I'd suggest having different error tolerance levels: the excessive pressure on the gates is a problem that would stop the machine from performing its primary task, so that really does need 99.99999% reliability. The other issues raised are much less severe, they're just dropped marbles. From what I understand, the machine has a huge overabundance of marbles relative to what it strictly needs to run. Losing 2 in 5000 marbles due to "dumpouts" probably won't mean having to stop a concert half-way to collect them off the floor. Or, if that actually is too many, it's the sort of problem which doesn't need 99.99999% reliability, 99.9999% (an order of magnitude easier to achieve) is probably sufficient. Having different error tolerance levels should help identifying what Martin *actually* has to focus on.
I’ll bet that this will happen naturally. Martin will fix these flaws at their current reliability. Afterwards, the 10 000 marbles will drop fine (hopefully). As he pushes it to higher and higher orders of magnitude of reliability over the months to come, some errors will pop that he just wont mind being there anymore, and viola, you suggestion was realised.
Another addition: if you have problems 100% of the time when there are too many marbles in the pipe, it makes no sense to talk about 99.99X reliability. It doesn't just randomly happen that the gate fails, it fails at a specific friction level. Reliability here is finding all the conditions in which something fails, not just test the same setup for hours on end.
@@billkeithchannel yes, but unless they make a really short concert, that problem is something they'll have to with because there's no such thing as 0% error rate
Missing a note is a major error, but not catastrophic, the song could go on, and so long as the condition is known can be rectified at the end of the song. I would like this to a guitar string breaking. A catastrophic error would be a jam in the lift mechanism like happened with the conveyor. This would not just end the song but end the concert. Chasing such a high level of perfection is unrealistic. Develop a risk based approach to the build, recognise the areas where critical errors can occur, develop maintenance schedules around those parts to ensure they’re always at optimum. Where major errors can occur, like a stuck gate, have procedures in place to correct.
@@melkerclaesson6570 True, true :) . Its just an aesthetic thing/habitual left-brain type thing. I will try to implement something in as few lines of code as possible, because my brain doesn't allow me to do anything else and be happy. I know recursion doesn't have a huge amount of use at the industrial level, and anything implemented recursively can be implemented iteratively at (generally) a cost of fewer resources, but I still like using recursive logic sometimes, just for the fun of the challenge in non-commercial stuff. Even where it seems a dead alley in terms of system resources, you can find certain break condition(s), that can stop the dead alleys in recursive statements and improve efficiency to the same degree. I like to be more of an explorer when I'm writing than just an A to B type. But generally like you said fewer lines of code mean nothing, so most of the time I eventually leave it alone and just write the annoying code that I know will work (but offends me).
Martin truly matured, "my goal in life is to stop using the angle grinder", "I want to avoid friction fits", I know you use tap wrenches instead of the cordless drill for a long time now. Proud of you man
Tbh its easier to get a straigt tap wjen starting with a drill, and as long as its plastic or aluminum its fine if youre careful. You can also use the ratched mecanism build into the drill so it wont break. Id rather see him countersinking properly :3
13:12 "I don't care about it looking cool any more, I just want it to work." This quote says a lot about the future of the project. Probably in a good way. :)
I'm sorry, but the slow-motion marble machine with all the machine sounds and all... THAT'S THE BEST THING EVER, I love it so much, so amazing, I hope we have more of that in the future too!!!
Martin, you need to track "critical" and "non-critical" failures. Anything that can stop the music or cause damage is critical. Missing the funnel is non-critical. If you lose one marble per 1,250 drops due to missing the funnel that is an annoyance but has zero impact on making music (if anything, it adds a bit of charm). So if THAT is the difficult one to fix, do not sweat it. You need "5 9's" on critical failures. 4 9's on an annoyance is fine.
A bit of a tip. Nylon lock nuts are designed to only be put on once and then left. They don't work properly again if you remove them and then put them back on. So, for parts like the dropper brackets that are being taken on and off, you're better off using standard hex nuts and then spring washers to prevent it loosening off from vibrations. Love your videos, keep up the great work!
