BUILDING A MARBLE CLOCK Pt.4 - NOW FASTER!!

Check out PCBWay here: www.pcbway.com/?from=Ivan for a trusted one-stop solution to 3D Printing Services, CNC Machining, Injection Moulding and more!
As I promised in the last video I tried my best to make the clock update minute by minute. It wasn't an easy feat as I dealt with many (expected) issues. As many of you will notice I'm still having lots of issues with the marbles getting stuck on the machine as there's about a 2:5 ratio between good and bad marbles. If anyone knows about a good source of glass marbles please let me know. I had to do a bit of fiddling with the programming to get the timing right and I broke a few more things that what can be seen in the video but at least we have a clock that runs on time now. It is now louder and has less music to it, it is less zen and more stressful but learned a lot here. This project was a lot of fun and I may try some of the things that I could not do in this series in the coming videos. As always loving to hear from you in the comments.
Join this channel to get access to perks:
/ @ivanmirandawastaken
If you want to support this channel you can do it here:
Patreon: / ivanmiranda
Merchandise: ivanmiranda.space/store
PayPal: www.paypal.me/IvanMirandacom
For regular updates on my projects follow me on:
Instagram: ivanmiranda.space/instagram
Twitter: ivanmiranda.space/twitter
Facebook: ivanmiranda.space/facebook

[@Edward.Hillier]

This is one soundproofed glass case away from being in the lobby of a science museum. Excellent work!

[@thomasrosebrough9062]

And a significant increase in reliability, unfortunately

[@zekiz774]

@@thomasrosebrough9062 huh?

[@TheRationalOptimist]

Ivan or the marble clock?! Both seem like they would be good additions!

[@BTboxing]

@@zekiz774needs to run all the time for a long time, be relatively easy to operate, easy to fix and maintain.

[@nafitron]

Can I just say, your "ad" was exactly what more creators need to do on KZbin! Show how you used your sponsor... this was GREAT. I think it's one of the only ads I haven't skipped!

[@antonliakhovitch8306]

Except he didn't actually say the word "sponsor" or in any way mention that it was an ad. Otherwise, I completely agree

[@marcpym5251]

@@antonliakhovitch8306 He showed "Paid advertisement for PCB way" in the top right corner, which I think suffices.

[@nikkiofthevalley]

​@@antonliakhovitch8306It says on the screen "Paid advertisement for PCBWay" when he starts talking about them. And it was pretty obvious it was a sponsor, anyways.

[@Tasarran]

@@antonliakhovitch8306 4:48 it says 'Paid advertisement for PCBWay' at the top of the screen...

[@bernardorodrigues8848]

@@antonliakhovitch8306when he first mentions using it there is text on screen saying it’s a paid advertisement tho

[@TheLeadZebra]

I love the fact you managed to do all of this mostly without changing parts already on the machine, or at least not redesigning any part. Incredible stuff

[@kryvian]

in part 3 he redesigned most of the parts tho.

[@TheLeadZebra]

@@kryvian yeah true, but all of the improvements for improving throughput were software tweaks and mechanical timings

[@PTRRanger951]

@@TheLeadZebra In this video were software tweaks, but he had to remake the machine completely from the ground up, because EVERYTHING had to be redesigned.

[@guts60]

I would have timed the clock perfectly to work for 5 minute intervals and called it a day. Shows this guy has more patience than I do, and is WAAAAAY smarter than me lol

[@Starfireaw11]

Now we need a 24 hour video of it running.

[@bottle-]

Yup, viral fs

[@swappycars]

it would keep breaking it would need to be rebuilt with steel or aluminium to run that long

[@tonyhawk123]

I'm sure the neighbours would appreciate that!

[@fbelard]

@@swappycars I think you're underestimating this person's tenacity. at this rate by part 12 he'll have figured out how to make it wearable.

[@swappycars]

@@fbelard by part twelve the material used should be marble so it’s a marble marble clock 😂

[@TheBekker_]

This should really be in museum or something, so cool!

[@ionymous6733]

maintenance would be a killer

[@TrixityMcLight]

Agreed came to the comments for this. It's just one of those projects that could actually just sit in a modern art museum

[@Viniter]

I feel like very few museum would be willing to deal with the noise :D

[@JaredConnell]

Look mum no computer's museum might be a good fit

[@bortimer96]

Yes! Like a science museum.

[@wake11223]

Love it! Maybe upgrade: If the elevator links could carry unused marbles back down and reject them there (counterweight), it would be much easier for the motor to move (assuming that full elevator is quite heavy).

