A true breakthrough for the Marble Machine Project. :) Check out more Bouldering with me and Hannes from the awesome Trainerds Channel Here: kzread.info/dash/bejne/nHynmMyBeLatl8Y.html Also check out the "How much weight can i lose in 7 days" Trainerds video that was posted today: kzread.info/dash/bejne/jJifq82zZLmbetY.html Thanks everyone for following the journey, good luck with everything you are doing!

@halko1

6 ай бұрын

❤❤❤

@metern

6 ай бұрын

Combining the Huygens drive with the Governor and flywheel is the best solution i think 🙂.

@angellestat2730

6 ай бұрын

This is a recommendation that I am telling you in every video. *BUILD the music machine First, then focus on the power input and the tight music mechanism* Because only then you will know what would be the total friction and the energy input you need to apply to play a whole song. If you dont know that, then you can not start with the Huygen Drive design, because it will change all your parameters constantly forcing you to redesign the Huygen drive over and over. I also think that all this mechanism, plus all the other instruments you want to add in the machine, it will add a lot of friction and it will be impossible for you to have the strength and endurance required to play a whole concert. Besides, if this small prototype requires such heavy counter weight, imagine the mass that the final machine will require and how hard would be to set in each stage.

@thomasbecker9676

6 ай бұрын

So this channel is just an advertisement for your other channel?

@jacobstew7089

6 ай бұрын

if you want a different governor that adapts to rpm and isn’t a constant brake i would use a centrifugal governor like those on a stationary steam engine or a traction engine, old elevator/lift ect.

Calculating precision in milliseconds is bringing in some confusion here, as you end up making it looking more precise at higher speeds. You have to remember that the intervals get smaller and smaller as you increase the speed. Imagine a full rotation at a given tempo being 1 ms as an example. The deviation will always be less than 1 ms, however it might still be deviating 50% off the optimal speed. You need to figure out how precisely it’s hitting the same spot relative to the speed of the full rotation. This means a percentage deviation of the full rotation. This will become apparent since the deviations will scale with the gearbox. This will essentially be re-introducing more deviation (in milliseconds) at lower speeds.

@romi5505

6 ай бұрын

Came here to say this, error measurement should be relative on this case, not absolute. I understand that more speed means more energy, but because an absolute error of 1ms at 10 bpm gives the same relative error as 0.1 ms at 100bpm, It does not necessarily mean tighter.

@JoramAgten

6 ай бұрын

yes this, your calculation of standard deviation is not enough to determine if your music will be tight

@TheScalpel1

6 ай бұрын

Thank you, i have wanted to and started writing several comments on this topic after he started using standard deviation but couldn't formulate it quite as concisely.

@NoOne-dj1ou

6 ай бұрын

This exactly!

@johannesr8709

6 ай бұрын

I don’t want to sound like a broken record but: Yes, this ! 😅

Wintergatan next week: "So I didn't account for irregularities in the earth's gravitational field which means that it will play 1/100000 x slower if I tour the machine next to Mt. Everest. This is an enormous problem and we're back to square one."

@WeArePharmers

6 ай бұрын

dude fr, it's honestly getting so tiresome now. we will never have a world tour at this point

@STAG162

6 ай бұрын

the first one played music, the second played it better, this one shouldn't be a strive for perfection because sometimes good enough is all you can ever get. right now it's looking like an exercise in which mechanisms play tighter music and content based off that. I know people don't want this series to ever end, but there has to be a final product - I just hope I don't get much greyer in the process.

@user-yw9mw9hv8o

6 ай бұрын

He went 180 after MMX, which, in his own words, failed miserably. I think just let him do and gain more experience with this approach.

@numbr6

6 ай бұрын

Yes. I've unsubscribed and totally lost interest in anything MMXYZ related. Martin should have finished the MMX, with its known flaws, but at least that project would have delivered on the promise of the patrons, engineers, machinists who helped him over the past 3 (?) years. After completing MMX, then announce: I'm doing complete redesign using strict engineering practices, etc... MMXX is now our project. I could back that. Both me and my wife were interested in the MMX project, learning as you go. But just "trashing" the entire project when near completion was insulting and a betrayal to everyone who has backed Martin over these years. I don't know, and do not care, what is end goal is today. I have no confidence he will have a completed anything, ever. I'm a software engineer, and I know you at some point must "just ship it" with all of the known flaws and do overs in the next release. You pull this with your employer, and you are looking for another job!

@DavidJohansson

6 ай бұрын

Wintergatan next year: So I managed to account for the gravitational irregularities due to geo position on earth by using a sextant coupled to a gearbox. But forgot to account for the moon pulling on earth which will play 1 billion less tighter music depending on the tour dates. And the air humidity that affects the air govenor, temperature differences that affects the bearing inner tension, static electricity, radiation from earth, dark energy from outer space and the state of god. Will be solved with marble machine ultimate with the aid of super artificial general intelligence in year 2030. Stay tuned!

Remember that everything you tie to that power output is going to add resistance, so the gearbox will need to be tuned after all the power outputs are added!

@Alexander470815

6 ай бұрын

The question is: Is the added resistance constant? Likely not. So just putting a constant power into the system won't cut it.

@Ed.R

6 ай бұрын

Made me think the gearbox actually needs to be a CVT such as an enviolo. A control system of some sort to adjust its ratio to maintain a constant output rpm would be needed.

@cranemon

6 ай бұрын

Yeah I’m thinking he’ll end up hitting a wall if he continues on without addressing this. Unless he can come up with a very stable governor, the tempo will be all over the place.

@lucasherrera1989

6 ай бұрын

Yes, constant resistance is key.

@JBFromOZ

6 ай бұрын

I was sitting here wondering if the system is that twitchy with changes in the grease on the bearings, how are the differing power demands over time from actually lifting and releasing marbles at different rates e.g breakdown low demand, full chorus high demand? Not sure how the power requirements would be stable?

so we'll end up with a crank, stabilized with a flywheel, feeding a huygen drive, with another flywheel, slowed by a continuously variable air governor, that in turn runs a cvt, which then goes to the programming wheel. and we want to minimize the number of total parts. what a project

@thomasbecker9676

6 ай бұрын

Just wait until he tries to reinvent the transistor so it can be programmed like a computer.

@FLics-po5oc

6 ай бұрын

I really hope they go back to the escapement mechanisms that he used in the MMX, those would be perfect for this application (basically a clock) and use way less parts than a CVT transmission for your flyball governor that still will be off.

@barb0za0

6 ай бұрын

no progress will ever be made if he wants perfect timing. just program everything in midi at that point. nobody but him cares about this “issue”

@anotherdrummer2

6 ай бұрын

​@@barb0za0how else will he milk the project as a stable income. This is no longer about creating a live action animusic machine, but creating content.

@Elmo6855

6 ай бұрын

MMv1 is still my favorite. It's quirks, it's sounds of the mechanism, it's flying marbles. I was fine with v2 because I know v1 required lots of maint after each run. I saw the need to make something reliable for shows. But it didn't need to be perfect. Perfection is digital and digital has no soul. Sadly I've lost interest at this point. Would rather have seen some new music machines, not yet another version of the same.

Part of the problem that you have is that a governor should be dynamic, changing shape and therefore resistance as the speed increases or decreases, to try to regulate that. Notice that the ones you showed from the museum had springs that change the aerodynamics of the wings, making it have more air resistance when it's going faster.

@SilvaDreams

6 ай бұрын

Yup he needs a centrifugal governor that will adjust its "throttling" with the speed. But at the very end it seems he may already know that but hasn't implemented it yet.

