sorry about the comic sans 💀

Mark the tab you bent with paint, or something, so you don't use it again. Bent tabs will tend to break off if re-used.

Following the livestream today, I think one of the reasons it is so hard to make this feel tight is that you only have control of input, for half of the time. On the MMX it was possible to control the input of power for the whole circle movement and thereby 100% of the time. It was mentioned and quickly disregarded to use bike pedals, because it would be very limiting on stage. But I think it's important to recognise why bikepedals work so well. Having two pedals allows for one foot to take over, when the other is not able to input energy into the system. Imagine trying to get a certain tempo while biking, but only using one pedal. This would be extremely hard and I think this answers the primary reason why this design is causing trouble and would do so even if it was more well tuned. Remember on a fixed bike you would also break when applying force in the wrong half of the circle. Using a strap as shown today would be slightly better, but I do think you need to be able to distribute an even amount of force on the up and down movement of the pedal for it to feel natural.

Please add a hinge to the pedal (or allow the linkage to fold) so it's never forced down by the flywheel. The sides are a good start for saving body parts, but if something wedges itself in the gap, you will tear the machine apart.

At the end of all this the Marble Machine tells Martin that he doesn't play tight enough to go on a World Tour with it and starts a solo career.

You should look into the governor and other control mechanisms after the flywheel, so your goal with the pedal is not to match BPM but to keep the flywheel above a threshold. You would control tempo via another system further down with more refined control.

The pedal would be even stronger if you make a triangle instead of welding just another one. Also in a triangle you could support the covers.

A "continuous variable transmission" can fix any timing issues! Thank you for everything you do!

Martin is such a performer at heart. Creating things in front of an audience clearly gives him energy and motivation, even if it's the design and build process rather than making music. It's really heartwarming seeing how happy it makes him to be working in front of people again.

I stopped in to the live stream yesterday to lend you a little advice regarding TIG welding and arc welding in general. 1) the tighter the plasma shield and shorter the arc the cooler and cleaner the weld will be. This will help to not overcook and weakening the material around the weld. 2) using a square wave pulse setting at a lower voltage will allow you to produce a much cleaner weld and help you control your filler feed timing. 3) increase post gas cfh and flow time. Hold position until the post gas process has finished. These 3 things should help significantly. Best regards, great work thus far.

In reality you would not need to keep the BPM synced to an external metronome, but only keep the BPM synced to a specific number, if you understand. What you need is a "speedometer" that monitors your own speed and outputs that number to you.

I remember when LTT first built that PC. Nice to see it in its natural habitat

Oh it appears the visitor did read the comment section being full of: Martin, don't forget about bending the tab

Remember that you dont need to match the beat of the metronome, youre making things way harder for yourself than you need to. What you need is a rpm sensor so you can know your rpm/bpm, and try to keep that constant.

I love seeing the ideas develop with an audience, it's such a good idea to challenge yours and develop faster! Keep up the great work!

Keep in mind that when there's a full machine being powered by the flywheel, it'll be much harder to maintain tempo since you'll need to add a lot of energy at just the right rate, which will add to the already challenging task of keeping with the beat

You've built yourself a Phase Locked Loop, but inserted yourself as the Loop Filter. Your undershot / accelerate / overshoot / decelerate pattern is a well-known phenomenon when a PLL is under-damped.

my guy, it is so fun to see these sort of spontaneous crowdfunded knowledge and expertise show up randomly like this. it really speaks to how large this project and your ambitions have grown, expanding well beyond just yourself to a whole global community of rad people

I have been following your journey since you posted the first video of the first marble machine over 7 years ago. It's truly incredible the community you've built behind the journey, the amazing engineering and manufacturing groups you've brought together, and sparking the interest of builders worldwide. Well done, friend. Can't wait to see where you journey takes you in the future! I wish I could have been there to meet you.