[@eulachonfish]

My suggestions for more speed: (1) Double buffering so you don't have to stop the conveyor. You don't need to store an entire set, just enough to keep the elevator running. (2) Instead of the critical timing on the solenoids, use the same ejector comb as the return, but use the solenoids to DISABLE the ejectors for all marbles not selected. This is fail-safe in the event of a timing mishap, only ejecting marbles instead of damaging equipment. This type of design is fairly common on old mechanical equipment. Otherwise, great build!

[@f308gtb1977]

That’s brilliant. Great ideas.

[@celestewilliams5681]

This is one of the coolest things I’ve ever seen. I hope you release a seamless looping 12 or 24 hour video of it!

[@BillyPearson48]

Supporting this idea!

[@FoxHoundUnit89]

​@@RepentandbelieveinJesusChrist-didn't ask mfer

[@NathanaelNewton]

@@FoxHoundUnit89if enough people flag their comments as spam they'll get blocked by youtube eventually

[@natthasaranwiangnak1066]

@@RepentandbelieveinJesusChrist- who care about this?

[@BillyPearson48]

@@RepentandbelieveinJesusChrist- TF?

[@IcthioVelocipede]

Now that you have made it faster, I think the next challenge should be to make it LOUDER. Compressed air jets instead of the combs, perhaps?

[@fbelard]

no chill

[@fbelard]

maybe connect a second machine that alternately makes a tower of champagne coupes and champagne flutes (the rejected glasses from the sorting chain could be used to mark the seconds on the clock, or to chime the hours.), then fills the champagne tower and runs it under the path of the falling marbles. but that's not very sustainable, so you're gonna need a third machine, something like a combined kiln/3D printer to melt all that glass and make new marbles. edit: i'm just not sure about all the sugar in the champagne, it's a bit of a fire hazard. I guess you could just use sparkling water. if you prefer chill over no chill.

[@DijonFromage]

I was thinking compressed air to pop the extra marbles, too. The air compressor would add an ambience...

[@ssFaxes]

Air horns instead of air jets, we are engineer and so can you!.

[@DijonFromage]

@@ssFaxesTHERE it is. Truly no chill.

[@sskgmail]

For a Pt. 5, you should do sorting at the base - maybe via two stages at the base one white, one black, and then the thing up top can be a secondary check. Would also be cool to see release of new marbles shortly after release of old marbles.

[@ronsijm]

Yea, I was going to suggest the same. Though I'm curious what percentage of the marbles is send back down... Like if 50% of the marbles are discarded back down, you could potentially speed up the machine by 50% if you had perfect sorting - or if you're already under a minute, run the machine 50% slower and thus making less noise

[@HarmGeerts]

Having dedicated separate loading of white and black marbles to the lift would be a massive improvement and fix color starvation. I think it would simplify the clock. You would need one sorting section. A section to optionally load white based on the clock display. And a section for black to fill the open spaces. The lift could run as fast as is needed for marbles to fill a spot.

[@ComradeTiki]

@@HarmGeerts For really easy sorting, I'd have 2 sizes of marbles (like having the black ones slightly smaller) and pass all disposed marbles over a rail to sort them by width.

[@wusulus]

You could also speed it up if you wouldn't 'rewrite' lines that would stay the same. Or you could reuse parts of the line by only dropping for example the first 10 marbles of a line.

[@thiamath]

Yeah, picking the marbles in a random fashion has been the elephant in the living room during all the chapters...

[@TOPSHOOTAS]

That last swipe at the for the stuck marbles was hilarious! Good job, that was a fun project

[@computingnerd7005]

Holy I can't believe you managed to optimise it to make it sub minute refill (on average), impressive work!

[@Titanblade17]

You might be able to make the clock quieter by padding the return funnel for the rejected marbles.

[@Kalvinjj]

In fact it could be just a plain pipe with a rubber pad at the bottom where they land.

[@gayMath]

if it were a clear pipe it would also look better than current

[@Lodit24]

@@Kalvinjj I think the problem with that would be marbles landing on top of each other and getting damaged

[@lupeters213]

Just use a rubber hose with a big enough diameter.

[@kingcosworth2643]

And painting the rear of the main track with sound deadening

[@Schwuuuuup]

The code could be optimized to only replace the lines that actually changed from one minute to the next. But this would change the visual appearance of the change

[@ha69]

This would defeat the whole purpose

[@Ninjaguy1b]

This would also involve a redesign of the physical channel releases so that you could retain some channels while purging others. And it isn't clear that this change actually saves time overall, since there is a dedicated elevator lane for each channel, meaning that on average the clock will not be waiting on the unchanged channel anyways, so you don't gain anything for all that work.

[@erikjohnpotvin]

I was thinking the next prototype could have verticals channels, and only one number would have to change constantly, the rest at 10 min and 1 hour intervals.