@drewmanzara5731

6 ай бұрын

Came here to write this!

@drewmanzara5731

6 ай бұрын

Do you think the governor would be tighter if the aerodynamic braking had two bi-stable configurations. I’d bet money it would have more authority to confine the RPM’s within an upper and lower limit. I’m imagining such a contraption would flutter under steady state then quickly flip into one mode when the RPM’s drifted out of spec. Perhaps using two of such bi-stable governors would be required?

@carlchong7592

6 ай бұрын

There are also issues in that the fixed air governor (more like a retarder than a governor because of it's fixed vanes), cannot put energy back into the system. If the downstream mechanics draw more torque (say a lot of pegs triggering a lot of events), RPM of the system will slow and the air retarder cannot put much energy back into the system to help it catch back up to tempo. Instead, recovery of RPM can only come from the inertia of the flywheel which will respond slowly because of it's high mass. A flyball governor, like the one James Watt used, can rapidly put energy back into the system if there is a sudden increase in torque. The spinning flyballs are a flywheel which can rapidly change their moment of inertia which means that they can store energy and return energy while maintaining the same RPM. It would be important to connect a flyball governor to the output shaft with very little backlash in it's coupling to the output shaft so it can act quickly to return energy to the output system. The air governor, and air retarder, are drag only devices which cannot return energy to the drive system. Their vanes are too light to store much inertia and they can barely alter their moment of inertia through the movement of their vanes. The upside of these air drag devices is that they have unlimited capacity to burn energy while the flyball governor can only accept energy until their masses fully swing outwards. If the input drive, for some reason, continues to overdrive, the flyballs will slowly swing further outwards increasing stored energy, until they reach the full extent of their swing arms and no further energy, and capacity to govern ends. I could see a flyball linkage which engages an air retarder to provide some unlimited energy burn though.

@DevinDTV

6 ай бұрын

even without changing shape/angle with speed, it'll still act as a governor because air resistance is velocity squared but yeah his governor design is dubious. why even use an air governor if you can use a fly ball governor? air governor with introduce a lot more noise

You're slowly but surely building a giant clock that plays music.

@FLics-po5oc

6 ай бұрын

Yes, next video will be him discovering clock escapement mechanisms even tho he has used them before. It was called out 4 months ago when he first showed off the Hyugen drive.

@thomasbecker9676

6 ай бұрын

He's not building anything. He's inventing new ways to procrastinate.

@ColossalZonko

6 ай бұрын

@@thomasbecker9676 hahahahahahah, best comment I read all day. spot on

@axiom1650

6 ай бұрын

It was called years ago

@Nooticus

6 ай бұрын

unfortunately so. @@thomasbecker9676

FLY BALL GOVERNOR FTW! I'm so glad you included that hint at the end. I didn't even know what a governor really was until I tasked myself with fixing my mom's antique gramophone years ago. Such a simple but genius mechanism!

@boonjabby

6 ай бұрын

I really like the use of them in clocks and elevator safety devices !

@Edmar_Thorn

6 ай бұрын

Honestly always thought they were for decoration...

@IvoTichelaar

6 ай бұрын

Its name is a winner as well 😂

@Oosh21

6 ай бұрын

Another one that uses gravity.

@fabianruckstuhl1351

6 ай бұрын

Thats definetly the way to deal with the speed increase

You mentioned that you want to quantize the output using a governor, but it is still delivering a continuous output. To truly quantize the power, you need to add an escapement mechanism leading to the programming wheel so that it ticks like a clock. You could then adjust the escapement (changing the tension on a balance spring, or changing the length of a pendulum) to deliver whatever tempo you want. Adding a centrifugally governed clutch to the flywheel to prevent the weight drive from "overwinding" it would also prevent overloading the escapement.

@FLics-po5oc

6 ай бұрын

This should be way up

What are you going to do about uneven mechanical load? Adding or subtracting various instruments will change the energy needed to run the system. Also tempo changes will be very difficult. The first one will be quite interesting to deal with, the second can probably be dealt with using a continuously variable transmission and a crank/dial to adjust tempo.

@lennytheboss100

6 ай бұрын

yesss!!!! Why is nobody else mentioning this? As the load varies (adding/subtracting instruments) the bpm will surely change. The huygens drive creates a perfectly stable input torque but thats not gonna result in tight music if the resistance of the system is all over the place. At least i imagine it to be like this. Am I mistaken here?

@TlalocTemporal

6 ай бұрын

A governer can change load based on speed, so it should be possible to have a purely mechanical system that balances at a consistent speed despite variable load.

@thomasbecker9676

6 ай бұрын

@@lennytheboss100 People have been mentioning it for months and months, but Martin doesn't want to hear from anyone that isn't in lockstep with his mindset.

@StuffandThings_

6 ай бұрын

@@TlalocTemporal That might end up wasting a LOT of energy depending on how much the load varies, resulting in an even bigger weight. It also will not be an instant process, as the governor will need to gain/lose angular momentum which will take some time to happen (this can probably be nicely modeled with some differential equations). Even worse, a larger governor will be slower to respond. You could maybe put a series of governors of different sizes to balance this effect out, so that smaller governors can respond to finer adjustments and help keep up with the slow response time of larger governors, but this could get complicated and messy fast. And that isn't even taking into account all the vibrations that governors could cause, with such variability its not far fetched to accidentally end up with resonance (VERY BAD) under certain conditions. The thing is, mechanical instruments with such precise timing and complexity are... difficult to say the least. Most automatically playing instruments actuate things based off of air or steam or something iirc, which would have far less resistance, and even carillons, which are huge and take a ton of energy to work, are a very simple mechanism (basically just a rotating drum). What Martin is trying has the worst of both worlds, and I'm really not sure if there truly is a purely mechanical solution here. Complex machines that fine tune precision like this within a complex system under heavy and inconsistent load, like cars, tend to be computer controlled for a reason. But needless to say there's no way that Martin would ever consider this. Honestly, a thought that I've had is to perhaps have a gearing system that works in conjunction with the levers - so, pulling a lever to activate an instrument not only activates that instrument but also changes the gearing from the mechanical input. Once again very complicated but I genuinely don't think there is a simple solution for this, at this rate he needs an entire engineering firm to make this work lmao

@GBR6000

6 ай бұрын

No. The difference is simply in the number of marble gates opening. If the marble gates are made well they will require minimal force to operate. I don't beaver that the difference between 10 and 20 marble gates opening will matter for tempo. Especially since they are going to be actuated using a bowden cable which are extremely efficient at moving force. Additionally if really necessary an extra part can be added to each of the gates to make turning off an instrument independent of gate actuation.

The first Marble Machine is what blew up, went viral, captured our imaginations and interest. Even with all of its flaws and imperfections. Thinking of it as a pure failure and something that needs to be moved AWAY from is spitting on the thing that brought us all together! The first Marble Machine was incredible for a thousand reasons but for me the main 2 were that it was an incredible feat of engineering and secondly it blurred the line between an instrument and a machine. Hand cranking the machine to keep the music on time, flipping switches, playing the base, all of it felt time I was watching a guy play the most complicated instrument in the worl and it was amazing. The current Marble Machine is slowly drifting farther and farther away from an instrument and more into a machine. Watching a guy flick on a machine then hearing it play music is nifty but only interesting for a few moments.

@Bbeaucha88

6 ай бұрын

The MMX genuinely felt like an amazing step in the right direction and still kept the soul of the MM while vastly improving on the design and function. I wish Martin would pick up the MMX again after his team spent so much time tweaking/tuning and getting it running smoothly.

