Worst Bearing Housing Ever - Marble Machine 3 Ep.3

[@irukard]

You don't need concentric flywheel. You need balanced one.

[@FizzicksDude]

Very concise and well said.

[@dexterm2003]

Well balanced is much easier if you focus on concentricity. Otherwise you have to custom balance everything.

[@Jrakula10]

you mind explaining why you don't need concentricity?

[@michelhv]

That's what I was thinking as well, and that's why car wheels have a balancing procedure.

[@k1nk1ne]

@@Jrakula10 vibrations are the problem. Vibrations caused by weight disbalance. Look at how automobile wheels are balanced

[@recurvestickerdragon]

The more individual parts you use to solve a problem, the closer to MMX you backslide, Martin. They rattle loose, they fail sequentially, they take time and effort to futz with.

[@nilsdock]

the best part is no part, because it cannot fail. having a lot of manually built housings would add complexity and a lot of new parts. I do think Martin would be more sane in the long run if the design has fewer parts in total.

[@InventorZahran]

@@nilsdock Yes. And pre-made parts are almost always better than bespoke solutions, since they've already been tried/tested and are proven to work well.

[@Nooticus]

this is happening again isnt it... jesus...

[@williambartholomew5680]

​@@InventorZahran not to mention far better to diagnose issues or fine-tune specific aspects. Good luck diagnosing a rattle or shake in 10k moving parts vs 1k

[@tbillington]

Couldn't agree more, I thought Martin wanted to basically order everything and just put it together. Trying to reinvent the bearing housing is a total waste of time.

[@xxportalxx.]

I like how the whole point of the mm3 and ditching the mmx was to do things differently and more results oriented, yet here Martin is reinventing the wheel again hahaha

[@mewmew32]

you must give him credit at least: very seldom do you get the chance to use that phrase literally.

[@keonfrey5560]

No Martin! The bearing load ratings are not what you should be worried about! The threads on the bolts WILL crush and that will allow movement and will get worse in a hurry. Even solid steel pins/rod would be better than anything with threads. The contact patch of the threads on the bolts is very, very tiny compared to the contact surface area the outer housing bearings were designed for. I have tried a threaded rod for a shaft inside a bearing and the threads crushed within a few minutes of runtime and had terrible vibration. I assume you dont want a lot of vibration in MM3!

[@kk-wl2yg]

I totally agree. The steel that is used in bolts is generally plasticine. It's not serious.

[@JZStudiosonline]

Not to mention that he's hammering them in and already mangling the threads to begin with.

[@keonfrey5560]

Yes the only way I can think of that the housing wouldn't move would be if the tension was really really high between the laser cut plates but that probably won't be enough friction to keep the outer bearing race from moving.

[@CristiNeagu]

It's weird how he is looking at posts discussing the bearing housing, all in response to a video on the bearing housing, and he thinks that the thing people say will fail is the bearing and not the housing. But I see he's not the only one, given the comment at 11:21...

[@bleepbloopblahp]

@@kk-wl2yg They make bolts out of all grades of all types of materials. It is very easy to find bolts made out of high grade steel. I do not believe bolts are the right solution for the bearing housing but I do not agree with what you have written at all It's easy to verify what I have written to be true. Here are some grade ratings for metric bolts made of steel or steel alloys: 4.6, 4.8, 5.8, 8.8, 9.8, 10.9 and 12.9. Nobody is making bridges using bolts that could be compared to plasticine.

[@BagManBlog]

Local man rejects a flat disk with a hole in it for a flywheel. A design so simple, that they didn't even bother with it when I was in machinist school because it was too easy to make. Martin. My man. My guy. You are building a piece of rotating equipment. Embrace the lathe. Don't fear it.

[@aquilux-vids]

He even built a rudimentary one into his mill already. Just mill the pocket to press fit a bearing, then throw the thing on the rotary axis and just... rotate the stock to cut it concentric.

[@jfolz]

9:50 *NO!* Do *NOT* ask ChatGPT to calculate anything for you. It does *NOT* handle calculations well and is confidently wrong about it.

[@aldasundimer]

Yes. It is true it read all texts about physics and can generate texts that look a lot like physics texts, but it does not understand the underlying logic and laws.

[@jameshogge]

+1 for this. It doesn't perform the actual calculations. When you ask ChatGPT a calculation, its response is best thought of as a ballpark guess as to what the answer could be. And that guess can be *very* far off

[@jfolz]

@@aldasundimer this goes for all interactions with large language models. You can't trust their output. If you can't verify what they're telling you, you're better off not using them. Even if they lack information they will simply "hallucinate" information to produce an answer. This is an active area of research and nobody has found a satisfactory solution yet.

[@jameshogge]

That said, its explanations tend to be better than its numerical answers. *IF* ChatGPT is going to be used, you have to at least follow through its logic using a calculator and substitute in the correct numerical values

[@DominusFeles]

I saw somewhere someone working on connecting chat gpt and wolfram alpha, that could turn out interesting 🤓

[@raedev]

Martin, there is a saying in engineering: "don't try to reinvent the flywheel."

[@TiagoJoaoSilva]

please don't look at audiophile turntables, you'll cry yourself to sleep... and have nightmares.

[@HANIMEME]

​@@TiagoJoaoSilvawhat does that mean

[@FireFish5000]

@@HANIMEME No clue but the word audiophile before any product name typically means its trash

[@Lilly-Lilac]

@@FireFish5000 not necessarily, but it does mean it’ll be so overpriced you’ll think it was in the wrong currency

[@bruno_semi]

@Anna Hanham The audiophiles care a lot about the "perfect experience", so it's a goal for the turntable to turn very regularly.

[@tomeshec1122]

The real solution is just balancing the flywheel with some weights like a wheel on a car I think. Then it can be slightly off center and the precision can be lower

[@AndersJackson]

But it has to hold the radial position. But you are right. It is always possible to balance the disc reduce the forces from the rotating discs. But you should hold that force.

[@someguy5334]

You'll still need a way to measure the balance, though. Much easier to make it as concentric as possible in the first place.

[@jordanshilander7402]

A good example of balance being more important than shape is the way many of the BattleBots spinning weapons are very odd and asymmetric shapes but spin smoothly because they are balanced with the center of mass in the middle of the rotation axis.