[@JustinHixson]

It also simply wouldn't speed up the process very much. Each channel updates completely independently from the others. Turning off some lanes doesn't make the others go faster.

[@pabo26e]

i think there is an argument here, the bottom and top channels never use white marbels but they get them nonetheless so if you make the feeding mechanisma for those 2 independent not only they will get faster but they will increase the probability that a white marvel gets to where it should go, also i would cut the ammount of marvels in every line those blacks at the beggining and end and also in the middle mostly waste time

[@v5r223]

It was much more interesting than all the TV shows in the world. Congratulations on your success.

[@denziljoe]

Your clock still has waaaaaaaaaaaaaaaay more chill than Martin's quest for "tightness"

[@sapincher]

is gravity actually tight idk

[@cooperised]

Yeah that's driving me up the wall. If he wants "tight music" he should just use a synth. The art is in the imperfections - the only criterion is that the imperfections must be small enough to reduce the chance of critical failure to an acceptable level. Once again he has extravagantly failed to capture his requirements properly.

[@jonasthemovie]

The marble machine is dead.

[@yinlu3610]

The guy you’re referring to is the Wintergaten guy who’s been trying to make a marble machine “perfect” right?

[@denziljoe]

@@yinlu3610 Yeah, that's the guy

[@TooBarFoo]

I think, if starting over, the marbles should have always have been pre sorted into black and white pools and then loaded. It's a simpler design without having the mass reject and return. It would also reduce the volume as less marbles are in motion at any one time. Very very cool though. What a crazy journey. wonderful!!!

[@MrFozzie52]

I like that it has that element of chaos though

[@trevonrose3909]

Then he would have to find a way to get the intended slot and then a way to sort the marbles back into black and white after they get released, which is probably more work tbh

[@RedstonerD]

​@@trevonrose3909 I was about to reply the same when I thought of a solution, oops. If you have a channel of black marbles and one with white marbles that meet a loading channel perpendicular to the belt, you could use solenoids or a wheel to add the marbles in the right sequence. You'd have to wait for an entire row to be dispensed before loading them onto the belt however. Similarly you could separate the marbles at the top with sorting channels. I'm however not sure whether this would actually slow the whole system down too much or not. Plus it would make the design a lot less fun

[@xenicat]

@@trevonrose3909 he could just use a bunch of splitters like the marble machine to have a constant stream of each color available at every lane which then can be released as needed for the display. the conveyor belt can stay as it is and the two rows of solenoid valves just push out the colors into separate funnels which go into the black/white pool that feed into the splitters.

[@trevonrose3909]

@@RedstonerD yeah I thought the same thing. That would probably be more complicated and the chaos of rejecting marbles is nice lol

[@nicxmoon]

this really needs to be in a science or art museum or something, like someone said behind a some kind of noise dampening glass but with information on how you made it. This is absolutely fascinating, i could watch it work for so long and silly be amazed

[@gaurishpassi4812]

Thank you to the patreons and channel members who pay him enough that he can make these videos and spend on these crazy projects. Truly a great guy and an engineer in true sense. I hope for your success!

[@Khofax]

The journey was so fun to follow and the end product is truly incredible

[@joltingonwards2017]

Yeah, shame the comments are filled with bots, like seriously this is unbelievable

[@pmcquay1]

air blast instead of solenoids so that theres no chance of getting caught, and you can probably completely eliminate the stepper drive, in favour of a constant speed dc motor. That would save program time as well. Add some sort of triggering element to each link in the chain so that when it passes a sensor the read head and air blast can trigger. You could do air blast instead of the comb to kick marbles out of the chain on the back side as well, and then you dont have anything rubbing the chain or potentially getting caught, and can probably run the motor faster. Some way to sort the marbles even a little after the fact would probably help as well, though without a redesign I have no idea how that would work.

[@Blue84Stang]

The concern for me with air solenoids is consistency. The delay between the triggering and the actuation might have too much variance to maintain its accuracy.

[@samhnky]

@@Blue84Stang I don't know about the air solenoids you could have access too for a home project, but don't major industrial factories use Air Solenoids all the time to remove bad products (I think mostly Produce) from their lines? Seems those have to be very accurate and work consistently to be able to do that...? Right?

[@Gearjerk5]

Wouldn't air solenoids require an air source in order to function? He'd either have to add an air line inlet, or add an entire air compressor. Not insurmountable by any means, but it would add more complexity/cost/noise. That issue aside, they would be an excellent solution to the solenoid hang risk and comb issues.