Keep up the good work, great to see this Vid Doc series day by day! We will see a working marble machine!!! 😊

I really feel like the falling weight design would be a lot better suited for this, for several reasons, and here are just some of them: - In your comparison video you mentioned how the gravity powered design is used in clocks and even for telescope turning which are very precise scenarios. This will give you constant control over bpm without having to focus your leg movement to balance between overshooting and undershooting, which is both distracting and tiresome - Flywheel systems are meant to store energy and release it, no one uses them for precision timing. It's kind of like a pulley (and additionally like a short term energy storage), where you input bursts of less energy quickly, to do a much more difficult job slower. Similar to a flywheel, where you pump in bursts of smaller amounts of energy, and can later release that energy for a more intense quick job. Sure a pulley is not a direct comparison but you get the point.

This new video format is great. Really enjoying this series!

We finally got to see the LTT computer! I was like "that's not is it? No don't recognize the names. Did he say Jansen or Linus?"

You should be so proud of yourself. It was great to see all the other projects that you inspired.

that looks fun...lots of shared creative energy

You being in essence an organic PID controller you‘d need an even lighter flywheel to perfectly match an unchanging external beat. Why don‘t you try playing 80pbm with a bpm measurement of your mic instead of trying to match an external beat that you continuously chase and overshoot, you‘ll see that you will be MUCH tighter than you think you can right now. Don‘t throw your design/idea because of bad measurement methodology. Also isn‘t it hard to match with your foot down? I thought you‘d be easier to time your step if the beat is on the ‚up‘ position

Awesome creations!

What an amazing group. Meeting of minds.

Thanks for the videos and the creativity !

Always surprising when you see other people you follow around, like Neil! Glad this event was held and looks like a huge success!

It has always been so inspirational watching Martin's battle

I never clicked on a video so fast, i love seeing the progress

its so cool, you started a entire movement :D its bigger then just the marble machine now, amazing :) !

The activity and involvement of and from everyone blows my mind ❤

reengaged with the build now things are getting built. great stuff

So glad to finally see that PC in one of your videos haha :D wish I could have made it to the event, maybe next year! :)

Musician, Engineer, and Educator chiming in here. I have a lot of experience working in the marching arts. Drum corp international to be specific. I played in the quad line, which was a good experience, because you're not exactly the center of pulse, but you are close. With drumlines, a similar kind of inertia forms around the pulse. I've also seen many drumlines oscillate around the tempo (something I call slinkying). When addressing this problem with students, it's helpful to mention that there are two adjustments you have to make when syncing with a metronome. The first is to start drifting yourself toward the metronome, and the second is, once you meet the metronome, you need to return to the original tempo. The problem you're encountering is that ass you add flywheel weight, it becomes harder to make the second adjustment quick enough to stay in sync. But the benefit you have with a heavy flywheel is that pulse will not be affected by changes in load on the machine. I think that is the more important requirement you need to meet for what I would call 'tight' In my experience, syncing two clocks like this is a losing battle. It's very difficult to keep two vinyl records in sync for a short crossfade transition. And it brings me to the dumb requirements question. Is it a dumb requirement to try to keep the marble machine in sync with a fixed click? For one, it would be easier to configure a midi device through an arduino that would turn the rotation of your clock into a midi sync signal. Then show elements like lighting, visualizations, and digital effects could be synced with the machine, rather than trying to keep it in time with some external click. Maybe a simpler solution would be to have a 'speedometer' that reports the BPM you are running at, and then you would be free to organically play with the instrument, yet reference a clear tempo at points where the music demands a specific tempo?

Martin. I hope you may one day achieve a marble machine that meets your musical standards, Until then, know that your efforts have never been wasted, as they have inspired legions of inventors, musicians, and tinkerers to reach higher and dream larger. DREAM!

None of the MarbeleMachines are a failure at all! Look at the inspiration and beauty were created only because you've build them! I am touched and sad again not to have been there ... Perhaps there will be a meetup next year again, where I am able to be also :)

Isn't inertia the opposite of what you want? If you're manually controlling the BPM, it must be incredibly difficult to get a system with this much inertia to stay at one BPM level as you have essentially no fine control at all and the feedback is very slow

Maybe you need some electronic support to maintan the same speed. I have someting in mind like the cruise control funktion in modern cars. You have full manual control of you speed but if you like to hold the speed push a lever and the machine maintain the speed by herself until you want the manuel control back.

