Faster than the Eye Automatic Marble Sorter - Marble Clock Seconds Display Pt.3

[@obant]

You already know we want it incorporated.

[@ofens2001]

He wants it too, he was just looking for an excuse.

[@robertdascoli949]

He's already incorporated it.

[@EEEEEEEE]

E‎

[@vinny142]

​@@ofens2001 He was just asking so we'd "engage" for the algorythm.

[@darrennew8211]

I'm impressed that the power of lazy hasn't convinced you to put a chair next to your table.

[@mrdzin1209]

What's lazier, just stand there Vs. walk to find a chair, pick up the chair, carry the chair, walk back, put the chair down, adjust the chair, sit on the chair, stand up when needed?

[@Whatsinanameanyway13]

@@mrdzin1209 No, need to 3D print chair components, assemble the chair, have the chair fail multiple times and re-design, etc.

[@raafmaat]

@@mrdzin1209 true, but lazy people tend to work hard in short bursts, if it means they can then be lazy for the long haul! ;)

[@EEEEEEEE]

E‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎

[@shadowisp911]

🤣

[@azukar8]

Me, several times in the video: "Wow that's pretty fast, that seems like a good sorting rate, he'll probably stop th- oh."

[@EEEEEEEE]

E‎ ‎

[@althejazzman]

Me too. The final result was mesmerising.

[@dutchletters]

He needs at least 15 per second because each second is represented by a 3x5 image.

[@yashvardanvkr53]

i was happy at 3 marbles per sec

[@scribius5555]

Build two sorting machines. One for black and one for white. Then you need only... 10/sec

[@Hiddenpower]

Ivan, you need to make the chain in white so the solenoid doesn't fire on empty links!

[@fredward5728]

Or just switch to knocking the silver ones out instead of black.

[@safetysandals]

​​@@fredward5728That would work, but there are way more silver/white marbles than black marbles, so it's less work for the machine to sort the black ones out.

[@sethpolevoi4027]

Or rotate the angle of the color sensor. Instead of putting it above the balls, put it to the side

[@mastercko]

@@sethpolevoi4027 in that case he would have to make sure that there is a background or something to prevent it from firing on something being behind the slot that gets judged as "dark".

[@makers_lab]

@@fredward5728 Yes. Something I encourage is the approach of if something is proving tricky, change what it is that you're trying to do. It's a very useful design and problem solving technique.

[@hacksmith]

This is very satisfying, especially with the solenoid.

[@mantrachhaya6835]

Yo when is the liquid oxygen light saber coming out ??? I'm waiting for it to come out.....

[@Siro_Simo]

I design high speed industrial control for work and I'd like to share a few suggestions You could use pulse compressed air to push the marble out of the belt. It can operate way faster and will not create a physical barrier that collide with the belt. Also the trigger solenoid can be directly connected to the infrared sensor so you sort the marble while they are being evaluated (direct connection without logic will work fine). Last, using a ramp and gravity would also simplify your setup. The only electronic would be for detecting/sorting (speed may need to be tested though). I designed several high speed sorting systems (1000+ items/s) for all type of products, and using sensor directly coupled to the actuator (pneumatic) is always the most reliable and fastest solution.

[@mastercko]

yeah, 1) compressed air 2) on the downhill 3) with the sensor and "pusher" at the same spot seems like it would solve a lot of problems.

[@vosechu]

Thanks for sharing! I love hearing from pros!

[@SpaghettiEnterprises]

I think that he is using an IR proximity sensor, which looks like the ones designed for use with Arduino or other microcontrollers boards. As far as I know you can't directly connect a sensor like that to an actuator because the analog value representing the sensor's light level has to be evaluated somehow to trigger a control output. This is all just speculation though.

[@emiliaolfelt6370]

@@SpaghettiEnterprises and this, my friends is exactly when to bodge

[@Siro_Simo]

@@SpaghettiEnterprises You are most likely correct. Another type of sensor would need to be used. Something a bit more industrial where you can set a level of detection and provides a dry contact where you can use whatever voltage you want (typically 24V). They are a bit more pricey ($100 - $200), but they are reliable and no logic needed. Also the solenoid for the compressed air would need to be a pulse type, basically provide a pulse of air on a rising or falling edge. That way even if the detection last a long time, only a single pulse would be provided. Industrial world has all these use case covered ;) But it takes $$$. Hobby electronic is cheap but take time time time....

[@PhantomRaptor1]

I know this would be a _lot_ of work on top of what you've already done, but what about building a second sorting machine? You don't even have to change the design, just have a second sorter running in parallel with the first. That would allow you to reduce the speed (maybe to only 15 marbles / s?) and still be able to sort _really_ quickly, since you're effectively multiplying that rate by 2.