[@Blue84Stang]

@@samhnky it depends on the process but automated sorting machines will sometimes use air solenoids. As I understand them though, the products being sorted arent usually moving fast enough, to where a variation of a half-second or so would make a difference. In the case of this ball clock, a variance of even an eighth of a second could be the difference between ball and belt (either too early or too late). I could be completely wrong, however. I know air systems can run into issues if the pressurised supply has too much variance (like, if you are running too many solenoids too quickly for the compressor to keep up).

[@fbelard]

couldn't the tips that actually push the marbles be made of a softer material, like felt or a brush? something that could hit way off and not snag or damage the chain. don't know about the durability, but it sounds less complex (and less noisy) than air blasts.

[@Erkaut]

In case you are really going for Seconds some day, here 2 major improvement ideas: -Make the channels vertical, so you can exchange single numbers. -sort black and white already at the bottom, use 2 transport belts and a timed release mechanism from that two to create the pattern in order. That way you really could reach the seconds😅

[@HrKosmoshosen]

It's been so satisfying to watch this come together. I can't imagine the immense effort, time and general dedication just to this project alone. Thanks so much for enthusiasm and sharing your creation!

[@Jozoman1995]

Incredible! As an engineer myself, I loved seeing the whole process, including all the issues you ran into and your solutions to each one. And then of course… The issues that came from the solutions. Ha! Great work!

[@paulphillips3783]

Crazy. Stupid. Genius. Mesmerizing. Awesome. Pointless. The world really needed this. Thank you Ivan for sharing this journey. Food for the soul.

[@sc0or]

You'll see what the World will really need in 5 years. Start thinking now how you will live without tiktok, fb, yt, and comic shows.

[@jay-tbl]

what@@sc0or

[@khodis2002]

​@sc0or , can you elaborate what do you mean "without ... in 5 years"? Do you mean when you grow up or in general?

[@DivineKnight_115]

Seeing this project come this far was a fun journey to witness. My only idea to perhaps make it more efficient is to do the marbles falling vertically into place rather than build up with a massive track. Perhaps in the future when you’re bored and want to streamline this idea even more and make it more practical, it’s a thought.

[@sanketb1613]

This was an awesome journey to follow all along Ivan. Gladly awaiting to see more such short series!!

[@Peter_A1466]

By adding a second buffer, or a partial buffer, you do not have to stop the lift. Which probably means you can let it run a bit slower while displaying faster. Happy New Year! 🎉

[@stevebabiak6997]

Most of that one minute of time is spent filling the buffer. There wouldn’t be any / enough extra time to fill a second buffer. And the existing buffer loads by passing thing through in a serial fashion. A second buffer would have to somehow load in parallel, and then you have the added complexity of switching the display to load from one buffer and then the other. This design is already too complicated as it is.

[@OiDepp]

​@@stevebabiak6997 I wondered at the first attempt why he didn't just sort the marbles all at once by color. Then he has two buffers from which he could directly fill the individual lines of the clock. But this method would probably be too simple to turn it into a project with exciting challenges.

[@Catheidan]

​@@OiDeppI was wondering why he wasn't sorting in the lines before the elevator but I think your idea is even better

[@ChucklesTheBeard]

@@stevebabiak6997 the second buffer can be inline with the first; it would only take a third set of gates. Currently marbles flow like Lift -> B1 -> Disp -> Hopper Adding another buffer: Lift -> B2 -> B1 -> Disp -> Hopper The lift would still have to pause (or reject all marbles) for a moment to allow all marbles to flow through B2, but that should only take like, 1-2 seconds of downtime vs the current 10 seconds. Currently: T=B1_full: lift halts T=x:55: Disp opens T=x:60: Disp closes, B1 opens T=x:65: B1 closes, lift resumes After adding second buffer gate: T=B1_full: lift halts (or starts rejecting all) T=B1_full+2: B2 closes, lift resumes T=x:55: Disp opens T=x:60: Disp closes, B1 opens T=x:65: B1 closes, B2 opens

[@omaristephens2143]

​@@stevebabiak6997 a second buffer could easily be in series. Just add another stop that closes after the main buffer completes, and opens after the main buffer is shifted to the display. Then if the clock is running behind, you automatically skip it (since the main buffer never fills) and the behavior reduces to the current behavior.

[@ToddRafferty]

This chapter shall be forever known as, "Ivan and the race against time."

[@funkydozer]

A marble machine that works. And was built in a fraction of the time. And a creator who didn’t milk KZbin monetisation. Watch and learn, Martin.

[@MichaelGreenway]

This has been an epic series, I have loved following you on your 3d printed engineering adventure. Keep it up!

[@martinrpolloc3709]

I would really love to see a commercial mini version of your creation!

[@jonasplima]

Wonder how small that monstrosity can get.

[@ionymous6733]

small enough to wear on your wrist?