Instead of aligning to an external metronome click, it'd probably be more useful to feed the mic into a delay effect and synchronize against the appropriate delay -- in whatever ensemble the MM will be part of, it's going to be setting the time, so it just needs to synchronize to the correct interval length, not synchronize to some external click.

Yay🎉 so good to g hear all that positive optimism 🎉😊

I just thought of a perfect application for the Huygen chain weight drive. Add a small one to your machine (could even be the size of your lego demonstration), with a small weight that both sets the speed of falling when not powered and is the indicator for the following: When powered by the pulley, it will do one of three things, If you are lower than your target RPM, the weight will start to fall. If you are higher, it will rise, however, if you match it, it stays in place. It is basically a differentiator to give you feedback on how well your timing is in real time. This way, when you are playing without a metronome, all you have to do is keep the weight as steady as possible, no need to overshoot the RPM and have the governor do extra work for you. An actual mechanical differentiator or RPM reading device would probably be better, but figured the huygen drive deserves a chance to have an actual use after seeing how much you liked it in your video

wow! overall I thought the test was very encouraging for a first go. you have only the pulley, pedal and flywheel atm. there are so many things you can add to make it work better. still you were able to nail and sustain the beat. yipee!

Is it an idea to use a stroboscope to measure the thightness? If, in your current case, you play exactly 80 bpm but out of phase, you need to accelerate and decelerate to correct the timing. With a stroboscope you can ignore the phase issue for now.

fabulous! just fab!!!🦌💌❤️🙏

Wow! So cool to see Neil! I have played his auto glockenspiel online!

Whats about a Tachometer. With an electronic one (Bicycle), it migth be even programeable to BPM....Magnet on fastest moving part, Flywheel.

You should make the vertical pedal linkage be able to buckle under compression. Then, even if something manages to jam the pedal (foot, a marble, etc). It will safely just buckle instead of instantly crashing the machine to a stop and likely damaging things. Just using a single screw instead of 2, and something to keep it from going perfectly straight would probably accomplish this.

Just my observation that your ‘power’ stroke is only supplying a force 50% of the total revolution. If you consider a steam locomotive the piston is powered by steam pressure on both sides of the piston or equivalent of a 100% power stroke with no interruption. This is also the case with pedal powered bikes, with power stroke’s alternating with the right and left foot to create uninterrupted power strokes. Unilateral powers 50%, bilateral power strokes produce a more consistent rotational force for the power. ❤ love your videos 👍

Nice to see all that positivity.. we all know the world needs some .. 👍

The makerspace is awesome! :)

DAY 6! LETS GOOO!!!

Great event

For some extra safety, you could ad the broom like thing that you would usually find on escalator around the hole for the pedal

just looking at the ring locking mechanism for the bearing, I personally would've thought the tabs spoke for themself when "locking down" the assembly, otherwise they're not needed. Sebastiaan's PC takes the cake for me, and such a cool dude with some knockout answers (wasn't ready for the feature-creep, sprayed my coffee).

Tight and in sync aren't the same thing, though. The problem with the fast acceleration is that when the machine force is higher. because f ex it might drop a lot of marbles at once, it might slow down.

Instead of microphone, add hall effect sensor, hook it to arduino or raspberry to display live bpm. I am sure a crowd of makers can help you do that in an hour

I don't think syncing to a click track is important... you will always have problems since you've put so much inertia into your system. If you need to sync other electronics to your machine... give them the click it generates! You have plans for a regulator? Perhaps adjustable? Like a spinning ball (steam engine) or wind vane (music box)? Music tempo feels better when it breathes anyway... don't be a slave to the click Martin! Yes, the locking tabs are important, I had not realized you did not understand them, it's what happens inside a car wheel hub's locking nut.

Definitely make this a yearly or so event and make it into a competition of designing and building music machines in different categories.

Looks so nice there, I had been planning to come, but then time passed by, too busy to organise, and in the end: Covid stroke... (luckily almost recovered now) Hope to be there next year!

Love that you did this weeks videos, would love to be able to travel to see in action/person, unfortunately i dont see that happening for myself😢

🎉Such a fan bloody tas tic meet up, love it, I want to make my own Yonis

I occasionally work on punch presses with flywheels. They typically have a brake with friction material on the drive shaft. That may help with controlling the overshoot.