This needs an upvote, as he is entirely correct. Generally speaking, lock nuts are meant to LOCK something in place - if a lock nut fails, that means you've not taken your tolerances into consideration. Such uses tend to have a lot of redundancy to reduce failure. Been watching Martin for a while, and it doesn't look like he's... really doing redundancy that much.
@@SmilingDevil Deforming Schnorr washers are good, and Nord-Lock are marketed for "extreme vibration applications". What makes my brain melt is when he puts Threadlock on a Nylock nut! There's something just conceptually wrong with the engineering philosophy of that combi...
next video: Martin rips the machina apart and rebuilds it from fiberglass and aluminium because it weighs so much that it creates its own gravitational field and the marbles fall into its center
Hey Martin, I hope you will see this because I just want to give you a couple of tips for TIG welding. #1 Do not wave the torch around over the entire weld when you finish it, you can contaminate the tungsten and if the weld pool hasn't fully solidified you will allow oxygen to enter the pool and cause porosity which will weaken the weld. #2 When welding thin bars you can clamp a piece of aluminum or copper or brass to the opposite side and it will act as a heat sink, also it wont bond to the steel. #3 If your machine doesn't have a pre-flow setting you can hit the trigger before you get into position to weld and start the gas flow a little early, doing this will cover the area you are about to weld with argon before the arc lights thus preventing contamination. If you have any questions about TIG welding post a reply or send me a private message and I will be happy to help you.
It's very good to see this starting to run consistently. Even the fails that happened during this test (at already like 99.99% success rate) were only marbles falling out of the machine, which though that is not great it's something you can fix and most importantly even if it ever does happen it will not stop the machine from playing.
16:05 On the subject of friction-fit things with bolts, I'd like to point out a neat aspect of bolted connections; friction is the primary aspect of how they hold things together. The shear strength of a bolt is usually much less than the friction between the two surfaces that it is clamping (if properly torqued). EDIT: A lot of people are saying this is false. In the context of plywood and plastic, maybe this doesn't apply. But I will back up my 'claim' by citing one of the most revered books in machining, "Machinery's Handbook". "High preload tension helps keep bolts tight, increases joint strength, *creates friction between parts to resist shear*, and improves the fatigue resistance of bolted connections."
This comment dose not factor surface area of object in contact for friction over area calculations vs kinetic mass and angular mass there for is false and a bunch of talking in a circle with no facts presented.
@@s56505 generally if the part is going to experience a high shear load you design it to avoid the load path being through the fastener by using a boss or location feature. It's safe to say that in most heavy duty applications the bolt itself is not used for resisting shear. With rough cut metal and plywood there is a significant coefficient of friction.
All predictions will end as the carbon graphite used between the wood slowly invades all friction fittings as the marbles transport the carbon throughout the machine. I cringed when I saw Martin dump that stuff on the new plywood parts, to reduce friction, of the redesigned marble drop gates. It is not healthy to breathe graphite nor will it provide reliable endurance without consequences over time. Furthermore, what if the graphite induces a static charge on the marbles or parts of the machine? Martin will need to ground all parts of the machine to eliminate ESD events. Any time there is kinetic friction, there will be issues with ESD events. Rather than friction parts sliding together for the marble slide gates... a better design might be to incorporate materials that last for trillions of drops. Fork truck wheels take billions (or more) of rotations under load before being replaced. Use captivated mini fork truck wheels for the movement of the marble gate droppers. Fork truck wheels take billions of rotations under load before being replaced. These too would require anti-static grounding unless designed with non-carbon material wheels.
The failure in the marble drop is not only due to pressure of the marbles, but by their friction on the bolt threads. Replace the bolt with something threadless! (You could even use an angle so that the pressure creates a force that helps to open the gate, counteracting the increased friction.)
Just make a P-Trap for the Marbles, this will reduce friction at the point of drop. In other words a slight angle up after the down bend should relieve the friction.
