatronn]

10:35 Hey martin as much as as love the choice of using marbles for the clutch wouldn't it be simpler to use a plate with bumps on it? Alot of the complexity i see here is because of the use of marbles and by having a plate it could reduce complexity and compress the design down. It could also would make the disengage mechanism of the clutch simpler since you would on need at most one spring for the whole set of detents. since they are on a plate you could place a separate plate that is pushed by the spring that separates the spring from the plate with detents you could pull back on that spring and release the clutch that way saving another spring.

[@semerhi]

I think the marbles add "continuity" that the bumpy plate does not. Unless you break really fast and high force, separating the plate might be a jerky action with high friction between holes and bumps, thus degrading the bumps faster than marbles.

[@MH_Binky]

Machining a plate with bumps to a high enough precision to be effective would be a lot more complicated than machining a plate with holes and putting springs and bearing balls in them. And I imagine the metal-on-metal screaming that would occur during an unplanned disengagement (or a re-engagement in general for that matter) would be awful.

[@adam6806]

Fewer parts. I agree.

[@foldionepapyrus3441]

Define simpler? A plate with bumps on it is a relatively serious machining operation to create if you can't find a good one off the shelf, and even if you can could be tricky on the world tour if its not a super common part. And as the bumps are wear areas it will need replacement sometime. This is simple constructed method to get the result with parts that can be made almost anywhere, with very limited tools, and if you pick the right materials you can be sure it is the cheap and relatively easily replaceable balls that must be replaced most often.

[@jex0]

I think that would save a lot of parts but it could also limit the usefullness of the torque limiter since one of the main benifits of using marbles in the desighn is the low friction when the torque is too high since the marbles are free floating and mostly free to move. A plate with detents would allow more torque through the limiter since the unmoving detents would cause more friction. It would also wear faster while being less esily replaced and also heat up more and maybe distort depending on the way it's made. that being said I'm an electrical engineer not a mechanical one so I can't tell for sure.

[@kingofcastlechaos]

Love your R&D process. I am of a vintage where CAD, Laser Machining, etc. were just science fiction. So thrilled that young people such as yourself are jumping in and thinking outside the box. I do have an idea of what the ratchets are for, but I have so much glory banked from my glory days (kept in multiple 55 gallon drums buried in the yard) that I will pass this time and let the cool ideas of others take the prize. Keep up the good work, and remember to take all free advice with a grain of salt.

[@ArrakisMusicOfficial]

Love to see your progress Martin! I wonder, have you thought about the torque on these bolts? If I understand this correctly, all the torque goes through 6 bolts, right? I understand they are clamped well, but there will be a lot of shear forces and also general twist forces. Or am I reading it incorrectly? Also, when engaging the manual brake, is it possible that the moment when the two discs come into contact could create some issues? Like some weird thing with friction/lockup/too much torque on the brake system itself/...

[@fletchro789]

All good points. It's a good thing this is a prototype!

[@gnlaredo94]

About the second issue, maybe he use a bistate mechanism. Full engage / Full brake. Then when transition to Full engage, for an instant engaging the force will be strongest to not going in those friction/lookup problems. But anyways I'm worried about the 6 bolts managing all of those torque and energy without bending

[@ArrakisMusicOfficial]

@@gnlaredo94 thanks for your insight, I think you're right about the second issue, since Martin had these bistable mechanisms on MMX already. So if any issues occur, he can always put a chunkier spring on there, to get a shorter transient. Good to know I'm not the only one who sees the first issue, I'm looking forward to see how Martin addresses it in his design.

[@steveszp]

You did a pretty good job explaining something very complex while also providing applicable visuals. All in real time! I applaud you.

[@Reeuwijk78]

Those half moon connections are really something. A little peg holding onto that shaft wouldn't stand the tork. Just perfect modular elements interconnecting to make one solid struture. Well done Martin! About the bolt baring housing I'm a little more cautiously enthousiast as I really enjoy this idea, but a little worried about the point load ont the baring housing. Try to get some info on that at SKF (if you have not allready). Are those ratchets on the wheel perhaps for the future marble lift? You keep up the good work and don't forget to enjoy the ride.