[@bluecraney1]

on that line he shoul build them off height so there isnt any colisions between lane 1 and 2 2 and 3 and so on

[@minecraftersunited8565]

I'd say stick with like 10 marbles/sec. It was around that point that the solenoid was actually consistent in getting every black marble out from the shiny ones. The machine only needs 17 marbles/sec, with two sorters at 10 marbles/sec that gives you 20 marbles/sec, more than enough.

[@GoffyOne]

Just add a second solenoid and alternate between them to hit the marbles.

[@associatedblacksheepandmisfits]

Why stop at two ? 😊

[@BrianUnderwoodCodes]

Yeah, I came down to the comments to say this. You may be relatively "reliable" at 15-18 marbles per second, but you're probably not as reliable as you think and being in the range of 10-15 will probably help you keep it from going wrong one out of every five hundred marbles (for example)

[@angelusmaker]

Someone is going to see this and say "Uhhh, this is dumb, it serves no purpose, make something practical!" and completely ignore that Ivan just demonstrated how an industrial sorting machine works. Some other kid is going to see it and think "Hey, so that's how they do it in a factory! Much simpler than I thought, I can do that!" and thus start on their path to becoming an engineer.

[@ch1pnd413]

❤ yep 👍🏻

[@thearkmecha4637]

@@ch1pnd413 Helo fellow trans human ^w^

[@dave4882]

Instead of a solenoid, can you use a blast of air? No chance of that damaging the chain.

[@111111222223]

That'd make the thing even louder 😂

[@zeprahzzz]

@@111111222223 i thought that was one of the design goals.

[@TheHellis]

Or perhaps a magnet. A well placed magnet that pulls the marble of the belt

[@Dolmeca_the_Knight]

@@TheHellis The belt (at least the last one) has metal parts, and a magnet could attract the other marbles when pulling the right one so I think using a magnet wouldn't be very reliable in this case.

[@bruceleenstra6181]

If you speed the chain up until the marbles are launched in an arc then they could easily be deflected at the start of the arc by an air blast or the solenoid.

[@AlexandruVoda]

Two ideas for possible improvements: 1. Eliminate switching on empty slots. Either switch on the light marbles instead of switching on both dark marbles and empty slots. Or make the chain white and continue switching on black. 2. Make use of paralelism. Use more solenoids to kick multiple marbles at once. With two solenoids you can do odd and even slots. Etc.

[@dragoncoder047]

Also another improvement: use a gear motor, not a stepper. They can go faster and with more torque for the same size, and since Ivan has a disc with slots and a photogate to time the solenoid and sensor, the step by step precision of a stepper motor is unnecessary.

[@Li-Nuss]

I think it would be less work to just put the color sensor on the side.

[@confuseatronica]

@@dragoncoder047Yeah, I'm noticing a lot of maker type people use nema stepper motors for EVERYTHING, even when they're not using them as stepper motors. Maybe it's a coder thing? They're just used to dealing with things that move in discrete steps.

[@AlexandruVoda]

@@Li-Nuss If you put it on the side you would need to build a wall in order to have a consistent background. Otherwise that sensor might have environment dependent behaviour.

[@tekbox7909]

Adding more solenoids doesn't really work cause the chain is moving constantly and the solenoid reached the limit and having the chain move any faster means it can't punch out and retract within the time a single link passes by meaning it'll interfere with other marbles. More solenoids won't change that. Instead you'd have to duplicate the entire mechanism

[@coreymartin9630]

This series really reminds me of Microsoft's insistence on not putting seconds on the taskbar clock, in Windows 11 they finally added it but the setting still has a warning about using more power. If one of the world's largest companies had that much trouble with seconds, it's no wonder that you are too. I absolutely love watching you press on through every problem you hit, it's super inspiring

[@omegahaxors9-11]

They must be seriously incompetent programmers if they can't make a performant 1s timer, even lua can do that no problem.

[@ETXAlienRobot201]