2:48 „less anglegrindable“. We thank you, Martin. So I don’t have to wake up my neighbors by shouting at the screen to not use it once again 😉 Greetings from Munich, Tom.
I really loved the slow-motion shots on this one. The closeups and sounds made it seem like an absolutely gigantic machine that came out of ancient times. Love it! :D
Hi, If the new marble path reduces static pressure on gate, won't friction also prevent the small upward marbles move, leading to an other way of blocking the gate?
It's really cool to see this. One of my first engineering internships was running reliability cycle testing for the products the company I worked for made. We had big test machines that would do this--come in, turn them on in the morning, check them every ten or fifteen minutes all day, turn it off at night. Document any failures or wear for the design or materials engineers to improve. Some of our tests would run for weeks. It's fun by the minute, dead boring by the hour, and critical by the day since it's a lot of the real business of making real machines, proving them in action over time. I think it's worth categorizing the failures by criticality. The marbles locking the gates is a failure of the machine, so any one of those is a failure of the machine, and needs to be fully diagnosed and dealt with. OTOH, if a few marbles jumped the tracks in a song, it wouldn't _really_ impair even a live performance as long as there's sufficient reserve. 1-in-5000 is still not good enough, but it's a pretty good start (it's enough to reliably play one song in the studio, for instance, per your numbers from the Oct 15, 2020 video which say a song needs about three to four thousand marbles). Still needs work, but getting better. On the other hand, the twenty total escapes from the machine in the course of over 5000 drops mean that in a show that plays 8 songs on the machine, you'd only lose about 160 marbles. The machine's reserve capability seems like it can definitely handle the 20 you might lose during a single song, and topping the machine off quickly on the fishladder or other reserve locations seems easy enough mid-show. If you can cut the wheel intro marbles from 14 to, say, 2, then you can cut it down to less than 100 lost per show. To be honest, I think the marbles ending up everywhere is a bit of the charm as long as it doesn't interfere with the operation of the machine--so I'd focus on the gates first until they're up to the required reliability, and then the largest source of marble-out events, and then the rest.
In fluids the pressure on a point is determined only by height of the liquid pillar regardless of the shape. I think this applies to marble tube and the marble gate too. Curly path will carry the marbles yes, but at the same time you will be adding marbles to counter that.
To further secure the delrin parts from sliding, use toothed washers. The ones that bite into the material when clamping down the bolt. Usually they prevent bolts and nuts from unscrewing themselves, but I think they will also do a better job at preventing any lateral movement, than the flat washers.
Put Water on the felt, this prevents burning it. PS: also Use as Short Arms as postible to open the gate,this prevents too high forces on the spining/Programming Pin
GREAT You maybe appear in the history books congratulations, I havent seen a mindmaster like this, this kind of people like you are the future of the humanity!!!!!!
19:30 Those link bars need to be removable. 1 or both ends should be bolted on instead of welded. Makes the whole 4:55 removable marble gates incredibly difficult since the marble gate sliders are now hard mounted to the top lever bars.
That would be my comment too. The gates can be removed, but almost all the 'convenience-factor' Martin has laboured to build in to the new design is negated by this one glaring issue! EDIT: I also worry about the stability of the gate clearances with that length of actuating arm, but time will tell..
My suggestion to this, would be to have the screw connection on the bottom, so that the "arm" still attached to the machine, this would have less of a change for it to bend, and miss-align the marble gate. There is the screw that hold the arm to the little metal bar that connects to the wood, so he could remove that screw instead as well. Over all I'm really liking the the re-design thus far! Keep up the great work Martin!!
11:27 Wintergatan making detachable gateholders to be able to weld them on the workbench, then proceeds to still keep on welding them on the MMX anyway.