[@atharvabedarkar]

​@@ionymous6733haha not possible 😂

[@CapN3Cook]

​@@atharvabedarkar possible.

[@Kafj302]

@@ionymous6733 how long can you hold your wrist in the perfect position for the clock?

[@protogenxl]

You know you would only need to update every 5 minutes if this was on display at a children's science and technology museum

[@zekiah2]

I think that would be great for a scaled up version using golf balls or pool balls Something big enough for kids can really walk around it and look into the machine

[@aiodensghost8645]

​@zekiah2 the only problem is construction materials... you'll need some really strong glass to use golf and pool balls.

[@tin2001]

I was thinking the same... 5 minute updates and intentionally slow it down a little so the clattering is more of a pattern that just keeps going like an old steam driven machine.

[@dolphin64575]

I wonder if plastic ball pit balls would work, though the mechanism would be enormous

[@ghosttwo2]

Had an interesting idea you might want to simulate and/or implement; a new operation. As the marbles go by your reader, there is a certain 'miss rate', where the color doesn't match and it gets skipped. You can optimize this rate by varying the marble color ratio, but there's one better. In the horizontal portion of the track, add the ability to drop a marble down into the next track, sort of a shift operation. Now, if two adjacent lanes fail to find a matching marble simultaneously, but the left one is what the right one needs, then you can redirect it to the proper channel after releasing it from the belt. It might also be possible to delay slightly for cases where the left one fails, the right succeeds, but the left is what the right needs on the next iteration; stagger the drop gates to introduce enough delay. Assuming each lane is biased towards one color or another, one color being much rarer, this could essentially double the rate orders are filled.

[@danielkatayama]

Or have the marbles being sorted directly at the bottom, with two containers. you load what you need from the beginning. Potentially, the clock channels could even be vertical as well to speed up the reload.

[@djyotta]

Yes, a smarter selection algo will help a lot. I feel there is one. Your idea is pretty good.

[@auselessdigit575]

I was thinking that to consider the rejects for use in the next frame, to get a head start while the current frame is being loaded. Instead of sending them all back to the reservoir, drop any that are in the correct position into a buffer, then the buffer can be released before starting to build the frame proper. Make sense? A better sorting algorithm would be better in the long run though. Also I would like to see it running with a bunch of coloured marbles and just see the random patterns every minute or so.

[@zeppie_]

there might be a simpler way to optimize: that would be to simulate the marble count using a basic computer program to see which ratio yields the best results

[@HandyKindaGuyUK]

Sound proof glass box. Like others have said, this deserves to be an art piece

[@SilentBeutlin]

If you’ve ever worked with PLA you’ll know that this thing will completely disintegrate after a few hours of operation.

[@vybusnyblogcz]

The comb could be part of top rotor. It would rotate into holes, pushing marbles out.

[@bunnyquark]

this would probably also be better long-term since the stainless steel comb rubbing against the plastic elevator is surely going to grind the elevator holes down in no time

[@warrenyoung173]

I think they’ll just fall out due to gravity when they go over the top.

[@MakerBees333]

Bravo 🙌 and thank you, this is inspiring! Now we just need a 12hr cycle of it running for a digital photo frame so you can share the joy of a marble clock with everyone all day and night… I sure hope the motor could handle that 🤞. It would hopefully get you channel some watch hours too ❤.

[@ChrisMX-ze5vh]

that would actually be awesome

[@NeoMatoane767]

Hehehe, you said, "watch hours"

[@DukeMaxwell]

You surely mean 24 h cycle

[@MakerBees333]

@@DukeMaxwell most people watching this utilize a 12hr AM and PM, 24hrs could be great but obviously twice the ask… unless the audience is mostly southern Europe, I would happily stand corrected 👍.

[@bigfatbill5187]

This is the most insane project I've ever seen created. I can't wait to see what you do next!!

[@Sidkain]

This was such a fantastic project for me to find you channel on. Such an awesome idea and I love that you were able to work out all the kinks and get it to update, every minute, like clockwork!

[@joshmiller887]

I was thinking of this being an enclosure and a small door that people could open to see how loud it really is . Amazing work.

[@ThomasGiles]

Well done, very cool! When I first saw a thumbnail of this series I was imagining vertical lines instead of horizontal. Which means it would be easy to reuse numbers that are already there and even individual vertical lines. And make it waaaaaay quicker to update, with no noise most of the time. Would be interesting to see such a build and compare the two.

[@schnitzelhannes6431]

go make it yourself if youre such a smart boy

[@fbelard]

i love that horizontal is conceptually simpler, but also exactly the wrong way to do it. it makes me giggle. vertical is potentially more efficient, of course, but more complex. one thing that makes Rube Goldberg machines funny is that they keep choosing simplicity or efficiency in all the wrong places, while still having to be reasonably deterministic (or else they're not machines at all).

