I can't believe that didn't cross my mind at all! I'm going to reach out to a spring shop to see what it would take to have something like that made for this application. I agree it would be much nicer to not have to deal with 4 springs per catch basin. Thank you for the input!
@jackbaumgartel maybe have an option in the programming for stepper motors over springs. Me, I'd rather have a more expensive but accurate lifter system.
As someone eagerly following this with the hopes of using it for another tcg rather than MTG (Flesh and Blood) - it looks like there's a slightly higher card weight in FaB - (1.8-2g/card). In addition to wanting to use it for a different game, I'm also an idiot and suck at physics -- Is this something that will be ultimately negligible in terms of performance, or would different games literally require different springs (assuming the end design goes with springs?) Anyway, I love that you're working on this and either way I can't wait to see the finished product.
That’s a great question! To be honest, I’m not sure if a weight difference that small would be negligible or not. I’ll add that to my to-do list of testing. My guess is that the cards would simply sink a little lower than MTG cards. I bet it would still work, they just might have to fall a little further as the stacks build. And thank you, I appreciate it!
You could have a spiral gear in the wall of the catch basin attached to a platform, or a collapsible rod in the middle of the catch basin. Then have this interface with a motor on the main machine. As it drops a card in, it'll spin up the motor to lower the platform by the amount needed (or once every 10 cards or so). Instead of 6 motors, you only have 1 new one on the machine proper. Plus it means that the parts to repair, replace, or expand the catch basins doesn't include specialty springs. Could potentially use printed parts if you can get the spiral/worm gear working well enough. I can't speak to other card sets. But I suspect that different games have different weights. At least when added together after 600 cards. So the springs might lock you into MTG unless you have quite the tolerance in height to compensate for how the beds might lower, or not. This does add a good bit more complexity. But I suspect it'll have more longevity than a spring that might get worn down after a few thousand cards constantly compressing them. As well as keeps the system more card-agnosric in case you're able (or willing... theres a lot) to expand into others in the the future.
Could the horizontal movement of the gantry itself do this? Spin the gear as it arrives to the correct catch basin? Would probably need a servo to control the interface so that you only spin the one you arrive at, not all the basins in the line. Unless this could be done with the release servo after each card is dropped. Drop card, hold servo below drop ramp / platform to be low enough to hit lowering gear, move gantry sideways, then raise servo to ready for next card.
@@WNighteyesW potentially could use the motor on the gantry to turn this as well. Since it's not in use on the gantry to move the catching basins as you're putting cards in it. This would require the gantry motor either gets relocated or a kinda obnoxious system to port it's power over to the catcher you're currently focusing on. Either way a way to switch what the power is going to would be needed as well. Or at least a method to prevent the gantry from moving while you're engaging the card stack. It would be more complex, but hey one extra motor was a huge negative with my idea so I'm all for it. Solid idea.
So I realized a potential annoyance with this system over springs. You'd have to reset every catching basin after you're done. So a program you can run where it'll just go around and set every card stack to near-top. Minor inconvenience. However this system would likely not be very fast so it has more force to prevent the card stack from forcing it downwards. Depending on the speed, multiplied by 6 stacks. It might take a good moment to reset lol.
I like this idea, I hadn’t thought about using a single gantry-mounted motor to adjust stack heights! Brilliant work, I’ll have to model something to see how it would look and if there’s enough space to do so, but I do think it’s a potentially nice solution
@@jackbaumgartel Happy to give more options to explore! Hope you can find a solution that fits the needs and goals you're going for. This or springs or whatever else people have come up with in this comment section lol
I think you might have issues with binding. The springs are going to have some variance to them, so it's very likely that sometimes the platform will essentially be lifted from one corner. Consider printing a full size card holder with no springs and attaching strings to the top of each of the spring interface lobes. Then do your best to get the platform to lock up by pulling on strings. Once you have the tolerance and platform thickness right, it should slide smoothly no matter how you lift it.
Thanks for the input! That's a great way to test the fitment and especially ensure that any variances in the springs won't cause issue like you said. I appreciate it
So rather than have moving parts on each catch bin, especially with temperamental springs and platforms prone to binding, I think it would be cheaper and more physically reliable to have the catch bins be static, with bottoms angled up at the back(towards the direction of entry so they don't fall out) and no rear wall, air flow slots on the front and sides, and then add a z axis to the catch chute(rotating up or down). You would have to track the stack height of each bin in software so that the chute would rotate its angle correctly, but this gives the best entry angle and it preserves the momentum of the card and doesn't force it to change direction until it is almost at a stop, and gives the best "falling" profile (narrow edge of card acts like a lever, or a closing lid)
That's a pretty neat idea! I hadn't thought much about giving the gantry "aim" and another degree of freedom to place cards more accurately. I appreciate the input!
Throwing my 2 cents in, albeit a little late. What about a pneumatic approach? It'd be easier to maintain consistency, and you could have the controller be centralized on the machine with hoses going to each collection area. Limit switches would guarantee proper zeroing. Other than that you could have the controller know how many cards have been sorted into each section and actuate the bed by the card width. This method also allows you to use software to easily change card widths for use with other games and troubleshooting. Since you only have 6 slots i think the tubes would be manageable, and from my research the controllers are relatively cheap (compared to the r&d on fabricating specialized springs, at least)
Perhaps line the bins with a different type of material? You could print the inner wall in TPE or something, while will exert a tiny amount of grip to the card?
Just found your channel and I liked what you are doing. You gained another follower/supporter :). I am excited for the final result. Do you have any updates also on the software you will be using in addition to the machine?
Springs feel like they'd be too fiddly to get right personally, given any friction or binding that may occur. An idea I do have though is to call into question some assumptions about simply dropping cards into chutes/bins though. In particular the fact that everything must be orthographically aligned with gravity. At least with the ones you've shown the cards lay parralel to thr ground and fall directly vertically. And as you mentioned that's unstable beyond a short distance. However my idea is for the bin to be diagonal such that it follows the path of the falling card more natually, perhaps with the card builng guided downwards by the far/top wall as it lands on the cards below. Additionally you can play with the angle that the bottommost surface is at. If you just rotate a rectangular bin then the lowest point would be at the back of the card, and with that the cumulative effect of cards falling into imperfect position would be them levelling out to be horizontal, which is inefficient. However if you rotate the bottom surface such that the lowest point is at the far end then as any imperfections add up the slope would get greater, which actually encourages future cards to more snugly fit into their perfect position and thus I believe perfects itself. In summary instead of having the bin have vertical walls and a horizontal floor, I'm suggesting it have the side profile of a parralelogram where the normal tangents of both the "walls" and "floor" are pointing away from the directions the cards come in from. It feels rather awkward to describe it in text without being able to draw a diagram for you, but I hope I painted a good enough picture. Also I'll most certainly acknowledge that this design will take quite a bit of testing to get just right. However I do believe that once dialed in this design will be much less sensitive to failures, cheaper, and easier to produce than any design using spings.
Interesting idea! I think that I can at least somewhat picture what you're describing, and I hope I'm visualizing it correctly haha. I think you're right about the benefits of matching the incoming angle of the card's trajectory and allowing them to come to rest on an angled surface. Do you think that difference alone would be enough for even the first few cards that fall to do so in a non-chaotic manner?