@andreaswiklund7197

6 ай бұрын

Hey guys, the journey is the goal here 😊

@justinnaramor6050

4 ай бұрын

The first marble machine "blowing up" does not at all change the fact that, as a musical instrument which could be relied on, it was shit. Period. the MMX was, well, only a bit better to say the least. The first machine was a mechanical failure. Get over it. Stop overglorifying it. The MMX was also a mechanical failure. No, I'm not saying this to be a hater. I'm saying this because it's just the truth. The first MM could hardly play that song we all know and love (or maybe hate, for some people). You're seriously gonna consider that "a beautiful instrument" and act like that wasn't a big deal? It was too a big deal! No musical instrument should function so horribly if a musician is to take it seriously, and in Martin's case the whole marble machine project was always intended to be a musical instrument, not just some beautiful visual art sculpture with the ability to play music as "just a cool little feature". No, the musical ability is the top priority. Marbles falling onto musical objects is simply the methodology it uses for playing said music, because, well, it's cool, and it's visually interesting, ETC.

@Bbeaucha88

4 ай бұрын

@@justinnaramor6050 multiple times you reiterated my main point. I'm not saying the MM was perfect and that he should just replicate it. Obviously it was a mess and needed to be dramatically improved. However current Martin IS just slowly creating an art sculpture that plays music and not an instrument. All of this however is just my opinion and speculation.

@justinnaramor6050

4 ай бұрын

@@Bbeaucha88 So because the new MM is no longer being hand cranked you consider it "not an instrument"? That's a bit of an odd way of looking at it, to be honest. People said the same thing when the MMX had an electric motor attached. I'm sorry but that statement is bullshit. The hand crank's function is purely as a power source, it does nothing at all whatsoever for the music other than maybe tempo. Removing the crank shouldn't automatically remove the ability to have tempo adjustment. Just look at DJs for goodness sake... they can freely adjust tempo by adjusting the speed of the turntable even if it's not manually cranked. Also, removing the crank allows both hands (and maybe even feet, if we're removing foot pedals as a power source as well) to do more stuff musically. The hands could manually operate the keys which open the marble gates, like what we've seen Martin do before, or the feet could even control drum beaters with foot pedals (when the drums are not activated by the marbles and the pins on the programming wheel). All this only adds more musical functionality... removing the crank/pedals as a power source shouldn't automatically mean it has to be yet another "boring" music box that really only acts as a "playback" device. Even just the ability to manually change the pitch of the bass strings, using the fretboard, was itself already a really cool feature that made the first MM so unlike most other music boxes that are literally just playback devices. I sure cannot think of any other music box that allows for that much a degree of musical control. The closest similarity I can think of would be player pianos or some mechanical organs, which have keys that can either be operated by pinned cylinders (or any other musical program method), or they can be played by hand. Adding a non-manual power supply to these instruments, including the marble machine, shouldn't automatically make them lesser instruments.

You're going to be powering a couple of marble elevators, and a programming wheel. The programming wheel needs to turn at a rate that is tight, and will have a tight energy draw. The marble elevators can be sloppy, and will have a sloppy energy draw (depending on the number of marbles in each lift at any given time). I'm not sure you need to care about how tight the flywheel is. The marble lifts are going to add slop to your power train. I think whatever mechanism you use for tightness ought to be as close to the programming wheel as possible, and it needs to accomodate for a sloppy power input.

@mrpraefekt6252

6 ай бұрын

Exactly. The machine could play 10 marbles or 1000 marbles per minute -> the power consumption of the lift will vary by a lot. Having a tight flywheel while testing with a constant power consumption isn't doing anything for the marble machines tightness.

@GordieGii

6 ай бұрын

@@mrpraefekt6252 I've said it before, and I'll say it again: There needs to be some kind of isolation between the ball lift and the programming wheel drive.

@XGigabowser51

6 ай бұрын

Just slam the marble lift to its max capacity all the time, lol. Recycle the overflow back to the bottom again

@larsk.6619

6 ай бұрын

You could have a flywheel that is powered like the metal prototype in germany, that drives the marbel lifts and use the output of this flywheel to power the new drive system, that only drives the programming wheel. With this you get a pre smoothed power input for the new drive (stop the wavenes of the output) and you dont have to wory about variating loads of the lifts.

@andrey5405

6 ай бұрын

Powering the lift with the crank would solve the issue with changing number of marbles in the lift leaving more constant power input to play of music.

The weight is not just filtering your input, smoothing it out but it's storing it like a battery.

@Lttlemoi

6 ай бұрын

It basically behaves like a mechanical version of a capacitor

@splynncryth

6 ай бұрын

It's kind of like a mechanical capacitor :)

@martinshoosterman

6 ай бұрын

It is 100% useless as a battery for this purpose though, as using it on battery power will sound terrible as the music immediately starts slowing down

@Ed.R

6 ай бұрын

@martinshoosterman That's making me think something like a CVT with ratio control feedback system based on output rpm is needed. Or basically the mechanical equivalent of a voltage regulator.

@bradley3549

6 ай бұрын

@@martinshoosterman The music won't slow down until the weight hits the ground. So as a battery it's great if you had enough vertical space to hang a weight for an entire song.

I have a very "Good enough" approach today for a lot of the DIY/Hobby projects I have. I've had to reduce my personal perfection aspects to make sure I leave room improvements or advancements... But also not overly complicate things or over scope my goals... I'm amazed how much your still improving things, but is it going to conflict with your perfectionist ideas? Always rooting for you and your projects, they're amazing content and music!

Non-perfect timing is nothing bad at all, it makes music sound human, and there is no point in sounding tighter than a skilled musician. I also never even noticed any off-timing in the first MM

@99xara99

6 ай бұрын

I think this is a common big misconception, but the first MM video is heavily edited and stitched together out of dozens of seperate takes for the sake of a clean outcome. It did not work properly. And neither did/does the MMX, it keeps breaking down and has to be fixed all the time.

@chrismofer

6 ай бұрын

there will always unavoidably be imperfections and randomness in the output of a rube goldberg machine, but to have any chance at reliability and durability it should be made to the most exacting design possible to begin with.

@richsackett3423

6 ай бұрын

Yeah, there were several spots where the MM slowed down unmusically.

@FreejackVesa

6 ай бұрын

So what though? That's what made it charming, at least to me, the non-perfection of it.

@justinnaramor6050

4 ай бұрын

@@FreejackVesa You obviously don't know shit about making a musical instrument or something, considering you take such issue with "tight music" (and hundreds of other commenters, for that matter, are just as ridiculous). The machine needs to be dead-on accurate on its own first. Then, if any musical imprecision does occur, it can be stated confidently that said imprecision is the result of whatever the operator does during the task of operating the machine, like changing the speed of the drive mechanism unintentionally (or sometimes intentionally, for any reason), or by mistakenly having misplaced pins on the programming wheel. All of this also applies to the issue of dropped notes or falsely triggered notes; this absolutely must be because of errors from the human operator, not because the machine is doing this stuff on its own (I.E, a flaw in the mechanical design). These are 2 completely different types of imprecision that, really stupidly, hundreds and hundreds of commenters are totally confusing. Literally anyone who's spent even a relatively small part of their life building a musical instrument would surely agree with what I'm saying.