I'm glad to see you making compromises for function over form even if you hate making compromises. This is a mechanical marvel and people already think it's amazing. It's worth the compromises on a few aesthetics if it means you getting it finished and being able to write and play music on it; I think it'll be the last thing you're thinking of when you're up on that stage playing for thousands of cheering fans, or when people are complimenting you on your masterpiece. Keep up the great work, looking forward to all the music to come :3
Martin, lets put aside your amazing intellect, superb problem-solving and critical thinking, as well as your talent as a musician; there is still the fact that your filming, sound and editing are top-notch. I love watching your videos and seeing your process. Much love from Fate Texas!
Awesome :) kind of sad and happy to see the end on the horizon. You’ve worked so hard for optimal success for the marble machine x, but also for so long, with us involved, it feels like a part of the community now, and a part of us individually :)
Martin: Instead of a sliding gate, why not use a metal rotary gate with a pair of small bearings? The marble weight issue would be dramatically reduced and wear and tear as well. Think how a six chambered revolver (pistol) ratchets one chamber at a time. Similar idea.
You know, it's funny... I started following the MMX build just as I finished school. I had some setbacks in college just like Martin had with the project, but I hope he can finish the machine before I get my PhD and grow my first grey hair.
Been watching for years from day 1. When everything gets back to normal, I would love to take a vacation when he starts doing concerts with this machine.
I stumbled across a marble machine video about a month ago... and I go what the hell is a marble machine??..I watched the video and was completely blown away... since then I have subscribed to your channel and ill be honest I have not had time to watch them all yet but I will. The ammount of work, outside of the box, under the box, over the box, and behind the box thinking and engineering that you have done and are continuing to do is absolutely incredible. Just incredible. Thanks for putting this out there for us to see
I am sure someone has pointed this out by now but I could not find it. In the calculation of the preliminary error rates you seem to have put two additional decimal places of 9s into the result. 5082/5096=0.997=99.7%, not 99.997% and so on. I hope this did not mess too much with some reliability predictions and so on.
@@thomasbecker9676 Still it's important for him to realize that, because the bad math is part of the reasons behind the bad methodology... Martin thinks he's 2 orders of magnitude closer to his (unachievable) goal than he realistically is...
@@stekikun6854 Unfortunately, it would seem he doesn't read comments, nor has his much-lauded "engineering team" pointed this out to him. I think this project is going to go the way of SV Seeker. Never finished, always something new to make content about, continue selling merch and advertising Patreon.
Love the “Marble Machine Rag-Time Blues” at 14:00. I’d love to be able to just knock out something like that, and to think it so inconsequential that I could just use it as a bit of background filler like this.
"If you're struggling with finding meaning in existence and you have some spare change, get yourself some ratchet tools and just put nuts on bolts." -Martin
my mind went to some kind of fantasy gnomish or dwarfen factory, with the slow-mo snare sounding like some ridiculously oversized powerhammer or something along those lines
There were some pretty mean comments from people who lost patience on the last few videos, after the redesigns. My guess is that he lost some of the more recent backers, who weren't prepared for Mr. Bones Wild Ride like the rest of us.
Nah it will be fine, also the marbles are shaving off a minimal amount probably. But in case he wants to cleen it he just needs to get a stronger magnet, put it in a bag and move it along the ones on the machine. Also where are they everywhere? The ones in the gears and the demagnetizer?
The marbles themselves are demagnetized often and regularly which means that there's no worry about dust settling in them. As for the permanent magnets they've been placed usually behind wood if I recall correctly which means the iron dust is just going to be pressed into the wood.
If only this was the primary focus in the beginning, it would be done by now. I'm all for a balance between function and form... but considering Martin designed this for a world tour---function should override form. Life on tour is rough even on the most robust parts.
I’ve said from about 100 episodes ago, this instrument as it’s designed is for studio use, not for touring. It’s far too complex and too many parts and adjustments. I’m glad to see some things are being simplified. As we are seeing black lines building up in the clear pipes, I’m wondering how much buildup there will be after 10m marbles going through. Are the pipes going to start clogging if they aren’t cleaned out?