[@falcon1209]

@@someguy5334 Yes, but its easier and more simple to balance the thing rather than to machine it to insane tolerances.=

[@falcon1209]

No kidding. Its literally a spinning mass. There's literally no other design parameters other than it being balanced. Its not interfacing with anything other than the axle. Its physical dimensions are irrelevant unless there's clearance issues.

[@spplS.]

16:24 no bearing in the world can prevent misalignment! That's just not how bearings work. There are different solutions how to clamp bearings, but they all require at least two bearings. Just take inspiration by the headset of a bike.

[@BrandonAGr]

Exactly, it's not the rotational rigidity of a single bearing that keeps the flywheel perpendicular to the shaft. It would be that there are two different bearings coupled together that enforce the flywheel staying perpendicular. A single ball bearing isn't designed to take rotational loads in that dimension no matter how it's attached

[@craftedbyorre]

I hope you will spend your time designing and testing the actual difficult parts of the machine (marble lift/divider/etc) and use standard pieces for things that are already solved. I believe!

[@slartibartfast2649]

Not only that, but Martin assumes that everything is perfectly concentric in the dynamic load calculations for the bearing, when his design is highly unlikely to be concentric. Sticking with proven and cheap designs that can be bought off the shelf worldwide is certainly a good recipe for a supposedly reliable touring machine.

[@craigwootton1114]

This is maddening to watch!

[@dylandreisbach1986]

@@slartibartfast2649 At least he should use shoulder bolts since they have good dimensions, but even this has problems used in this way. I just hope he goes with the standard mounts so he doesnt waste time trying to fix the issues the bolt mount causes.

[@xander1052]

I think the bigger issue is not the bearing failing, the housing itself could be more likely to fail than a standard tried and tested design.

[@matteodecastro8901]

I agree with you on this thought because the screws in the housing will be subjected to shear forces (which means that a force is applied perpendicular to the axis of the screw). Screws are generally not designed to withstand this type of constraint, and their shear strength is much weaker than their tensile strength. Furthermore, these shear forces can cause the loosening of the screws. It would be better IMO to use dowel pins or elastic pin for securing the bearing as they are specifically designed to withstand this kind of constraints.

[@apocolisp7773]

This is what i was thinking as well. He goes into depth about the bearing specs, but all of the housing parts need to be subjected to the same scrutiny. Yes Matteo, bolt strength, but also with vibration and side-loads, i can see the threads wearing into the plates, affecting their ability to center the bearing. Just my thoughts.

[@Nevir202]

@@apocolisp7773 Exactly, long before anything else, the threads of the bolts will begin to be flattened out, turning the interference fit to a sloppy one. Once that happens it may or may not have the freedom to flop around until the bolts give out. I kind of suspect not, just because in my old job I saw how much abuse bolts can often take in non-ideal scenarios before sheering. I think before you'd need to worry about sheering a bolt, the machine would have long since shook itself apart.

[@CristiNeagu]

The biggest problem is that the bearing itself is resting on the tips of the threads. I doubt we're talking about hardened threads here, so they will start to deform before long, developing radial play in the bearing housing.

[@abpsd73]

The bearing could also fail, although probably not catastrophically, since the outer race is only supported by a marginal contact area of the thread major diameter in some 8 points. A machined bore fully supports the race throughout it's width and entire diameter. I would also assume the load rating of the bearing also intends the outer race be fully supported.

[@w0nd3rlu573r]

It's not the stress on the bearings, that is a problem. The bolts are NOT designed for such perpendicular loads. This is the most scary part.

[@thomasphillips885]

Especially the threads. They'll deform

[@goober2964]

Not only that, the threads have to withstand the forces. Since the threads act like small stands holding up the bearing and the weight of the flywheel, more pressure is on the tip of the threads, thus blunting it, causing the bearing to sink.

[@Robert-sq7bp]

Also belting the bolts through with a hammer? It deforms threads and bolts, breaking the precision. So many parts…

[@charetjc]

Shoulder bolts?

[@spplS.]

13:37 no?? Those holes are oversized for a reason, to ADJUST them. Take a dial indicator and a hammer, turn it a few times and after a little tap here and there, it will be perfectly concentric.

[@awesomo9k]

He'd probably know that if he had any practical engineering or machining knowledge/experience. He needs to go work a shit job in a machine shop for a few years, rather than play around in CAD all day.

[@spplS.]

@@awesomo9k haha I (m19) just a student, but I've got a little lathe in my basement. That and a few KZbin videos is enough to learn a lot about engineering. And HE's been dedicated to this kind of unprofessional trial and error for years, but has still not learned a thing.

[@foldionepapyrus3441]

And then you can drill out that 4mm registration pin location in the plates so it goes back perfectly every time and can't shift on you with the shocks and bumps of world tour transport!

[@CristiNeagu]

My mechanical design professor in college told us pretty early on: "Bolts and screws are fasteners. They are not to be used for alignment. That's what alignment pins are for."

[@DevinDTV]

​@@awesomo9k oh yeah that sounds like a really efficient use of his time nice comment good job

[@HorizonMakes]

For the live axle design, i think you can jusy use a keyed shaft to lock the flywheel rotationally to the shaft and maybe use 2 shaft collars to lock it axially. Another option is to use a hexagonal shaft instead of the keyed shaft. It's popular in robotics and gives you access to ecosystems by VEX, AndyMark and REV revolving around 1/2" hex shafts.

[@tomerkarpovsky3421]

you are forgetting about the built-in tolerances of these shafts. They are designed to be a clearance fit for all of the other parts in the ecosystem that won't help with the balancing ,perpendicularity or the vibrations

[@comeradecoyote]

Keyway is the less labour intensive solution, but it also may give some backlash if the tolerances are loose. A splined shaft would pretty much eliminate backlash.

[@cinavik]

At 16:40 you are discussing the need to perfectly align the shaft perpendicular to the frame. This is not strictly neccessary and the bearings are designed this way because it isn't a critical dimension, even for your application. As everyone is saying, don't overdesign what already exists.

[@RegebroRepairs]

In fact, the bearings are designed like that because things will flex. Trying to make alignment perfect sounds is a recipe for failure. I think Martin got that in the end.