@@omegahaxors9-11 my favorite part, i do a search, find an article, and it includes "(minimally)" win 11's central theme sure seems to be "too little, too late" it's *astounding* how many things they "couldn't do" before 11... my new laptop is not running windows when/if i get a new desktop, it also won't be running windows maybe a VM for, hopefully just a few stubborn apps/games i'm *so* done with the BS you know what's really sad with windows is every good feature they haven''t axed, they've managed to ruin. task manager, for example, was specifically optimized to run under a possibly-might-crash dire scenario not anymore... it also can't shut-down "protected" processes. [and a handful just auto-restart] but yeh, i doubt displaying seconds causes a notice-able increase in power usage. i'd be actually impressed if they wrote their code *that* bad. i can't even think of a way to do so without adding insane levels of bloat. probably intentionally, at that... something akin to a certain infamous DRM. clocks/timers with seconds have been present on *games* on *old* systems. systems that probably couldn't run windows [or not without major compromises] M$ has a history of not being truthful, though, it's entirely possible some other reason exists for why they "couldn't" add seconds to the taskbar clock. "power use!" sounds like a cop-out, it's too absurd even for windows. i'd guess it's either something they somehow managed to back-burner for an outrageously long time, or they intentionally held it back to eventually make the old version "obsolete". all of win 11's QOL features seem to follow this trend. 10 isn't getting them, because that gives people a "reason" to switch to 11. "consumes more power", though... man, i want to see some real figures yesterday on that one!

[@JoshuaBost]

@@omegahaxors9-11 My understanding of the issue is that it's not about the power draw/CPU cycles from the clock per se, but rather optimisations elsewhere of anythign not doing anything right now going to sleep - and now it can only sleep up to 1s at a time instead of up to 60s at a time.

[@coreymartin9630]

@@omegahaxors9-11 The issue afaik is that the machine has to redraw the taskbar 60x as often when tracking seconds

[@sharp14x]

@@coreymartin9630 Which is absolutely not the case if you have a sane GUI system that doesn't repaint everything. It's a problem that Microsoft has invented for Windows 11.

[@shangerdanger]

amazed by how precise the organizer is. it looks like it's making a mess and then the bucket only has black marbles. amazinjg

[@billkeithchannel]

There were some black ones in the plain bucket but I think they were from bad bounces not a sorting error.

[@sigstackfault]

local man puts more marbles on the floor trying to make the marble sorter fast than the speed will ever save

[@uvbe]

lmao im sure he'll add a roof and some walls in the beginning part of the machine

[@JakobBerglund]

Time to build a machine to pick up the marbles from the floor.

[@bluecraney1]

@@JakobBerglund they make those the are called magnetic floor sweapers

[@zalibecquerel3463]

Well hello, handsome!

[@sigstackfault]

@@zalibecquerel3463 mine's custom :P

[@adamjb8165]

I love that rails are the last thing you ever add. Whether it is to your workbench or to your marble sorting machine. Marbles everywhere!

[@EEEEEEEE]

E‎ ‎

[@amarinetto2]

Absolutely!!!!!! it makes total sense to include marble sorting to make a clock that is totally self managed. Vamos Ivan!!!!!

[@KPalisar2009]

My STEM inspired daughter and I love your videos. The editing is brilliant and the way you show your failures and the thought process behind overcoming them is inspiring. Keep up the great work!

[@jankokert653]

Nice project! Two suggestions for easy improvement: (1) Activate the solenoid on silver marbles, not on black ones (and swap the buckets). I saw the solenoid activating also on empty chain positions (2) use a damper on the solenoid to get rid of the vibration/oscillation and thus reduce the interference with the chain.

[@TheHellis]

There is no need to swap places of two empty buckets

[@clementita54]

There are more silver marbles then black one, not really a good idea to switch the selection. An other option would be to print it white or to paint it

[@PJeBenn]

If he printed the chain out of white material, I think that would solve the problem as well.

[@NealFosterHD]

@@PJeBenn He only has black or that bright red

[@tekbox7909]

Don't really see the issue as the solanoid needs to be ready to fire again in the same time even if on average it has to fire more.

[@soviut303]

This is some "yak shaving" that I can really get behind. Next you should build a robot vacuum that can identify marbles and collect them from the floor for you.

[@thealmightysnoood]

Or, a machine that buys them off Amazon, removes the non-round/wrong-size marbles, dyes half of them black and then feeds them into the clock. Easy ;)

[@SpaghettiEnterprises]

Ooh strong roomba 🦾

[@beyondee]

i find it interesting that you traverse the marbles up instead of using gravity to queue the marbles, like placing a funnel on the top and letting the marbles flow through a funnel where the bottom would have gates that open for the correct collector bucket. that way you don't have to deal with chains and motors but only servos and possibly solenoids

[@Dolmeca_the_Knight]

I thought the same but I also thought that there is a big chance of the funnel getting clogged, and I think the gravity isn't "fast enough" to meet the required 15+ balls sorted per second.

[@jolish85]

The marbles will be down after they are pushed off the display area, and they need to slide down a ramp to push the number before. So it seems logical to sort them on the way up the ramp.

[@Slide100]

First line of the video: “Building a marble clock that shows seconds is not as easy as it seems” Trust me, *nobody* thought it was easy. That’s why we watch your videos, you try so we don’t have to. 🙂 Thanks for the awesome videos.