I hope you see this Martin. Your new connections between the drop mechanism and the activator at the top won’t hold up over time if you’re relying on welds. Look into a sheathed cable such as a motorcycle clutch cable where you can adjust it to a very fine degree. It will mitigate the issue you’re having with the friction pressure on the marble gate by transferring more force easier. Please boost this post so Martin sees this! This is such a simple solution and I want to see him succeed!
@@stephss I've done so, thank you for the suggestion!
4 года назад
An advice I wanted to give for some dozens of episodes already : as an engineer I've been educated to design complex machines as modular as possible. Each module should be easily removeable so one of the choices is to fit things together with bolts. Seeing the angle grinder and TIG welder used to remove/replace a sub-assembly made my cry each time. Happy to see it's now bolted ! :-) That's the way to go, for sure. Why not the step beyond by allowing adjustments to be made while bolting ?
8:00 "[random humming...] If you're struggling with finding meaning in existence and you have some spare change, get yourself some ratchet tools and just...put nuts on bolts. Every little click feels like life is getting better." - Martin
It's been at least 5 years since I've found your channel and I'm still following your videos! I can't believe Ive been here since I was a sophomore. Keep up the good work! I look forward to the advancement and progress you make next!
I imagine that in one place in your brain you have developed a perfect MMX that is working every moment of your day, and this MMX whispers to you the necessary improvements it needs. A fascinating evolution!
So many engineers and machinists are commenting here about technicalities (hopefully Martin and/or his team will get some useful advice), and I totally get it. I'm a violinist and every time somebody asks for feedback or posts a video, I go all technical. "Yes, but your little finger; yes but move your elbow/wrist/head this way or that, just a little bit; yes but pressures/contact point." Etc etc. For me, as a non-engineer but DIY problem solver who's worked with (or aspired to work with) so many of these kinds of tools, it's extremely satisfying just to watch him work through the problems and, well, use the tools. I keep coming back to these videos because they're so ... calming. Thank you, Martin. (My favorite thing to do with my Dad was walk through a tool store. :) Good times. I miss that, I don't know anybody else who's into these things.)
Thank you so much for just being you. You're an incredible guy; funny, smart, charismatic, and so passionate about this machine. Ever since a few months ago when I found the video of the marble machine making music, my life has gotten a little bit happier. These videos always help to make me smile no matter what. Keep up the amazing work Martin, and thank you again.
Everytime I see Martin in the navy I Believe hoodie i think “he's wearing my hoodie. Oh no, wait a second...” - its one of the greatest things in my merch collection.
Being in here since the beginning, I really feel like both you and the machine have grown so much. It does really feel different, and it makes me smile that with the redesigns you're able to troubleshoot more easily! Greetings from Costa Rica~!
None of that math really has anything to do with how mechanical reliability is calculated in the real world, though. It's like saying if you have a 50/50 chance of flipping a coin and getting heads; if you flip it twice, you'll get two different results. That doesn't hold up in practice.
@@thomasbecker9676 Your prior in a fair coin toss never changes. You always have a 50/50 chance, it doesn't matter what the past results were. How is it implied that the past results matter for Martin's calculation? The most common mechanical reliability estimation in the real world is mean time between failure, which is derived essentially like this - you find out how long it takes for a failure to occur (over many, many iterations, to get a meaningful sample size), then you take the mean of that. Here, marbles dropped at 120 bpm is time.
"At last! I've set things up so I can weld at my bench instead of on the machine!" - (leaves the snare marble drop attached to the machine the whole time he's welding it)
Martin, have you considered marking the mables that missed the funnels with a bright colour and then re-using them to see if the same mables miss again! i suspect that they will- i expect that you could seperate these non confomist marbles during your test runs!
Hi Martin. I actually work at ITER, and yes, it IS a COMPLICATED machine, but I have to say that I'm amazed about your dedication and obsession to build and improve this wonderful machine.
Really really happy and exciting with this video. The best "Martin moment" for me was when blowing the welding flame in the plywood with the mask on. Absolutelly pretty distraction from a genious mind. Thanks again master for every minute of your live posted forever in this videos. I just can't love you more.😏🤗