[@cinavik]

@@RegebroRepairs Exactly. Forcing a perfect rigid tolerance will result in binding and reduced performance. Martin has to see the advantage of not redesigning a perfectly good wheel. Literally, in this case.

[@CristiNeagu]

I don't see why he isn't using those bearings as he should? Just mount the bearing housings on the frame and fix the flywheel to the shaft. You know... like any sane engineer would do.

[@codahighland]

​@@CristiNeagu Did you watch the video yet..? That's exactly what he said he's considering.

[@CristiNeagu]

@@codahighland As a last measure, sounds like.

[@theonlyalan731]

You should not be using bearings in such a way that moment load on the bearing is what keeps everything perpendicular. If you have two bearings such a distance apart, it will be the radial loading on the bearings that keeps everything perpendicular. As a general rule, ball bearings do not handle moment loading well

[@spplS.]

8:34 great! SKF really know what they're doing, there's a whole scientific field just about mounting a shaft in specific applications, and the guys from SKF are the experts of the experts when it comes to this.

[@billkeithchannel]

They probably have not gotten back to Martin because they are still laughing and can not get off of the floor to sit in front of the computer.

[@spplS.]

@@billkeithchannel and everytime they hide another component in fusion and discover what's underneath it, this cycle starts over and over again.

[@kennyduncan7]

​@@billkeithchannel Your comment is not helpful.

[@Touwtje]

@@kennyduncan7 To be fair, yours isn't either.

[@d_vibe-swe]

Svenska KullagerFabriken. ;)

[@hoej]

Martin: perfect is the enemy of good enough. Also Martin: 'we need millisecond tightness'

[@christianvanderstap6257]

Milliseconds do matter for music

[@thomasbecker9676]

@@christianvanderstap6257 You're kidding, right?

[@OHomemquecalculava]

​@@christianvanderstap6257 when you reach [a delay of] tens, dozens, of milliseconds, yes they do. 3ms do not.

[@NewtoRah]

His targets are faster than the speed of sound. The time it takes for sound to travel from one side of the machine to the other is ~3ms, so standing on the left side vs the right side will change the timing of the sounds reaching your ear more than the actual error between notes

[@christianvanderstap6257]

@@OHomemquecalculava i am not

[@jordanshilander7402]

Your demonstration of the bearings pivoting inside the housing prove that they always stay perpendicular to the axle even if the pillow blocks are not.

[@dexterm2003]

Without uniform pressure on the outer bearing race is important for a decent life. At 2000 RPM at an hour 120k cycles per hour. You will easily start to deform the bolts and lose the "press" fit that you think you have. Also the bearing race deforms and will wear out prematurely. All being said you are welcome to try whatever you like. The SKF application engineers will tell you the truth and set you on a good path.

[@human_brian]

My thoughts exactly, these bearings aren't designed to be held with 8 point loads on the outer race. The race is likely to deform over time and cause vibration or worse yet complete failure of the bearings. They're designed to be held equally around the outside circumference. Press fitting them into plates and then bolting those onto the flywheel would solve that issue and allow for easy swapping if needed (aka the press fit flanged bearing holder).

[@elindred]

I don't understand all this mental energy spent reinventing parts that will at best work as well as parts he could already have in hand and enable him to move on with the project. Not to mention that the commercial parts have far fewer components which I thought he had decided was a priority. Either way, whether it works or doesn't it'll be good content for us.

[@Owl90]

He's milking it for content. He wants this project to last as long as he can.

[@Wilczylak]

@Owl at this point, you might be right, the other possibility is that he's just turbo-stubborn

[@RhynoD2]

@@Wilczylak I think the problems with the MMX show that he's turbo-stubborn. He can't shake the artist in him - which is good, don't get me wrong. He's a great artist. But he gets a vision in his head and he wants that vision to succeed instead of being flexible enough to pivot and conform his vision to practicality.

[@StuffandThings_]

@@RhynoD2 There seems to be a fundamental struggle between the artist and engineer mentality, and each time he lets the artist mentality win. Eventually, he'll have to come to the realization that this is an engineering project (something he has stated of course, but deep down I don't think he has truly accepted yet) and let the engineer mentality win almost every time. That, and he needs a pretty sizable team of professional engineers working with every major mechanical component, just like with any complex mechanical design. I think he still has a long way to go to get to this point, but eventually it will become the inevitable, unavoidable conclusion.

[@PCFixer]

Prediction: Martin will get stuck on something, something won't work the way he wants, he'll scrap this one as he did with MMX and then it'll be Marble Machine 4,5,6,7,8,9... and then SOMEONE ELSE will build a Marble Machine with standard off the shelf parts and it will do its job. It will work. It will play music--and THAT person will have achieved the dream where Martin got caught up in his own 'genius' and failed. Sorry for the harshness, but this happens with a LOT of creative people; they obsess over designs and minutiae and ultimately fail, or they criticize their own work far too much, are never satisfied, etc. At this stage I'm like, just... come on, man, just... build the Machine. Take it on tour. Done.

[@JonathanKayne]

11:00 they weren't talking about the bearing being the failure point, they were talking about your housing being the point of failure. The issue is that those bolts pretty much have to be perfect to properly hold a regular bearing in place

[@Vendavalez]

Let alone the wear and tear on the bolts and the bearing where they connect!

[@JonathanKayne]

@@Vendavalez the moment they wear even slightly, the whole assembly would go out of whack. Grip would be lost and the concentricity would be too

[@francescosirotti8178]

A tight bolt can hold a car. There is just no way Martin can put enough energy on the flywheel to move it.

[@Vendavalez]

@@JonathanKayne honestly, the weight of the wheel when it is not in use might be enough to deform the threads.

[@JonathanKayne]

@@francescosirotti8178 you are correct but those bolts are connecting the tire to the shaft, not the bearing itself. In that regard they are doing a completely different job, and are doing exactly what bolts excel at- joining two flat surfaces through tension. The ultimate issue is that Martin is trying to use the sidewall of the bolts to hold a bearing captive, which isn't good for a bolt, and as he showed in the previous video, you need them to have the correct radius pattern to hold properly.

[@mrspecs4430]

Martin: "what is 2 + 2?" ChatGPT: "easy, it is -5" Martin: 😁👌

[@0-Kirby-0]

I'm sorry, Martin. I loved your journey so far, seeing you learn and solve problems, but this is feeling more and more like you're making problems to solve. When everyone tells you you're drunk, maybe it's time to sit down.