@@jackbaumgartelI cannot say with 100% certainty but I think it would likely be good enough at this scale. If you would like to perform an experiment before investing in designing my proposed bin might I suggest you record video of a number of cards being allowed to freely fall out of the chute. Then you can compare how tightly they follow the same path on their descent to get an idea of how chaotic it is. Additionally, this experiment would provide you with an approximate curve/angle to guide you when designing the bin if you decide to proceed.
I suspect very strongly that a mechanism like you are describing is going to jam up/be unreliable in practice. Its a clever mechanism to be sure, but i just keep imagining the little taps from the card falling farther and farther causing a multiplicative uneven effect on the springs, and eventually causing jams.
It's certainly a possibility, but I am hoping that loose enough tolerances between the parts and smooth sliding surfaces will help prevent that. Definitely something that will need to be tested though!
@@jackbaumgartel I would think you could use one stepper motor to continuously drive a screw, and then use the movement of the sorter to engage/disengage the screw to lower the platforms, so you could drive 6 platforms with 1 motor?
haha i’ve been procrastinating for a while during those 8 months, but yes progress is being made and i’m hoping to put out a video shortly! y’all have been extremely patient and i appreciate it!
The custom springs are probably going to be an issue, though they are cool and will absoutly work the price will probably cause issues if only doing small batches of them and the height may cause other issues with them like the springs possibly not fully going back to their proper size. Just a thought but would the reverse be a better idea? instead of compressing they streach? Another possible idea would be to use only one motor and some gears to spin the screw that controlls the height. You will have height changes while moving back and forth while sorting but as you should probably keep your cards and inch or two below the top anyway to prevent the cards blowing out or other issues that shouldn't actually be a problem and it gets rid of custom hardware you would need, though it does add some fancy movement code or at least good gears and motors (just let them almost free rotate but stiff enough to hold the cards)
i think you should just have the catch basins be curved a J shape, such that when the stack is tall, the card just falls onto the top, and when the stack is low, slides along the curve and then slides onto the pile when it hits it
wait i have a better solution! a magic card is 2.5 inch x 3.5 inch right - instead of making the catch basin these dimensions, you make it 2.5 inch x 1 inch and the load the cards in diagonally. after a card enters the top of the basin it will not be able to flip over entirely, and since the bottom stack is angled, it will slide onto it when it hits the bottom. then it can be arbitrarily tall you may be worried about damage to the card but i think like a falling elevator, the air column in the basin will slow the fall so it should be fine
shorter springs are going to be easier to manufacture, and replace for maintanance and testing. Given that, and that springs do not behave ideally at the extreames (in your case the start and end of the stacking) a few shorter springs in serires with a washer between will work much better. These shorter springs (maybe 2 or 3 in serires in the existing design channel) will behave much more consistently since the manufacturing tolorances will be lower (you need a "take the average" approach to manufactured goods unless you plan on some large percentage of rejects). I think 3 is probably the number to start with to minimize moving parts and open up design tolorances. The major downside will be making sure the washers don't flip or shift (hexagonal channel could fix that and still maintain enough contact on the springs). I love seeing the work progress and will want a prototype when they are ready!
If the card moves smoothly to the catch, a spring or any way of lifting the cards may not be needed at all. Similar way to a coin sorter works. You would just need something to prevent the cards from over shooting the catch.
Springs are cool, another possible solution rather than add motors is to have the card drop servo pull double duty. as the finger drops the card into a bin have it continue to rotate to advance a mechanism on the back of the current bin/stack. This could be a sort of ratchet/pall system that uses gravity to advance the stack by 1. If gravity isn't enough (especially true early in the sorting process) can have system use a heavier plate or have the plate in tension pulling downwards by a spring with its motion constrained by the ratcheting system. This setup would require a manual reset of the bins prior to sorting but not too big of a sacrifice.
For something like this, I would highly recommend using extension springs instead of compression springs, or variable force spiral torsion spring or vspring if you want to go with springs. Otherwise I would highly recommend a continuous rotation servo motor. It would make for more electronics, but at ~$10 per stack for servo, switch and maybe some other small stuff, it's comparible in price. (A 9g servo has 1.6kgcm of torque)
@@jackbaumgartel the vsprings are the same ones that are used in key card retractable lanyards, so are pretty low profile, and you can buy some casing mechanisms so everything is properly enclosed. For extension springs, you don't need a center post to stabilize the stack. And since there will always be a small horizontal component to the force it will self stabilize to a point in the middle somewhere
Hey there, interesting video as always :) I'm not in any sort of product designing or engineering type field, those these thoughts are more speculation- - Would there be a concern of the springs losing strength over time? If so, how precise would the spring coefficient have to be to function? - Perhaps a more elegant solution could be having the basins recieve, catch, and hold cards at an angle. The basin walls would still be vertical (just as the example one you have is) but a "wedge" at the bottom makes the cards sit at angle and the basin walls along the top and the bottom of the card appropriately closer The hopper has walls and is at an angle, so I feel like cards will be coming out of there decently consistently (possibly using some sort of wide slit/gap to ensure the entrance angle of the card is decently consistent). This feels hard to explain via text so I'll use airplane terms lol. For this, the "long ways" of the card would be in line with the roll axis If the card is going hopper to basin, both of which are about a card in width, there shouldn't be much trouble with Yaw. If the card falls off the hopper ramp at an angle, this will mean its going faster and falling "bottom edge of the card first." This would mean the card has less fall time and isn't falling flat, so there would be less general "tumbling." Notably, I think this should eliminate issue with rotation along the roll axis. As for the pitch axis, since you have the card falling "bottom edge" rather than "card back" first, this would be the biggest issue. Making the bottom of basin hold cards at an angle could guide cards that have have ended up at varying pitches. Flat falling cards would simply fall onto the pile. Cards that have ended up more vertical be guided end up settling face side up. Sorry if that an incomprehensible mess. Taking a step back, my main conceptual point is that the other machines keep their falling height shallow because they drop the cards cardback-first, making the cards more likely to tumble and tumble along random axis. Since you're using a ramp to guide cards into your basins, your cards will start off moving edge-side-first. Speed + aerodynamics would reduce fall time and randomness. Utilizing this, you could reduce the severity and randomness of tumble. An angled bottom of the basin could handle a wider tolerance of rotation along a singular axis.
Wow thanks for the long response! I appreciate the interest! - The spring I specified is able to compress fully without any degradation, so it should last a very long time. - And I like the idea of angling the cards within the stack, that could be helpful! You’re right about the nature of this design meaning that the cards will have more predictable paths in the air. Again thanks for the input, I appreciate it!
@@jackbaumgartel following this comment, if you lean the whole card holder back, towards the sorter and have the cards stack at an angle, they will fall down vertically then since the whole holder is leaning they should fall into place. one possible issue is the card gets stuck against either wall standing veritcally. another approach is a stack leaning forwards, away from the machine and just give the gantry the ability to move up and down (at the same angle as the forward lean of the stack), and have the back of the card stack (sorter side) open so cards can be inserted at any height, just keep track of the stack in software or visually or with a feeler every 5 cards inserted to double check, and you can just use the gantry that you already have moving to test the height only one servo or something to tilt down a momentary switch. obviously this requires adding extra complexity to the sorter but it means your deck holders are dead simple dumb pieces of plastic.
@@joshieeee20 Thanks for the input and ideas, a few others have similarly suggested angling the catch basins, so it's definitely and idea I'll be looking into more!
