I feel there is a strong correlation between marble projects and madness. Based on a sample size of 2 youtubers I love...

Someone please buy this poor guy a clock.

Now do ms🎉❤

There are some smart corners that could be cut for this project. Firstly, there are 5 marbles that will always stay black or white (chrome) through all digits. The two black marbles in an 8 will always stay black, and three marbles in the right most column; the top, middle, and bottom, will always stay white. Secondly, not all marbles change from one digit to another, but many marbles stay either white or black. The numbers of changing marbles for each transition are: 0 → 1: 7 marbles 1 → 2: 8 marbels 2 → 3: 2 marbels 3 → 4: 4 marbels 4 → 5: 4 marbels 5 → 6: 1 marbels 6 → 7: 7 marbels 7 → 8: 6 marbels 8 → 9: 1 marbels 9 → 0: 2 marbels This has an average of 4.2 marbles per transition, instead of 15 marbles for a full update.

Man, when this clock is finished, it needs to be on display somewhere for public viewing; it is fantastic.

Grande. "I lack the knowledge to stay alive" is a great line...

8:07 "That's not a process, that's a lifestyle choice." I haven't laughed that hard in a long time. Bravo.

lol that coloring process... The level of sanity this man is willing to sacrifice at his viewers' behest!

how can we make the marble clock even louder? ah yes, electromagnetic solenoids and metal balls lol

I'm getting some Wintergatan PTSD. 🤣

I can't wait until he realizes he needs way more than one number.

"And that's why I shouldn't be trying to do thi- (cut) And that's why we need to find another way..." Your dedication to making this project work is ridiculous in all the right ways. I love it

One advantage of the design you discussed around 2:14 is that you could build it without making any sorting mechanism. Just keep the marbles in the same channel and loop them back up to the top. It would effectively work like a flip clock.

Imagine having this as an alarm clock 😂 you would never even fall asleep

Whenever I get fed up and bogged down in a project, I think of Ivan and his inspiring perseverance.

This is the level of procrastination that men achieve when they're supposed to be doing the dishes or taking out the trash.

I love the way you think "That's impossible to do" "That's impossible to do THIS way"

So, I've started painting minis again after many years, and I was feeling discouraged at how long the process can take. Then I found out how long it took Ivan to make black marbles. I'm no longer discouraged and am going to go back to painting, now. Thanks, Ivan, for your .. genius.. madness.. perseverance is probably the right word, honestly. I wish you luck with this, as I know soooo very little about engineering. I'll keep watching, though!

Just when I think my own projects are getting a bit difficult, I come here to be reminded that there are projects out there that are orders of magnitude higher than my own, in terms of insanity-making capacity. Genuinely floored by this project, and your stubborn determination to see it through.

It gets crazier and crazier with every episode. Give me more!

If you can control the lip at the end of the final ramp, you could have only marbles in the positions of the digits. Hopefully that idea helps prevent you from having to blacken any more of them!

You're a maniac, mate. Love it. Idea: since you are now using steel balls, could you use the electromagnets directly? In other words, dispense with the actuator rod and spring on the solenoids and just use the magnet directly as a stop/release mechanism for the balls. Magnet on - ball stops and is held in place; magnet off - ball is released. Cheers from Geoff in France.

16:19 wow this is so satisfying! it feels like a physics simulation, but this is even better because it's done in real life. It's truly beautiful!

And when I thought the marble clock project is done, this video pops up randomly in my feed. You earned my subscribe 😁

Mark Robber eat your heart out too. This guy actually builds it. Not a fancy team sponsored by us.

Engineering the rotary selector would have been pretty easy with solenoids. The neutral position is kept with an spring in both ways. now add a solanoid on the right and one on the left. On activation of one of the solenoids you push the selector to one side which selects the corresponding marble. Now use a spring to bring back the selector to the neutral position and release the marble. Greetings from a swiss mechanical engineer ☺️

I admire your persistence in designing and building these monstrous projects! Keep going, you inspire lots of people.

Dude, I’m half way through this and the insane understanding and skill to accomplish all this is top notch! Incredible!