[@Anya-B.]

"When everyone tells you you're drunk, maybe it's time to sit down." Such a great slogan for this season of "FeatureCreeping like a Champ"

[@nolansykinsley3734]

The failure mode they are referencing with your design is not with the weight capacity of the bearing, it is the support of the bearing in the clamp.The contact area keeping the bearing concentric is the threads you already deformed by hammering in the bolts. Think of how little contact area there is on the outside of the bearing being only the threads of bolts vs an entire solid housing. This will cause it to deform when flexed, any deformation each flex as it rotates will eventually lead to wear and failure, or running out of concentricity with prolonged play.

[@scottbedard5095]

as a mechanical engineer i can't help but cringe... please listen to the comments, or read some design handbooks at least... or worst case ask for the help of an engineer... that's what we're here for. if machined parts are a problem, there are rapid machining prototype services available that will work much like your laser cutting service. Flywheels hold a lot of energy and should be treated seriously.

[@francescosirotti8178]

Martin isn't just falling into multiple useless rabbit-holes. He's farming rabbits and digging holes for them.

[@CristiNeagu]

I haven't done mechanical engineering in almost 10 years now and I'm cringing too... He really should have considered off the shelf parts first and foremost. Not only are they a solution proven to work, but if anything fails, it's a phone call and an overnight courier away. What is he going to do if one of his highly tuned, custom built parts fails on tour?

[@leviwilder1205]

@@CristiNeagu it's called industrial lego for a reason, he needs to use those pieces instead of bespoke parts

[@randomdude1rd1]

Don't apologize for a long video! I love long form videos and how organic they feel

[@billkeithchannel]

Yeah Martin did weeks of 6 hour test videos before he gave up on MMX. SMH! Those were awesome to watch. I put them in 1.25x speed so they flowed better.

[@kennyduncan7]

​@@billkeithchannel You're a grown adult, get over it.

[@kennyduncan7]

​@@billkeithchannel I didn't realize that a community full of adults could create such a toxic comment section. It just proves that maturity has nothing to do with age.

[@billkeithchannel]

@@kennyduncan7 Grownups like to criticize and cajole each other. Get a thicker skin or get off the internet. Beta males get butt hurt over social media comments.

[@baileythesleepless9966]

@@kennyduncan7 Im confused, where is the toxicity?

[@braiansingh9730]

Money isn't an issue. Perfect solution presents itself. Chooses unnecesary complicated solution instead.

[@thomasphillips885]

End up spending way more money in the form of time on the bolt cage

[@YogiTheBearMan]

Did he ever say why he choosing this design if it isn’t money? Pretty sure his stated motivation in a previous video was cost but maybe I’m wrong

[@thomasphillips885]

He thinks he can design something with better concentricity and more parallel

[@DjesonPV]

There is no need for perpendicularity. In fact having thoses bearing rotating is better IMO, gyroscopic forces will do the job. As you said you don't want/need perfection, and therefore use coupler at the flywheel output to correct the shaft alignement.

[@recklessroges]

Spends ages looking at the bearing tolerance. Forgets that the bots will be rubbing and straining against the outside of the bearing. Either bold or bearing are softer, and will wear the other destroying concentricity and perpendicularity.

[@Vendavalez]

This is precisely my concern and I find it strange that it has not come to the forefront of the discussion enough for him to feel like he should address it.

[@G5rry]

Agreed. The bolt threads are supporting the force from the bearing. Threads are not made for that purpose and will deform, thus loosening their grip on the bearing.

[@Lelentos]

The outside of the bearing doesn't move. It's the steal balls inside that are rolling. The outer ring is stationary with respect to the wheel, the inner ring in stationary to the axle.

[@JoshJones720]

"The most precise solution is the one where I have to hammer the bolts in" -- wut??

[@dazley8021]

added to that the threads will be ruined and taking apart those housings to fix them, will be a pain.

[@lagule]

I think you should read about Hertz Theory : this helps modelling and understanding why everyone hates your housing. The constraints (force over surface) on each contact point bearing/screw are enormous because of the minuscule surface, thus the deformations induced are huge (because of the very small half radius of the ellipsis modelling the screw threads crests). This yields poor concentricity under load and, because this holds a flywheel which exerts more load as the concentricity gets worse (less concentricity = less balance), this may be the worst design for this application. In the last video, we can see user "remove_the_kebab" partially adressing the issue, suggesting to use shoulder bolts. This transforms the minuscule half radius of the thread crest to an infinite half radius and really helps with the amplitude of the deformations. Furthermore, it helps with the precision of your bearing location. Also, a word on the perpendicularity requirement you brought around 15:30 in your video : any shaft rotating should be held by at least two bearings. Also, if they had strictly perpendicular bearings, the whole assembly would be hyperstatic and hard to manufacture, and the bearings would "live" less longer than their actual isostatic design. If you really want perpendicularity, I would recommend not laser cutting the holes for the shaft and drilling and rectifying the holes after it has been laser cut. Although, because the plates are tall and thin, the perpendicularity is weakly ensured (hammer it on the sides and it's gone). An easy way to do it is to drill and ensure good surface quality on the sheets before laser cutting them, and then to laser cut them as they are centered on your laser cutter origin with a weld circular pin exactly the size of the drilled hole - then you make sure your program laser cuts the sheet concentrically around the pin.

[@ScamstinCrew]

The point of pillow blocks and that "slop" is to allow for the reality that we dont love in a perfect universe to creep in. You don't need perfect parallelism. Especially if you are using belts to transmit the torque.

[@illiteratebeef]

5:36 Martin fundamentally misunderstanding "perfection is the enemy of the good". Doing it the professionally accepted 'safe' way isn't the same as being a perfectionist and changing every facet to your application.

[@rabidcabbage7230]

he'll understand that when he's making MM IX

[@illiteratebeef]

Fingers crossed MM 42 is the one that actually gets finished.

[@bl4cksp1d3r]

Better idea: MMX 69

[@diobrando2160]

@@bl4cksp1d3r ZAMN! LE HECKIN FUNNY REDDIT NUMBER!