Has anyone suggested rubber bands yet? Rubber bands on each corner could suspend the platform and would remain linear (more or less) rather than springs which will definitely bind if they're 9 inches long. You can also use pull springs, but I think that rubber bands will be easier since you can get a bunch of uniform rubber bands that are relatively weak and just use multiple per corner. This would allow you to fine tune the design by adding/removing rubber bands until you get a good stiffness. Also keep in mind that the upward force doesn't necessarily need to exactly match the weight of the cards, you just needs to make sure that the card fall is a height that you're comfortable with. I would say that as long as a card is falling no more than 2 inches you should be good. Sorry if this has already been suggested, but I skimmed some of the top comments and didn't see suggestions for rubber bands.
Probably be a bit too complex but what about raising snd lowering the basins themselves. Basically have a motor on thr gantry that interfaces with some form of track on the end of the basins. You could then have a switch at the bottom of the basin that'll send a stop signal to the robot to allow the operator to empty the basin.
I think springs would work well, you would just need to take into account weight of the platform, springs nonlinear characteristics and friction. Some alternatives would be constant force springs with the ability to wind them tighter or looser to adjust tension, or a ratcheting system where when the dispenser moves over a collection bin it goes down one "click". Each one of these "clicks" would correspond to one car in size, or the machine can have coarser clicks with the dispenser triggering it every 5-10 cards. The ratchet system would only need one servo/solenoid and the rest can be 3D printed.
Thanks for the input! I had thought about an incremental ratcheting system, and I do like the idea. I couldn't figure out a compact way to implement it though. Would the strategy that you're describing still be pretty space efficient?
@@jackbaumgartel I don't think it will be as space efficient as a set of springs, but I think it could get close since you would be moving space taken up from the front to the rear. My thinking is something similar to a clicky pen, but instead of the mechanism shifting the tip in and out, it will shift it down in steps.
I think investigating plastic compliant mechanisms to replace the springs could be interesting. They might do the same job as springs, but also be cheaper to produce, and more customizable for this purpose. 🤔
@@jackbaumgartel the mechanism could be made of plastic or metal and tuned to bear weight of the cards the same way the spring does. However, compliant mechanisms can be designed to only move in certain dimensions and minimize defeclection in all other dimensions, so if properly designed it could be more reliable and less demanding on the design of the container.
One random idea I had, could you catch the card with an air cushion? Given the miniscule weight of the cards even a very small pc fan would be enough and they're very cheap.
I have tried using springs in similar circumstances before, they are doable but ornery. I am concerned that the spring constent is very low yet needs to be very precise both to the magic cards and each other. If the spring constant is too high the sorter will not go down enough each card, this can be compensated for by having buffer at the top but the more buffer the more risk of flipping. If the spring constant is too low each card will depress the platform too much this is not as large of a problem but will eventually leave too much of a gap, again leading to flipping. You seem to have a lot of confidence in these custom springs in terms of accuracy, precision, and durability. Maybe this is warented, I have not worked with custom springs but with how low the spring constant is, how large the range of elastic compression must be, how resistant to degredation, and how many environments it may need to be used in, having a spring that meets design specs sounds very hard. I wish you luck.
Thanks for the insight. I admittedly haven't worked with such precise & thin springs before, so your concerns are certainly valid. I appreciate the good luck!
What about a basin with a vent like attachment at the top so it controls the slide into the basin. Might be a risk of jamming maybe but if tweaked right could be effective?
Yeah! I like the idea of controlling how the card enters the stack a bit more, but I still wouldn’t want the card to plummet ~9inches or so. A vent is a good idea though
@@jackbaumgartel another option are short or slightly bigger than short brissels to guide the fall instead of a spring? Or using the dynamic of the falling card to fall exactly in place? Like if it falls on one end at an angle that the other end will fall and keep it in place. Would require a cushion as to not damage the card
@@shaper_i-o Oh that’s an interesting idea! You’re thinking bristles along the interior wall edges of the stack to slow the card’s descent? I imagine the card might have trouble staying level while falling through them, but maybe not?
@@jackbaumgartel it's a very unique take, my guess is it would muster a 40% success rate? But that's a guess, I edited my previous message and added another solution if you want to reread it
I wonder how many compressions the springs can take before needing replacement. Also need to account for the base that the cards will lay on preferably so the basin is somewhere between .1"-.5" deep to prevent first few cards from sliding off.
Great considerations! Yes I didn't do a great job of showing it in the video but the platform should be able to hold a few cards at it's "resting" height. As for the springs, the ones I specified are able to compress fully without deforming, meaning that they should last a very long time. Thanks for the input!
springs looks like a realy good solution. Reach it the goal of cheap price when they are always need to be customize create? Maybe you can find a seller in china which can help youat this topic.
I think a motor underneath each stack would be the most reasonable option. They could be basic little motors that lower each time a certain amount of cards are added to that stack. I.E. 25 cards drop in bay 3, then bay 3 motor drops .5 of an inch or whatever the calculation may be. They only need to go down with the weight of the cards, not up. Think about a force fed paintballgun hopper. They stop when a paintball is in place with sensors, instead of stopping with cards in place, just have the sensor at the very bottom when it is full the machine stops until it is reset.
It's a good solution! It would take some more hardware to get things moving properly and I suspect from prior experience that the cheap amazon servos likely wouldn't have enough power or precision, but it could be done. The added electrical power and wiring is considerable for 6 motors as well, but still I appreciate the input and will keep it in mind.
Look up conical compression springs. They nest into themselves. Have you considered using springs in tension instead of compression? Im envisioning a rubber band fastened to the base, up over a pulley and attached to the platform.
I'd definitely look at how a garage door works. Add two torsional springs on rods to the long edges of the card-chute and have the platform suspended from cables that wind up onto spools attached to the rod.
@@Fredward0216 Thanks for the input! I do think that even a conical spring would risk buckling at 9in, but it's an improvement! And the garage-door torsion style would be interesting! A few others have suggested pulley systems as well on here. I would be concerned with the additional space & complexity required, as well as the additional hardware. Do you think the garage-door style could be accomplished in a small form factor?
a rectangular spring would probably be better , and springs from music wire are pretty easy to make, see if there is a small machine shop or even a hobbiest with a lathe they could probably make them
Since the ramp is at the pile that's getting more cards, and you have a servo there now... Could you have some sort of ratchet/belt system which can lower the tray in the column when the servo arm goes down next to the pile? Then you don't need springs or extra moters...?
That's a really neat idea, it would be awesome to use the servo already there as you describe. Figuring out a proper mechanical system to accomplish that will be a task for sure, but is definitely promising. I appreciate the input!
If you expect the stacks to be about equal height, you could have all of them on one side of a sea saw with the input stack on the other. This way every time you take away a card from the input stack all of the sorted stacks will lower by 1/6th of the card height. Then once the card i dropped by another 1/6th. Moving the pivot point of the sea saw you could adjust the fractions to be a desired target. By doing some clever lever mechanics that lock and unlock whenever the gantry selects the stack you could have a perfect system that lowers the stack by the exact size of each card no matter it if weighs more or less due to foiling, warping or lther factors. Though this design would be very complex and the mechanism needs to be developed thoroughly.