[@andy4an]

the break also being a clutch looks totally boss. spending your complexity points on this part seems like a good choice

[@Escamilli]

Hello, I think there are a few constructive errors in this part. First, it is important to separate high energy parts (only the flywheel!) from low energy parts. As long as it is possible that a lot of energy can suddenly get into the construction, you will have problems. It is best to separate the safety clutch from the normal clutch and also separate the brake. The safety clutch is placed directly between the flywheel and the belt, this is where the greatest jump in energy is possible. Then there should be a V-belt. With an adjustable tension, you automatically get a manual clutch. And then a normal bicycle brake is sufficient as a brake in the low-energy part. Lots of love and keep going! ❤

[@madeintexas3d442]

I remember when I saw the iconic video of this machine I thought it was just a super produced stunt for a band as someone who appreciates 3d printing and engineering I now realize it goes far beyond that and really appreciate you showing your process because it truly is inspiring.

[@TS_Mind_Swept]

I didn't even notice that there weren't any edits, your delivery was that good (plus I tend not to be overly persnickety with little hiccups anyway since they're just part of regular speaking, almost makes the video feel more personal)

[@thomasdamiana9815]

That is some beautiful CAD, I love the look of the cut away of the clutch/ torque-limiter

[@dexterm2003]

Martin that is not how bearings are designed to work at all. Bearings are designed to be pressed into a housing and they actually compress when they are pressed. You should never locate on holes for bolts only locate with reamed pins. You also need to make sure to use either an angled contact bering that are opposed or you need a deep groove X contact bearing to handle the thrust load. Most bearings are only radial bearings and will die a horrible death if you use it in that manner. That is so much more important when using it on the flywheel. The bearings on the flywheel are the most important and critical for safe operation in the whole machine. Please please please use a single piece machined flywheel with integrated bearing races. Also please have to balanced by a contactor. You do not want your flywheel blowing out your bearing and exploding your machine!!!!

[@sebbes333]

*@Wintergatan* 16:13 I would add an outer "ring" around that "controversial bearing housing", it goes between the flywheel & the "lid" of the housing (extend the "lid" more outwards to cover the new ring). The reason is leverage, you might bend the "lid" with the necessary clamping force, but by adding the ring around it, it turns from a 1st Class Leaver to a 2nd Class Leaver (or 3rd class?), because the ring adds stability & "a point" ("a line" around the ring) for leverage for the "lid" against the bearing.

[@CezaryAkakios]

Not a fan of the D-style axle flat. You get a tremendous torque multiplier from the large flywheel down to the comparatively small diameter of the shaft, and all that torque wants to force the two split halves apart. Maybe I'm reading the diagrams wrong but the only thing constraining them from separating are those 6 or 8 concentric bolts in their laser cut holes. Those are going to have slop because they have to be clearance holes (you need precision ground shoulder bolts or dowel pins in reamed holes to eliminate that) which means they will move ever so slightly. All it takes is those two split halves opening by a tiny amount and now you have play and slop against the shaft, and slop on a shaft multiplies over time because every time it starts and stops it wears a little more -- especially when it's mild steel of the clamp against (hopefully) a ground polished hardened shaft. Shafting isn't an area that I'd get creative with, there are plenty of established solutions for almost every need. For example you might want to use a taper lock system, which attaches to a plain round shaft by friction alone and is very strong.

[@constantinosschinas4503]

He is a genius, he doesn't care. Been digging his genius grave for 7 years now.

[@SpringTime4]

Amazing. Speechless. The Art of Music. That's it!