[@bl4cksp1d3r]

@@diobrando2160 hey, somehow your caps lock got stuck, you should try to fix it before writing your next comment

[@descendantdan]

Innovate on what makes you unique, you don’t need to create a new solution for every problem and a bearing flange does not make you unique. Every time you try and design something new, rather than picking up an existing solution, that has an opportunity cost, it takes time, thought and money to create something new, far more than the cost of buying something off the shelf that works for a given application so the choice to design from scratch should always be taken only after you have decided that it is truly worth the investment.

[@rustyreckman2892]

Just to make sure I understand, Martin has on this quest for a bearing housing design to avoid machining the flywheel?

[@454Casull]

It's to troll people. That or he is really not learning anything.

[@grunions9648]

I'm 99% convinced that these recent videos are trolling to prove some kind of point.

[@chicken_person]

Gotta love how Martin keeps talking about how great it is that people upvote comments on the reddit page to show what's the consensus, then ignores the most upvoted advice and features a comment with 2 upvotes (literally just the poster and Martin himself) that agrees with him

[@conorstewart2214]

Two of those bearings on opposite sides of the flywheel would hold it perpendicular. They are only allowing misalignment because the bearing blocks are allowed to move in regards to each other. If they were mounted perpendicularly at the start, they would stay perpendicular.

[@NateD77]

This! Also he's stressing over getting the bearings aligned concentric with the flywheel and thinking of offsetting the mounting holes off center of the holes in flange. All this will do is ensure misalignment. That's what adjustment and bolt clamping force is for to get it aligned and held in place.

[@typewriterplants]

The problem is not that the bearing will break, it's that there is not a solid concentric fit between the bearing and the housing. It will slowly slip from position over time and you will not be able to use a hammer to adjust the concentricity like with commercial bearing housings, you'll have to disassemble it. As for the perpendicularity, if it is with a flexible belt, then this will not be a problem with the pivoting, and the frame will flex slightly instead. As for the gym machine design, it sounds much better.

[@amberjude2886]

It's the kinetic fingers all over again.

[@dm45lm]

I said it in the last vid but by hammering the bolts in you are mashing their threads which makes it impossible to keep the bolts tight without stripping threads and also will cause them to become loose as the threads continue to mash the longer its used.

[@hyperphrog69]

8:13 I'm preatty sure that in a matter of minutes the bearing will have compressed the threads of the bolts and become loose. You should thus use bolts that have threading only in the end

[@roippi3985]

If you want inspiration from existing machines which spin stuff fast, look at lathe operation. Your options on a lathe for workpiece holding: use a collet (only for round machined OD), or use something like a 4-jaw chuck that must be "dialed in" by the operator. There is nothing wrong with having to manually dial in a setup.

[@thomasbecker9676]

Collets can hold material that isn't round or machined.

[@Kyrazlan]

My thoughts exactly.

[@Nyky95]

Bolts are not designed to take the loads that you will put on them, they can fail any moment.

[@foldionepapyrus3441]

They are also much much wider in tolerance than the laser cutter - you can get more precise thread sizes but 99.99% of the time a bolt only has to tighten, so the OD of the thread is almost entirely irrelevant, as is the shape of the thread really - it just needs to have the right pitch to thread properly with enough thread engagement to clamp down hard... When you want a positional bolt you have a shoulder of unthreaded proper sizes...

[@w0nd3rlu573r]

Please everyone upvote this, so it's seen by Martin.

[@userPrehistoricman]

I lick the bolt bolts are not designed to be licked so it can fail at any moment

[@mateuszbugaj799]

​@@w0nd3rlu573r he doesn't care. He sees the criticism and suggestions, acknowledge it and moves on with his idea. When it fails he just starts to build another machine.

[@thestormcraft4012]

That depends on the type of bolt and the material it is made of, for example every single bulkhead on a Boeing 747 airliner is in fact bolted together truth be told bolts have much higher tolerances than using something like welding it's why aeronautical engineers avoid welding at all costs.(for reference the load of the flywheel will be divided among each bolt so there are 8 bolts so each bolt will take 1/8th of the total load.)

[@markmetzler9108]

Have you considered a dual conical bearing design like used in the front wheel axels of nearly all auto/truck designs? Off the shelf availability, strength, and automatic centering positioning.

[@kumbackquatsta]

pulling an elon, ignoring engineers, taking people's money, delivering vaporware

[@TilTisback]

sad because its true

[@kumbackquatsta]

@@TilTisback he's openly worshiped elon in many videos

[@PedroCoelho98]

Sadly, today I am certain that we are not seeing a finished MM3... Seeing Martin defending the same decisions that lead to the previous failures is heart breaking... Hope I'm wrong, though...

[@NieJestemWikingiem]

Yep, I was very optimistic but it seems hopeless now

[@Nyky95]

The numbers on chat gpt are wrong, you should do a fatigue analisys. Static loads are rarely the limit.

[@thomasphillips885]

ChatGPT is also not a calculator

[@oskioskioski]

Yep. Chat GPT can't do math, it's not what it's designed for. It's just generating text.

[@crazy_dan420]

I loved the beta iteration of the marble machine 2, before you started to time it to the 1/100th of a second. There is no band/orchestra on earth that has perfect timing to 0.01s. Trying to perfect timing made it more robotic, it used to have an artistic humanlike slight timing variation throughout the songs. I also would have been lazier on the balls falling out. Having them hit in the same spot every time once more makes the sound more robotic. I would have made a catch tray for the strays with an elevator up. Having them fall in an inch diameter target at the old timings every time would give a more human variation in sound than hitting the exact same spot every time with a 1/100th of a second drop robot. There is a reason a painting of something is worth more than a picture.

[@creageous]

I'm glad that your content on mundane engineering issues is still entertaining. Much gratitude for your sense of humor and keeping it fun.

[@Nyky95]

You don't need to flange, if you machine it you can use keys

[@shawnhale2631]

Those pillow blocks seem like the most reasonable option. Definitely the most "used as designed," concept.

[@Scriven42]

"We don't need space grade flywheel" ....... Says the man that got microsecond+ tightness on the electromechanical parts, even when other musicians were reminding him that slop is what makes a drummer a drummer and not a perfect drum machine...? Martin, really?

[@maartenbeute6742]

You should definitly go for the active shaft concept. this is usually used for heavy machinery.

