Faster Flywheel Plays Tighter Music

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.net/#/shop
SPOTIFY ► bit.ly/2oKxXWd
ITUNES ► apple.co/2ntWNsZ
ENGINEERING DISCORD ► / discord
COMMUNITY DISCORD ► / discord
-
- As expected.
Follow the Trainerds adventure: / @trainerds
The prototype is designed to answer the age old question: Can I play tight music using a very strong flywheel? The new flywheel have 20x more moment of inertia compared to the MMX Flywheel.
We are recording at the wonderful location of Siegfrieds Mechanical Music Museum in Rudesheim Am Rhein, Germany. Check out their youtube channel: / @musikkabinett
Edited By the Glorious Hannes Knutsson From the @Trainerds KZbin Channel
Trainerds: / @trainerds
-
MUSIC DOWNLOADS ► wintergatan.bandcamp.com
WINTERGATAN RECORDS ► www.wintergatan.net/#/shop
SPOTIFY ► bit.ly/2oKxXWd
ITUNES ► apple.co/2ntWNsZ
MERCH ► teespring.com/stores/wintergatan
DISCORD ► / discord
SECOND OFFICIAL CHANNEL:► / wintergatan2021
-

[@fusionaddict]

1) I realize the longer handle on the hand-crank was difficult to manage but the shorter one is why the cranking is so difficult. The more leverage you have, the easier the crank will be to turn. 2) Consider relocating the hand-crank to be accessible from the pedaling platform. That way you could easily and quickly shift between cranking when your hands aren’t needed and pedaling when they are. On a song where your hands are needed the whole time, you could use the crank to get up to speed much more quickly and then switch to the pedal to maintain. 3) As mentioned above, the flex in the pedal is absolutely costing you leverage and therefore is going to make your timing less consistent. Making the pedal a truss rather than just beams would greatly help. 4) Playing faster is definitely making things tighter so I think it’s clear at this point that a gearbox is a necessity. It’s going to be far easier to crank 120 and gear it town to 60 than to try to consistently find the right pressure to crank 60 consistently. 5) At some point you may want to reflect on your standard for tightness and whether, in this case, perfection is becoming the enemy of progress. You are measuring the deviation margin on your machine in single-digit milliseconds - that’s *thousandths* of a second. In many cases, these are variances that are going to be inaudible and undetectable to the vast majority of listeners. Indeed, the only reason we are even able to hear it is because you are letting us hear the click track. As it already stands, your machine will likely have more consistent timing than any human rhythm section and because all of the attached instruments will be working off the same timing, any inconsistencies will be pretty much inaudible. This is a mechanical contraption, not a computerized drum machine. Parts will stretch, parts will bend, parts will expand and shrink. There has to be a point where you accept that any mechanical device will have inherent limitations and that those limitations are part of what makes that machine unique, just as how no two guitars or violins will ever sound truly identical, even if made by the same hands with wood from the same tree. These variations are what give your machine its character, its uniqueness, its soul.

[@kidgrit]

This is by far the best comment. agree with all 5!

[@stevenwagner1236]

Please everyone bump this comment up

[@coreymcardle1916]

If you look at the safety box on the pedal, there is a large amount of flex. This HAS to be a portion of your consistency issue. If you extend the box to cover the entire length of the pedal bars, you can quickly strengthen the entire assembly. Thank you, enjoying your efforts!

[@hjalfi]

The crank shaft is bending too, when putting force on the crank, presumably because of the unbalanced nature of its operation. That can't be good. ...it occurs to me that a way to combine the kind of power you get from your legs with the simplicity of a crank in a balanced fashion is via bicycle-style pedals. This is particularly useful because (assuming it's made from a spare bicycle!) you typically get a shifting gearbox for free.

[@Steppenkater]

That's what I was saying

[@NOLAfugee]

Or some sort of metal truss to make it more rigid. He already doubled up some of the square tubing, which was a big improvement.

[@JWSpradlin]

@@_fishythat’s actually a good point. They should make the pedal be a balanced seesaw instead. It has the added benefit of allowing two people to power the machine via two platforms.

[@AdelaeR]

There's also a lot of energy being lost in flex. Extending the box to form a large triangle is indeed the way, but It'll be better to extend it with a piece of metal.

[@Chittebengo]

It's fun watching you prototype a musical exercise machine.

[@michalgrbk]

Collab with Flake of Rammstein is coming.

[@NaisanSama]

Hopefully they add brakes.(that doesn't wear the flywheel)

[@Knowbody42]

@@NaisanSama Maybe he could make a disc brake for it.

[@okieo2k958]

I’d like to point out that the increase in speed should naturally correlate to an decrease in the standard deviation because there is just less time in between the beats for you to get off.

[@doismilho]

each video that passes I feel like martin and hannes' lifes are more endangered by this machine

[@kidgrit]

the speed of that fly wheel is terrifying honestly

[@thetruthserum2816]

not to mention that it's going to cost a fortune to ship to each stop of the World Tour + the additional backup machine.... ;-)

[@amarpersaud2950]

A correction; "Moment of Inertia" is the angle equivalent of mass. By increasing the angular speed, you are increasing the angular momentum of the wheel. You get L=Iw where I is moment of inertia, w is angular speed, and L is angular momentum; higher speed or moment means more momentum. More moment of inertia does = tighter because the "mass" is higher, and this means that the same force (friction) results in a smaller change in speed (loss of speed). This also means its harder to turn. You can also increase the speed to increase its momentum which has the same effect.

[@AveryIII]

Correct. Moment of Inertial is a physical property of a rigid object that is the same whether at rest or in motion. When you added weight to the flywheel you added moment of inertia, but spinning it faster did not. The spinning skater pulls their arms in and changes their Moment of Inertia which in turn makes them spin faster. Spinning the skater faster (don't ask me how) without pulling the arms in or out does not change the Moment of Inertia, but the angular momentum does change.