I have a few suggestions: 1. Just use a single colour of the marbles, so you do not need to combine/seperate them. 2. Use several layers of the externally printed marble gates for presetting the next numbers to come up. Every second all gates are moved and the new number is displayed in the front. 3. Just exchange the digits that change from number to number. 4. Use some kind of a release (drop, poke out etc.) mechanism to remove the marbles on the display level 5. Think about which digits change from number to number, so you just have to remove/add a single marble (5 to 6 or 8 to 9) 6. If you keep using the glass marbles, you will not have any magnetic issues.

Well, that's the idea I just told in the Pt4 from the previous clock! You are way ahead of me! lol

If you pay close atteattention to, you can see him slowly losing his marbles as the project continues. Joking aside, this is absolutely incredible. I love watching these videos. I can’t wait to see how this project turns out. Hopefully you can maintain your sanity as you make all those black marbles.

Tenacity. OCD. Self flagellation. Borderline genius/insanity. This guy has them all. Kudos - fascinating, next level engineering, educational and entertaining. I would've just dropped the ball bearings in sharpie ink. :)

Amazing! A suggestion for coloring the marbles: Baked-on electrostatic powder coating. A suggestion for the magnet-sticking issue: Brass marbles, if they even exist. Also, an idea for a future project, now that you are dealing with multiplexing: A marble display using colored marbles.

Hi Ivan, very inspiring work, love it. I am wondering, since the tracks are already black and your steel marbles are shiny, why not just send marbles to the tracks where you need them to show the next number. Use the aluminum bracket you have, paint it black and use it to stop the marbles at the front. With a solenoid move it down to release these marbles when changing seconds and a second to arrange the next number before releasing them to the front. This way you wouldn't need any black marbles at all. You wouldn' have to change change much an the marble selector, just put plug in the track so no marble can be released in the black rows, so your work so far isn't wasted. In case you fear the environment being to bright, print a cover to be installed around the frontmost 10-15cm. Just thinking. Two thumbs up for your enthusiasm.

I love how 3d Ivan and human Ivan tilted when the machine tilted at 1:30

You can cut down the number of black marbles by not using the centre column 2nd and 4th raw. Non of the digits use them. Also, having 2-to-1 merging channel at the digit release step, takes away one solenoid valve movement (so 1 instead of 2 for each change) thus giving you more time to select the right marbles and queue them properly. One the other hand, why not have a separate selecting solenoid valce for the left-right selector? This way you could have a separate color marble queues. Great work! The front to back digit change is a great idea!

I'm sure a million people have said this already, but you'll never need white marbles in the middle of rows 2 and 4, so these channels could be left out completely and save yourself a bunch of black marbles. That said I love your dedication to the cause - great work. Looking forward to seeing more :)

Ive been trying to think of a simpler way to sort the black and white marbles other than optically as was done with the other marble clock. If you could use a slightly smaller marble for the black marbles, you could sort them mechanically via a sieve.

Peanut gallery observation: The "digit" is now 3 pixels wide by 5 pixels high. Ejecting the digit's marbles scrambles them, requiring re-sorting dark from light to reload the feeder channels. Consider forming each column of pixels as a "closed loop" of 10 x (5 + 1) marbles, sequenced much like any hobbyist (Arduino) LCD library of letters & numbers encoded with bits to be fed into a shift register... Feed each column of 5 + 1 pixels from the bottom thereby "pushing" the top marble into that column's collection chute to be reused in 10 seconds (or 60 seconds for the digit to its left.) (The "+ 1" is to allow a row of 3 dark pixels to separate the bottom of one digit from the top of the next.) With more digits, this would look just like an odometer whose display rotates "from the bottom to the top". When you realise each of the 6 digits of HH:MM:SS would only require 1 solenoid to elevate 3 separate stacks of 7 marbles, you don't need 6x8 solenoids, or the complexity of sorting coloured marbles. Further, tens-of-seconds (00,10,20,...50), tens-of-minutes (00,10,20,...50) and tens-of-hours (00,10,20) don't require as many marbles in their "closed loop" circuits...

I think I have an idea for feeding the marbles to the reserve channels. You'll still have the down ramp to a recovery pool, but a) lifting involves something that's basically an Archimedes screw to lift the marbles to the distribution point to three different chutes - one for black, one for white, one for return to pool (to account for the possibility of all of one color's channels being full at any moment). From there, just have the black and white runs set up to lock out full channels and round-robin the reserve channels with accommodations for channels with fewer marbles - and counts of marbles can be had with a few optical sensors. This will simplify the distribution of marbles, but the idea I have is going to create more than just a bit of complexity, and more than one sorting elevator to maybe help speed things along.