[@Nyky95]

You also should machine it. It is a flyweel, it has a decent amount of energy in it when it's spinning. It can be very dangerous for you and the machine if it is not well dinamically balanced.

[@MrJamesBanana]

Just because it's machined doesn't mean its balanced. The density of steel can vary somewhat throughout.

[@slartibartfast2649]

@@MrJamesBanana but that issue would also apply to plate steel only more so, since the thickness of the plate will vary across a sheet as well as its density. Martin should also increase the radius of the flywheel and include spokes/ lightening holes and a solid rim. This would result in identical moment of inertia (and therefore rotational energy at the same rotational velocity), but reduce the mass of the flywheel, reducing load on the shaft and bearings as well as making the machine lighter and thus easier to transport.

[@jonasstahl9826]

​@@slartibartfast2649 Spokes are pointless if you cut the flywheel out of a plate or machine it, you want a flywheel go be heavy the material is already there why remove it. Also spoke create aerodynamic drag and can produce noice, what could be a problem for his application. Weight saving doesnt matter, you need multiple person or a forklift anyway to move this machine, 20kg more or less dont make any different.

[@tjtobin86]

Martin after MMX failure- "We are not going to over complicate things. We are not going to build things we can buy off the shelf. We are going to simplify the design to eliminate failure points" Martin on MM3 - "We are ignoring something off the shelf and instead over complicating the design by making something custom to fill a role that an off the shelf item has been doing for centuries.."

[@Bbeaucha88]

See what people aren't understanding is that Martin is going to use those indestructible bolts where the threads NEVER deform or bend. To be honest, I can't think of a single time that any threads have ever gotten damaged, stretched, or dented even on normal bolts. Threads are also perfectly suited for side loading with the contact points being the thinnest part of the threads! I am SHOCKED that many other engineers don't use threads as a mating surface with another surface under load. Note: There MIGHT be some sarcasm in the above.

[@billkeithchannel]

That design should be using hex head shoulder bolts.

[@xander1052]

lol, it is really clear that Martin is still thinking like an artist at some level still.

[@awesomo9k]

Hammering bolts in with a mallet is the industry standard

[@Bbeaucha88]

@@billkeithchannel did you not see that he will be using INDESTRUCTIBLE bolts? Shoulder bolts would be much more useful if we were worried about threads getting damaged. But we are NOT worried so there is no need for such a logical alternative.

[@Bbeaucha88]

@@awesomo9k obviously... What other options are there?

[@OrbitalCookie]

Every part that can be reused is a part that was designed and tested by someone else. The part may not be completely perfect for the application, but, if we are to believe you, that isn't the goal. There seems to be a misalignment between the stated goal of a non-perfect design and the reluctance to use off-the-shelf parts that may not be perfect. If the channel shifted its goals from building a fascinating marble machine to exploring unconventional solutions to common problems, there wouldn't be an issue (except for disappointment, but who are we to dictate what you should do?).

[@CristiNeagu]

Sounds like he's trying to order everything from that Tailor Steel supplier. It's fine to order as many things from a single supplier, but you can't compromise the integrity of the design because of that.

[@Rykvp]

At this rate, we will will probably see Martin on the moon before MMX3 is anywhere near finished

[@leviwilder1205]

@@CristiNeagu and the fact that if something were to fail while being on tour (like that'll ever happen) you can run to nearly any bearing supply and pick up a new pillowblock and be on your way with only minor adjustments made with as many bespoke parts he wants to use it's no wonder it's taking so long to do anything at all. Coming from an engine mechanic background, nothing can be made concentric straight off the forge (think crankshafts) and therefor have to be balanced to a degree, I think he is shooting for something that he either doesnt understand, or like some other people have stated "to drag on the show"

[@CristiNeagu]

@@leviwilder1205 I said as much in another comment. What's more, I'm pretty sure big companies like SKF will overnight parts to you. I'm sure they're used to customers whose downtime is measured in thousands of dollars per hour.

[@tymoteuszkazubski2755]

​@@CristiNeagu Yes and they charge accordingly

[@reclusingrecluse4835]

The biggest issue is not the engineering of the flywheel, it's that he's using Reddit in light mode

[@chrismathewsjr]

or using reddit at all. anyone who posts on reddit re: engineering should have to post a picture of their bedroom and/or something they've built that doesn't suck

[@slartibartfast2649]

@@chrismathewsjr why the bedroom? Why not your workshop or work desk?

[@chrismathewsjr]

@@slartibartfast2649 because i don't respect bad bedrooms

[@slartibartfast2649]

@@chrismathewsjr Right, I agree that a chaotic room is a sign of a chaotic mind. However, that doesn't make them any less skilled or knowledgeable, only disorganised. You don't need to respect a person for them to be right about something.

[@chrismathewsjr]

​@@slartibartfast2649 i simply could never take advice from a virgin with a mattress on the ground that also has never designed anything successful

[@piotrk2639]

Its not a good idea to take numbers from ChatGPT ( 10:00 ). Chat gpt "guesses" an answer and the answer is "most likely" so if you give him some complex Math problems probability that nubers are corect is realy low. That problem is well know so WolfranAlpha share plugin to ChatGPT and with that combo answers will be OK

[@jfolz]

Exactly. ChatGPT does not understand math. It was trained on text that contained math, so it can throw together digits and symbols in a convincing manner, but it's almost always wrong.

[@TheUnrealPownament]

5:54 there It is, Martin. This Is the part of the video you will need for future montages 🤗

[@slartibartfast2649]

I really don't understand Martin sometimes. It is as if he can't taste the blood in his mouth as he repeatedly sprints face first into the same wall for 3 years non-stop.

[@gregorykaznacheev]

Best part is no part! Press it in to flywheel

[@Sembazuru]

I was thinking of something similar. Heat up the flywheel and then press the shaft into it. Recognize that you only have one chance to get it right, so maybe don't have the shaft machined to final length and have a stop-block so you can push the shaft in until you hit the stop block and then you should be within mm of correct placement.

[@conorstewart2214]

Physics may not have changed since chatGPT was trained but that doesn’t mean it’s output is correct. It has no understanding of anything, including physics, it is just a language model, it just works on the probabilities that words follow each other.

[@ramous5182]

I don't think you need absolute concentricity here. You can adjust the position to get close enough and then add/remove weight to balance the wheel.