MM1 - Plays fun song with tons of views on youtube MMX - Almost finished, but lost in chase to perfection MM3 - devoted to perfection before it's even started ... MM20X6 - "I have studied the ways of the temporal shaman and through 7 years of meditation I plan to adjust the gravitational constant in my workshop. After which we will redo the marble drop tests with my quantum nano tester I developed in my trip to space during MM2001 prototypes"

Just wanted to say that the BPM precision battle is not an uphill one, but an unreachable one. Once you're in the milisecond range, you might not need to worry any longer, seeing that contact with the programming pins will make it look like the number gets worse when you install the midi wheel. All of them will behave as micro impacts that slow down and overall destabilize the wheel by the amount of notes being played in the moment. If your intro has a few notes, in comparison to the chorus of a track, the chorus will play slower, by a miniscule amount, but slower. Depending on what is being programmed, some may even slow down the machine's spin too. Be it weight or added friction.

Two thoughts: 1) You're measuring accuracy in absolute time (ms) over a relative time (period between transients). Measuring that way, the higher bpm will always be "tighter". What you want is the relative tightness. 2) As others have already mentioned, when playing songs the load will vary, which is equivalent to your wind governor growing bigger or smaller. The Huygen drive is, as you said, a way to filter your input, to achieve (near) constant force. But you want constant velocity(tempo) under varying load, which becomes a trickier problem, some kind of feedback/compensation is needed.

@Pootie_Tang

6 ай бұрын

agree, we all should reserve absolute tightness to some kinky stuff

@dvl973

6 ай бұрын

He could use an escapement, that would give him the perfect tempo no matter the load. It would just tick very loudly 😂

Great to have you back, Martin! Love that you are getting excited for finishing the project!

I'm very happy, as a long time follower of the entire marble machine project, that you haven't lost steam and still plowing through. That said, it feels like this channel became a mechanics channel than a musician's channel. You've been venturing in the unknown territory of mobile, compartmentalized semi-automatic mechanical muti-instrument for so long, and like many of the viewers have been telling you, feel like it has lost the artistic side of it. There were many times you've sit down talking about engineering failures, but in terms of music we haven't gotten any "real" ones from MMX. Accuracy at this scale with marbles would prove and have proved insanely challenging, and borderline impossible to maintain between tours. I really hope that instead of fighting with the nature, you choose to embrace it. Even if tight music is attainable, you can put that effort into creating a better "marble machine" in terms of goldberg machine, or simply finish it quicker. I think people would understand even if your machine never leaves the workshop if we could hear a proper album coming from it.

@justinnaramor6050

4 ай бұрын

There's a reason why he's talking so much about mechanical ideas. It's just a part of building any musical apparatus. Sorry, but these viewers complaining about how "the artistic side has been lost" kind of shows how little they know. Bet some of them have never built a mechanical music apparatus in their lives, to be honest. If you're gonna build a musical apparatus you need to discuss mechanics.

Two quick suggestions : - as people said, calculating deviation in ms at different speed doesn't really tell you if it's tighter or not. Halfing the deviation in ms by doubling the speed is actually not an improvement. That's because when you then add a gearbox to divide the rotation frequency by two (doubling rotation period) you will double de deviation in ms as well. A more useful metric to measure the tightness would be the deviation in percent of the duration of one fly wheel turn - maybe try running some longer tests to see if it stabilize over time. Like an hour minimum. Maybe use a electric motor to turn the crank to make long tests possible.

The new machine seems way more complicated and will end in tears again.

@danielwilson5102

6 ай бұрын

I agree. This is going to be the pattern from now until eternity sadly.

@br52685

6 ай бұрын

The goal here isn't any completed device. It's simply to sustain engagement on his KZread platform ($$$).

Always appreciate seeing progress in your videos, frequency of posts and length of videos isn't necessarily important. Just happy to see you pursuing your passion

@thomasbecker9676

6 ай бұрын

What progress are you seeing?

Martin, every time I see your persistence on quality, I get amazed. Please be well and continue your beautiful project.

This channel feels like the Groundhog Day.

instead of air speed governor, try an eddy current brake using magnets. They are adjustable on the fly if needed

I'd like to point out, in case nobody else has, that you're going to need a device in the transmission that can maintain a constant output speed, because once the machine actually starts playing, you're going to have a constantly variable load put onto it. Without something to compensate for that, it will have the potential to end up sounding like a music box at the end of the spring winding. (Plus you have to compensate for the energy it will take to maintain all the marbles at a given tempo)

@drewmanzara5731

6 ай бұрын

If that governor fan had centrifugal weights and springs to modify the damping as it’s speed fluctuates it could compensate for the varied load. Perhaps some sort of bistable spring and lever arrangement would allow the same mechanism to /quickly/ adjust both over speeds and under speeds. I’m imagining a best case scenario where the governor oscillates or flutters between the high and low drag configuration at steady state.

Thank you so much for the detailed explanation of how this drive system works. Can’t wait to see you playing your new machine. Please keep at it. 😊

I still can't believe that after hundreds of episodes between MMX 2 and this one that the input is still correlated to the output instead of the flywheel just holding power and something else tapping into that energy at a steady BPM that it would ask for, a bit like the flywheel bike where the flywheel is just stored energy and the accelerator takes a little bit of energy or a lot of energy at a completely steady rate which also leads to another thing often commented. That means it still keeps the machine human, it's still Martin adding the power into the system by method of a crank or pedal but the output would be perfectly accurate as long as there's enough energy in the system to keep providing the asked for amount.

@TlalocTemporal

6 ай бұрын

So you mean instead of a governor managing the speed of the drive shaft, the governor should manage the gear ratio of the gear box? If there's a CVT that can reliably handle these loads while also being simple and cheap to repair, that could be a more robust system.

@dvl973

6 ай бұрын

Have you ever seen a mechanical clock? You're kind of describing the escapement of a mechanical clock. In a big clock that is a pendulum ended with a pallet fork and escapement wheel going back and forth always letting go a little bit of energy from the hanging weight limiting the beat of the clock, usually to 60bpm or 1Hz. In a mechanical wristwatch, it's a balance spring with a circular weight being driven by a power spring or a set of them, connected to a pallet fork with escapement wheel which goes back and forth, limiting the balance spring oscillations to 4hz or 240 bpm always allowing just enough energy to "escape" the power spring to run the clock. These forces are tiny but imagine the forces needed to run a huge multiinstrumental machine like this one as tight as a wristwatch. Holding a beat so tight mechanically, accurately and reliably is difficult no matter the design, because it's always an incredible balancing act of the thing that makes it tick - the pendulum or the balance spring. The more accurate your pendulum/balance spring is manufactured, the more accurate your beat will be. This system is more or less the same in that regard but with larger forces and I also think it's simpler than an escapement which needs to be very precise and inherently has a limited set of oscillations. With a governor you can control the amount of energy continuously, you can easily correct for new loads in the system, being able to handle higher loads, and it should have a higher BPM range while also being easier to change it. Escapement is very undesirable in clocks because it TICKS. An escapement big enough to handle the load of a machine like this would have a beat of its own 😂 which is why all the music instrument manufacturers probsbly opt for an air governor or pendulum governor instead. But it's true that escapement could make the machine run incredibly tight.

@deathofallthingspotato9919

6 ай бұрын

@@dvl973 The best way I can see of using an escapement would be having a relatively small escapement driving a shaft, then taking the offset between that speed and the output of the drive system, and feeling it into some kind of mechanical PID. You could get the derivative term from taking the difference in speed (literally use a differential), attaching it to a flywheel with a clutch, and comparing the speed of the flywheel to the speed of the output - again using a differential, then measuring the speed of that with a clutch to a sprung wheel. Get the proportional term from a clutch on the difference in speed on a sprung wheel. Get the integral term from wheel attached to the speed difference geared down really far, and set with e.g. a torque limiter and a pair of end-stops. Then add them together (mechanically easy) and attach them to the input of a CVT (or even an IVT, but that might be hard), or just attach them to a regular governor, or maybe threshold them with some hysteresis to a gearbox and a governor - though that would be very complex.