[@BLBlackDragon]

Bits to add: - A cover over the pickup chute, to keep them from bouncing out - Side walls along the track (not on the chain itself), to keep them from bouncing out - Covers on the receiving bins, to (you guessed it) keep them from bouncing out - Convert the plunger mount to metal, so the solenoid doesn't get twisted I think you can beat 20 mps with these tweaks, as they should solidify the marble path, and get all marbles through the sorter more smoothly. This looks pretty amazing. Can't wait to see it hooked up to the new display mechanism.

[@aveabutchers]

"I know where north is." I love it! Well, I love the engineering more but you being hilarious doesn't hurt. Awesome channel!

[@mcabualee4998]

Instead of a solenoide, you could try a controllable valve and a air compressor ro blow the marbles of the chain.

[@theavaliengineer]

Yes! this is what's used in industry. A solenoid valve would work pretty well.

[@dan-nutu]

So, replacing one solenoid with another solenoid and additional parts?

[@robertdascoli949]

@@dan-nutu It's solenoid's all the way down

[@theavaliengineer]

@@dan-nutu The key is it has to actuate over a much smaller distance, so it can be much faster.

[@TobiasLloyd]

And a jet of air won’t get hung up on the chain.

[@TomachoGajardo]

"once they break, they are... broken" That gets me

[@timderks5960]

I think the main issue is using something that goes forward AND backward to kick the black marbles off. It would be easier to have a gear shaped object over the conveyor. The marbles pass through the valley between two teeth when they're silver, and the gear rotates by 1 tooth if a black one needs to be kicked out. That way you only have to move in one direction, instead of moving to kick the marble out, and then retracting it. This gear could be tight to the track, so with teeth on either side pretty close, but that would probably result in acceleration issues, so you could also have a very wide gap between the teeth, so it has time to get up to speed. In theory, this solution would be able to kick off marbles at any speed, it would just require a larger gear wheel and larger motor.

[@LiteralGray]

Summoning Salt needs to get on this for the history of marble sorting speedrun world records if it hasn't started already.

[@gameingtermite411]

Why not use a wheel with arms on it. Like 8 of them. Whenever a black marble is detected, it rotates a little and follows through with the movement. No stopping and reversing.

[@OldManBOMBIN]

Bro, not only is this project still just really fun, but your ad reads are the most consistently watchable ad reads I've ever seen.

[@BaselineNL620]

YES! it should definitely be integrated! love it!

[@Reanimateryt]

Yes so happy a new episode is out

[@Shilorius]

Marbles on the floor!! *cries in Wintergatan-tears *

[@BassBoostedDuck]

YES SOMEONE GETS IT !!!

[@billkeithchannel]

Martin Molin: Let's go shopping and visit the cake decoration aisle...

[@fifty-plus]

The simple piece of rope around your table and your engineering iteration speed speaks volumes compared to another channel that has been picking up marbles from all over the floor for years.

[@niemlodafoka5441]

I have an idea. In your current design for your clock, you can make, that only marbles that need to change, are changing. This would add some work on your program, but would really help reduce speed you need to sort marbles, and 2 marbles wouldn't change anytime. Of course it wouldn't be that spectacular. You have to decide what is more important in this situation.

[@jolish85]

I like this idea. No need to replace the marbles if they are the right color.

[@Michael-xq7vj]

i haven't read any comments yet, so i have no idea if this was mentioned before: in the industry, they use air pressure to separate good from the bad (and the ugly, if necessary). it has the advantage of not having any moving parts (except the valve, of course), and therefore it is much faster and less prone to failure...

[@Brainstunt]

Another machine? You need to build a machine that sorts machines!

[@ETXAlienRobot201]

why stop there? a machine that shorts machines that sorts machines!

[@GetTheFOutOfMyWay]

I thought this was supposed to be about organizing ur marbles?? 😂

[@emiliaolfelt6370]

@@GetTheFOutOfMyWay Today, we organize the marbles. Tomorrow, we organize THE WORLD.

[@IntoTheCrowsNest]

the bucket needs a shroud or tunnel at the start of the chain to stop the marbles from shooting everywhere

[@TheMadManPlace]

some ducting and shroud work and it should work well. also, mount the whole machine in a tub that will catch all the marbles that "escape" and return them to the unsorted hopper for reprocessing. in fact, when the marbles get released from the display let them fall into the tub right away for resorting. and if you use ping-pong balls the noise will be much reduced. OK, the display will be BIG but that is ideal for a makers fair or a shopping mall.