I have an idea for turning the bearings black. Run them through an abrasive tumbler to remove the chrome, then dip them into blueing fluid.

Thank you so much for making and sharing your journey with the marbles. Utterly fascinating 🙂

4:04 That is really satisfying to watch 😀

Haha, you presented the solution in the first few seconds. When the marble has to much impulse it clears the stage. No marble no dot in the matrix. Single color marbles fine. Just two main registers shifting all at once an one solenoid + guided wire for stopping individual marbles on the higher position for more impulse. :)

I LOVE watching these videos. Thank you for showing your thinking, fails, adjustments, and successes. I can't wait to see the next steps!

Couldn't understand a word but guess I didn't have too. Well done sir.

You put so much work into this designing each track multiple times and programming, not to mention coating all the marbles. I appreciate it as it inspires me and keep up the awesome work!

I'm thinking somewhere down the line you can set up your clock to have a buffer, so that the next second is always primed, and the second before that pushes the previous second into the primed position, since it looks like you'll be using all 15 spaces continuously. I do not envy the work you have to do, but it is very cool, I can't wait to see what you come up with next time :) Glad I found your channel!

Just some ideas: The first style of marble selector gate which is three position could be held the center position with spring force and a mild detent. A spring attached to the armature of a push-pull solenoid holds the armature in approximately the center position. A mild detent (such as a ball detent on the armature shat) can provide positive centering if necessary. The armature can be actuated as pull to move the gate one direction, and actuated as push to move the gate the other direction. Two regular pull solenoids can perform this push-pull task by arranging them back-to-back with the armatures connected. In this case the control circuitry is simplified by connecting the two coils together as a single center-tapped coil. The center tap is connected to the positive supply and each end of the coil is connected to ground through a switch (relay, MOSFET, etc.). The mechanism can then be actuated in 'pull' mode by activating one switch, or in 'push' mode by activating the other switch. A snubber diode is necessary to control coil inductive kickback when the switch is disconnected. The solenoids can be connected to the gate with a light-weight Bowden cable (the thin sort used in RC model aircraft) so that the solenoid size would not affect the geometry of the gate mechanism. This setup requires one push-pull solenoid (or two pull solenoids) and one or two control lines per channel (with a few external components per channel one tristate output can handle both push and pull switches). For a 3x5 matrix that is 15 push-pull solenoids (30 if using two pull solenoids as a push-pull) and either 15 or 30 control lines. If the matrix is simplified for 7-segment display then 11 push-pull (or 22 pull) solenoids and 11 or 22 control lines are required. If control lines from the microcontroller are scarce four 74HC595 8-bit cascading shift registers can be used to convert a serial bitstream into a 32 bit parallel output (some cheap AVRs with a simple program could also handle this task). That's a lot of solenoids. It would probably be cheaper, smaller, and require less electronics to use hobby servos. The servos can be controlled with an ATMega or a PCA9685 16 servo control board. As a bonus small hobby servos are compact enough that they could probably be mounted directly on the gate. The main advantage of this would be speed. All elements of the matrix can be actuated at the same time.

Every step is a little more amazing than the last. You are brilliant, sir. And your tendency to consider the advice in the comments just adds to the value of your videos! Thank you!

Very good sponsor integration! I viscerally hate advertisements, but this was good.

Just sharing an idea. Your design for the seconds is a differential gate. It looks like you're preparing to to queue up more than 1 second in the buffer. Relying on the kinetic energy to remove the current marble. If you want to minimize the work required, you can just map the difference between digits in a pre-calculated pattern. You know that it will always go from 0-1-2-3...9-0. You can use this information to instead of changing the whole number for the next whole number, to just process the difference. For example, In a 3x3 design, the middle marble only changes on 0-1, 6-7, 7-8, and 9-0. This should reduce the number of marbles required, and increase readability for the watch viewer. You can add additional gates to increase the buffer size if required.

Get exited to see his videos pop up. The best parts are the trial and error process and his explanations. Plus he is a great teacher

The amount of dedication to this clock. I like the process of trial and error.

Muy bueno, Iván! Deseando ver la continuación del proyecto...

I love Ivan's Engineering design and testing!

I love how you find all those great solutions! Nice video as always Ivan!