Also worth noting, this is a design decision that might constrain the build to just MTG, since card thicknesses and weights vary from game to game. Unless this is not an issue if there is a tendency for other cards to have their weight and thickness to have the same ratio? Worth looking into the weights and thicknesses of Pokémon, One Piece, Flesh and Blood, Digimon, and Vanguard. Those seem to be the games I've seen played at my card shops that have the same 2.5x3.5 inch size
Yes thanks for pointing that out! A few others have noted that too and it's definitely something I'll be testing out. I'm hoping that the weight difference will be trivial but that remains to be seen!
I'm no engineer, but what about a spring in a similar shape to the container (e.g. rectangular instead of circular), so that it would be the same size as the platform and possibly not require four of them per container but still hold the same weight needed?
I would try having them manufactured as rectangular but have them nest like conical compression springs. I'd imagine you could get some thin stainless plates cut with a water jet then stretched and tempered , but I'm not a spring expert.
What about a slide and the card holders can have about 4 inches of space to catch the cards. And instead of landing flat the bottom is slanted on a 22 degree angle. And a push out hole to take out the cards. Add holes on sides to let air flow out to prevent pockets to flip the card
@@jackbaumgartel I would say 4 inches because anything higher would likely flip but with enough air flow and at a slant it may be possible to go 9 inches
I love the idea! A chain or similar heavy rope would be able to match the increasing card weight just like a spring. I do think that the addition of pulleys & "chains" for each stack would be a mess. Any thoughts on how it could be implemented in a space-efficient way?
maybe you could use two pulleys and ropes on either side of the short part of the card stacks to avoid racking and save space in the width? if you then also use a wide ribbon for a rope you could have increased stability. the ribbon could ride over a rotating pin on the top and bottom of the card stack. The whole catcher assembly would get essentially twice as heavy this way though.
What about instead of just springs, a combination of springs/pulleys and weights that would instead allow you to use more commodity parts and not need to rely on expensive custom parts?
Thanks for the input! I like the idea of using pulleys to avoid any specialty/custom parts. I think it might be difficult to make use of such a system while still maintaining a small form factor though. Do you have any ideas on how that could be done?
Since springs are designed for magic card weights, wouldn't this limit the potential use for the sorter being used on other card games? Pokemon and yugioh cards have different weights on their cards due to their rarity as well.
Another way instead of screws is something a youtube channel called "Stuff made here" in a video called "A simple human task that's insanely hard for a robot". The robot is a puzzle solver and he had an issue for storing those pieces. At around the 11:45 mark he shows what he did. You could have one motor that is fixed to the base and have your card holders move side to side from it. here's the link! ruclips.net/video/Gu_1S77XkiM/видео.html
Thanks for the input and the link! I'm a big fan of Stuff made here, and the multiplexer option is a great one that I hadn't considered! A few others have suggested similar ideas so I'll definitely have to see how that would fit in!
I had thought about that but didn't see any real benefit, and it would take up more space. Would you see a benefit to a tension spring over a compression style?
@@jackbaumgartel I just thought of it when you shared the 4 spring design. Might be better reliability (less likely to hang up maybe). Would be able to utilize the entire receiving tray capacity with a tension over a compression spring as well. However not an expert, so I would use my judgement sparingly
Great project Jack, Bravo. I am eager to see additional progress! Are you interested in a sponsor of sorts? I could financially help grease the gears sort to speak...I don't want to speak out of turn, but I am in the midst of sorting my own 20k collection and this would be a lot of help. Let me know...
Hello and thank you! I'm glad to hear you're enjoying the project and I really appreciate the offer! Right now, the best way to support the build is by using the link in the video description. You can make a general donation or help fund specific parts. And 20k is a lot of cards! hopefully I can get you something to help :)
Cool. I know I talked to you in another video. I have my python code identifying most normal cards in under 1 second but the code is not optimized yet. The game has been out for 30 years. There's oddball stuff I am working on now. I have figured out how to detect foils. I cannot build my machine right now as my chop saw is dead :(
I am thinking 6 bins doesn't seem like enough. How do you plan to sort the cards? I would think 8 is a minimum needed for most common sorts for MTG. 5 bins for each color, 1 for colorless, 1 for multicolored, and 1 for unidentified/junk cards (like those rules reminder cards). I suppose you could sort by color, everything else in the unidentified bin. Then sort that bin by colorless and multicolored. If you could add a couple more bins though you could be a little more efficient with sorting time. I don't think short output stacks is a bad thing. Think about this. Let's say your input bin holds 2 boxes of MTG cards. If you sort by color on average there will be equal amounts of each color. There will be less colorless. Multicolors can be all over the board though depending on set. I think your output bins don't really need to me more than half the height of your input bin. Also, what do you think about pushing unidentified cards backwards or in some dedicated direction? That bin could be tall then and not match your regular output bins.
@@SirPoonga Awesome to hear from you again and impressive that your code is so quick! I talked about different sorting strategies a while back and what I learned is that everyone wants to sort their cards differently, so the ideal number of bins really is personal preference. This machine will have as many as it can fit without being too large that it's no longer a desktop device.
@@jackbaumgartel Right now I am doing a perpetual hash search to narrow down the card match. Look up ball tree image search, combined with phash, can narrow down the search quicker. From there I take the top matches and do a ssim comparison. I think mse works also. These are more traditional computer vision techniques. However, with a robot the position of the card in 2d/3d space will be near identical each time which allows for some shortcuts. You could go Tenserflow and neural networking but I think that is better for a phone app. For source/training material I have taken the default cards list from scryfall's bulk data api. I filtered out unplayable cards like the rules reminders you find in boosters, oversized cards, artwork cards, etc. I then download each card image, check if it has the identification text in the lower left. If so throw the image away and mark the card in my db that I can find it by the id text. Otherwise, keep the image and phash it. There's some more details to it than that as over the 30 years of MTG they have done some strange stuff that I have to account for. So when searching I first check if there is the id text in the lower left corner. if so I identify it by that. This requires some OCR cleaning of the data as it may think an I is a 1 or a 1 is a 7 because it thinks the flag is long enough. If the card does not have id text I am then doing the phash/ssim comparison. Right now I am finding the patterns of how the OCR fails. I am randomly picking a card newer than 2013 and not in brother's war. Grabbing the image. Giving it a slight contrast and noise filter so it isn't a perfect scan. Then run it through the identification and see what the OCR comes back with. Anything that isn't a match for which card was picked I note in a log. Once I get enough data I will start looking for patterns of how the OCR fails. Note, if you also do the OCR thing they changed the format of the id text with MOM and newer. ChatGPT helped me out with making a regex that can identify the parts of the id text in either format.
„Euler‘s buckling cases“ should be something you understand thoroughly if you plan to go with a spring based design. I‘d advise against it though. If the part isnt readily available and has to be custom manufactored then you‘ve lost one of the original goals for the project: for people to built it (easily) at home. Unless you wanna become a spring distributor and have ppl but them off you.
Thanks for the input, and yeah that's a fair concern. Avoiding custom parts is a major design goal, but so is cost & complexity! I've got some great feedback here so this is something I'll be working on more
Why use springs at all? You already have an actuator poking down each well every time a card is added: why not have the base of the wells start near the top and slide/ratchet down? If you do actually need them to be automatically reset, then suction the bases back up. If the depth is an issue, add a spring underneath with a clasp that releases the ratchet from up top. It's not strictly passive but it adds 0 active components and doesn't rely on specific and delicate spring force
This machine doesn't use any suction or an actuator to drop the cards! Sorry for the confusion, I showed a few examples of other machines which used such systems, but this design does not. A few videos back on my channel I did an overview of the mechanical design decisions and whatnot if you're interested!