Ansioso para ver o resultado final... máquinas musicais interessantes, divertidas.

Maybe I'm missing something, but when you showed off the Huygen weight drive last month, and then the flywheel made a comeback in this prototype, I assumed you were going in the direction of using the drive to produce constant rotation for the machine, and then use the flywheel only to raise the weight -- meaning it wouldn't matter how tight or consistent the flywheel was at all.

I think the position of the clock hand influences your "feel" of the time a lot. It seems to be aligned so that the clock hand hits the microphone exactly when the pedal hits the lowest point. If you feel you're to slow, then the hand should hit earlier than at the lowest point. I am not sure if this will actually help but I can imagine it's hard to feel the beat when the movement doesn't match the click.

There is an actual value in letting people from "outside" the project, you get someone who spots something you missed. This is more than ever an open project.

put a freewheel clutch on the input of the flywheel so the peddle isn't driven and use a spring to raise the peddle you then wont crush your foot and can turn the pulley at any down stroke position.

A reminder to everyone posting suggestions in the comments here that Martin has said that he does not read the YT comments, but he does read the comments over on r/MarbleMachine3.

I was wondering what happened to that PC. Glad it finally made it

definitely take a close look and do tests with the weight chain drive in place of the pedal system

Applying a force to an object to align with both reference speed(tempo) and reference position(sync) is a really difficult problem. It will always overshoot if you aim for reference position. Google control theory, PID regulator, wikipedia. Love this channel, can i make a wish for more ondophone? Love both the scarry ondophone song and the one where you were just playing to some nice playback beat. Keep up the awesome work, i believe in this!

You may be forgetting that the whole machine behind will also add a massive inertia, that's why the MMX had much less trouble staying on time despite the much lighter flywheel. You could simulate that by adding a big weight on the time counter.

oh nice its back

There is something you'll have to consider for the inertia : here you only have the flywheel, but the machine itself will bring a lot of inertia to the whole system

lets go!

Nice

Martin, there's something I was saying in the Chat on the livestream. "BPM" and "Tightness" are not the same thing. 1) You can get the mechanism Super-Tight, but still play your 120 BPM song at 73 BPM. 2) You can play at your desired 80 BPM _perfectly_ but still lack Tightness. I want to introduce you to a Physics Concept: Jerk. Acceleration == Change in Velocity over time. Jerk = Change in Acceleration over time. 0 Jerk == Constant Acceleration == Smooth. Nonzero Jerk will create non-constant acceleration, which can create surges in Velocity changes. *Remember:* You've been defining "Tightness" as "the _variance_ in the delta-T between Marble-strikes on the instrument". The lower that variance, the "Tighter" the MM3 is playing. This is has been your definition of Tightness. So if you have any Angular Jerk, your BPM _will_ have discontinuities. Jerk will introduce Delta-T changes between marble drops. You need Zero Angular Jerk.   Yes, zero Acceleration, once you're playing, is nice, but ultimately what matters for your Tightness is _fixed [Angular] Acceleration. You, as a musician, know that tempo of a song can increase or decrease. So, let any _natural_ Angular Acceleration and Deceleration that you need to adjust to your desired BPM just be part of the song! Let it be part of improvisation in the song. As long as you have Zero Angular Jerk, your Tightness will be fine. But now you're redefining "Tightness" to mean "Constant BPM" And as I said above, that's _not the same thing as your earlier definition of Tightness. You're changing the metric mid-measurement, so you're going to get Invalid results and draw Invalid Conclusions. The Flywheel is storing Energy in the form of Angular Momentum. That's its job. Smoothly releasing that Energy *_IS NOT Its Job,_* nor can it be! Plus, you have _another_ Energy Storage component: _The Programming Cylinder._ That's Yet Another Thing I brought up in the livestream chat: You _have to_ account for the Angular Momentum of the Flywheel *_and_*_ the Programming Cylinder_ when examining the power-train. Oh, BTW: A cylinder has a higher Moment of Inertia than a disk. You can make a cylinder that _has less mass_ than your flywheel but will store the same amount of Kinetic Energy at the same BPMs as your flywheel. This is Yet Another Reason why the Programming _Cylinder_ is actually also part of the drive train. Let me put it another way: How fast does your phone use power? Does the amount of power your phone uses in an hour depend on the capacity of its battery? Or does its power use depend on, say, what you're doing on the phone and whether or not the screen is on? It's the latter, of course. Your battery will determine how long you can use your phone, not how much power your phone is using. The flywheel is the MM's _battery._ But you're treating more like the phone screen or the phone's GPS circuits. Sorry if this paragraph is harsh; I'm typing fast with a cat clawing my forearm! I'm not trying to be negative, I just want to make sure I get through to you. The flywheel is your battery. It is therefore not designed to keep a fixed beat.