[@stevebabiak6997]

@@fbelard - your mention of being deterministic caught my eye. Adding more white marbles to make the machine go faster to me implies that there is something about this machine that isn’t deterministic enough. He should be able to calculate the ratio of black to white marbles needed, and use that to provide more than the minimum of each to be certain of success.

[@fbelard]

@@stevebabiak6997 hm, maybe i used deterministic wrong, i meant that on a higher layer of abstraction it will show the time correctly every time, even while the underlying “code” may take different paths to arrive there.

[@numbr6]

This is a great example of analysis and optimization of a problem. Congratulations! Great success.

[@badsamaritan8223]

This is spectacular. I have a suggestion. You could speed this way up, and eliminate the black marbles if you replaced the white marbles with steel bearings, and had little electromagnets for all the number positions. The electromagnets turn on or off according to the time, and as the steel bearings roll over them, they're captured until the EMs cycle off, to set the next time. You could probably actually do seconds with a setup like this, though you might need to switch to a vertical track, so the seconds have their own set of track. Or maybe you could even use the conveyor as the seconds display, either by having magnets behind it to only pick up bearings in the number pattern, pushing bearings away from the sections you want to keep blank. OR you could go nuts, and have two separate tracks, with magnets on them in the number patterns, and the track moving at the correct speed, so that the displayed number travels by every second. One track for each digit, so you can change the two digits separately. Otherwise your conveyor would have to be more than 30m long, lol. It would also probably be much quieter, because there won't be as many balls.

[@voidnull9438]

This has been a great series! Congrats on the clock finally working! :)

[@quantumm313]

you're already looking to see what color each marble is, but then you're sending the ones in the wrong position all the way back down and just hoping for some to go to the correct spot. The machine would be way more efficient if you used another two sets of solenoids to push black marbles into one return and white into another. You could then bank them and use them to structure another update without relying on the lift itself. Another way to boost efficiency - You actually also know what the next sequence you'll need is, since its just counting up. Once you punch out 31 marbles on a specific track, you know that track is done for that time. Having another stopper come down so you can start preloading one more minute in advance means less marbles are getting sent back to the bottom of the lift while it waits to dump the new time in

[@4kirb]

Part of the designs beauty is the mixed marble pool, It wouldn't look or function as cool if you had two banks of colours.

[@teehlfx5238]

What an enjoyable and satisfying series! Thanks for taking us along on your process 😊

[@unairamos74]

No he visto nunca a nadie hablar con un acento español tan tan tan marcado, y a la vez un perfecto inglés!! :D

[@audithisworld24]

One thing I couldn’t help but notice is that there are fewer white marbles on the left side of the elevator. I am unsure why that is, but I wonder if fixing that will help save another 5 to 10 seconds.

[@simeonsurfer5868]

maybe it's because there are more white marble that are rejected at the top than black?

[@Pystro]

@@simeonsurfer5868 Yup, I am pretty sure that's it. An easy fix would be to dump rejected marbles all the way on the left of the tray, where the lines dump into it.

[@Tyrope]

@@Pystro Or evenly among the tray, perhaps with an overengineered marble divider? Couldn't possibly go wrong.

[@jeauwyte5151]

"If I increase the motor speed a little bit it will explo--- It will be able to do all the updates under a minute!" had me laughing.

[@johnpheth]

I bet your neighbours love you testing this at 07:19 in the morning!! New follower here, great build thanks so much for sharing!

[@djroth85]

This was probably the only video series where I eagerly awaited each new episode. I love kinetic art, and this to me is incredible kinetic art. Bravo.

[@GodOfKAPOW]

This is amazing, Ivan! Can't wait to see it in a museum a few years from now.

[@Abrikosmanden]

That's so impressive! I really like how you've taken us through all of the thoughts and design changes that led to a succesfull solution.

[@joshuahensley9395]

I love that this entire video is about getting a clock that counts by the minute, running fast enough to update in less then a minute.

[@Grantonioful]

I literally got goosebumps when there were 5 seconds to spare for even the slowest one. CONGRATS! Sell it or donate it to a museum cause this is so fun

[@sybilpops9070]

I've really enjoyed watching this series, thank you! It's been really interesting to see the whole project go through several iterations and see the whole process - and congratulations for getting it working!!

[@ricardoseoane1980]

Ivan, you madman, you did it! It may not have been the most elegant way of doing it, but you stuck to your plan, and bruteforced it into success. I can respect that.