[@John_Weiss]

As someone with a doctorate in physics, I approve 👍 this message. 😉

[@TheLivirus]

Another way to look at it is that kinetic energy E of the wheel is proportional to the square of angular velocity w. So, the higher w is, the more work is needed to change w. E = 0.5*I*w^2 or dE/dw = I*w

[@yobgodababua1862]

This. it was only 0:22 and I had already started looking for the comments...

[@daemn42]

Came here to say the same. Also, because rotational moment of inertia depends *only* on the sum of all mass at all radiuses squared, you can of course get the same moment of inertia with less total weight by simply distributing most of the mass near the rim. I ran the calculations last week for like a solid 80Kg wheel with a 50cm radius and found you could get the same moment of inertia at only 40Kg with all the mass distributed between 50-55cm radius.

[@Elias_Ainsworth92]

It would be interesting to see the tightness of you playing just a normal kick drum as a control.

[@gavinbarrett7890]

I would love to see this too. I understand Martin's desire for tight music but struggle to understand why this instrument (powered by a human) should be expected to perform better than any other instrument played by a human. I don't think there is a live band on earth that would meet Martin's standard for tight music.

[@jrgallagher1]

@@gavinbarrett7890 He probably can't play a kick drum as accurately as this but two things - the response time of changing the tempo of a kick drum is effectively the time between one beat and the next. If the flywheel falls behind it could take several beats to get it back on track. That is the cause of the "waves" Martin describes. - The whole system can't be better than just this subset of the system which is why he's working so hard to reduce variation now.

[@BensMiniToons]

@@gavinbarrett7890 For playing a instrument, 5ms tolerance is the best case within human ability. We are imperfect. That's why all these tests are redundant. They only demonstrate that mechanical assistance is superior to human skill alone. The flywheel weight should only be the minimum required for the stump pedal to work smoothly. Practicing with a reasonable flywheel weight would yield more benefits. At this flywheel mass. Altering the machine's speed/tempo is not reasonable and not feasible for a world tour.

[@Maric18]

i kind of like the imperfections of a life performance a music box with a few very slightly detuned/dampened/metal fracture having prongs slowly losing spring tension as it plays is so much more magical than a midi of perfect sine waves

[@jrgallagher1]

Spinning the flywheel faster increases its momentum. The moment of inertia of a solid object is a constant.

[@anteshell]

That's false. Solid object may still change its shape, also changing its moment of inertia. You meant to say "rigid" object or body. Just a reminder that when you confidently try to assert your intellectual dominance, the minimum effort you ought to do is to make sure you're actually correct in order to not have an opposite effect and make yourself a fool.

[@thecheeseman1662]

@@anteshellshut up

[@magingace5199]

That was bugging me throughout the video XD

[@petrathespacerock4764]

@@anteshelldon't be pontlessly rude to people, plus, the amount of expansion is so negligible in this case that it is essentially nothing, while you are technically correct, the whole thing expands so little that it likely can't even be measured by conventional means, my point is, don't rely on negligible amounts of change to justify being rude to people

[@BetaKeja]

🤔Then what term describes the increased inertia of a system due to gearing changes or changing the moment of inertia?

[@Jellooze]

a gearbox mechanism would be a great idea. especially if you add a neutral gear, would be nice to separate the pedal from the flywheel on the fly if needed

[@FreejackVesa]

I agree. The gears could shift to standard flywheel speeds that equate to tempo. So if you are pumping foot at 1 cycle a second (down, up) and you shift gear to low it would go to approx 80 bpm. Other gears would be other commonly used tempos for genres of music. Or you could go completely "automatic" and the speed you pump your foot is the speed of the programming drum and music

[@xander1052]

100%, another idea for a gearbox could be to use a CVT to maintain an ideal flywheel speed across all BPMs possible, albeit they do come with downsides themselves.

[@Porglit]

Huygen drive makes so much sense in my mind. The whole problem of trying to exert yourself exactly the right amount to keep the wheel going exactly the right speed sounds insane when you apply it to a world tour context. Separating the power generation and power output with a Huygen drive sounds like a dream come true both in skill required and tightness of the music.

[@bairdtownsend6546]

If we're abandoning human power as the power input, why not go for something less complex that can apply consistent force over a longer period of time, like electricity?

[@NormanTiner]

And if we're going to simplify it further why not just buy a midi keyboard.

[@kidgrit]

@@NormanTiner We could just listen to a recording honestly. Just put the machine next to a laptop and we good - see you in 2 hours

[@DeFaulty101]

Archimedes said that. He was at the heart of almost every mechanical principal which makes your machine possible.

[@mrasmus]

My impression is part of the drift you see with the foot pedal input (and honestly any input into the system) is coming from your feedback system -- the flex in the pedal creates a disjoint relationship between action and result (beat), and so you have an awkward offset between "foot at bottom of travel" and the actual beat, so you end up just having to listen for the trend of fast/slow, instead of being able to be your own metronome. In practice, the machine itself will be its own click track, and the drummer for any other accompaniment, so syncing up with a metronome/drum track means you're chasing to "catch up" when you're behind, which results in overshooting your target BPM/speed, and inducing an oscillation. If you were trying to accompany a drummer outside the machine, this would make sense, but I don't think it really matches the ideas involved in performing with a MM. Have you considered adding another feedback mechanism to help measure your speed/BPM, other than just listening/comparing to a metronome or drum track? A photointerrupt and an arduino would be a very cheap solution that could give reliable, live feedback of your true BPM pretty easily, I would think.