[@helipad4050]

so I think the design would work, but do you really need to do it with so many parts? would it be possible to decrease the number of parts? the more parts (not just the moving parts) the more chances there are for part failure. I see places that could be one solid piece that is split into multiple, when manufacturing for that part would be easy in one piece too. more specifically where the marbles for the clutch are held you could make 3 of those disks into one part, with simple threaded holes in them. maybe you chose to split it up to make it easier to tap them, but an issue would be that it might not line up, and you would only need to tap part of it and have the rest not engaged to the bolt. even accounting for the keyway, this part shouldn't have much stress on it (the marbles will disengage before there is substantial stress) meaning that at least for most of the clutch you don't actually need a keyway, the bolts should be strong enough to hold it together aligning with the part you show near the end with the keyway (and possibly a small part of the clutch with one too). this would drecrease part count, decrease production cost and time, and while it is "cutting corners" it is one were the corners cut are into the copious margin of strength that part would have. to ensure you don't over load the bolts however I suggest doing a strength test on them to see how much force it takes to bend, shear, or snap them to get an idea of what safety margin you should have in this part. as youve said in previous videos decreasing part count helps with reliability and manufacturing. the old fly wheel design I think has less parts, now the breaking feature with how dangerous the flywheel can be is a need, so parts should increase, but the clutch part itself should be about the same. im just curious if there was a design descision made not explained in the video to why it is like that. maybe it was for using a large metal plate to cut out all the disks, though with a bar like that it could be less time consuming to lathe it to shape.

[@XiraArien]

I agree. Also no breaking material. Would have done a cycle brake setup and a correct bearing with flange housing

[@TijmenZwaan]

The 3 disks aren't actually disks. They're all split down the middle, and the split has been offset between each disk, so together they form a solid whole. This is so it can properly grab on to the slot that martin discusses near the end of the video. So they're separate parts for a reason.

[@XiraArien]

@@TijmenZwaan yeah. But it is a bad solution. You should use a woodruff key solution instead, which is much better for transferring torque and alignments.

[@helipad4050]

@@TijmenZwaan I updated my comment to contain my reasoning to reject that. though that does seem to be the reason he did, it isnt the best solution.

[@boom5474]

this is very exciting, we don't often see 'industrial' music machines! I can't wait to see what's next for the marble machine

[@PabloEdvardo]

you'll need to shield that pedal below the surface, at the moment it looks like someone could slip something inbetween the pedal and the board and it could crush it (like a foot!)

[@Spiceylizard]

I love this! It’s always so exciting watching your videos, and there easy to follow. Absolutely genius design on the clutch/limit bearing, it’s a great safety design turned practical

[@alexteplyh]

Hi, Martin! Couple ideas that came to mind for what you might use the rachets: 1. You might want to add a some kind of display, to see how much energy is stored in a flywheel, so you can get the rotational speed from those ratchets for that. 2. You might want to see the current position of a track/song that is playing like a progress bar. Yeah, most likely neither of those ideas, but I tried :D

[@circadianrebel]

I'm sure lots of people will/have comment that bolts are bad at locating, and that threads do not take lateral load well so will wear out and not locate the bearing properly (something shoulder bolts would help with a but), but there is another side to consider as well: when the bearings torque sideways they will have a tendency to jam the bolts, making it difficult (approaching impossible) to disassemble. As the threads wear (which will happen both from the bearings being torqued sideways, and just from normal sideways rubbing) this effect will get even worse, but even with shoulder bolts the rotational torque will cause binding.

[@constantinosschinas4503]

He does not listen. Don't bother.