[@HenryLoenwind]

Put a lever between the solenoid and the marbles to speed up the movement of the kicker. The solenoid has plenty of power but moves too slowly. However, industry standard for this is to deflect the objects while they are falling with an airstream. The advantage is that the objects don't need to move the whole distance at the deflection point. They get an impulse and will move over by themselves while falling afterwards. The other advantage of using the falling method is that you wouldn't need a belt, just a container with a marble-sized hole...

[@deadcxap755]

Then it is necessary to provide protection against blocking the hole by colliding balls. In industry, this is solved by vibration of the container, but in our case it can be too noisy.

[@POTThaesslich]

Great work, I was screaming “solenoid” from the beginning ;) I would also eject the silver marbles instead to save the black ones from the mechanical impact. (It would also release some stress from the solenoid as it currently seems to eject the “no marble” slots as well.)

[@emiliaolfelt6370]

I think you could definitely program out the dry firing, he probably just went with a quick and dirty "if sensor is less than setpoint, eject marble". There's almost definitely some difference between readings from the black marbles and the chain, you would just have to add another case to the decision logic. Or you could just print the chain in something that's not black. A can of white spray paint might also do the trick.

[@billkeithchannel]

He is using 8 solenoids in the actual clock. He should have just used that to sort them.

[@GooberLeague]

This is probably an overly complicated and impractical solution but you could make a wheel of electro permanent magnets above the conveyer belt that only turns the magnets on when it detects the right marble. Put it in phase with the chain links

[@mikeydk]

Have you tried using a H bridge to drive the solenoid? It both react faster than a relay, and the rate you need to trigger it would be easy for an H bridge, where a relay isn't really made for that.

[@mikeydk]

Another idea for pushing the marbles out, could be to use a servo motor for that too, and a wheel with openings in. That way it only needs to rotate enough to push the marble out, and move a new hole into position. Depending on the diameter of the wheel, it would only need to rotate a few degrees, and at the same same movement move the next slot in the disk into position, ready for the next marble. To make it very reliable, you could have a look at the stepper motors with driver and encoder build in too. I have been playing around with some, and they have been very reliable and VERY fast in acceleration and deacceleration, and even if it is missing some steps, it will correct it immediately.

[@Tifu77]

This is awesome. Maybe it is easier to add a second lane instead of make it faster. I think the noise is there anyway because the other part was also quite noisy ;) But i really love the project

[@djgummikuh8895]

Instead of the solenoid pushing the marbles directly, wouldn't a sprocket gear make more sense? So that you only have the time of the gear moving through the chain once, without it having to get BACK (like the solenoid has to) to clear the way of the chain Alternatively, you could just go for pressured air, which is how sorting is done in recycling centers. It would also save you the timing of any backstroke and would have the additional advantage of not having to break mass like you do with the solenoid prong, which also causes noise and uses energy (=heat)

[@emiliaolfelt6370]

I think the major downside of pressurized air is the noise from the compressor. They're very loud, and not the charming click-clack of marbles being shuttled around, but a deep growl. Compressors also tend to be bulky, making them hard to integrate into designs. Both of these issues make them decidedly not maker-faire/youtube video friendly. I'm also not sure Ivan is particularly familiar with compressed air, and incorporating into a large scale project like this may be more work than it's worth.

[@0000000000323120]

How about slightly smaller black marvels? You could maybe use a set of rails with the right spacing and all black marbles would fall through and all shine (bigger) ones would roll onwards. It would be a simple setup, easy to scale, fast, quiet and cheap.

[@sbragaglia]

The elephant in the room: 20 marbles per second but half of them end up unsorted on the table…

[@LogicalNiko]

Compressed air solenoids are the way to go for marble rejection. They are extremely responsive and have nothing that will physically get in the way of the belt moving at high speeds.

[@Diazjake]

You don’t need to change all 9 marbles every second, only swap out the ones that actually change colour?

[@emiliaolfelt6370]

I think the reason he's doing it that way is for the visual effect. Right now the matrix has a total of 3x5=15 cells, each needing to be updated every second, as well as a ten's place that needs to be updated once every ten seconds. The minutes/hours add ~.28 balls/second for a total of around 16.78 balls a second. If you only replace marbles that change between digits, you'll need on average 5 balls per update for the one's digit, and 5.1 for the ten's place. Accounting again for the minutes/hours, this leaves your total consumption at around 5.7 marbles/second. Definitely more manageable, but I believe his goal is overall impact, and more marbles means more noise, more movement, and more impressive mechanics.

[@DoNotPushHere]

Maybe it would be faster (or catch less) if the solenoid moved a one way revolving wheel that pushed away the black marbles. Sort of like a revolver. Even better, the revolver wheel would work with two solenoids, for alternating stages. One pushes while the other one retracts, while the wheel keeps moving with any of the two.