[@josephhfry]

I would use a timing strobe. Basically put white tape stripes around the drive wheel exactly 20 degrees apart. Then have a flat panel close to the edge of the wheel with a rectangular window to look through with a needle across the middle and a strobe behind that flashes according to the desired tempo. For example at 80 bpm, it would need to flash 24 times per second to flash once per 20 degree segment . Then as you pedal you will see your white line every time the strobe flashes and will see if your ahead or behind the tempo based on if your line is ahead or behind the needle. I hope this makes sense? It's basically the way a mechanic sets the timing on a car using a strobe.

[@abitofabitofabit4404]

Also, the down beat click falls at the very bottom of the pedal stroke. That is not so for the crank, and I think that makes the pedal tempo less stable.

[@mrasmus]

@@josephhfry Not familiar with exactly that style, but it sounds sort of similar to the strobe that some turntables use for helping to tune platter speed, which is actually really intuitive (you can see the tuning dots "drift" one way or the other when you're slightly off-speed, and they stay still when you're spot on; they usually also have multiple rows of dots for different target speeds). I suspect the same effect could be achieved with a variable strobe frequency, which may be similar/exactly what you're talking about. Regardless, some system for measuring the rotational velocity more frequently than once per rotation, regardless of what it is, seems like it'd be valuable for reliably correcting RPM without overshooting. You're basically creating a human PID controller, and you need a higher sample rate, heh.

[@mrasmus]

@@abitofabitofabit4404 Absolutely a good point, I hadn't noticed that the crank is out of phase with the pedal, though it's hard to say exactly where it should be; I'd think at the end of the "push" action, probably? I'd assume that'd be something Martin would be able to intuit from the feel of playing it/his past machines as well.

[@mikeciul8599]

What about running the machine's click through a delay? For example, at 80 BPM you would run it through a 750 ms delay. Then you would hear the previous click at the exact time you expect to hear the next click.

[@wabash9000]

This video reminded me of a girl at my university. She would ride a scooter everywhere even when she didn't have anywhere to go. I saw her walking one day without her scooter and her right leg was a lot stronger than her left, and as a result her gait was off. I feel like that is going to happen to you with this foot pedal. For endurance you may want to extend the platform you are standing on so you can use either leg rather than just the right. You might also make some sort of CVT on the belt drive so you can make it easy to accelerate.

[@grahamthomas9319]

This is a great point and a hilarious story. Although I do wonder if it’s slightly exaggerated lol. 💪😂🤳 maybe she hurt her leg was limping and that also explains why she wasn’t riding her scooter.

[@foldionepapyrus3441]

I think the sensible option is to use a CVT to define/refine the timing anyway - pedal at a rhythm that is comfortable and set the CVT to correct the difference between what is nice for the human to pedal/crank at and the music, which also has the added benefit of letting you accelerate all the music playing drums of the machine slowly too. Which really will be required as if you just dump the clutch on that monster flywheel you will bend and break something in the machine, as it isn't sane to even try and build the whole machine heavy duty enough to take that level of acceleration and if you have to stop that monster flywheel to stop the drums to change songs all the time the audience will be watching Martin and Sir 3K exercise for a few mins to get up to speed on every song change.

[@abitofabitofabit4404]

Note the 180 at 5:47

[@wabash9000]

@@abitofabitofabit4404 I saw that after I left the comment, but he isn’t going to want to do that in a concert and turn his back to the crowd.

[@BanyanBirb]

This makes me wonder, how tight can the older mechanical players in the museum play? I'd love to see some of them put to the test!

[@Gefionius]

Clearly the higher the momentum, the less the variance based on human input. That is easier to keep on beat. But at the same time, once you get significantly off beat, it takes a lot more work to bring it back.

[@Nullpersona]

Not only that, but "expression" in the sense of real time variation of tempo, is suppressed by the consistent drive provided by the flywheel. Without a way to vary the energy drawn from the flywheel, the tempo will be artificially smoothed. Less variance by input means less expression.

[@rudie2902]

Hi Martin, as you said yourself the axles are currently too thin at 20 mm. NKF suggested at least 30 mm. I suggest to go to 40 mm when you eventually go back to Rüdesheim to test the governor. I imagine the increased axles diameter should improve the timing tightness a lot (as well as balancing the flywheel). Rudie

[@DominusFeles]

SKF* 😇

[@jrgallagher1]

Not as much as you might think. The axle, being at the center of rotation adds relatively little to the moment of inertia of the system. He's doing better by increasing the mass, but would do better by increasing the radius of the flywheel, even if he kept the mass the same.

[@Tobi-zd9zp]

@@jrgallagher1a bigger radius means more space, an space is a rare good on a machine like this

[@scaesic]

But more mass means less portable. He already needs lifting aides, which is not good for a world tour.

[@jrgallagher1]

@@Tobi-zd9zp The trade space includes flywheel mass, flywheel radius, flywheel speed, and desired timing precision. Pick any three and optimize from there.

[@angellestat2730]

At the end, when you finish adding all different mechanism and gears from the marble machine that will add friction, the power input would be high enough that would present stamina issues to keep up a whole concert at the same time you keep focus in all other aspects of the marble machine. That is why, I repeat for 5th time my recommendation that you should use an electric motor for your first marble machine version, once you already done with all other issues, in a next tour you can see if it worth the effort to add manual power input. Remember, all church organs use electric motors to compress the air, power input does not have nothing to do with the music that comes out from the machine.

[@quentinb8011]

I wonder how the loss of energy from all the systems you're gonna attach is going to impact the tightness ?

[@jdmjesus6103]

This is a valid point. It needs a weight and a governor or it'll be far too much work just keeping it in time. The more attention paid to that, the less he'll be able to use for the music.

[@garysmcdermott]

@@jdmjesus6103 BIGGER FLYWHEEL!