@rasmushaversen

6 ай бұрын

Couldnt agree more, it feels like he has misunderstood the purpose of the flywheel. Sure, it can smooth out jitter, but more importantly, it's a battery which can be uncharged at a steady rate.

@deathofallthingspotato9919

6 ай бұрын

@@rasmushaversenA battery is an alright analogy, but it actually behaves as a capacitor, as it discharges/charges proportional to the torque applied. (Note that this is true to such a degree that the physical flywheel effect of a motor in a circuit is effectively that of a very large lossy capacitor).

I'm always so motivated after getting some good wishes from you at the end of each Video! 😊 thanks a lot for beeing so inspirational❤ Carry on for Glory!

Love the precision involved. Congratulations!

Spring loaded fins as airbrakes. The faster, the air break spins the longer the fins get the slower it goes the smaller the fins get The strength of the spring can be finally tune to get the exact rotations per minute you want

@ChaosUnit178

6 ай бұрын

I just posted this, but it doesn't matter which comment he sees, so long as he gets this information, so I'll upvote your comment too

Nice to know that the gap between videos was because of the moving, and Martin has not been run over by his flywheel yet

Thank you for the update Martin! I am glad you are doing well. I am super happy you have finally settled on a mechanism. Sending Love and Happiness your way! BTW, I wore my MMX shirt yesterday but felt it was still "clean enough" to wear today... What a coincidence that this update found me while wearing your shirt while promoting my love for your work! hahaha

Glad you’ve been able to do so much despite the chaos in your life. Praying that you can feel good with balancing it all!

I've said it like 20 times and I'll say it again, Martin, if you keep looking for tight music, just use MIDI sequence, you will never be able to calculate and account for human error, and I remember that once you said "I'd like this machine to play as tight as a human can" and that's exactly where no matter how "tight" the machine plays, if there's a human operating it, there will be error, and that's what makes it better, if you keep working towards this unreasonable goal, you will never get absolutely anywhere, the MMX was the best option, and IMO you should've stuck with that version.

@justinnaramor6050

4 ай бұрын

And I'll happily say this again: that error must be the direct result of the operator's actions. Not because the machine is erroneously causing the tempo to drift with nothing instructing it to do so. So, sorry, but your comment is invalid because this whole "tight music" thing, as it relates to the marble machine, literally has everything to do with the machine accurately following the instructions given to it, and nothing to do with "hey, let's make it perfect, like a robot, where the tempo never, ever, ever deviates, because tempo deviation is such a huge flaw in music even if it's intentional". Literally every single video of Martin talking about tight music on the MM involves him doing something with the mechanical components. This fucking proves my point. If you don't think it proves my point then I've no idea what to tell you.

Looks very interesting! What could be difficult in the end: When you have to rotate the crank at a given speed which is not in time with the music, that could be very difficult for your brain.

@Robbedem

6 ай бұрын

That's why organs and such do not have direct link to the input and output. When you crank or pedal on an organ, you are increasing air pressure. That air pressure is then regulated to come out as needed. This allows the machine to play tight without having a precise energy input.

@dannykodicek9081

6 ай бұрын

Yes, I was worrying about this too. But if you think about, for example, a bagpiper, they seem to be able to cope with this. They have to re-inflate the bag at a speed that's largely independent of how fast they're playing.

Doing great man. Love the journey.

We love Wintergatan Thursdays

I'm not sure why we need super tight music and at this point I'm too afraid to ask. With the reliability improvements I got it but the speed of the music can fluctuate and not cause too many issues surely? Great video and engineering!

@FLics-po5oc

6 ай бұрын

Yeah, Martins main issue is that he has a very vague requirement “System shall play tight music”. The reality is that he should define “tight music” as in “BPM variance shall not be bigger than 2ms” stick to that and move on.

@InvitusCode

6 ай бұрын

I think, that he is loosing the whole Point of an mechanical Instrument / his journey. More reliability -> seems like a good idea.

@MammaApa

6 ай бұрын

He lost the plot a long time ago.

@justinnaramor6050

4 ай бұрын

Tempo fluctuation should happen under the control of the operator though. I firmly believe that's really the whole point. This has literally zilch to do with making the music sound "robotic" with zero tempo deviation whatsoever. Same is true for the issue of dropped or falsely triggered notes... it has zilch to do with "perfectionism" as folks here are calling it.

One thing you could consider doing is hooking up the governor to a cvt or other variable transmission system which would output the same tempo and torque reliably regardless of the speed of the huygens drive.

@coolfrost6

6 ай бұрын

I agree that a CVT could be a good option for Martin to have the option to change the tempo on the fly, if he want to for artistic reasons during a song

@Ed.R

6 ай бұрын

@coolfrost6 I'm thinking enviolo bike hub and a mechanical feedback system to adjust the ratio based on output rpm. Then he would have the mechanical equivalent of a capacitor and a voltage regulator.

Thank you for your kind words, Martin! You really are inspirational to me! Wish you the best of luck in your projects! Best Regards from Brazil.

I 100% understand the stable work life problem, and hope things settle some and work out. Excited to see what comes next!!

You could still use the foot pedal driven approach to lift the weight up... just use a smaller flywheel to smooth out the force applied to the lifting sprocket. This way you can still keep your hands free to play other instruments, maneuver levers, etc.

@Wintergatan You should really test with variable load. If you can play tight music but only if the load is constant it will not be useful for a marble machine! And I don’t think that all governors are equally good against variable load.

Good to see you on work again! Good test you make there

Thank you for this excellent video! We're looking forward to good things to come! 😃⚙

He’s alive!

@DominusFeles

6 ай бұрын

Now in a brand new hat.

Wait what 😅 ME engineer here, your drawings and explanations are really wrong. Pulling on one end of a chain will not accelerate the machines if the weight is not fixed on the floor. If you pull fast there is the just the moment of inertia of the lower right gear to overcome, but for this application it's almost negotiable. In the end you just pull the weight up. So it doesn't matter if you pull fast or slow, it does not change the tempo. The only thing that drives the machine is the gravitational force of the weight. And the machine will continue accelerating until there is enough friction in the system that balance out the input force of the weight. So in a system without friction this would accelerate infite because of the weight. By adding this air govener you create a lot of friction which slows the system down and brings it into a stable state. I thinks it's extremely important to understand what's going mechanically when using these mechanics.

@glenmcgillivray4707

6 ай бұрын

Actually. Pulling the track applies an upward force on the weight, but tension within the chain will be conserved. If the upper gear was static, then the force is multiplied thanks to tension. As a result of it being mobile, we do actually expect some force variation applied to that gear. That said. Martin would see less variation in gear force if he increased the pulley ratio to his primary weight, and require less force but longer motions to reset the counterweight. This will still multiply the input energy, but spread the acceleration of chain through more linkages and even out the forces. A chain or belt in tension is under equal tension only till it hits the loading elements. But you don't expect a tank track on a T34 to have variation in tension above and below the forward idler, it'd just rotate till both sides are under equal force. And that is why pulley loops multiply force input as you add lengths of rope, rather than merely adding a singular force input. I'm pretty sure you understand the physics. But Martin's interpretation isn't actually that far off. Personally I would have gotten Martin to graph out his flywheel based on fin length to create a chart of size vrs RPM in his setup as a guide for further experimentation. Also might suggest Martin go for a gravity powered governer. As RPM increases the governor would expand wider against gravity, and with some smaller fins add increased air friction to the system and help stabilise his flywheel with his purely manual power input.