I was playing Where's Waldo (er Martin) in the Wintergatan Meetup photo. I spotted Martin peeking out of the back row just before the video cut away

The test you conducted doesn't test if the flywheel prototype can play tightly, but rather how well it can follow a metronome. If the beats aren't synchronised you need to change tempo to one which doesn't match the metronome's, which causes you to overshoot. This is because the flywheel makes it harder to speed up or slow down, due to it's moment of inertia. In essence the mass of the flywheel is working against you if you want to synchronise the beat with the metronome, a smaller flywheel would do better if this was what mattered. However it isn't, the prototype should be able to play at a consistent tempo, although desynchronised from the metronome. If you measure the rpm of the timing tester you can then use a display to monitor your current bpm.

crazy

Isn't art inherantly optimistic? When we get to flex our creative muscles we really really want to make something happen, pull an idea into existance, and maker spaces are so fantastic for that. You get to meet other people who will push you, offer support and advice, because if you love your idea, they will too.

I think if you matched the phase of the bpm to the machine rather than trying to get the machine in phase with the already existing beat it would be a lot easier

Wait, LTT? wow can't wait to see their eventual video on that PC build!

There was a fabulous device made to help musicians get in sync with clicks, called the Russian Dragon, by Jeanius. It indicates if you are draggin’ the beat, or rushin’ the beat.

With some practice, I think you can address the overshooting issue by applying weight on the upstroke (braking force) and removing your weight on the downstroke. Alternatively add a toe hook so you can apply braking force on the downstroke

Trying to control tempo and phase together is why the "overshoot" if not trying to match the phase of the click-track, you could reach an exact tempo with much less overshoot.

How about adding a toestrap, which will help you match timing because you'll have double-action control (up as well as down).

I heard pain in your voice, Martin. Don't give up, you are very close.

I suggest a lattice structure for the pedal. You have plenty of space underneath for it. With a lattice structure you can arrange braces in different orientations to counter sideways flexion and torsion also. You could go with 4 angle irons jointed with welded 6mm rods.

The timing was always my concern with this design because of how much the speed could vary. The concept is impressive, using a foot pedal to generate rotational force that is largely smoothed out by flywheel, but without the level of control either an electric generator or something similar that can provide a constant RPM you will always be varying the speed by some amount. You go too fast, then too slow, then too fast, and so on. Unless you get real good at keeping the pace but that seems like it would take quite some time. Maybe you could make two separate sources for the machine, use this foot pedal idea for the parts where timing doesn't matter like the marble lift and use a separate source that keeps time perfectly for the timing wheel which tells the instruments to play. Then you give the impression of using the foot pedal as a power source but don't have to worry about being in time so much.

Look up what's called a sprag clutch for your pedal power input so you're only applying power one way and the flywheel freewheels so the pedal does become a flywheel driven leg breaker. Sprag clutches are used in power transmission systems for just this purpose.

Had a feeling that bending would cause an issue.. glad it's starting to be addressed; hopefully the timing thing and get worked out, would be curious to see what difference a governor would make

12:18 Nice!!

Where you step on the peddel will put more or less energy. It is just a lever transfering energy. So in theory you would have more control over the input.

I think putting the pedal link at a slightly bigger radius on the wheel will give you more power. And also less inertia in the fly wheel will give you better control over the tempo.

You could add a digital indicator that displays the tempo and is updated once per rotation. It would be a simple Arduino project if you could not find one on the market. This would help with the overshooting

To properly test the tightness you need to measure sub-beat timing intervals too. I suspect the weight of the pedal assembly makes the first half of every beat faster than the latter.