You'll get no criticism from me, I just love seeing your awesome designs and I appreciate all of the work you put into each build. I also appreciate that you don't sugar coat any of the obstacles that you run into. It is great to see you overcome them so you can "make something"

Hey I got an idea for the vertikal marbelclock. You don't need to discard the displaced marvels. Just use them again in that exact order. And build a loop. But your more chaotic system approach might be more fun :)

Art takes dedication. Keep burning metal balls. You got this....also we need a tune playing while all these marbles are playing ;)

You are bonkers, but in an awesome way. Love it. A couple thoughts; 1) I think vertical could still be fine, 1/3 of a second to see the time seems fine to me. At worst, they have to hold their glance for 2/3 of a second to see the digit. Most people are going to look for a couple seconds anyway, or even if they don't want to, within 2/3 of a second and a couple ms, they will pick up the image. If it's easier, might want to revisit that. 2) everything behind the digit is like 4 times the size of the digit. You're going to have to do something else to get them close, side by side, or if you dont want them close, may as well make the digit size bigger. Can't wait for the next one!

Thanks Mate from Sydney Australia loved your explanation and delivery! Keep it up! Cant wait for the rest of this series and your future challenging projects. Carl

If you intend this to work every second, I think you might want to add a second set of gates so you have a second buffer. You have the displayed digit. The next digit which is ready to swap. And a staging digit after that which is being loaded by the ramps. This will give the marbles an opportunity to dissipate their kinetic energy coming down the ramps, so that they’re not bouncing and ready to immediately slide through the gate into the buffer position.

Awesome project! The main thing that bugged me is the fact that you approached as if it works the same as the big clockface, where ALL marbles have to be refreshed on every update. Here, only some marbles have to be updated, and ALWAYS from white to black or vice versa. So technically you wouldn't even need a colour selector for each "pixel", as long as you make sure in the (to-be) sorting process that the marbles in the queue have alternating colours. Then you only have to select which "pixel" to update each second, possibly removing a lot of complexity from that system. As a bonus, fewer updates each second means fewer marbles to sort in a second as well :)

THIS ! is the wildest thing on this planet !

Genuinely an inspiring amount of effort goes into these projects, love watching these updates

I have an idea that can help the clock go faster. Prebuild all digits from 0 to 9 with maldes. Like that you can change the digit every second and you dont have to wait until the the next number is made

I do like 3D metal printing just as much as anyone else, but that aluminum gate could have been laser cut, which would also slide better because of the better surface finish. If you continue to have issues with magnetized balls, laser-cutting a steel "gate" between the machine and the solenoids should help and keep the machine more compact, which may become important if you want to add more digits. I can't wait to see the mechanics that sort the balls by color and feed them back into the machine. Happy printing!

You're an absolute chad for even trying, really cool project.

My poor guy, you're insane. I understand. And i watch everything, enjoying it.

You should ask around at some local metal working shops about the marbles. They could probably upscale what you're doing and get what you need done in a day.

Bro - awesome video, awesome machine, and another slick ad segue. Can't wait for the next installment.

Fantastic video. I love the creativity and imagination, combined with the engineering problem solving! Amazing.

The entertainment value you bring while guiding us through your thought and build process cannot be understated. Thank you for your efforts and creations!

This is INCREDIBLE! You're amazing!

I thought it was done BUT YOUR ADDING SECONDS! YOU ABSOLUTE MAD MAN! Mad respect

A couple things to improve efficiency: there are six slots that don't need switching at all; and since you're ejecting from the rear, ejecting the entire matrix isn't strictly necessary - just the delta.

Ciao! First af all, good job! Impressive work! Two considerations: -all marbles are replaced in under a second, and there is a big buffer of marbles, but in the end this buffer needs to be refilled in under a second to get the clock work indefinitely. Right? -each single pixel could che black or white so maybe you can "cheat" by using only two marbles per pixel, flipping them when needed. But maybe is not the point of the project

you could shorten the time it takes to blacken the steel marbles by doing it in batches: get a drum, drill two holes on either end, attach greaseless bearings(very important), the shaft should be a pipe witch is capped at one end with holes drilled in a straight line, insert the pipe through the drum with the holes pointing down and attach a propane line to the pipe, the drum will need holes for ventilation and a way to rotate it to both suply the fuel with oxygen and move the balls for even heating. you'll also want to add some sort of gate or window to allow extracting the balls. you could also mix the propane with air before injecting it into the rotary heater, but that has some safety implications, alternatively you could also just use a smelter/kiln to heat the balls, but that wouldn't be as even.