@fletchro789

6 ай бұрын

​@@glenmcgillivray4707yes, like a Lily clutch on a mine hoist drive. Or on a steam engine. They have arms with weights on the ends. As speed increases, the weights swing outward. This increases the moment of inertia and also air resistance. It stabilizes the rpm better.

@axiom1650

6 ай бұрын

It's a dynamic system, pulling up the weight will definitely influence things.

@onemadscientist7305

6 ай бұрын

Intuitively, if the weight pulling the chain down has infinite inertia, it will behave exactly like the ground, and tension will be conserved along the chain, yanking the the top gear down exactly as hard as the crank pulls up. Conversely, if it weighs nothing, then it is an empty chain and you are simply pulling the individual links up, which can obviously transmit zero force through, and the chain is simply accelerated upwards. With a finite mass, there will be some additional tension passing through, but not all of it, and you will also use up part of your force balance into acceleration, especially if this happens quickly. It’s unclear exactly how stable this is unless you calculate through it, I think, but clearly, from experimental data, it is sized properly, and smooths the signal over time as required. Two separate problems I identified: first, he states there will be constant moment of inertia. Technically, in absolute terms, this is true, but after going through reduction, this will not hold on the axle downhill from the power chain! The driving flywheel will « see » less resisting moment of inertia if the speed is stepped down at the output, or more if the speed is stepped up (by the square of the reduction ratio), and this will mess with the timings. Second, the higher the inertia of the flywheel, the longer it will take to reach equilibrium with the governor as it takes more energy to get to the same speed and the power loss rate through the governor goes up with its spin rate, so assuming power input through the crank stays the same, the target governor rpm and thus flywheel rpm will be the same for equilibrium, and so it will take longer to « fill up the tank », so to speak.

This, this is the kinda thing I love. Entirely mechanical, with a useful purpose. This is absolutely amazing. Please keep up the amazing work man you got this!!! And sorry to hear about your relatives 😔

I'm glad to see you found a way forward with the Huygen weight drive - really nice prototype! It still blows me away to see how much the community built around the marble machine has shown up with their time, brain power and resources. It truly is a thing of beauty :) Hope you and the family are doing ok - hugs to you all

@justinnaramor6050

4 ай бұрын

It still blows me away how the community built around the marble machine is falsely accusing Martin of being a "perfectionist" and criticizing his goal of "tight music", with absolutely zero understanding of what that really truly means in the context of this machine. Lol!

Combining the Huygens drive with the Governor and flywheel is the best solution i think 🙂.

@FLics-po5oc

6 ай бұрын

Best solution is to use the Hyugen drive with an escapement clock mechanism, way simpler and more reliable. Governors will still change BPM depending on the load of the machine, and will be more complicated to implement.

@Robbedem

6 ай бұрын

As I see it, the Huygens drive and flywheel perform the same function => storing energy So one of them is redundant and should be removed.

You can add a variable ratio gear between the crank input and the gear that turns the chain, that changes based on the height of the weight. This way you can create a mechanical close loop control system with negative feedback that speeds up or slows down your input depending on the height and thus helping you keep steady height even easier and with big margin for error. Finding the correct mechanism to implement this can be tricky but since you are on the path of the tightest machine it can be worth it.

@Ed.R

6 ай бұрын

I was thinking of something like a CVT such as an enviolo bicycle hub. The control of the ratio could be some sort of mechanical feedback as you say.

@panakon366

6 ай бұрын

@@Ed.R I am not familiar with that mechanism. What I was thinking was something like a ball at the end of a rod running against a wheel and depending on the distance from the center the speed changes. If you make it move with the height of the weight you have a system that can adjust its speed to keep equilibrium.

Love how happy you are with your good engineering! This has been a long time coming, can't wait to see it all come together :)

This video is fantastic you’ve taught me some incredible mechanics!

If you keep this flywheel at a constant speed, why not just use a motor? I figured the manual input was for haptic feedback in actually playing music, if you are abandoning that for “tightness” just use a motor. Seems like this solution to your problem makes the problem less relevant. I really question whether tightness should truly be the highest priority in general. A drummer isn’t exclusively ranked on their tightness, of course there is problems at extremes, but as long as everyone in the band follows the drums (aka you input timing) then the listeners probably will never know any timing deviations. Much less the ms you are measuring in these videos.

@RoyaltyInTraining.

6 ай бұрын

I honestly don't understand what Martin is trying to achieve anymore. The original Marble Machine and the MMX had all the charm of a human performance, but this is getting further and further away from that.

@sarabanks971

6 ай бұрын

SAME! I thought the charm of mechanical music is its imprecision. Maybe that's more about time period than about mechanical music. Just make a clock, or a metronome! I guess the idea is "what if digital music never happened. What would contemporary DJs look like in that alternate history."

@katherineannerivera1926

6 ай бұрын

Agreed, I truly don’t understand why he doesn’t just use a motor at this point. This system already loses the charm of it sounding like it’s being played by a human. It feels like an MMX issue of being attached to the aesthetic of having a crank rather than the practicality of being able to change and control tempos, have constant power output that isn’t dependent on what instruments are loading it, etc. Love the discussion of the mechanism, but I truly think it’s a distraction that won’t work long term.

@lo-firobotboy7112

6 ай бұрын

@@RoyaltyInTraining. Yup. It's a ridiculous exercise in stumbling through amateur engineering instead of creating a beautiful and inspirational music making machine. The original machine had charm, elegance, and whimsy. Subsequent iterations have lost all of that.

@lutherblissett8824

6 ай бұрын

If the flywheel has to operate at a constant speed while the music tempo is regulated by a gearbox - how are you supposed to play this thing? Cranking the crank with one hand at one constant speed while playing your instrument at another tempo...?

you can add a second huygen drive for pedal input so the forces equalize even more before the flywheel (a bit more friction tho). and wondered if you are considering bike chain drive for the transmission. its designed to be man powered so its quite efficient and as added benefit you could change gears without using a clutch or stopping the flywheel.

Somehow I've lost track of you and your machine. Looking through the old videos it looks like it's been around 5 years now. Last I saw you were in France. I don't know if I can catch up, and I may have to skip a few, but time sure flies.

Great time to be alive. I love this. Hope everyone is well now Martin vs the machine vs ww3.

I Wonder if this mecanism Can power the whole machine while keeping a constant bpm. It would be interesting to test with a varying load (a simple motor + breaking resistance should do the trick) to check that out

A little off topic to the video, but I'm so glad that Martin has been taking full advantage of 3D printing in the past year or so. It seems like he's SOOOO much more efficient in creating prototypes now than when he was just using the CNC.

@thomasbecker9676

6 ай бұрын

3D printers *are* CNC.

@TlalocTemporal

6 ай бұрын

​@@thomasbecker9676-- Same technology, yes, but they're never refered to as CNC printing machines. CNC here is refering to the CNC router.

@thomasbecker9676

6 ай бұрын

@@TlalocTemporal Yes, it's a router controlled via CNC. Just like a CNC lathe, or CNC mill, or CNC plasma cutter. It's sort of like how the general public calls remote-controller quadcopters "drones," even when they aren't.