[@D3nn1s]

Probably not, the levers it will activate take a misicule amount of energy that is pretty constant as well compared to the amount of energy that is in the flywheel. Its something to be considered for sure, but i mean its exactly the same as the MM and the MMX and it wasnt a problem there.

[@BensMiniToons]

its unsustainable parasitic drag all ready. The machine its self will have a lot of Inertia without the flywheel. This test is redundant to practicing with the real thing. Practicing with no weight would be more beneficial then adding assistance in the form of a over sized physical capacitor to smooth energy output. Twirls finger around ear at

[@BensMiniToons]

@@D3nn1s the problem he is solving, wasn't a problem before. I'm so lost. is this just to smooth timing or to add the stomp petal? either way practice is more useful then more flywheel assistance smoothing. humans are in the 5ms range. All the weights tested have been superior in timing to human ability. perfection will kill this project like the last 2. I've been here since the beginning . He needs a realistic tolerance for imperfections.

[@maunaowakea777]

you need to develop safety features for that flywheel. Bro! consider the energy stored in this at higher revs. There is a small chance but very high risk.

[@paxinum]

can we please make a baseline comparison for 'tightness' of music? That is, how tight can Martin play on other traditional instruments?

[@JonathanKayne]

This! Martin keeps using the word “tight” without giving some kind of numerical value to compare to. I get that it’s subjective but the word “approximately” exists for this stuff! What is tight? 1ms? 2ms? 5ms? 20ms? Right now we have no clue and having a human metronome would give us a baseline for that.

[@abitofabitofabit4404]

Arpeggiators and sequencers are also part of a live Wintergatan performance. Phase lock isn't optional.

[@daar1113]

I'm a bit late to this convo but I really think the marble machine will have to set the tempo rather than try to match it. If he wants to mix marble machine with sequencer machine then he will have to come up with a way to send a speed signal from the marble machine to the input of the sequencer so the sequencer can play along with the marble machine rather than the other way around. @@abitofabitofabit4404

[@bstiggy]

Martin, if you want to get some realtime feedback, you can put a rotary encoder on the system and with an arduino or RPi, be able to run a display with realtime BPM and StDev. If you dont want a proper encoder, then you can put some reflective tape strips spaced evenly on the flywheel and with an IR illuminator/receiver you can make you're own encoder. Just a thought, keep it up!

[@petervantilborgh5385]

A thought crossed my mind: You could replace the single pedal by a bicycle pedal and chain system. As a result, you get a safer input system (as it could freewheel and will not crush your feet) It will also allow you to double your inputs (2 pushes per cycle), which I suspect might make the input tighter and therefore the output as well. Furthermore, using two feet might be less exhausting . Also, as you could use gears on the inputs, it should make accelerating the flywheel easier as you can start in a lower gear. If you use a tandem bike as the basis of the system, you could even get somebody to help you pedal for startup. Another thought would be to increase the size of the lever for the hand crank, which should make it less hard to start/run.

[@Nn-3]

And it frees up both hands 😃

[@sswpp8908]

I think the problem with any freewheeling drive is that it appears Martin wants for the bottoming of the pedal stroke to sync with the beat. If the pedal doesn't move with the wheel then it loses that alignment.

[@nycdotnet4751]

yeah a recumbent bike feels like a great solution for this - keeps hands free, safety/freewheeling, double strokes, well understood gear-shifting, etc

[@EclecticMystic]

Problem with a chain drive might be noise. He's made having a quiet drive system a priority, and chains aren't great for that.

[@tigerzero5216]

@petervantilborgh5385 Your powering the MMX idea sounds good. ... I'm assuming rotating pedals like on a bicycle. Think about how it would look on stage. Bicycle powering the MMX? I quote "Furthermore, using two feet might be less exhausting ." That's the answer! One foot/leg on one pedal is going to get tired before the song is over. Make it a two leg pedal stair climber design.

[@Tvngsten]

This is without a doubt my favourite youtube series at the moment

[@Ragdollmaster333]

I think changing to the drum beat is a good idea but it changes a variable which also skews the later test to look better.

[@davidshiloh]

You should try distributing more of the flywheel's mass toward the outer circumference. That will increase its moment of inertia without having to increase its speed or overall mass.

[@imblackmagic1209]

although imbalances in the weight will be accentuated that way, this will introduce vibrations which are probably unwanted

[@timebomb418]

Assuming he's willing to make it thicker or bigger diameter sure, but an equal diameter circle will always have a bigger moment of inertia than a ring. Which is better just depends on if you care about mass or are resigned to use a forklift as part of stage setup.

[@maximilianheinrich1869]

You could use a variable Transmission to be able to make Sure that each BPM seting has the same Moment of inertia

[@morpheusmemnoch4160]

it doesn't even have to be variable, a gear for each tempo. I'm thinking about drag here.

[@maximilianheinrich1869]

@@morpheusmemnoch4160 I would think adding more gears would increase friction while a variable transmission could be implemented just like the belt reduction in the current design

[@morpheusmemnoch4160]

@@maximilianheinrich1869 Any belt driven system has friction and a belt driven CVT has more by design. or are you talking about something you would find on a pillar drill with different varying sized pulleys for different speeds?

[@maximilianheinrich1869]

​@@morpheusmemnoch4160I was thinking of a CVT and as of the friction aspect I'm not sure but I don't have much experience in transmission so you could be right

[@morpheusmemnoch4160]

@@maximilianheinrich1869 I suppose that 2 cones inverted to each other with a belt that could be guided from one to the other would be a compromise but difficult to get right.