[@bogusF]

If the solenoid actuates a fork at the end of the chain, you don't have to worry about it getting caught in the chain. Also, you don't need a chain anymore, because you can just let gravity make the marbles fall through a funnel and put the fork at its exit.

[@MDBuy91]

1. Make the marble's impact point higher than the tooth of the chain (to prevent the chain from jamming and causing destruction). 2. Create a wheel resembling a fan (with 4 wings) that rotates 1/4 turn each time marbles need to be moved. This will ensure a smooth motion (unlike back and forth motion like a piston, saving half the time). 3. Design a larger cover for the inner bucket. 4. Form a funnel shape to catch the outgoing marbles.

[@norrinaguirre3395]

1:05 bro is losing his marbles 😭

[@KingOfRedPlays]

This would put on quite a show with two lifts rejecting opposite marbles and sending them down a slide or tube. It also means you can get away with not pushing everything so hard - if you need 17 per second, you can have both lifts run at 12 per second and be well beyond the required speed. You then also get the wonderful aesthetic of watching silver marbles go down one tube/slope and black marbles go down the other tube/slope, and two buckets at the top which also could maybe also have a tube/slope that terminates in its destination splitter at the back of the clock.

[@masonl87]

That thing at max speed is TERRIFYING!

[@ivanmirandawastaken]

Indeed, everything flies everywhere the machine slides on the table... mayhem.

[@mikeselectricstuff]

A problem with the servo and solenoid is they need to change direction and retract after pushing the marble. Maybe a motor with a 4-bladed propeller shape could work better, as it could keep going in the same direction after pushing. The speed of the motor could be scaled with the belt speed for a consistent movement relative to the belt. Compressed air would be another option

[@ssskdsh]

Never underestimate "the power of lazy" lol. 🙏🔥

[@ETXAlienRobot201]

nah, too lazy...

[@simona625]

Love the design, but I would double the chain and slow it down to a more reliable speed. Two chains, two solenoids, half to two thirds speed (for reliability) equals same number of, or more, marbles per second.

[@officialshockzon]

MORE MARBLES 😫

[@pauljones9150]

😈😈😈😈😈

[@paulbort6371]

Slow the sorter down to 15 marbles per second so it is more reliable, and run five of them in parallel as part of the clock.

[@maxdarkdog5051]

Hmmm... I would have taken another path... Like a perforated disk at the bottom of the tray taking one marble at a time and dropping it in a shoot with the solenoid to change paths

[@itisorisit]

Awesome video! One minor thing I was wondering was whether it would make sense to push the white ones to the side, so the solenoid wouldn't fire when there's empty spots

[@darianleyer5777]

3:23 That's basically what I had planned for one of my High School Engineering Projects, except A) I had to sort between *three* different kinds of marble initially (the first sort was determined using only the branch of physics known as mechanics), and B) the second sort would have used visible light, tested opacity, and controlled a tilt mechanism. Still, glad to see my advice taken!

[@darianleyer5777]

Modify the sorter to have less Marble Spray, and then add it - but at EXACTLY 19 marbles/second. After all, you still have to do the tens place.

[@MarkusSeidl]

You could dynamically adjust the speed of the chain. If there are multiple black balls in a row, go a little slower. If there are multiple "white" balls in a row: increase the speed slightly.

[@greg4367]

YES! Please integrate the sorter. I can't wait to see it in operation. Hope to see you in San Francisco

[@Darkassassin09]

Of course you should incorporate it! You'll need a steady supply of sorted marbles. :) Two suggestions: - switch from the solenoid rejecting black to rejecting while/silver. Then it won't trigger on empty slots. Less runtime + heat. - mirror the design to run two chains rejecting marbles into a central bucket. Then you can run it a little bit slower, have more room for error, and throw less marbles around at the loading end of the chain(s).

[@rods87175]

Always interesting to watch your videos. In industry, sometimes an air blast jet is used instead of a plunger to push items off of conveyors, the advantage being that nothing is protruding into the path of the conveyor cleats.

[@sgsax]

Good proof of concept, but it seems the hopper and point of feeding needs a bit of tweaking. Always fun to see you problem solve. Thanks for sharing!

[@feuby8480]

That's awesome. No lie. I have few suggestions : -Put some white/reflective color at the bottom of the belt. It will prevent the solenoid to be triggered when there is no ball. -You could put some walls for the balls starting preventing them to fall -You could also put some kind of dampener on the chain in order to limit ball movement. -I'm pretty sure you coul try to get 2 of theses sorters in order to ease the requiring speed, and make it less noiser, with less vibration. -That's awesome, no matter what. You're really good.