This is super cool! Great work!

Este tío es un crack, adoro sus vídeos, aquí te has ganado un follower español desde uk

Really awesome. First video was amazing and each video you kick it up a notch!

This is so cool! I’m wondering two things that might make it difficult to actually read the digits though. 1. The black vs silver marbles have low contrast 2. The marbles bounce back a little bit once they land in their spot

Just another idea in case this one doesn't get to the speed you require. Thinking about a "mechanical flip calendar"-like system, where flipping the front block immediately shows the one behind. Marbles would be filling from the top, and once they flip, released in the same way, from the top, now inverted. The numbers would be drawn as they are on the queue behind the shown one, with plenty of time. Theoretically you wouldn't need to release them and just show the proper one, but then it wouldn't be fun or a marble clock. In order to draw the numbers, what about one line at a time while in motion? Like an assembly line, moving to the next station one second at a time. I'm sure it could be somehow more compact, and given you'd be working in vertical, you wouldn't have tracks that delay the process.

Engineering and humour all in the one video!

Great idea. Love the video. You started off strong but as a mechanical engineer, I could safely say there are a lot of inefficiencies with this design. Looking forward to seeing how you solve them in the next video.

Towards the end of the video I noticed a lot of the actuators are firing in sequence, when the clock mechanism could operate more rapidly if the solenoid gates upstream didn't wait for the final gates to clear. Also, In the display that loads minute by minute, one could conceivably mechanically program each channel to spit out the black marbles into a black marble bin and the white marbles into a white marble bin, by using analogue memory. If the seconds run through a repeating cycle of 0-6 in the tens place and 0-9 in the ones place, you could mechanically design a cam that kicks out specific marbles in the cycle. So in a cycle of 0-5 using a 3X5 display, the top row would look like 11111, 00001, 11111, 11111, 10001, 11111, where 1 represents an "inked" marble and 0 represents a "blank" marble. So every cycle, the first round of marbles are all accepted into a receiving bucket, the second only rejects the fifth marble into the secondary receptacle, the third and fourth rounds accept all marbles, the fourth rejects the first and fifth marble, and the fifth round accepts all marbles. In this way, you can make it so that you do not have to manually pick out the black and white marbles every minute. If you use this technique, I suggest making each digit place in the display run on its own timing and cam so that you do not have to design a way for the cycle to include 60 unique rounds, and only need to design a 0-9 cam and a 0-5 cam for the minutes places as well as a 0-2 cam (unbalanced!!! 0 has ten rounds, 1 has ten rounds, and 2 has only four rounds) and a 0-9 cam for the hours. That's if you're using 24-hour timekeeping. By the way, as I can conceive of it, you'll have to design one cam for each row, or one for each column. If you then put a belt that raises all the marbles to a selector that uses a similar cam system, you can make this clock run indefinitely, assuming no black marbles go into the white marble bin and no white marbles go into the black marble bin. Most of this wouldn't be necessary if you continued in the vein of using solenoids, but I'm fond of analogue machinery, so that's where my thoughts were heading.

This man has grit, I appreciate the level of patience and determination it takes to not give up or rage quit.

Haha just have this do this 5 more times and figure out how to conveyer belt the marbles back up to the top again each time. Can’t wait to see how you do it! ❤

I appreciate you documenting your decent into absolute madness

This is sooo much work, I hope you find solution to reduce the time it takes you to maintain, fill and build everything !!

I love that the marble clock series is continuing like this! I'm curious to see how you will fit the digits close to one another when the unit is so wide...

I liked and subscribed just because of your commitment and dedication to excellence... and also because you're informative, entertaining, and a damn good engineer!

Very fun! Thanks!

wow the dedication you give this project is insane!

I have to like your video for doing a single job 11 time... dedication hatts off

love your craziness, keep up the good work.

Love the series, thanks!

Losing your marbles have never been better Illustrated.

I would have stuck with what I knew with addition of 'Keeping it simple', ie, black and white glass marbles. No need to 'Blacken' the steel balls, no magnetizing issues and thus a lot of time saved and a way less hassles... But.. Still... A VERY interesting project! 🤔😏👍 You just got a new subscriber! 😏 😎🇬🇧