@@jackbaumgartel aah! I think the z axis was just a huge assumption and the suction was stuck in my head from one of the other examples! I was recommended the video without having seen the design of this one. I'll definitely go watch Sorry for the confused suggestions ;p
OK, but what is the point of these machines? Simply curious. Clearly there's a niche here for machines and MTG cards, but what do they do? Is it for like building a randomized deck, or sorting cards by color, or what?
I also sure if the software is open sourced after the machine is created there could very well be someone that could run with it to create random decks.
Great question! The main use case for these machines is card shops that will often buy and sell cards by the thousands. It's expensive and tedious to pay employees to sort and organize the 100,000 cards you just bought and now need to sell. Some serious collectors and hobbyists with large collections might want one too just to help organize their collections.
Yes, definitely could! But because the stacks will fill up at different rates, I’d need a separate motor for each one, which would be pretty expensive compared to the spring strategy
![VICTOR Denmark Open 2024 | An Se Young (KOR) [1] vs. Gregoria Mariska Tunjung (INA) [5] | SF](http://i.ytimg.com/vi/Y3QEyCih85s/mqdefault.jpg)
You might find a single fitted rectangular spring avoids your buckling issue. As a user, I'd much rather have one spring over four.
I can't believe that didn't cross my mind at all! I'm going to reach out to a spring shop to see what it would take to have something like that made for this application. I agree it would be much nicer to not have to deal with 4 springs per catch basin. Thank you for the input!
@jackbaumgartel maybe have an option in the programming for stepper motors over springs. Me, I'd rather have a more expensive but accurate lifter system.
Darn, I wanna see the final result lol. Well Im subbed. I hope you figure out the kinks and it works well without damaging the cards ^
Thank you! I should be putting out a new video shortly!
Springs are cheap and cool so I am with you on that decision. Just be aware of some possible friction that might occur if you build it to tight.
Glad to hear! And yes I will be sure to keep the tolerances a bit loose so the parts slide easy!
As someone eagerly following this with the hopes of using it for another tcg rather than MTG (Flesh and Blood) - it looks like there's a slightly higher card weight in FaB - (1.8-2g/card).
In addition to wanting to use it for a different game, I'm also an idiot and suck at physics -- Is this something that will be ultimately negligible in terms of performance, or would different games literally require different springs (assuming the end design goes with springs?)
Anyway, I love that you're working on this and either way I can't wait to see the finished product.
That’s a great question! To be honest, I’m not sure if a weight difference that small would be negligible or not. I’ll add that to my to-do list of testing. My guess is that the cards would simply sink a little lower than MTG cards. I bet it would still work, they just might have to fall a little further as the stacks build.
And thank you, I appreciate it!
You could have a spiral gear in the wall of the catch basin attached to a platform, or a collapsible rod in the middle of the catch basin. Then have this interface with a motor on the main machine. As it drops a card in, it'll spin up the motor to lower the platform by the amount needed (or once every 10 cards or so). Instead of 6 motors, you only have 1 new one on the machine proper. Plus it means that the parts to repair, replace, or expand the catch basins doesn't include specialty springs. Could potentially use printed parts if you can get the spiral/worm gear working well enough. I can't speak to other card sets. But I suspect that different games have different weights. At least when added together after 600 cards. So the springs might lock you into MTG unless you have quite the tolerance in height to compensate for how the beds might lower, or not.
This does add a good bit more complexity. But I suspect it'll have more longevity than a spring that might get worn down after a few thousand cards constantly compressing them. As well as keeps the system more card-agnosric in case you're able (or willing... theres a lot) to expand into others in the the future.
Could the horizontal movement of the gantry itself do this? Spin the gear as it arrives to the correct catch basin? Would probably need a servo to control the interface so that you only spin the one you arrive at, not all the basins in the line. Unless this could be done with the release servo after each card is dropped.
Drop card, hold servo below drop ramp / platform to be low enough to hit lowering gear, move gantry sideways, then raise servo to ready for next card.
@@WNighteyesW potentially could use the motor on the gantry to turn this as well. Since it's not in use on the gantry to move the catching basins as you're putting cards in it. This would require the gantry motor either gets relocated or a kinda obnoxious system to port it's power over to the catcher you're currently focusing on. Either way a way to switch what the power is going to would be needed as well. Or at least a method to prevent the gantry from moving while you're engaging the card stack. It would be more complex, but hey one extra motor was a huge negative with my idea so I'm all for it. Solid idea.
So I realized a potential annoyance with this system over springs. You'd have to reset every catching basin after you're done. So a program you can run where it'll just go around and set every card stack to near-top. Minor inconvenience. However this system would likely not be very fast so it has more force to prevent the card stack from forcing it downwards. Depending on the speed, multiplied by 6 stacks. It might take a good moment to reset lol.
I like this idea, I hadn’t thought about using a single gantry-mounted motor to adjust stack heights! Brilliant work, I’ll have to model something to see how it would look and if there’s enough space to do so, but I do think it’s a potentially nice solution
@@jackbaumgartel Happy to give more options to explore! Hope you can find a solution that fits the needs and goals you're going for. This or springs or whatever else people have come up with in this comment section lol
I think you might have issues with binding. The springs are going to have some variance to them, so it's very likely that sometimes the platform will essentially be lifted from one corner. Consider printing a full size card holder with no springs and attaching strings to the top of each of the spring interface lobes. Then do your best to get the platform to lock up by pulling on strings. Once you have the tolerance and platform thickness right, it should slide smoothly no matter how you lift it.
Thanks for the input! That's a great way to test the fitment and especially ensure that any variances in the springs won't cause issue like you said. I appreciate it
@@jackbaumgartel no problem, thanks for the reply! I appreciate your approach and I'm excited to see where this project takes you :)
So rather than have moving parts on each catch bin, especially with temperamental springs and platforms prone to binding, I think it would be cheaper and more physically reliable to have the catch bins be static, with bottoms angled up at the back(towards the direction of entry so they don't fall out) and no rear wall, air flow slots on the front and sides, and then add a z axis to the catch chute(rotating up or down). You would have to track the stack height of each bin in software so that the chute would rotate its angle correctly, but this gives the best entry angle and it preserves the momentum of the card and doesn't force it to change direction until it is almost at a stop, and gives the best "falling" profile (narrow edge of card acts like a lever, or a closing lid)
That's a pretty neat idea! I hadn't thought much about giving the gantry "aim" and another degree of freedom to place cards more accurately. I appreciate the input!
Throwing my 2 cents in, albeit a little late. What about a pneumatic approach? It'd be easier to maintain consistency, and you could have the controller be centralized on the machine with hoses going to each collection area. Limit switches would guarantee proper zeroing. Other than that you could have the controller know how many cards have been sorted into each section and actuate the bed by the card width.
This method also allows you to use software to easily change card widths for use with other games and troubleshooting.
Since you only have 6 slots i think the tubes would be manageable, and from my research the controllers are relatively cheap (compared to the r&d on fabricating specialized springs, at least)
Perhaps line the bins with a different type of material? You could print the inner wall in TPE or something, while will exert a tiny amount of grip to the card?