[@vizionthing]

Swap the chain for a wheel, have two or three rows of holes around the perimeter, wheel diameter of around 500mm, the wheel should be horizontal, you now have three times the speed with a single moving part that won't jam like a chain, expect much less noise, have 'trap doors' under the wheel holes that can be opened, these can be longer that the ball diameter to allow time for the ball to drop, it may help to have a light spring lever or brush to help the balls drop, the holes in the wheel would benefit being spaced around 2 to 3 ball diameter apart, have a set of guides on top of the wheel surface to keep balls one deep and feed balls to the three sets of hole lanes, take a look at the contraption Adam savage just unboxed, notably the way the balls move in the base - good luck Senpai

[@BrianRousseau]

Yeah, multi-component chains seem like such an obvious liability, especially given the material/tolerance limitations of 3D printing. I'd think that an archimedean screw or double-auger would be more robust and reliable. Might require a different way to reject marbles, but others have already suggested a blast of air, which seems ideal.

[@EmilieBlueBerry]

That’s beautiful machine! Definitely needs to be on the other marble clock.

[@strrrnr]

Hey! Usually those solenoids come with a rather soft spring built in. You could replace it with a sturdier one to make it retract a little faster and safer! ✨

[@AceKaci]

Tbh, I’ve been waiting for you to make this sorter since the very first video. If you can keep the marbles organized directly after clearing the clock, then you can fill the channels faster. Definitely gotta incorporate it.

[@chillaxter13]

Would like to see 2 of these running side by side, shooting into a shared central pipe for one color, and a shared dumping box at the end for the other. They could each run at a slightly lower speed that way, increasing reliability and also boosting noise... Perfect!

[@KolMan2000]

That sorting servo is playing the hardest and most random version of rhythm heaven

[@zebdavis2269]

Super cool and useful time saver for this project! I think it would be cool to see it integrated in the clock, however if you aren't able to get it running fast enough with reliability, you could also consider using 2 sorters to increase the marbles per second.

[@emiliaolfelt6370]

I think if you modified the chain a bit, you could avoid the solenoid crashing while still being able to carry the marbles, perhaps by trimming down the outer edges of the fins or reducing their height. Raising the solenoid up and moving it slightly farther away from the chain may also help, and mounting it on a spring loaded or flexible bracket could help with clearing and recovering from crashes. (who doesn't love 3D printed compliant mechanisms?) Another way you may be able to attain higher consistency would be to do something to tame the bouncing on retraction. A modification I would very much recommend if you choose to put this into the final clock would be a rubber tip for the solenoid, to avoid marring the finish on your black marbles.

[@ceerstar851]

Finally I've been waiting for this. I need closure. Damn, I thought I would get closure on the clock, which didn't even appear in this video.

[@Acaykath]

I would suggest altering the hopper to get them single file before they hit the chain and maybe add a top to prevent them from flying all over the place.

[@PaulStevensonPinball]

As someone who builds custom pinball machines, I was happy to see you swap out the servo for a solenoid, nice! Too bad you used a toy solenoid, instead of a proper pinball solenoid. That said, the solenoids spring appears too weak for the speed you are running it, and possibly the activation time seems too long, though a stiffer spring may fix that as well. It's all very cool, though, and I'm impressed you could push this setup to almost 1200/min. Further improvements could be had if you were able to lower the belt wall height and raise the solenoid off-center, so there's no possible collision between them.

[@dastoex]

Congratulations! With the sorter now working, you are .00000000000000001% of the way towards the ultimate goal of building an automatic Dyson sphere construction machine. I suggest going for the auto resource mining network next (plus a machine that can color the marbles for you). Keep it up!

[@mightygrom]

HECK YEAH, put that thing into the clock! great work!

[@easyBob100]

Thought: Use gravity instead of a breakable belt, and use the solenoid/servo to direct left or right, that way, you only need a small movement of a little pin to deflect a marble. Damn... now I want to build one lol. Great video :)

[@johnfrian]

Yes! Please make it part of the second-clock! What about a double conveyor at half speed? (less stress and maybe less noise) Maybe it's possible to find a conveyor geometry that allows pushing balls without hitting the conveyor belt?

[@tmriskl739]

Maybe switch between the black and silver marbles in the sorting That way the solenoid will not push empty spaces

[@DarkCloudXII]

Seeing this man spend the entire video on a marble sorting solution that's supposed to SAVE him time makes me thankful he's the one suffering through this for our sake. It also made me think, "Why not have multiple sets of marbles and leave one set to sort while using another set to test the clock?"

[@emiliaolfelt6370]

I think the reason he went for speed here was because he was hoping to include this machine in the final clock project, provided it was capable of keeping up with the ~17 marbles/sec the clock needs.