[@deathofallthingspotato9919]

I have a fairly serious concern about how the braking force would be transmitted through the rotating component. Taking the cross section at around 12:04, as the pink disk is pulled back by the Bowden cables, it will press into the rotating cyan disk, and will apply forces backwards and against the rotation. The only thing keeping the disk rotating is the red rods, which are only solidly anchored at the yellow disk. These will likely twist under the force, bending the yellow disk, and crushing the spring into the channel in the layered disks. This will get particularly bad if there is any kind of disturbance on the disks surface, as this could catch and dig in. Then as the disk is pulled further, the braking force will engage. While the flywheel is not attached at this point, there may still be substantial momentum, and this will act to try and rotate the cyan disk against the rest of the system - again, a weak spot. Some of this force will be absorbed by the green disk, but the green disk has mounting problems too (bolts don't like being loaded like that), and if there is more friction between pink and cyan than between cyan and green (which is likely, due to the springs), then there will still be substantial forces acting to twist the red rods, which provide a massive lever against the spring on one end and the yellow plate on the other. There is also the issue of securing the pink plate well enough to provide the braking force, as it relies on a similar connection, although its rods are further out, and connected to a thicker plate. To resolve this, you could take several steps: 1: add some solid rods from the plate stack to the yellow plate, say 8mm, and have linear brearings connected to the green plate. 2: remove the green disk, and make the bolts on the pink disk further out, add some bearings or at least grease between the cyan and pink disks, and add another disk (the brake disk), firmly mounted to everything else, and wider than the cyan disk. Coat this disk in friction material. Then, as the pink disk is pulled back, it first pushes the cyan disk back, disengaging the clutch with minimal friction, then it pushes further and makes contact with the brake disk independantly. The brake disk has independant connections to the rest of the system, although has hols to allow the connections to the yellow disk to pass through it. This preserves the same functionality, although it needs somehwat longer bolts for the yellow disk, and avoids any substantial force going through anything moving in any other direction. As a seperate set of improvements: Make the pink disk be moved by levers so as to reduce forced through the bowden cables, add another disk the other side of the pink disk, so the rods holding it are mounted at both ends, and put some cheap linear bearings on the pink disk. I much prefer option 2, and combined with he additional changes, this yeilds a system where no part has to excerpt force across a moving connection - everything is either lubricated or has a bearing. Forces on the cable are reduced by the lever, etc. Edit: I didn't see that you planned to use the green part to transfer power onto the shaft. Please make this disk solid. The green has almost nothing holding the two halves together, meaning the turning force put from it to the shaft will act to split the halves apart. Splitting the halves like this requires far less movement than is needed to twist the bolts, so you're just putting move leverage against the bolts. Having it play this role doesn't even help with the braking, as it still relies on the bolts back and forth to hold the halves together, and the braking force would likely hit the cyan ring, which is connected without much strength - so the braking force hits the springs and the yellow disk. Also, the half disks in the stack are more aligned than I think they'd need to be - I'm prett sure it would be optimal to have none of them line up, as the thing hat matters is the strength of the connections between the halves, which is dependant on how many splits line up. Although this doesn't really matter, as it's not a weak spot anyway, and having them all be the same is probably better for ease of assembly

[@deathofallthingspotato9919]

Feel free to ask for clarification on this - I can hack together some cad for the proposed changes if you want.

[@constantinosschinas4503]

Why do you even waste your time and brain with this "project"...

[@deathofallthingspotato9919]

@@constantinosschinas4503 Because I can. My time, my brain. I mean, its more constructive than adding even more LAMS to my current FtD ship - the thing doesn't even get hit by projectiles any more honestly.

[@constantinosschinas4503]

@@deathofallthingspotato9919 I have spent a huge amount of time jn this project, Discord too, for the very same reason you mentioned. Doesn't worth it, nobody listens, all martin groupies (which ruin him more than helping him). Regards.

[@deathofallthingspotato9919]

@@constantinosschinas4503 Fair enough - I've deliberately avoided the discord because I don't feel its worth it, but I do like analysing the problem. Honestly though, I quite like looking into failing projects, so I might give it a look.

[@AaronKay]

I'm guessing the ratchets are for a small motor which can spin the input shaft at a specific RPM. The teeth are grooved so that if the flywheel runs faster than the motor, it will just slip past that tooth.