[@droko9]

You should attach a motor to the flywheel shaft, not to spin it up, but to brake it electronically. If you hooked it up right then you could break quite fast and also measure how much energy was stored in it

[@colbylippincott7173]

If you want a foolproof way to keep it tight and not add tons more weight, add an assisting electric motor with a position encoder. It can kick in and adjust the speed slightly to tighten it up. It doesn't have to be a large motor either because you can rely on manually getting it up to speed, and it just maintains that. Add in a switch to turn the assist on and off, and you can change tempos easily then lock in the tempo again by turning it on. I know you want to shy away from electronics, but this would make your life way easier and the machine much simpler, safer, and play tighter music.

[@bossoyok.officiel]

You know what ? Since the first MM, i follow you, and i would thank you, not only for your work, but because you are my first anglophone chanel, and your passion and interesting content openned for me the huge door of english content. + im a musician too, and your job motivates me a lot ❤❤ Edit : i know this comment is not really related to this video, but i hope you and your team will see it

[@Ass_of_Amalek]

you really have to use the internet in english, as you will find more quantity and quality of almost everything than if you use a different language. that's how I largely taught myself english. streaming english language american film and TV content because it's much better than german dubs was also a very large part.

[@mdhebert]

It sounds like a good candidate for the application of a electric motor like those on ebikes, where it assists without taking over the rotation.

[@Guds777]

This is THE Most Complex way to play the bass drum. I like it...

[@user-yk1nh8jm1l]

Martin! Loving these iterations! One thought: instead of using standard deviation as a measurement of tightness, start using mean normalized standard deviation. This way the metric correctly scales with the bpm as a value that shows consistency of repetition

[@douglasburnside]

My wife claims that I am OCD because I carefully align the labels on the cat food cans in the pantry so they look nice. It takes me an extra 30 seconds once a month to do that, and they really do look nice with all the little kitty pictures lined up. God only knows what she would think if she ever watched this video!😀 Sometimes I wonder if you are not going beyond the point of diminishing returns in your everlasting quest for tightness, where you have to expand the waveform on an oscilloscope to see the difference between tight and an infinitesimally less tight performance. Eventually will you reach a level of precision where you can match a computerized robot playing a midi file? What will happen to the "soul" of the music then?

[@martas9283]

I feel the same doubt creeping in.. I once helped to restore a 17th century turret clock, programmed to play a tune on bells on the hour. The master blacksmith who created it made it both functional and 'steampunk' beautiful, without obsessing over infinite variations in its performance - if he had done, he would have starved I'm sure. I feel that if the goal is to take a functioning marble machine on a world tour, the machine should not be tasked to be every instrument in the band playing tighter than any band. If it's a visual feast with a tune to play and it loses the odd marble or split second, that will add to the joy of watching it do its thing

[@dranorter]

Looking at the "waves" in the performance, you can see what's going on is that a tempo which is very slightly off the target BPM will drift ahead or behind the beat until Martin notices. As soon as he notices, he puts effort into bringing the tempo up or down, and then of course he has to overshoot the target tempo in order to get the beat back in sync; and the new tempo drifts the opposite direction. Point, being, we can directly see what scale an audible tempo mismatch is, on these graphs. But Martin's giving himself the worst case, trying to sync it up to a click track. In real performance, the band will treat the machine as the metronome. I bet the tympo graph will look completely different.

[@siriany]

It's always so incredible to see you guys building this ❤

[@curtisnixon5313]

So you basically built a giant drum machine with the "stick" hitting directly onto the mike, being spun on a disc. Brilliant!

[@randomnickify]

Love the idea with the drums, much easier to figure out whats going on

[@davidchidester5463]

I get the drive for tighter and tighter music. But please remember that great music and great music machines don't need perfect timing to accomplish the goal of bringing joy to musicians and their audience.

[@Urd_Voiddaughter]

I agree. Furthermore the music does not have to be "tight" in comparison to an external source only compared to itself. Various instruments drifting apart is an issue, a wobble of a few bpm does not have to be an issue. Anyone playing alongside the MMX will be able to match the rythm even if the tempo drifts a bit.

[@pawekaleta4149]

Governors work better on vertical axes (gravity affects each weight equally). In the design it is mounted on a horizontal axis (gravity pushes weight on top towards axis and weight in bottom away from axis). Also, running it at the same speed as the flywheel may not be the best idea (too fast). Keep those in mind if any issues arise. Good luck! :)

[@Norden_Treee0127]

Always with the epic videos, Thank you Martin!

[@vaalrus]

I look forward to seeing the effects of adding a governer and non-rigidly connected power input.

[@YourN4m3]

There is more to learn from trying new things than by adding more weight to an already very heavy and potentially unwieldy system.

[@BensMiniToons]

I want to see what can be done with no weight, I can twirl my finger around my ear to the test at

[@AdrianHiggins83]

The wheel can be made lighter as the mass at the outer diameter is what gives it energy.

[@metern]

The BPM sound is stuck in my head after hearing it in all the test videos 🤣

[@htmagic]

Martin, the crank is also a lever. Put the handle on the end of the crank and it won't take as much force to turn. Cheers!

[@hsep09]

Don't you think the machine's significant weight might be a problem for stage use and touring?

[@raharuko]

ehhh once you get to the point of having to use dollies and shit to move stuff around a hundred pounds or so isnt that much more, he wasnt moving any of his other machines around by himself so this isnt any different

[@TarasMazepa]

Maybe have two pedals which will be in alternating state (one going down and one going up)? With crank you have an option to speed it up or slow it down at any given point of time, but with pedal you can only do so half of the time. Having two pedals in alternating state would probably make it better. Also I like how someone else suggested to get rid of the flex on the pedal.

[@EndlessPlane]

Marble machine X: The Stairmaster

[@MrJamesBanana]

I like the idea of two pedals, but must also say it sounds a lot like a bike. Not the worst way to transmit power, but I'm getting a feeling this machine might use more power than a human can deliver in the end.