@nandakoryaaa

6 ай бұрын

@@thomasbecker9676 A drone is an aircraft without any human pilot, crew, or passengers on board. Remote-controlled quadcopters are therefore drones.

@TlalocTemporal

6 ай бұрын

@@nandakoryaaa -- Here I was, going to correct you and say "drones must be unpiloted, not just unmanned", but turns out you're completely right. Drone refers to the sound they make, not their behaviour. Calling non-flying autonomous vehicles "drones" is the incorrect usage here. TIL.

Sending much love. Caring for family is more important than anything else ❤

Absolutely awesome. I love it. It's amazing seeing someone dedicate ~10 years of their life to do something no one has ever done before and do it amazingly well. The final result a couple years down the road will be without a doubt within the top 100 most amazing feats of engineering a single human has ever done and that is worth the grind. You'll be a part of history Martin!

@br52685

6 ай бұрын

The fanboy delusion is real!!

Have you considered using a one-way clutch bearing rather than a ratchet on the crank? There may be downsides, but it would at least reduce noise in this mechanism

In the case of the bearings I would be interested in the effect of oil based bearings like SKF's or even a dry based IGUS bearing.

I wish you and your family all the best! I hope they have recovered

Fan since 2017 coming back to see how Martin is doing. Don’t drive yourself crazy buddy.

Hi Martin, One thing you may not have considered is the air drag may change depending on where the venue is. Altitude could cause the air to be slightly thicker or thinner meaning that you may be better off going with a different governor.

@chrischris1722

6 ай бұрын

The only solution I see is enclosing the machine in an airtight container and applying a vacuum

@Bwilkinson036

6 ай бұрын

​@@chrischris1722 but it needs the air resistance to be a govenor, why put in vacuum?

Will Martin be sticking to the air governer in time? I can think of issues using the air around air density/pressure based off the time of year or altitude if your on tour.

Martin, great to see that you are still motivated by the project! From my mechanical engineering background I agree with your assessment of the source of the speed drift. The goal should be to have the flywheel spin as slowly as possible, while the governor spins as fast as practical. This supports that you intend to have a heavy (High rotational inertia) flywheel to store the energy required for "buffering" the Huygens drive force input. The plan to have the rpm constant makes a lot of sense in regards of tuning this system but adds the need for a complex assembly down the line. Maybe you find a governor design that can easily be tuned for an rpm range which suits your needs. I have a wind up music box at home from a cheap childrens toy at home, which uses a governor of fast rotating weights bound to a shaft with springs. As rpm increases they extend outwards and at some point begin rubbing on a surface which brakes the system. This might be the principle of the device you showed at the end. This could be easily adjusted by changing the distance to the surface which the weights rub against. Take care of yourself and keep having fun and making art! Paul

@-Tele.gram_OfficialWintergatan

6 ай бұрын

I GOT SOMETHING SPECIAL FOR YOU-

Been following this project for years now, but as time goes by I have started to feel like I am watching these videos out of a sense of obligation, rather than a place of genuine interest. I do hope you get this machine finished one day, and if it happens I will certainly be there, but for now the obsession with perfectionism is sucking all of the heart and soul out of the whole project, and it just isn't fun to watch anymore. All the best Martin, but this is where I hop off the train

Nice, for the gearbox maybe look at a continuous variable transmission (CVT)

@Ed.R

6 ай бұрын

An enviolo bicycle hub came to mind.

have you considered using a machined Olympic style barbell plate as a flywheel? You can get them from 2.5 kg to 25 kg, they have a ~ 27 mm hole and have a moment of inertia of about 10X for the same size disk.

@autobr_

6 ай бұрын

ooh that sounds like it would work great... where did you get such precise dimensions??

@SuperFredAZ

6 ай бұрын

@@autobr_ the hole in the center can vary from about 1 inch or 25 mm to 70 mm depending on manufacturer, but this can easily be adapted with a wood or aluminum plate.

I immediately clicked this notification. I love this series

Great video, two things: 1) You could print an adjustable governor wich can extend its wings or retract them for different tempos 2) If you already have electronic mics and amplifiers, why don't you just use an electric motor and finish the machine? later... really later, whe the tour ends, you can build whatever mechanic power system you want.... just kidding Martin, it's always a pleasure seen you!

I think a huge problem of the Huygens drive is that it always want to give the same power, but I dont think that the marbel machine always will need the same power, so for example the music will speed up if you are using less marbels wich need to be lifet and solw down if you are using all the instruments.

@MikeKrasnenkov

6 ай бұрын

Thats why governor is needed to burn excess power when not needed, and stop burning it when its needed.

So much of this is just... nope. Flat out wrong. Here's the thing: a flywheel is not a device to maintain constant speed, for any load or input; an air governor is ALSO not a device to maintain constant speed for any load or input. A flywheel will not keep a rotation speed constant - its only job is to make CHANGES to that rotation speed SLOWER, ie. to prevent fast jittering of that speed, smoothing out the change and slowing it down; but if either the driving force or the load changes to any degree, the flywheel will just follow and change speed - only MORE GRADUALLY. An air governor will ALSO not keep a rotation speed constant - all it will do is RESIST the change, by opposing more force to a driving force that attempts to speed up the shaft; this is because an air governor has a fixed resistance for a specific speed, and it NEEDS the shaft to turn faster in order to increase resistance to counteract an increased driving force. Any change in driving force OR load will NECESSARILY settle an air governor in A DIFFERENT speed that it requires to come to balance with the changes in its drive or load. Basically, it has a thing called "droop". But governors that settle to equilibrium at the SAME speed for all of their regulating range DO exist - they're just called "isochronous governors". Unfortunately, most of them are of not much use, because they are effectively regulators with infinite sensitivity - so they jump to one of their extremes as soon as speed deviates to any degree from ideal. But this is control theory at this point - and the point is that a "governor" (regulator) that has purely a "P" (proportional) component and nothing else (which mechanical governors tend to be) cannot ever practically keep anything constant without SOME persistent deviation from ideal for ANY deviation from ideal drive or load conditions. The only thing that CAN do that is a regulator that also has some "I" (integrating) component as well - as in, if drive increases it applies a progressively increasing opposing force until the speed returns to the desired value, keeping it applied even though it wasn't needed before to keep the exact same desired speed when drive force was less. There are mechanical hydraulic systems for example that, when out of equilibrium, let extra fluid in or out of a "helper" cylinder that compounds to the action of the main governor, until the desired speed is restored - what the extra fluid does is precisely to "integrate" the amount of change of control needed to keep the SAME speed for a DIFFERENT drive or load. Unfortunately, I cannot offer an example of a simple mechanical system to do this. But the main idea to take away is that NO control system - no matter how complicated - can keep any quantity PERFECTLY CONSTANT all the time; it can only strive to return it to the desired value as quickly as possible with as small a deviation as possible... but there will always be SOME. Which is just a complicated way of saying that chasing perfection into crevices of diminishing return is a fool's game. All there ever is real, is GOOD ENOUGH and NOT GOOD ENOUGH. Draw your line, live with the remaining imperfection.

@8O0I0O8

6 ай бұрын

Great explanation from a fellow ME... Martin has gone way overboard striving for perfection for YEARS now. Remember the Marble dropping craze? The Machine will never get built sadly.

I like design of the air governor in the museum: it is spring-loaded and automatically expands as it speeds up and narrows as itslows down. Such design would be more stable even in situation with non-constant force, especially if you could reach smooth changing shape of governor

@-Tele.gram_OfficialWintergatan

6 ай бұрын

I GOT SOMETHING SPECIAL FOR YOU-

It's coming! The return of the legendary marble machine!