[@TheTetrapod]

Have you considered having multiple solenoids in-line with one another? You could program them to cycle through which solenoid is used, theoretically multiplying your throughput.

[@DavidTaylorM]

That's exactly what I came here to say 😅

[@nonfam4594]

You might be able to change the spring on the solenoid to make it close faster :) Awesome video as always!!

[@jayzo]

For the marble sorting mechanism I'd design the chain so that the marbles want to fall out but are stopped by a wall, and instead of the solenoid pushing them, it is a gate in a hole in the wall that stops them falling or moves away to let them fall. That way if it lags behind slightly when retracting it doesn't catch the front or back walls of the chain link. This would reverse the logic of the solenoid with it needing to be OFF to eject a marble rather than ON, however I suppose you could have it drive a gate piece that slides away when ON to allow a marble to roll off the chain. You might also be able to move back to using a servo this way too. You could also drive more than one chain in parallel to increase the throughput, you'd just need a hood under the chain to collect the marbles falling through the gate.

[@Nutch.]

What if you made a sorting machine kind of like how you have the original marble elevator, where it has multiple channels with a solenoid behind each one, sorting by color, but catch the marbles it pushes out instead of putting them back to the bottom? I imagine that would be much faster

[@chairmannemo]

A thought regarding whether you should add the sorter to the clock - at the moment you eject the marbles from the clock's display, you already know which ones are black and which are silver, it would be great to find a way to take advantage of that. Ejecting the black ones separately from the silver would add a bit of time required to to display the next number but it would save you the trouble of sorting the marbles afterwards.

[@AdamPNelson]

Couple suggestions to consider 1) a double acting solenoid will get rid of the bounce and be faster than a spring return 2) Look at a step feeder as you can get the same part feeding rate result with things moving much slower and no possability of jamming 3) Take a look at type of parts feeder called a drum feeder, which you might find inspiring.

[@Richard0915]

this is my favorite youtube saga in a long time

[@Superhirn]

Hey man, I just wanted to thank you for being so inspiring. I just love watching your videos. Continue what you are doing.

[@viewerpast924]

Yes please use this for your permanent build. It's perfect!

[@TheHighlyannoyed]

Couple potential ideas.... Elecromagnets to yank the marble out. Air to shove it. Create the belt with an opening in the middle/bottom, and poke it from below. Also, you likely already have a plan, but if you're making a full clock like this you're going to have six of these side by side, having gear hanging off the side is going to make that difficult, unless you make a bunch of longer fancy ramps to deliver marbles, but then you risk your stops at the end not working.. Your Marble Return ALSO has to manage at least 15mps as well (probably closer to 20 when you factor in the rest of the digits) Certainly an interesting technical challenge!

[@errantpursuits4249]

@Ivan Miranda Consider: You could add a cone facing back towards the feed reservoir to reflect excess marbles getting lifted from the basin back into the basin. Small quality of life improvement. ❤

[@MCsCreations]

Excellent work, Ivan! 😃 Stay safe there with your family! 🖖😊

[@t3hd0t]

Definitely need to incorporate :) I reckon you could get faster by optimising the solenoid motion by a) adjusting the timing so that the retraction of the pin is matched to the "springback" of the solanoid when it extends, and it can effectively speed bounce it out of the way (rather than oscillating a couple of times after extending), and b) adding some sort of dense foam sort of material behind the solanoid so it doesn't oscillate after retracting, but gets to a stable position sooner. Fixing b) in particular I think is the key to stopping it getting it out of sync with the chain and colliding like it does at 9:27.

[@ObserverChat]

when i'm using PLA or PETG gears to be mounted on a stepper shaft, i use a thin sleeve of PTFE tune for seperating the shaft from the printed gear. works like a charm.

[@deadcxap755]

Ideas on how to optimize the sorter: 1. Make 2 rows on a chain drive. The left row is for light ones, the right one is for dark ones. The solenoid above the left row throws dark balls to the side, and above the right row, vice versa. As a result, dark balls will fall on the left, light balls on the front-left, dark balls on the front-right, and light balls on the right. You just need something like a tunnel to direct the side balls into the corresponding basket, and you will get a 2x acceleration - that is, you can slow down the conveyor and it will become quieter. 2. Why a conveyor chain belt at all? Raising the balls up is extra work, you will still have to raise them later to the level of the number dialing mechanism. Let there be just something like a gear wheel inside the corner of the diamond-shaped container, where instead of teeth there are seats for balls, on the sides of it there are 2 containers for already sorted balls, and above it there are 2 solenoids for resetting to the left or right. Fewer moving plastic parts = less noise and greater reliability.

[@predue89]

Looking really fancy ;) Since the clock is already so loud, you can just add the sorting machine imo.