[@EndlessPlane]

Some sort of electric assist to get it up to speed will be a minimum I suspect.

[@MrJamesBanana]

​@@EndlessPlane Yeah, something like that will be needed. If we are going for cool aesthetics I would probably suggest a water wheel, but direct drive electrics will be simpler and can be regulated more easily.

[@flipflow1909]

I came here from This Old Tony some years ago for the machining/CAD/DIY content. I am not disappointed that this is going to continue. :)

[@Scoots1994]

Transporting this thing is going to be insane. It's going to be like taking a more fragile large lathe with you on the road.

[@thelasttankmage]

i definitely would like to see how the flywheel design performs with the governer BUT i agree that its worth taking the time to small scale prototype a huygen drive as im very interested in what that would look like and if its viable. Loved this build series and can't wait for the next one!

[@cybermanne]

Also remember that it's possible that the ergonomics and other aspects of how the pedal and the crank are constructed might be factors in the results you got. It's totally possible that the best possible design of pedal might give better results than the best possible deigned hand crank. Maybe a gearbox on the pedal could be one such factor. But maybe there are other more subtle changes you could also do.

[@SciPunk215]

Welcome to Wintergatan Wed... uh... Sat... uh... whatever day it is, thank you Martin! Keep up the good work !!!

[@morkovija]

that thing playing violins you showed at the end - super cool machine!

[@nickfosterxx]

Well done Martin, must be very satisfying, and kudos to that software design. I like the idea of Huygens - more 'mechanical machine' and could look nicely steampunk. (Not that aesthetics are important, lol) But to get closure on the flywheel, it should be tried with governor and gearbox.

[@stephenbrown5333]

I may of missed it in a previous episode but are you planing on using a motor like you did for your previous version? Would that not making playing tight music really easy as you can set it to what ever speed you want?

[@im-leia]

Can't wait for the weight drive! I'm super excited to see a real-live version of that design. You're doing awesome!!

[@tomjameshh]

Loving the videos! I'm really excited to see you further explore Huygens' chain drive.

[@drewbiedo2239]

Love your videos! Thank you for letting us follow along with your creation as it is being built.

[@mhelvens]

Perhaps you could build it like a multi-speed bike? Low gear ratio to get the flywheel going, and then dynamically switch to a higher ratio for the tighter music? Just a thought. The mechanism would probably be too fragile (if my multi-speed bike is any indication 😅). Edit: Ah, I guess that's takeaway 8.

[@tronique5736]

Martin has the right intuition about speeding up the fly wheel, but he's mixing up the concepts of momentum, kinetic energy, and inertia

[@lifelexx]

martin it is so clear to all of us just how much youve learn as a designer over the past few years, this was an amazing demonstration of the importance of all the testing and prototyping you've done including all the work on the og machine and the mmx, cant wait to see the next steps

[@tuomasperttula3651]

I'd guess a large part of your difficulty to play tight comes from trying to match the drum beat, which forces you to constantly decelerate and accelerate to catch up with the beat. If you were just directly measuring the rpm and trying to keep that at desired value, it should be a lot easier to keep the tempo tight. That's assuming you won't be trying to match some external beat source in an actual performance, but instead using the machine itself to set the tempo.

[@Nevir202]

Well, it MIGHT be, That would all depend on how good his internal chronometer is.

[@tuomasperttula3651]

I don't mean that he should do it by ear, but with a measuring device. Trying to match the drum beat means he needs to have the right tempo _and_ the right phase, with an rpm meter it would be enough to match the tempo. The phase difference doesn't matter unless he's playing with a drum machine like he is right now, and trying to match the phase is very difficult when the machine has so much inertia.

[@josephhfry]

I would use a timing strobe. Basically put white tape stripes around the drive wheel exactly 20 degrees apart. Then have a flat panel close to the edge of the wheel with a rectangular window to look through with a needle across the middle and a strobe behind that flashes according to the desired tempo. For example at 80 bpm, it would need to flash 24 times per second to flash once per 20 degree segment . Then as you pedal you will see your white line every time the strobe flashes and will see if your ahead or behind the tempo based on if your line is ahead or behind the needle. I hope this makes sense? It's basically the way a mechanic sets the timing on a car using a strobe.

[@RandStuffOfficial]

As a musician, I would also think that the band would match MM's tempo (after the flywheel as gained up to speed and that Martin activates the levers). So measuring the real tempo would probably be more accurate than comparing the difference with an external beat.

[@Apollost]

Hey, is this solution feasible for a concert? It looks like you have to put in a lot of energy and focus to keep the tempo.

[@isaacgraphics1416]

Not in it's current state, it looks like. It looks like some kind of governer will be necessary. He could just use an electric motor but I think that probably goes against the spirit of the project

[@rudolphvanrooyen2655]

120bpm Martin: It's hard, but it's tight . 🤣🤣

[@mcg6762]

This whole channel is a demonstration of how "perfect is the enemy of good"

[@ivanlindmark]

To get a better understanding of what 2ms deviation "means". It would be interesting to see what the deviation is when you play on a drumset. E.i. is 2ms a world class drummer or is it what an intermediate drummer would have.

[@user-qu5mm3ei9l]

It's about what you would expect from a professional drummer. Google "Synchronization Error of Drum Kit Playing with a Metronome at Different Tempi by Professional Drummers"

[@DarthJadus95]

Extremely thorough approach (I mean, with all of this data, you could definitely write a scientific paper and publish it) ! Regarding takeaway 8, be careful with the addition of gears, as they are usually source of (quite important) parasite noise (depends of the gear’s tooth shape), which might be problematic for your overall functionality of the marble machine 😊 Congrats again for your achievement ! Hopefully this one is the good one !

[@TheQxY]

What would this paper even be about? How are any of these findings relevant outside this specific prototype?