That’s a good idea! I could see that being helpful for making sure the cards don’t get damaged as well
Just found your channel and I liked what you are doing. You gained another follower/supporter :). I am excited for the final result. Do you have any updates also on the software you will be using in addition to the machine?
glad to hear it! and i am trying to get a video out shortly with some updates!
Springs feel like they'd be too fiddly to get right personally, given any friction or binding that may occur.
An idea I do have though is to call into question some assumptions about simply dropping cards into chutes/bins though. In particular the fact that everything must be orthographically aligned with gravity.
At least with the ones you've shown the cards lay parralel to thr ground and fall directly vertically. And as you mentioned that's unstable beyond a short distance. However my idea is for the bin to be diagonal such that it follows the path of the falling card more natually, perhaps with the card builng guided downwards by the far/top wall as it lands on the cards below. Additionally you can play with the angle that the bottommost surface is at. If you just rotate a rectangular bin then the lowest point would be at the back of the card, and with that the cumulative effect of cards falling into imperfect position would be them levelling out to be horizontal, which is inefficient. However if you rotate the bottom surface such that the lowest point is at the far end then as any imperfections add up the slope would get greater, which actually encourages future cards to more snugly fit into their perfect position and thus I believe perfects itself.
In summary instead of having the bin have vertical walls and a horizontal floor, I'm suggesting it have the side profile of a parralelogram where the normal tangents of both the "walls" and "floor" are pointing away from the directions the cards come in from.
It feels rather awkward to describe it in text without being able to draw a diagram for you, but I hope I painted a good enough picture. Also I'll most certainly acknowledge that this design will take quite a bit of testing to get just right. However I do believe that once dialed in this design will be much less sensitive to failures, cheaper, and easier to produce than any design using spings.
Interesting idea! I think that I can at least somewhat picture what you're describing, and I hope I'm visualizing it correctly haha. I think you're right about the benefits of matching the incoming angle of the card's trajectory and allowing them to come to rest on an angled surface. Do you think that difference alone would be enough for even the first few cards that fall to do so in a non-chaotic manner?
@@jackbaumgartelI cannot say with 100% certainty but I think it would likely be good enough at this scale.
If you would like to perform an experiment before investing in designing my proposed bin might I suggest you record video of a number of cards being allowed to freely fall out of the chute. Then you can compare how tightly they follow the same path on their descent to get an idea of how chaotic it is. Additionally, this experiment would provide you with an approximate curve/angle to guide you when designing the bin if you decide to proceed.
I suspect very strongly that a mechanism like you are describing is going to jam up/be unreliable in practice. Its a clever mechanism to be sure, but i just keep imagining the little taps from the card falling farther and farther causing a multiplicative uneven effect on the springs, and eventually causing jams.
It's certainly a possibility, but I am hoping that loose enough tolerances between the parts and smooth sliding surfaces will help prevent that. Definitely something that will need to be tested though!
@@jackbaumgartel I would think you could use one stepper motor to continuously drive a screw, and then use the movement of the sorter to engage/disengage the screw to lower the platforms, so you could drive 6 platforms with 1 motor?
There hasn't been an update on this in a while I was wondering if you were able to complete this yet or where your progress is at
Apparantly the latest video is being worked on... 8 months of tinkering, can't wait to see what Jack has cooked up!
haha i’ve been procrastinating for a while during those 8 months, but yes progress is being made and i’m hoping to put out a video shortly! y’all have been extremely patient and i appreciate it!
The custom springs are probably going to be an issue, though they are cool and will absoutly work the price will probably cause issues if only doing small batches of them and the height may cause other issues with them like the springs possibly not fully going back to their proper size. Just a thought but would the reverse be a better idea? instead of compressing they streach?
Another possible idea would be to use only one motor and some gears to spin the screw that controlls the height. You will have height changes while moving back and forth while sorting but as you should probably keep your cards and inch or two below the top anyway to prevent the cards blowing out or other issues that shouldn't actually be a problem and it gets rid of custom hardware you would need, though it does add some fancy movement code or at least good gears and motors (just let them almost free rotate but stiff enough to hold the cards)
i think you should just have the catch basins be curved a J shape, such that when the stack is tall, the card just falls onto the top, and when the stack is low, slides along the curve and then slides onto the pile when it hits it
wait i have a better solution! a magic card is 2.5 inch x 3.5 inch right - instead of making the catch basin these dimensions, you make it 2.5 inch x 1 inch and the load the cards in diagonally. after a card enters the top of the basin it will not be able to flip over entirely, and since the bottom stack is angled, it will slide onto it when it hits the bottom. then it can be arbitrarily tall
you may be worried about damage to the card but i think like a falling elevator, the air column in the basin will slow the fall so it should be fine
shorter springs are going to be easier to manufacture, and replace for maintanance and testing. Given that, and that springs do not behave ideally at the extreames (in your case the start and end of the stacking) a few shorter springs in serires with a washer between will work much better. These shorter springs (maybe 2 or 3 in serires in the existing design channel) will behave much more consistently since the manufacturing tolorances will be lower (you need a "take the average" approach to manufactured goods unless you plan on some large percentage of rejects). I think 3 is probably the number to start with to minimize moving parts and open up design tolorances. The major downside will be making sure the washers don't flip or shift (hexagonal channel could fix that and still maintain enough contact on the springs). I love seeing the work progress and will want a prototype when they are ready!
It's been 10 months. I hope you are still making progress. Is that a frame from a 3d printer?
I know sorry about the delay! and the frame is made of extruded aluminum, which is very commonly used in 3D printer frames
Has this project continued? I'm very curious on it's current progress.
yes it has! working on getting an update out as soon as i can
If the card moves smoothly to the catch, a spring or any way of lifting the cards may not be needed at all. Similar way to a coin sorter works. You would just need something to prevent the cards from over shooting the catch.
A stretch goal could be catch beds that can be removed and replaced automatically as an add on.
Or even just quick manual release
that's a great idea! would make swapping stacks out super easy
Springs are cool, another possible solution rather than add motors is to have the card drop servo pull double duty. as the finger drops the card into a bin have it continue to rotate to advance a mechanism on the back of the current bin/stack. This could be a sort of ratchet/pall system that uses gravity to advance the stack by 1. If gravity isn't enough (especially true early in the sorting process) can have system use a heavier plate or have the plate in tension pulling downwards by a spring with its motion constrained by the ratcheting system. This setup would require a manual reset of the bins prior to sorting but not too big of a sacrifice.
For something like this, I would highly recommend using extension springs instead of compression springs, or variable force spiral torsion spring or vspring if you want to go with springs. Otherwise I would highly recommend a continuous rotation servo motor. It would make for more electronics, but at ~$10 per stack for servo, switch and maybe some other small stuff, it's comparible in price. (A 9g servo has 1.6kgcm of torque)
Thanks for the input! I'm curious what advantage you see in using springs in tension or torsion compared to compression?
@@jackbaumgartel the vsprings are the same ones that are used in key card retractable lanyards, so are pretty low profile, and you can buy some casing mechanisms so everything is properly enclosed. For extension springs, you don't need a center post to stabilize the stack. And since there will always be a small horizontal component to the force it will self stabilize to a point in the middle somewhere
Any further updates on this? Has this been completed?