[@GregN456]

I really enjoyed this series. You have a ton of perseverance! If you ever decide to have another try at this, what if you had individual white and black marble reservoirs sitting up at the top and do the marble sorting at the bottom before lifting them up? It may not make any difference but I think it has potential.

[@gamemeh]

Simply outstanding. Please.. don't break this thing apart... this deserves to be on display somewhere.

[@shizzling1286]

I've been watching this series 5 hours after the first video was released. I can't believe it's already done.

[@ragtop63]

Great device! Next version should eliminate the 50/50 chance of each pixel being the correct color and employ a mechanism that injects the correct color every time. Guaranteed that will make it update in a matter of 20 seconds or less.

[@AlyssaNguyen]

What I was thinking of was sending alternating rows up the elevator.

[@BrianDominy]

It's not exactly a 50/50 chance - the ratio of black/white marbles is fixed, but not all digits have the same ratio. If you stylized the 1 and 7 to include more white, the ratio wouldn't vary as much.

[@ragtop63]

@@BrianDominy The exact odds aren’t the issue. The issue is the wait time for a white or black marble to make it to any given row. It’s essentially random right now with only 2 options (50/50). Creating a mechanism that directly injects the required color for each pixel removes that unnecessarily unpredictable nature of the current configuration.

[@sadafule9008]

This is so sick and the enthusiasm overflows more than the marbles!

[@anToha_UA]

I am so happy to see it working as planned. Congratulations!

[@ColinLittley]

Last issue may be a vibration injection on the top ramp just as the release happens to encourage the stuck marbles to start moving when released. Great work looks awesome

[@juusoleinonen]

Thats is insanely cool! I would see this at an air port and just watch the time pass

[@francisdeighan6040]

I saw the first one when it came out, and I've been watching as you went along, this thing is awesome!!! Thank you so much for putting the time in to make it work!!

[@Lastered]

it's truly a treat whenever you upload

[@SwitchingPower]

another option is to filter the black and white marbles after they are used, then you can have 2 feel trays with all the same color and use solenoids to allow the needed marbles to enter the elevator at the right spot and time

[@SrFrozen]

Hey Ivan, I calculated the average number of white marbles your clock uses per display, considering the chance of a certain digit being a certain number and I came to the number of about 76 white marbles per display. That would mean you need 19% of your marbles to be white in order to maximize the chance of the correct marbles being chosen by the elevator.

[@TheManChild]

Crazy how much time you’ve spent on this project just for fun because you had an idea to reinvent the wheel. Or clock. This kind of stuff is what drives innovation. 👍 great job.

[@pookie1307]

Amazing project so far I must admit at first I didn't know how you where gonna pull this off but this truly is a great achievement!

[@Chimpster21]

Great idea, if slightly crazy. If the mechanism that rejects the wrong colour marble was lower you could save a lot of time.

[@notahotshot]

And reduce the nose level as well.

[@1234fishnet]

I thing the steel thingy that pushes out the marbles will wear down the plastic quite fast

[@gyuhooo]

I love it. I've never seen a watch that I can't wait to see one minute later like this!

[@guillesl2304]

Simplemente impresionantes, como estudiante de ingeniería, y tras haber trabajado tambien con el mismo compilador, es asombroso tanto el nivel de programación como el de elaboración y diseño!! Simplemente alucinante

[@kilrahbi]

this is such a crazy cool concept! def needs to be displayed publicly. so curious as to how you came up with the idea for this though 😂

[@danielcbraga]

Have you watched his previous videos? There are some insights there

[@kilrahbi]

@@danielcbraga no but I will soon!

[@thecarbonprop]

I love seeing these projects come together. What a beautiful piece of engineering this clock is. Ivan I wonder if you’d consider doing a video maybe talking about you life outside of KZbin videos? Like what your career is/was before KZbin. Maybe some of the other things in your life that you enjoy besides engineering. Love your stuff!😊

[@IBAndreas]

I have an idea for increasing the speed. In the last example when the time incremented from 07:14 to 07:15, only ONE of the digits actually changed. In these cases, you could send the three static digits into a special “high speed lane” which would fast-track them directly to the top again. That way you only flush the digits you actually need to change. If I were you I probably wouldn’t do it though since there is a really beautiful elegance in your design that appeals to me.

[@TheCorkins]

I dont think that would work quite the way you are thinking as the lanes run horizontally and not vertically. you cant "save" the marbles used for the 0, 7, or the 1 because its not designed to keep those marbles together. it would be like writing , but instead of finishing a letter before moving on to the next you wrote out the top of each letter, then moved on to the 2nd row and so on before finally finishing the bottom row of the letters.