[@Zedandlev]

Always a good day when Martin uploads

[@psy0rz]

wow that revolute trick really opened my eyes. i would normally had created the more obvious sketch (circles) and then end up with a horrific stack of skeches and extrudes with offsets etc

[@huub8]

Since you spend so much time in CAD, might I suggest buying a 3D/SpaceMouse? I only recently got one, and it is amazing. (It will allow you to move, rotate, pan and zoom AT THE SAME TIME. which will not only greatly improve your speed but also results in very smooth motions for your CAD video’s. It does take a little bit of time to get used to, but after a day or so you’ll never want to design without it. )

[@Robinlarsson83]

I was thinking the same! I got myself a great deal on a lightly used Spacemouse Pro last autumn, and it's just fantastic! I can't believe I've been working in CAD for close to 20 years without one.

[@JaxonJJB]

Man this is some crazy engineering! I'm in highschool FRC Robotics and we do a lot with bearings, machining, and motion. We also use a lot of CAD, and I'm impressed with the detail you have designed with. I really like the ingenuity of your hybrid clutch-brake assembly, and even though it has a lot of pieces its very cool. But one thing I have learned from our engineering mentors and building a whole robot, and I'm sure you've learned in redesigning the marble gates, is no matter what, simplicity is better. Always. I had a concept for the gripper of our robot that we managed to make work, functionally, but in practice was not effective. But I kept engineering it and trying to make it work, because I thought it was cool and inventive. But simple is always better. Always. Even if you lose some small nice function, or have to leave a brain child behind, do it. I think in your quest for radical pragmatism you've possibly strayed from that path. If you wish to really make it work, I'm sure you can, you have a lot of talent and drive, I'm really impressed!

[@maxvilocity]

Nice to see you again, Martin!

[@Sulphur_67]

you could put literally motorcycle brakes onto it, it’s a mechanism that is already built and has bolts for mounting

[@bricemackenzie8427]

just a thought in regard to the pedal and platform positioning (I may have observed incorrectly, if so disregard.). If the standing platform is level with the highest point of the pedal, pumping will use far more energy than if it is level with the lowest point because every push will require you to move your entire body up and down. If the platform is at the lowest point you can still push all your weight but when things are going, you only need move the one leg to pump. If the platform is at the peak to avoid crushing, you could just add a skirt that stops anything from getting between the two whenever the pedal is above the platform.

[@Benlf]

This. I seem to remember that the platform was there so you didn't get a foot trapped under the pedal. If so, you are going to use your right leg to lift and lower your whole body to operate the pedal, which will be very tiring. If you attach the pedal at the back hinge point and let it float on the cam point (rather than a hard attachment to the cam) that would reduce the crush potential.

[@grymkaft]

You can also still get crushed between the pedal and the underside of the platform, seems almost more likely than the reverse.

[@bricemackenzie8427]

If the pedal were just made taller than its range of motion then nothing could get beneath it. So long as the range of motion is less than the height of the lowest point so it won’t collide with the ground. Like an elliptical cylinder rather than the flat oval that it currently is. (this is what I mean by a skirt as it can just be a hollow wall on the bottom of the pedal)

[@thejatomis]

Organ pedal, design it like an organ pedal!!!!!

[@gringene_bio]

First minute: "an engineer's approach; form from function" Nek minnit: "A torque limiter using marbles"

[@francisdorval2533]

You might have done that already, but I'll ask the question anyway: have you put some thought into the REAL necessity of all the adjustments you want to include? With the amount of adjustment nuts you have in this assembly, my first thought is that this will be a nightmare to adjust and make sur everything is parallel.

[@PrixyPurple]

Martin I just love how happy this design process makes you! I too am a non-engineer who is getting into CAD and engineering style design and I often feel the same way! It is really delightful to see! (If I’m crazy then we both are and I’m in google company)

[@intocoasters]

The stepped bits are to prevent the flywheel from turning the wrong way. ***EDIT*** Thinking on this more, I wonder if there will be a spring loaded ratchet that will allow the flywheel to turn in the wrong direction at first when the pedal is pressed. Then when the pedal is released, it would push the wheel in the proper direction.