You should be playing the music off the Gear box not the Fly Wheel. You need to stop measuring beats off the Fly wheel. it will be speeding up and slowing down. While the Fly wheel is going to power the system, you need a gear box attached so that the gear box has a consistent tight music

I know I won't be the first to suggest this, but would a CVT gearbox be the best solution for the gearbox part? You could get a very exact and easy to change speed

@FLics-po5oc

6 ай бұрын

It is, but we shouldn’t be looking at gearboxes! The governor stuff should be dropped and escapement mechanism come back! That way you eliminate the force created from the user with the ratchet, the need of a gearbox, and for way better tightness than the flyball governor.

dont give up martin, its looing wonderful.

Yessss! Wintergarden Thursday!

It would be nice to get some updates on what you are doing. Given that you no longer have a full time workshop, it really seems like the project is being de-prioritized. Which is fine, but you should be communicating that

E

Learned a lot with this. Thank you

I can't wait to see the new machine!

Im confused why Martin didnt take the original peddle drive to its conclusion before jumping to this, im sure he planned to have a govenor for the peddle drive too, right?!

@King.-Arthur

6 ай бұрын

I missed a couple videos, so don’t know the whole pedal story, but I assume Martin hitting the pedal is responsible for keeping the tempo consistent. Adding a governor would not make him tighter since he is not accelerating the flywheel at a constant rate as gravity does in this design. I am concerned about how his current design will handle the significant load change between many/few instruments playing. He may end up getting better results from the pedal drive as he can directly manage load fluctuations.

I swear martin actively tries to misunderstand physics sometimes in these videos, or at least he hasn't gone out of his way to learn it until after he makes mistakes.

@samk2407

6 ай бұрын

The air governor is a very good example of this. Air resistance does increase nonlinearly with speed, it's a quadratic relationship though (x^2) not exponential (2^x) which means it increases more slowly than exponential, and because of the way the terms work it's pretty close to linear unless you increase the speed A LOT. Near-linearity is a problem because the drag won't increase fast enough to counterract the creep of the wheels velocity. That's why all the air governors martin looked at had springs or folding mechanisms that increase their surface area with speed, and help them increase their drag on a more aggressive curve than just a static governor like he's using.

@samk2407

6 ай бұрын

With all of the variables that he's dealing with here, it would be very very prudent to actually do the math on this situation, or ask one of his engineer friends in the community to because it would save a lot of time making incorrect guesses at what is actually important and what is not in order to get the balance right.

@samk2407

6 ай бұрын

Martin almost got there by saying that the drive is filtering his input through gravity but if you think about it, it's actually more than that. Gravity is trying to accelerate the mass downward, in order to do that it must spin the flywheel. In fact the flywheel is just way gearing down the acceleration of gravity and making it look close to 0. But it is not zero, it's just dividing 9.8m/s^2 by a big number. With no friction that flywheel would continue to accelerate indefinitely! The governor is actually the single most important factor keeping the system in balance. So NO WONDER that the system creeps up in speed over the duration of the test. That's not a bug that's a feature.

You could make an adjustable resistance governor with magnets. A magnetic break clutch.

@hyperteleXii

6 ай бұрын

Theoretically you could even mechanically link the clutch position to the falling weight, automatically adjusting resistance based on elevation...

@Wintergatan Hey Martin, here is an idea. In your new design you need to power at least 3 things: 2 marble lifts and 1 programming wheel. But out of these only the programming wheel influences how tight the music is. Do all 3 need to have the same power source? Why can't the marble lifts have separate power sources that do not need to be tight nor even in sync with rest of the machine? The marble lifts just need to be fast enough to provide a constant supply of marbles, if they are too fast they just run empty. Also I will reiterate my 2 earlier suggestions: 1. To use some sort of chained punch cards (like the Jaquard loom) instead of a programming wheel (the wheel restricts you to a single loop length). It might even allow live splicing of the pattern. 2. To keep the whole thing modular with easy to couple and decouple interfaces so you can exchange/upgrade power sources and instruments. At least a power source interface between the power source/s and other modules and a trigger interface between the programming wheel module and the instrument modules.

@ivanmoren3643

6 ай бұрын

Yes. This baffles me as well.

@ivanmoren3643

6 ай бұрын

But maybe I've misunderstood - is this already the plan for the third machine? The whole talk about having a big big weight to make it super tight just makes it feel like its emulating digital precision rather than exploring making tje human play as tight or loose as they like or can?

If "tight" music really is so important, why not use a computer to release the marbles?

@planisfo2317

6 ай бұрын

because it has to be mecanical

@fwanknmt

6 ай бұрын

@@planisfo2317 why?

@williamhanna3113

6 ай бұрын

@@fwanknmt its kind of the point of the marble machine

@fwanknmt

6 ай бұрын

@@williamhanna3113but tight music seems to be more important than a completely mechanical design, at some point something needs to give.

@Sorestlor

6 ай бұрын

@@fwanknmt So far the machine remains completely mechanical so I see no issues. He could just use a motor like he did before to turn the crank to keep tempo if you remove the mechanical requirement.

It’s Wednesday my dudes. 🐸

@jb76489

6 ай бұрын

You sure about that?

@Butter_Warrior99

6 ай бұрын

Thursday my dude.

@MartinWepe

6 ай бұрын

It's always Wednesday at Wintergatan.

@sopcannon

6 ай бұрын

They're creepy and they're kooky......

Congratulations on the video. It is really well explained and put together. And even more exciting, you seem to have turned a optimistic corner. I loved the positive energy!

@thomasbecker9676

6 ай бұрын

There's money to be made in prolonging the problem.

maybe with the final machine you could make the speed governor a fan that you can rotate toward you when it gets hot on stage;) Great to see you back Martin!

Martin, I cannot emphasize enough how much I love your experiments and enthusiasm. But I’m increasingly worried by your use of “tightness” to mean consistent BPM. In my understanding, “tightness” has more to do with musicians being in sync with each other, regardless of whether the tempo ebbs and flows. Tempo is almost irrelevant. If it does ebb, I don’t personally see this as a flaw, but rather an influx of the complexity inherent to living (and mechanical) systems. It makes music live. I’ve always understood the MM project to be a response to electronic music, leaning into the benefits of mechanical systems… reinserting some of the random magic of “real.” That is, treating the machine like a Musician (and feel its ebbs and flows), rather than a click track. If a perfect tempo is non-negotiable, I’m honestly confused by the motivation for the whole project. Why strive so hard to mimic what computers already do easily?

Oh boy, another procrastination video. Just think, only 3-4 years before he gives up on this attempt before starting again.

Welcome to a six year long case study on perfect being the enemy of done.

Awesome. Keep the videos coming.

Ive been watching this channel for years. I honestly can' t understand why the power input needs to be human powered. If tight music is such a high priority, why not just use an electric motor and a gear box? Is being human powered a design requirement?

@explodedstarmonkey

6 ай бұрын

he could also just plug an mp3 player in to a speaker.

@thomasbecker9676

6 ай бұрын

@@explodedstarmonkey at this point, he should, except that wouldn't generate as much ad revenue.

@GraphixJunkie

6 ай бұрын

​@@explodedstarmonkey Dude I get that the point of the machine is to play music with marbles, not mp3s. My point is what is more important? Playing music with marbles that has an electric power train or not playing music at all? Screwing around with the power input is just silly at this point. Focus on the music and instrument parts of the machine.