[@DarthJadus95]

@@TheQxY honestly, there’s a bit of everything in the current scientific community and papers. Some papers are very abstract whereas others are very specific and/or related to a unique system/product. In the experimental papers, most of the content is experimental setups, results, comments and perspectives of improvement. Which is (almost) what Wintergatan did in this video (the incomplete part would be materials and methods). Honestly, I was pretty stoked when they presented that rotation velocity would reduce the standard deviation of his tempo. Is that straightforward ? I’m not so sure. In the hypothesis of writing a paper on it though, going into the theoretical why would be a great addition to the whole work. And truly, are all papers truly relevant nowadays ? The answer is unequivocally *no*. They’re mostly to fill quotas for research lab public fundings. Some present major breakthroughs but very little. It’s also a great way to advertise your work to other people, which may attract partnership or fundings, which I think would be the most interesting for this project.

[@TheQxY]

@@DarthJadus95 This could never be published. It makes perfect sense that increasing the angular momentum of the flywheel would make it play more tightly, as it would require more energy to slow it down and speed it up, so relatively the wheel speed is more consistent. These kinds of papers were maybe interesting in 1700, but they're more common sense findings now. And what you're saying about that there are papers being published about very specific systems, this is only true if they do something novel, exceed a previous limit, or give some other relevant insight, none of which is true in this case. The breakthroughs in engineering and science behind flywheel systems have been basically perfected more than a hundred years ago.

[@DarthJadus95]

@@TheQxY thinking about it, yes, since energy is proportional to the squared angular momentum, it makes sense. Nevertheless, a little article (if not scientific) about it would be really cool for it !

[@TheQxY]

@@DarthJadus95 Yes, I agree on that. An in-depth article or blog post could be interesting. :)

[@bagelbandit6076]

The Wintergatan workout leads to marble-machine muscles 💪

[@jrgallagher1]

Martin, what you really want to look into is /momentum/. The moment of inertia of a solid object is a constant. Momentum varies with RPM. I don't mean to be picky but I think this distinction is important for your design. Increasing the mass, or radius, or both, of a flywheel increases its moment of inertia. Spinning the flywheel faster increases its momentum. The reason the music is tighter with a higher momentum is that tiny variations in input energy through the pedal or crank are much smaller in comparison to the kinetic energy of a flywheel with a large momentum. Making the flywheel heavier increases its moment of inertia, but the moment of inertia is a component of the momentum. The two properties are multiplicative. Moment of inertia, I, increases with the square of the radius, r, of a disk of mass, M: I = 1/2*M*r^2 Momentum, rho, increases with the angular velocity, omega: rho = I*omega The kinetic energy, KE, of a rotating flywheel increases with the angular velocity squared: KE = 1/2*I*omega^2 For tighter music, increase the mass, yes, but also consider increasing the radius of the flywheel. Doubling the mass doubles the moment of inertia, and doubles the momentum. Keeping the mass constant and doubling the radius increases the momentum by a factor of four. Doubling the radius and doubling the RPM increase the momentum by a factor of 16! At that point, you would want to consult with SKF about the capacity of the bearings and couplings to tolerate this. All rotating components of the system contribute to this, including the rotating parts of the bearings, the couplers, the rotating shaft, and the bolts holding the whole thing together. SKF can give you the values of moment of inertia for their parts. Fusion 360 can calculate the values for your design components, and consolidate all this into a composite moment of inertia calculation for the system (note: the distance from the center of rotation matters). Then you can play with the variables, M, r, and omega to optimize your system. Note that whatever you do to increase the kinetic energy of the system will mean a longer start-up time. You will still find using the pedal easier to start the system up, but with a higher kinetic energy, you may be able to maintain tight music also with the foot pedal because differences in pedal inputs will just be smaller compared to the total energy of the system. You've probably already planned for this, but installing a clutch would allow you to get the flywheel up to speed and disconnect it from the system to preserve the built-up energy while you're busy doing something else.

[@Raven-rv9jr]

I am very interested not just how well the gravity drive works, but how you specifically feel about it. It might feel too perfect since you don't have control over the tempo personally.

[@John_Weiss]

That's his problem: he's trying to keep _tempo_ using the *_POWER_*_ module._ What's the purpose of the *_POWER_*_ module,_ do you think? Keeping fixed BPMs? Or, I dunno, maybe something along the lines of *_SUPPLYING POWER_* perhaps? 😉 He needs to make his power module _store_ work-energy to _POWER_ the MM3. Tempo needs to be governed by a _completely different module/mechanism._

[@theodorechiou2886]

i just want to say something simple: I don't want to watch a machine play music like a machine Love this engineering journey that you have taken us on, but at the end of the day, i want to see a human performance :)

[@stedmenknox1169]

Lit..I’m stoked to see the old adage “third times the charm” play out

[@c.a.r.s.carsandrelevantspecs]

That flywheel looks so cool! 😃😎😃

[@lukasdon0007]

This is getting stupidly dangerous. You seem fully unaware of the sheer amount of energy you're playing with here, without any adequate protections in place. It's all fun and games until your shirt gets caught in the crank or the open belts. Please take a step back here and reassess the situation. It's not safe.

[@IsaacDaBoatSloth]

even watching at 2x speed its really hard to know if its out of time

[@evanrobison567]

I am so happy that you are going to try with the weight drive! It's like the drone on bagpipes; I think it's going to work well! Great job!!

[@MauroTamm]

The hand crank looks lethal at high speed. Feels like it needs some ratchet/decoupling (like a bicycle hub but likely noisy or a clutch to disengage it).

[@marksnethkamp8633]

Martin still prototyping on production builds ! I wouldnt have it any other way.

[@owensparks5013]

Production build? This entire assembly is just an evaluation prototype. There isn't a production design yet.