Do you have any more recent updates for this?
Hey there, interesting video as always :)
I'm not in any sort of product designing or engineering type field, those these thoughts are more speculation-
- Would there be a concern of the springs losing strength over time? If so, how precise would the spring coefficient have to be to function?
- Perhaps a more elegant solution could be having the basins recieve, catch, and hold cards at an angle. The basin walls would still be vertical (just as the example one you have is) but a "wedge" at the bottom makes the cards sit at angle and the basin walls along the top and the bottom of the card appropriately closer
The hopper has walls and is at an angle, so I feel like cards will be coming out of there decently consistently (possibly using some sort of wide slit/gap to ensure the entrance angle of the card is decently consistent).
This feels hard to explain via text so I'll use airplane terms lol. For this, the "long ways" of the card would be in line with the roll axis
If the card is going hopper to basin, both of which are about a card in width, there shouldn't be much trouble with Yaw.
If the card falls off the hopper ramp at an angle, this will mean its going faster and falling "bottom edge of the card first." This would mean the card has less fall time and isn't falling flat, so there would be less general "tumbling." Notably, I think this should eliminate issue with rotation along the roll axis.
As for the pitch axis, since you have the card falling "bottom edge" rather than "card back" first, this would be the biggest issue. Making the bottom of basin hold cards at an angle could guide cards that have have ended up at varying pitches. Flat falling cards would simply fall onto the pile. Cards that have ended up more vertical be guided end up settling face side up.
Sorry if that an incomprehensible mess. Taking a step back, my main conceptual point is that the other machines keep their falling height shallow because they drop the cards cardback-first, making the cards more likely to tumble and tumble along random axis. Since you're using a ramp to guide cards into your basins, your cards will start off moving edge-side-first. Speed + aerodynamics would reduce fall time and randomness. Utilizing this, you could reduce the severity and randomness of tumble. An angled bottom of the basin could handle a wider tolerance of rotation along a singular axis.
Wow thanks for the long response! I appreciate the interest!
- The spring I specified is able to compress fully without any degradation, so it should last a very long time.
- And I like the idea of angling the cards within the stack, that could be helpful! You’re right about the nature of this design meaning that the cards will have more predictable paths in the air. Again thanks for the input, I appreciate it!
@@jackbaumgartel following this comment, if you lean the whole card holder back, towards the sorter and have the cards stack at an angle, they will fall down vertically then since the whole holder is leaning they should fall into place.
one possible issue is the card gets stuck against either wall standing veritcally.
another approach is a stack leaning forwards, away from the machine and just give the gantry the ability to move up and down (at the same angle as the forward lean of the stack), and have the back of the card stack (sorter side) open so cards can be inserted at any height, just keep track of the stack in software or visually or with a feeler every 5 cards inserted to double check, and you can just use the gantry that you already have moving to test the height only one servo or something to tilt down a momentary switch.
obviously this requires adding extra complexity to the sorter but it means your deck holders are dead simple dumb pieces of plastic.
@@joshieeee20 Thanks for the input and ideas, a few others have similarly suggested angling the catch basins, so it's definitely and idea I'll be looking into more!
Has anyone suggested rubber bands yet? Rubber bands on each corner could suspend the platform and would remain linear (more or less) rather than springs which will definitely bind if they're 9 inches long. You can also use pull springs, but I think that rubber bands will be easier since you can get a bunch of uniform rubber bands that are relatively weak and just use multiple per corner. This would allow you to fine tune the design by adding/removing rubber bands until you get a good stiffness. Also keep in mind that the upward force doesn't necessarily need to exactly match the weight of the cards, you just needs to make sure that the card fall is a height that you're comfortable with. I would say that as long as a card is falling no more than 2 inches you should be good.
Sorry if this has already been suggested, but I skimmed some of the top comments and didn't see suggestions for rubber bands.
Probably be a bit too complex but what about raising snd lowering the basins themselves. Basically have a motor on thr gantry that interfaces with some form of track on the end of the basins. You could then have a switch at the bottom of the basin that'll send a stop signal to the robot to allow the operator to empty the basin.
Thanks for the input! Definitely not a bad idea and one that I'll likely be exploring a little further
I think springs would work well, you would just need to take into account weight of the platform, springs nonlinear characteristics and friction.
Some alternatives would be constant force springs with the ability to wind them tighter or looser to adjust tension, or a ratcheting system where when the dispenser moves over a collection bin it goes down one "click". Each one of these "clicks" would correspond to one car in size, or the machine can have coarser clicks with the dispenser triggering it every 5-10 cards. The ratchet system would only need one servo/solenoid and the rest can be 3D printed.
Thanks for the input! I had thought about an incremental ratcheting system, and I do like the idea. I couldn't figure out a compact way to implement it though. Would the strategy that you're describing still be pretty space efficient?
@@jackbaumgartel I don't think it will be as space efficient as a set of springs, but I think it could get close since you would be moving space taken up from the front to the rear. My thinking is something similar to a clicky pen, but instead of the mechanism shifting the tip in and out, it will shift it down in steps.
I would still go with 5v pancake stepper motors. They are cheap and with a worm gear you can prevent back driving
I think investigating plastic compliant mechanisms to replace the springs could be interesting. They might do the same job as springs, but also be cheaper to produce, and more customizable for this purpose. 🤔
I do love a good compliant mechanism! I'd be interested to hear specifically how you imagine one might be of use here
@@jackbaumgartel the mechanism could be made of plastic or metal and tuned to bear weight of the cards the same way the spring does. However, compliant mechanisms can be designed to only move in certain dimensions and minimize defeclection in all other dimensions, so if properly designed it could be more reliable and less demanding on the design of the container.
@@ZilchZilch11 Thanks for the explanation! Do you envision it similar to the passive fin-style catch basin I showed in the video?
@@jackbaumgartel I envision it most similar to the single central spring arrangement, and maybe a "scissor lift" shape
One random idea I had, could you catch the card with an air cushion? Given the miniscule weight of the cards even a very small pc fan would be enough and they're very cheap.
subbed! is there an update? cant wait
thank you, and yes update coming soon!
Any new Videos of the Progress you made so far? : )
I have tried using springs in similar circumstances before, they are doable but ornery. I am concerned that the spring constent is very low yet needs to be very precise both to the magic cards and each other.
If the spring constant is too high the sorter will not go down enough each card, this can be compensated for by having buffer at the top but the more buffer the more risk of flipping.
If the spring constant is too low each card will depress the platform too much this is not as large of a problem but will eventually leave too much of a gap, again leading to flipping.
You seem to have a lot of confidence in these custom springs in terms of accuracy, precision, and durability. Maybe this is warented, I have not worked with custom springs but with how low the spring constant is, how large the range of elastic compression must be, how resistant to degredation, and how many environments it may need to be used in, having a spring that meets design specs sounds very hard.
I wish you luck.
Thanks for the insight. I admittedly haven't worked with such precise & thin springs before, so your concerns are certainly valid. I appreciate the good luck!
What about a basin with a vent like attachment at the top so it controls the slide into the basin. Might be a risk of jamming maybe but if tweaked right could be effective?
Yeah! I like the idea of controlling how the card enters the stack a bit more, but I still wouldn’t want the card to plummet ~9inches or so. A vent is a good idea though
@@jackbaumgartel another option are short or slightly bigger than short brissels to guide the fall instead of a spring?