[@dmynerd78]

It's been really interesting how all the machines you've built go through so many design and engineering processes. It's fascinating to see you learn and grow as a designer - especially because I've had to learn many of these ideas myself. However, at the end of the day I always look forward to what you can create musically. Would it be cool to have a machine use marbles to make music? Absolutely. But you don't need a marble machine to make music. So while I have been really enjoying the building the next marble machine videos, I do hope that you will continue to make music too. From Detektivbyrån to Wintergatan, I always loved the music you and everyone else involved created.

[@kl1nk0r]

14:30 Big long time fan but I am sorry, I gotta say something. This problem is not a problem. The actual way you would go about this is to laser-cut the hole undersized and use a H7 or H6 hand reamer that costs literally 10 Euros. You ream the hole to spec and you cannot get more accurate than this. Certainly (!!) not with screws (their outside tolerances from the DIN/ISO spec is way worse) against a bearing housing - don't do that. 16:20 Thats wrong. These bearings are not inappropriate for your use case, they certainly are appropriate. You actually NEED this degree-of-freedom in an application such as yours. in contrast, your solution over-defines the mechanical setup of bearings+shaft. Imagine one of your flange-mounting-plate is tilted slightly. Depending on the bearing a very slight tilt might already be enough for the bearing to not ride freely anymore. Some bearings have tighter internal fits than others. Again, listen to the mechanical engineers (I do the same, since I am "only" an electrical engineer, but that is all stuff I have learned over the years. Use dowel pins or, as others suggested, shoulder bolts which are dowel pin and screw in one).

[@falcon1209]

If you started a drinking game for every time Martin said "balance" or "balancing", you'd be sober.

[@conorstewart2214]

I think whatever you do you should be prepared to balance it manually, that way the actual mounting of the bearing isn’t as important (although it is important to get a solid mount and being mostly concentric will always help), however you balance it you will have to add or remove material somewhere, either with a bit of grinding or drilling holes, which will probably effect the looks. In that regard grinding may be better as you can polish it later and as long as you don’t need to grind much it will still look circular.

[@jimbobur]

So much controversy over this. Everyone here saying the bolt cage won't work, or that asking ChatGPT to solve a physics calculation is like asking a parrot that can squawk the melody of Ode To Joy to explain Beethoven's compositional philosophy to you, is overlooking something very important: how funny it will be to see Martin's wobbly death spinner tear itself apart.

[@Argosh]

9:30 did you just half the load because you're using two bearings? Somewhere a FEM Software Engineer just died.

[@CraftingCake]

Isn’t the static load just the force pressing down on the bearing while not rotating? Wouldn’t that half when the force is split between to points? I would love an explanation, I’m genuinely confused right now…

[@Argosh]

@@CraftingCake that's assuming a list of things that is longer than my arm. For one, no rotating mass will ever be a static load. It'd need to be perfectly balanced for that. Like... Not well balanced. Perfectly. That perfect mass would have to be perfectly positioned on a perfect shaft. If the frame is somehow twisted, even slightly, there will be forces anyway, and don't even get me started on what vibrations will do to this system. If you look at the "standard bearings" you can see that they can swivel freely. That's a very intentional feature. It greatly simplifies the transmittable forces. That being said, the characteristics of the bearing he showed are orders of magnitude above what he needs. Thus "failure" in this case means: High and atypical wear, high friction, unusual behavior like vibrations, etc. It's just another symptom of him grossly oversimplifying an extremely complex topic. Then again he is in contact with a very well renowned company now, and they will hopefully give him honest feedback.

[@Finnspin_unicycles]

You might want to consider looking at common machines that have similar issues. Cars, motorcycles, washing machines, etc. A car's front wheelbearing assembly is readily available, should likely be suitable (do some RPM and load guesstimations) and has a flange to bolt to and a place to ensure concentricity for example.

[@dasWarsteiner]

I have used chatGPT many times. You should never trust chatGPT's calculations. The forms are usually correct, but chatGPT is extremely bad at basic calculations!

[@MidwestSirenProductions]

“The perfect is the enemy of the good enough” That quote speaks volumes to me. Coming from a perfectionist who likes to do it right the first time I can tell for a fact that these two are indeed enemies out for blood. I started my first engineering job a few months ago and that battle has been raging since I started, and I can happily say I have started to shift away from perfection and more towards my boss’s “80-20” mentality (80% of the work done in 20%) of the time. It’s still a great challenge, but I’m still trying day to day, and the fact people at my work see that speaks volumes. There are plenty of days where I sit at my desk spinning my wheels trying to come up with a great idea, but eventually when I get my head away from perfection and just throw something down, I find that usually works out best. Not quite the “don’t think just do” thing, but thinking a bit less can help get you back on track.

[@singarith]

You mentioned finding engineering ideas at the gym. If you have gym bikes, maybe look at how those are designed. Some of them come with flywheels that help power the bike.

[@RustyRen]

Ah yes, the final boss of design. Adversity

[@osculant]

Made it three minutes in. “I know all the comments said it was a horrible idea but I think it’s great” 😢

[@Owl90]

"Surely everyone else is wrong, not me!"

[@princecuddle]

I was never concerned by the bearing. There are so many bearing options that you are guaranteed to find the correct one for your application. What I am concerned about is the threads on the screws you're using over time being worn down

[@KyleGetson]

I’m not smart enough to have an answer here. But I am smart enough to know if it takes a 22 minute video giving excuses on why you are ignoring experts, you’re probably wasting your time. Either admit your going your own way, or rely on experts. Either way I’m excited for these videos :grabs popcorn:

[@NochSoEinKaddiFan]

I admire the smugness with which the musician lectures the engineers on mechanics, physics and design he self-admittently doesn't fully grasp while no engineer would ever dare to wander into the field of melody, tone or rhythm with the same unfounded confidence...

[@lucienwerner]

Perpendicularity issue is easely solved by adding a second barring

[@charetjc]

And yet the video shows an example of two bearings with a shaft that is not perpendicular to the flange of the bearing blocks. Do go on about how easy your idea is.

[@WesHawkins0]

When I saw the pillow block solution, I finally was able to relax. Maybe Martin wasn't a moron. Then he immediately starts talking about insane complicated hare-brained schemes to 'clamp the flywheel to the shaft' instead of just using a key. MARTIN, ARGH!