[@IBAndreas]

@@TheCorkins actually the tracks “run” neither vertically nor horizontally - they are purposefully sloped to keep the balls moving. What I am suggesting is adding some sloping tracks / gates which would “preserve” the state of the non-changing integers so that they can be directly fed into the next elevator. I agree that my explanation was over simplified. Nonetheless, I’m sure it’s possible to increase the efficiency of the system by preserving the state of the static digits! Peace!✌️

[@manuelipenza]

No se como es que KZbin me recomendó este video. Pero estoy impresionadisimo con el nivel de creatividad que tienes. Sigue así! Me encantó ver todas las partes del proceso y aprendí muchísimo! Nuevo suscriptor!

[@Lellba47]

If you add some felt or soft materials on the spiral and some weights on the bottom of the device you can reduce the rolling and vibrating noises by a lot! This is looking really cool! Congrats!

[@SuicideNeil]

What a journey- well done dude. Now, do one that tells the time with vertical channels instead of horizontal ( just make the digits smaller so it requires fewer marbles and channels ) :D

[@funkyjohnhuie]

CONGRATULATIONS! thanks for bringing us along on your journey.

[@fukit420]

NO.... this series CAN NOT be over. this project is so beautiful. i think you were joking about the seconds, but i really hope you werent joking about the seconds!

[@DragonsAndDragons777]

If you make the clock part vertical it would be even faster

[@TerminusVox]

You marvelous madman! Now it's ready to be a bedside clock. Is there an illuminated version with an alarm?

[@AlyssaNguyen]

marble-ous

[@r3n470f3rn4nd35]

It´s amazing how a video series like this makes us think and have ideas to improve even more the project. That´s what science should be about: collaboration I thought 2 main things (that probably was already detected): - the angle of the buffer should be slightly higher to get rid of the marbles stuck persisting problem (a huge change on the original desing, so probably not a good solution at this point) - I would create a white marble buffer for each column, to increase the time assemble speed even more, because I can see that what takes time is exactely counting on the randomness of getting whites fast enought at the proper columns.

[@user-tc4cz6ky1m]

Great to see it get to this point.

[@marksinnott4951]

Hello and what a great project! I wonder how difficult it would be to sort the black and white marbles before the conveyor. Seems to me this would really speed things up. (I see now this has already been suggested)😊 Anyway I have really enjoyed these videos!

[@AlyssaNguyen]

I think it's possible to come up with something that not only separates the marbles by color but also does alternating black and white rows up the elevator.

[@ChessVisionAIBot9000-xg6ru]

Either that or dedicated black lanes on the edges. You only need enough marbles for two and a bit times in order to take out two separate solenoids and fancy stuff.

[@user-ko5jk2il8w]

I think this would be very easy if there was a sort of "pre-inserter" before the main pool of marbles. You could fill up specific rows with white marbles, and then the empty rows and spaces will naturally get filled in with the current hopper system. You could insert an (almost) all white row every 4 or 5 cycles to make sure there is a consistent supply. Sorting is also very easy, simply use slightly different sizes of marble and sort them like a coin sorter. I think it would make more sense to sort the output marbles by color, but keep the reject hopper random like it currently is. I think the black marbles are getting rejected more than white marbles, so this would be the easiest solution.

[@PugsLongs]

If you wanted to go faster you could make a color sorter that puts the marbles in the correct lift spots, saving the time of the marbles lifting.

[@lukafricker]

Yes, i also thought it would be much easier if the marbels would get sorted in black and white depots after they come down from the display and afterwards the new time is constructed directly with the right colors. This would spare the lift of a lot of work and would possibly reduce the noise to a tolerable level to have it around constantly if this is another design goal. Anyways it's cool to see the clock working as it is now.

[@matthewcasagrande231]

He is like the evil villain right before he turns evil and still full of hope and optimism

[@natewoods4566]

It has been really cool to follow this series, thanks for making it!

[@111111222223]

If you could manage to fill the belt with alternating lines of black and white marbles that could significantly speed it up as well.

[@PleakeCrions]

Absolutely incredible! So cool to see it working

[@jacenjustice]

Been watching this since the beginning, and its crazy how far we've gotten.

[@alphagenius13]

This is the coolest thing I've ever seen.

[@OldManBOMBIN]

Been following this from the beginning, and the most impressive thing so far was that ad read, and that's saying a lot.

[@althejazzman]

This marble machine project is so much more satisfying than the other famous marble machine project on KZbin. At least it's completed and it works!

[@stevepowell2000]

8:20 ... truth. he's a mad man! And I'm here for it. Love it

[@GateFather]

I have thoroughly enjoyed watching this thing get built, improved, rebuilt, and so on. Little sad to see it completed. Agree with a previous commenter that behind some sound proofing, this could 100% be an art piece in a museum.