[@D3nn1s]

Could be, but its unlikely. It would add a lot of noise and vibration, its more likely itll interact with an instrument or so but really hard to tell

[@intocoasters]

@@D3nn1s I wondered about the noise as well. But there are ways to disengage an anti roll back once a flywheel gets up to speed.

[@sonefall6268]

A possible use case for ratchets on a wheel is to count the rotations or measure the speed. But in that case you would only need one ratchet. In a previous video martin said "ratchets make the world go round" and corrected that statement to "ratchets make the world stop". The ratchets might be there to prevent a rotation of the pulley in the wrong direction, eg prevent cranking in the wrong direction. They could also be for a musical purpose like the cymbal machine from the mmx

[@roboman2444]

Instead of 4 cables, look into how manual transmissions in cars operate their clutch with the clutch fork.

[@constantinosschinas4503]

or buy a clutch. they come in all sizes, specifications and configurations.

[@PhilWaud]

Your designs are getting so sophisticated, and beautiful! This is a fantastic design, thanks for sharing. X

[@troelsnielsen2848]

maybe you should do something that converts the 4 bowdencalbles into one. its 4 things to calibrate and 4 things to maintain and replace.

[@LReBe7]

Nah, that's one for the department of redundancy department.

[@mickpauline4630]

The main issue I can see with the bearing housing is with wear, not with concentricity (provided you can get the parts machined accurately). If you have some imbalance in the pulley, the hardened bearing will soon start to wear through the bolts. You could use off the shelf flange blocks that constrain the bearing axially but will still allow you to get concentric alignment if your machining isn't perfectly accurate

[@MV-vv7sg]

I can only suppose but I’m quite sure that Martin loves his conceptualising and philosophy of engineering and spending time in cad. I think he ought to do as much of it he enjoys and we all wait patiently until it arrives in the omnipotent form in reality we’ll call the Marble Machine. There’s no rush. We know this. I hope he continues to enjoy his work.

[@OutermostGold]

i now realise that you get more views the longer it takes & i'm concerned that you've bene a one hit wonder all this time, its beautiful to watch you build this but it is looking more and more like its too good to be true.

[@deisum]

Martin, you're quite literally reinventing the (braked, clutched) wheel.

[@disposable157]

That pedal looks so unsafe 😭

[@mackit]

Seriously, my immediate thought was “this thing is going to break your ankle or shatter your foot!”

[@srttechie]

My only concern with the flat spot on the drive shaft is over time with the rotational forces on it, it may round the edges of the flat spot. If that happens it will either start to slip/spin or just have a lot of wiggle room to move and damage it further. The clutch settings for torque limiting will need to take that into account. This is the problem the keyway was designed to get around.

[@WizardTrixx]

I need my fix 😢

[@tobyCornish]

Anyone else feel like Martin is sliding back into old habits of design for design's sake in this video?

[@harmonic5107]

I think its important to make clear the thought process. People may look at this and think it looks very industrial and not very Martin. But it's just because you are working twords the minimum viable product. Design will happen later. Just to reassure people who may think this is what it is going to ultimately look like.

[@jamport5973]

My guess for the ratchets, actually a ticking mechanism for the main wheel house

[@GameHut]

EASY SIMPLIFICATION IDEA - Feels like the ring attached to the cables that moves the plate towards the brake will wear very quickly. It effectively acts like a brake already. Instead of having a separate brake plate, why not just thicken up the plate attached to the cables so that can be the brake. Or am I missing something?

[@D3nn1s]

Thanks for wondering off, i was kinda expecting you do model the holes kinda like you did by revoluting twice but it was nice to be confirmed in that assumption :) As a machinist i actually think your bearing idea might work out, getting it centered will def be a challenge, however getting this on a lathe would either need a huge chuck or more holes, so i think your idea is pretty good. The "no key idea" would also be good for a single part, however since youre plitting 5 discs or so into halves your actually reducing the stability of them while adding more parts to the assembly. The best solution would probably be to simply extend the flat spot all the way to the end and slide the discs onto it. Since its getting machined anyways it doesnt matter how long the flat spot is from a cost perspective.