[@fsxnico]

All I hear is : "I don't have the right tool for the job therefore I will go out and beyond to get a somewhat satisfactory solution, even if it means going against all gods of engineering." Get the parts machined.... end of the story

[@32rq]

I used to look forward to every video. The last video I kept wanting to skip forward, this one I actually did. 😴

[@AB-Prince]

while perpendicularity is important, allowing some slack is important to making the bearings last. A lovejoy connector may be the solution you're looking for to make up for misaligned shafts.

[@SciPunk215]

Hey, it's only the flywheel... the heaviest part of the machine spinning at high RPMs, with a gigantic amount of kinetic energy trying to explode outward in a random direction with the ability to cause massive damage and pain to whoever or whatever is in its path. No big deal.

[@josuelservin]

If you have to hammer in bolts something have gone terribly wrong...

[@Lysou]

Hello, As a mechanical engineer, i think that when you will tight the bolt, it will compress the bearings and create a stress in the axial axis. At time, this stress will degrade the balls. Whereas the skf flange doesn't transfer any axial effort. By allowing rotation, they make the system isostatic. Blocking the two bearings will result in a hyperstatic system (which you don't want in your case), if it's not perfectly concentric (which is not possible) it will create radial effort in the bearings. I think the best way to lock the rotation between the axis and the flywheel is the conventional one : use a fitted key, you juste have to laser cut a notch in the center hole. You can also use a machined flange with a radial screw if the torque transferred from the wheel to the axis is low (should be the case if the pulley is attached directly to the wheel) Also, i think its more important to balance the wheel by drilling some holes or adding weight, than having it perfectly concentric Hope this will be useful, sorry for my imperfect English

[@warbossgrimjaw7352]

Not an engineer but I can spot I think the biggest concern with the design. It's not the alignment or load anything else you've addressed. It's the fact you are relying on 8 bolts contained in 2 plates to hold a bearing under strain and not shearing or shifting. The solution at the end of the video would make more sense since the load is onto the frame, not the shaft

[@MaxMakerChannel]

I am not sure if this is a April fools joke. He spent years already building a new marble machine and wastes his time with things like these. If he find it interesting and fun, sure go for it! But its not a good use of time if the goal is to actually build a machine.

[@StephenHughesJelen]

You don't need a flange to clamp the flywheel. You machine the shaft to have a "bulge" that the flywheel is pressed onto. Press-fit is standard for very large machines....

[@DaftFader]

Heat the fly wheel up, freeze the shaft .... badaboom badabing.

[@averagejoe5050]

Martin, the pillow blocks are the way to go. You can get a standard sized shaft with a key way. A matching key way in the flywheel will allow you to lock the flywheel to the shaft, and locking collars will keep everything in place. Added bonus is the shaft and flywheel should be perfectly aligned with each other, while the pillow blocks will have some adjustment built in to align the assembly to the frame. The type of shaft you need should be common for farm machinery, so look to a farm equipment supplier(think Fendt, or John Deere). Lots of heavy equipment and farm equipment are using these exact parts all the time under the most abusive conditions and would give you easy access worldwide for spare parts for the future. Cheers from Canada!

[@henryD9363]

Why do you need 7500 RPM? That will be incredibly noisy even if you have the most magnificent NASA air bearings. What is the purpose of this big flywheel to begin with? In addition, the mass of a flywheel should be along the rim and not in the center. The moment of inertia of a flat disc is very poor which means it can't store much momentum or kinetic energy. The mass should be distributed away from the axis.

[@codemonkey2k5]

Martin, just a heads up. ChatGPT is known for rounding errors when used for complex math. Trust, but verify. ;)

[@Wintergatan]

This video went a bit long but i had so many thoughts to communicate about this haha :) Leave your feedback on the design at the reddit page here: www.reddit.com/r/MarbleMachine3/comments/13dsv37/worst_bearing_housing_ever_marble_machine_3_ep3/?context=3 Huge thanks to Lucas for doing the test! Check out Lucas KZbin Channel: @Musikkabinett And thanks to hannes for keeping me sane! Check Out Hannes KZbin Channel: @Trainerds

[@F7c82]

I personally prefer longer videos because….. …..more content

[@Oxtorayk]

Hey martin, i've been a fan for many years, since detektivbyrån. The marble machines are really impressive engineering projects, but don't you think it is driving the music making to the second plane at this point? I'm still hyped for the concept of the second Wintergatan album you posted many years ago...

[@ЕвгенийБахметьев-н3к]

Hi Martin! Love your videos and music, but please don't use ChatGPT for any kind of calculations (especially ones you cannot proofread yourself), as it's not meant to do that. ChatGPT is a language model meant to string words together really lifelike, but it's not a true AI by any means, and there's a very good chance that its calculations are just random gibberish. Wish you all the best, hope you'll solve this problem and any others that arise in the future!

[@CullenLorsung42]

This video highlights what has caused you to fail twice already, but now with a touch of arrogance.

[@hunterra217]

as much as I think the threaded bearing housing is a bad idea, I can appreciate that you're designing it and showing it BEFORE you build and integrate it, I think that this sort of thing will definitely help with the longevity of the project. Though I think, like many have said, the benefits of tried and true outweigh the slight alignment concerns you have. Though the housings you showed are designed to allow for concentricity adjustment, designing with tolerance is a must in any active system, especially fly wheels, where designing for exact precision will only cause it to break once a slight deformation inevitably appears somewhere.

[@andymouse]

FFS if your stuck at a bearing housing you won't get this done in your lifetime...cheers.

[@SomeGuy-qd3li]

I want to cry watching this

[@jasondm]

I feel that the threads on the bolts will eventually chew into the bearings and cause slop. I don't know if that would take a month or 10 years of use to get to that point, though.

[@3Canibus]

Big Love.

[@ramirezkine5704]

This video is the vivid definition of "Let him cook".

[@SomeGuy-qd3li]

Let him char*

[@ssilk]

I ask myself: Why do the parts need to be lasercut? Why not lathing??? And no, big holes in the plates... for storing rotational energy it's not needed to make the plates completely solid. The stored energy is proportional to the radius; So it's much more efficient to compare this design with how real flywheels look and works. It will spare you around 50% of mass but looses only some 5% of the storage capacity.