[@landonkryger]

When you visit other machines, please measure their tightness to compare to yours.

[@edwardbentall5665]

I would still look into the Huygen chain drive, but it would be interesting to also have a flywheel in the system.

[@MrTurboturbine]

You're going to need a significant braking system for that flywheel.

[@Steeger007]

Since you're focusing on timing, why don't you play a typical set of drums and see what your timing comes out to be on them to compare the machine to? It seems like a couple of milliseconds wouldn't be a huge deal if drummers can make mistakes all the time and it hardly be noticeable. Between getting slightly off beat, missing a hit, or losing a drumstick, it doesn't seem like you're losing much being a millisecond or two off at any given time. Same goes with guitar players missing a note, or singing off pitch. Seems like you're trying to get machinated/computational perfection like creating a song in a program and not allowing yourself any room for human error. - Sound guy in a band

[@m00str]

This. Martin has ignored everyone screaming at him that it doesn't matter for half of the development of mmx now. Even with mmx, people were joking that he could skip a lot of effort by just letting a drum machine play preprogrammed music, if he needs it to be accurate to down to the milliseconds.

[@Steeger007]

@@m00str I can see how it would seem like a huge deal, but once you put all of the music together you’ll hardly know the difference. A trained professional can tell when the beat adjusts, but hardly anyone else can tell, and it’s just that, the beat changes and gets back on track. Even trained musicians can make mistakes. I understand wanting the machine to be working “precisely” but a major component of deviation is the human muscle and timing. For it to be off by a millisecond is… pretty spot on. I question how much is the way the machine functions and the way it’s played. If you hooked a motor up to it, you probably wouldn’t be able to tell much of a difference. It’s a simple mechanism

[@NickCombs]

The higher variance with the pedal does make sense. If you think about it, you only have control over the force input to the system on the pedal's downstroke, half as much time as with the crank. A possible solution would be to have two pedals operating at an offset, so one pedal is going up while the other is going down like we see on bicycles.

[@NickCombs]

Actually I guess a bike is more similar to the hand crank in design, so maybe one of those stepper exercise machines is more apt. Also, two pedals might also spread out the forces stressing the pedal's structure, so they could undergo less flexing and become more efficient that way.

[@levilukeskytrekker]

Always love your videos, Martin! Keep following your dream! You have a vision in your mind, not merely artistically, but also from an engineering standpoint. You want this to be a great machine that plays great music, not just a machine that plays great music, and the wondrous ambition of that is why I always keep coming back to follow you on this journey. I know people can give you flack, sometimes, for doing it your way with your own specific goals (hey, it's the internet), but I hope you know it's valid and worthwhile to chase this dream the way you want to chase it. Even the pursuit of art is art itself.

[@Physiotherapie.ThomasWerner]

- my opinion, as a physical therapist is: that a main factor in the tightness of the machine would be your physical capability to repeat the movements in a presize manor; you definitely need strength endurance paired with intramuscular coordination and intermuscular coordination. you can´t expect your body to just have the skill that is required to move that massive weight tightly. - i wish you all the best.

[@abitofabitofabit4404]

In your professional opinion, might he be better able to match a beat if the beat were to fall at a different point in the pedal stroke? Right now the beat falls at the bottom of the pedal travel.

[@gregkent8955]

Traveling at this pace you will be way into your 50's before you finish this if you demand this level of perfection in all the rest of the MM. I hope you prove me wrong. Keep working

[@massminer2343]

You should definitely do a chain drive it'll give you a good comparison point

[@chriscorvus1808]

I love that you haven't given up. I love the Idea and wish you well in completing it to the fullest, with a sucsessfull music-tour. Thanks to your videos i feel like i'm part in something big, like professional and impressive. And im glad that you share all the high and lows.

[@repalmore]

Maybe a crank handle on the flywheel shaft for starting up. For safety you can use what is called a sprag clutch. That allows the handle to drive the flywheel but the flywheel can't drive the crank handle.

[@kkupsky6321]

Archimedes said the lever thing. According to legend. He’s more than “somebody”. Good show

[@inthefade]

I just realized how dangerous a flywheel like this is. It is pure kinetic death just waiting to be unleashed.

[@metern]

You really need to install a Governor. It is a reason they use it on a machine with a heavy flywheel. It really helps to keep a steady speed. 😉

[@tomm21]

It was Archimedes that said that. I think the full quote is, "Give me a lever long enough, and a fulcrum on which to place it, and I shall move the world."

[@sbabbaro356]

So happy to see you (and Hannes) working again on these prototypes! About tight music...at this point you should consider a variable balance + clock escapement mechanism. There is no better clock than a clock itself...

[@isibboi]

Really impressive machine you are building there!

[@illagevidiot8254]

This thing is terrifying. That pedal system looks like a foot smasher. That's even ignoring the insane potential of that flywheel to destroy everything in its path.

[@mgeisert6345]

Please add another criteria : SAFETY. 120bpm * gear ratio = ~1000rpm on a heavy wheel, that is a very large amount of energy and you need to make sure this is contained. At this level, if a screw get loose, the axis get bent or something, things can become very dangerous very quickly. At least make sure the flywheel is always rotating away from you and the crowd as it is now. The axis that you said was a bit too small for the wheel, I would probably replace it. And probably build a strong cage around the wheel.

[@EskoLuontola]

Huygens drive! 🙌 Excited to see the prototype.

[@garrywhiting8398]

I was just itching to type "additional gearing" before you mentioned it at the end. I'm pleased that you did. It might be worth using a motor to accelerate the flywheel initially...

[@kronkolweg5917]

Well done Martin, keep experimenting

[@thetruthserum2816]

Just though of you, and decided to stop by! I'm so glad you have some content I can catch up on. 🙂