[@TheBradinator214]

I'm so glad to see you continuing the marble machine legacy! I've been along for the ride since the beginning. I noticed a potential safety harzard with the pedal. If the pedal is designed to go down into the platform when moving, then if your foot slips off of the pedal to one side then it could get chopped off or crushed by the platform. Perhaps you could put in some safety walls under the platform similar to those you put on the pedal itself. Food for thought

[@thejatomis]

Design it like a freaking organ pedal

[@haatch999]

8:22 well u litterally just solved a 4 month problem i had intracate internal lay out that master sketch ill have to practice as thats a stroke of genius always here for the awesome work and insane yet accomplishable idea hope u feel better soon as this is awesome

[@CapableCaptain-ahoy]

please make more cad videos like this 🙌! This is super informative to my cad endeavors; not to mention too that this is incredibly fun to watch.

[@gerrywalsh5766]

Very elegant Martin... Your design is very much like like the Slipper Clutches I have on my high power RC race cars. You have gone a step further by designing a brake into that system. The Slipper Clutches I use have 2 metal plates with bearings in between them. Nice, quiet and smooth power delivery. Awesome work there... You continuously amaze me.

[@dennisash7221]

Hi Martin - I suggest that you add a magnet to the springs so that when you engage the clutch the marbles will remain attached to the springs and withdraw in a more control. I like your bearing solution, it means a lot less complexity in the production.

[@advanceringnewholder]

your first flywheel is really good. real simple. maybe back to that

[@mattsoup4121]

I see a lot of "it's so complicated do x or y solution that will require a lot of work to accommodate" but in reality it's going to be much more efficient to just pick a solution and fix it if it doesn't work. Iterative design isn't the cheapest way to do something usually, but it's definitely fast and effective. You might have a laser cut steel sheet assembly that doesn't work, but you can easily get a new part designed, cut, and installed within a week. If an off the shelf assembly doesn't work you now have to put in a significantly larger amount of effort to replace it.

[@raphi25895]

18:00 I work as a design engineer, we have already mounted bearings in this way on a trade fair model. We only used fitting screws ISO 7379 for higher accuracy. We had the parts laser cut

[@mnoreke]

Another fun video to watch. A Fusion 360 tip for you: Look into doing a Motion Study to show how the mechanisms work. It allows you to animate your parts based upon joints, so when one joint moves, the rest are acted upon accordingly so that you can see the effect of one action on the rest of the machine. It will be much easier to follow than selecting a few pieces and trying to explain what moves where. :)

[@Heliforce]

Great to have you back. Love your videos. Not sure if some one else asked but, why not use a simple clutch from a moped or motorcycle? They just need one bowden cable and are easy to find.

[@christoney2491]

Right on! That's a great idea for a torque limiter. I'm an engineer (software/systems) and really love all things engineering. I especially love mechanical engineering. Love watching your videos. Thank you.

[@INSIDEHARDWARE]

Clutch regulator is identical to that found in many battery powered drills. When selecting the clutch strength, tightening or loosening, you are changing ball and spring tension by changing the distance of the backing plate with drill head. What Martin is achieving with the Bowden cable is done with threads

[@pointlessone3702]

The one in which Martin cashes out on all the saved parts from all the previous videos.

[@don.3s]

For the bearing housing you can just cut some extra plates and sandwich them instead of letting the bearing rest against the thread of the Boults

[@MrPhippo91]

Hi Martin, please reconsider your choice of using the torque limiter as a clutch. The torque limiter is not designed to re-engage when there is a difference in speed between the flywheel and the other side of the torque limiter. You would be better of using a proper friction based clutch for that purpose in addition to the torque limiter.