Or using the dynamic of the falling card to fall exactly in place? Like if it falls on one end at an angle that the other end will fall and keep it in place. Would require a cushion as to not damage the card
@@shaper_i-o Oh that’s an interesting idea! You’re thinking bristles along the interior wall edges of the stack to slow the card’s descent? I imagine the card might have trouble staying level while falling through them, but maybe not?
@@jackbaumgartel it's a very unique take, my guess is it would muster a 40% success rate? But that's a guess, I edited my previous message and added another solution if you want to reread it
I wonder how many compressions the springs can take before needing replacement. Also need to account for the base that the cards will lay on preferably so the basin is somewhere between .1"-.5" deep to prevent first few cards from sliding off.
Great considerations! Yes I didn't do a great job of showing it in the video but the platform should be able to hold a few cards at it's "resting" height. As for the springs, the ones I specified are able to compress fully without deforming, meaning that they should last a very long time. Thanks for the input!
springs looks like a realy good solution. Reach it the goal of cheap price when they are always need to be customize create? Maybe you can find a seller in china which can help youat this topic.
Thank you! Yes, they are expensive when only buying a few, but can be purchased for just a few pennies each in bulk.
You alive? Lol its been a year im hopping your well and made some progress
I think a motor underneath each stack would be the most reasonable option. They could be basic little motors that lower each time a certain amount of cards are added to that stack. I.E. 25 cards drop in bay 3, then bay 3 motor drops .5 of an inch or whatever the calculation may be. They only need to go down with the weight of the cards, not up. Think about a force fed paintballgun hopper. They stop when a paintball is in place with sensors, instead of stopping with cards in place, just have the sensor at the very bottom when it is full the machine stops until it is reset.
Amazon has 6 for $7 i believe
It's a good solution! It would take some more hardware to get things moving properly and I suspect from prior experience that the cheap amazon servos likely wouldn't have enough power or precision, but it could be done. The added electrical power and wiring is considerable for 6 motors as well, but still I appreciate the input and will keep it in mind.
Look up conical compression springs. They nest into themselves. Have you considered using springs in tension instead of compression? Im envisioning a rubber band fastened to the base, up over a pulley and attached to the platform.
Or a torsional spring like a garage door.
I'd definitely look at how a garage door works. Add two torsional springs on rods to the long edges of the card-chute and have the platform suspended from cables that wind up onto spools attached to the rod.
@@Fredward0216 Thanks for the input! I do think that even a conical spring would risk buckling at 9in, but it's an improvement! And the garage-door torsion style would be interesting! A few others have suggested pulley systems as well on here. I would be concerned with the additional space & complexity required, as well as the additional hardware. Do you think the garage-door style could be accomplished in a small form factor?
a rectangular spring would probably be better , and springs from music wire are pretty easy to make, see if there is a small machine shop or even a hobbiest with a lathe they could probably make them
that’s a good idea! i’ll see if i can find any local vendors who are willing to help me out
Since the ramp is at the pile that's getting more cards, and you have a servo there now... Could you have some sort of ratchet/belt system which can lower the tray in the column when the servo arm goes down next to the pile? Then you don't need springs or extra moters...?
That's a really neat idea, it would be awesome to use the servo already there as you describe. Figuring out a proper mechanical system to accomplish that will be a task for sure, but is definitely promising. I appreciate the input!
If you expect the stacks to be about equal height, you could have all of them on one side of a sea saw with the input stack on the other. This way every time you take away a card from the input stack all of the sorted stacks will lower by 1/6th of the card height. Then once the card i dropped by another 1/6th. Moving the pivot point of the sea saw you could adjust the fractions to be a desired target. By doing some clever lever mechanics that lock and unlock whenever the gantry selects the stack you could have a perfect system that lowers the stack by the exact size of each card no matter it if weighs more or less due to foiling, warping or lther factors. Though this design would be very complex and the mechanism needs to be developed thoroughly.
Been some time since last update. Hope all is well
Thank you, yes just handling some life stuff, I'll be putting out an update shortly!
no Update after a year? ; ;
Also worth noting, this is a design decision that might constrain the build to just MTG, since card thicknesses and weights vary from game to game. Unless this is not an issue if there is a tendency for other cards to have their weight and thickness to have the same ratio? Worth looking into the weights and thicknesses of Pokémon, One Piece, Flesh and Blood, Digimon, and Vanguard. Those seem to be the games I've seen played at my card shops that have the same 2.5x3.5 inch size
Yes thanks for pointing that out! A few others have noted that too and it's definitely something I'll be testing out. I'm hoping that the weight difference will be trivial but that remains to be seen!
I'm no engineer, but what about a spring in a similar shape to the container (e.g. rectangular instead of circular), so that it would be the same size as the platform and possibly not require four of them per container but still hold the same weight needed?
I would try having them manufactured as rectangular but have them nest like conical compression springs. I'd imagine you could get some thin stainless plates cut with a water jet then stretched and tempered , but I'm not a spring expert.
Great idea! Someone else also mentioned a rectangular spring and it's something I'll be looking into now
What is the sotware that you use to design your machine?
the modeling shown here was done in fusion 360!
@@jackbaumgartel Great! I will play with it
What about a slide and the card holders can have about 4 inches of space to catch the cards. And instead of landing flat the bottom is slanted on a 22 degree angle. And a push out hole to take out the cards. Add holes on sides to let air flow out to prevent pockets to flip the card
I like the idea, but maybe I'm not visualizing it properly. Would this catch basin still be able to hold ~9in of cards, or is it just 4in tall?
@@jackbaumgartel I would say 4 inches because anything higher would likely flip but with enough air flow and at a slant it may be possible to go 9 inches
@@jackbaumgartel if you want I can send u a pic of what I'm thinking. Let me know I'll sketch one
What about using a pulley system where the counterweight is a chain? the more cards you have, the heavier the counterweight gets.
I love the idea! A chain or similar heavy rope would be able to match the increasing card weight just like a spring. I do think that the addition of pulleys & "chains" for each stack would be a mess. Any thoughts on how it could be implemented in a space-efficient way?
maybe you could use two pulleys and ropes on either side of the short part of the card stacks to avoid racking and save space in the width? if you then also use a wide ribbon for a rope you could have increased stability. the ribbon could ride over a rotating pin on the top and bottom of the card stack. The whole catcher assembly would get essentially twice as heavy this way though.
What about instead of just springs, a combination of springs/pulleys and weights that would instead allow you to use more commodity parts and not need to rely on expensive custom parts?
Thanks for the input! I like the idea of using pulleys to avoid any specialty/custom parts. I think it might be difficult to make use of such a system while still maintaining a small form factor though. Do you have any ideas on how that could be done?
I'm waiting on your next video like CHIRSTMAS
haha im glad to hear it!! sorry christmas has taken forever!
Since springs are designed for magic card weights, wouldn't this limit the potential use for the sorter being used on other card games? Pokemon and yugioh cards have different weights on their cards due to their rarity as well.
It very well could! A few others have mentioned this as well, and it's definitely something I'll be testing out. Thanks for the input!
Another way instead of screws is something a youtube channel called "Stuff made here" in a video called "A simple human task that's insanely hard for a robot". The robot is a puzzle solver and he had an issue for storing those pieces. At around the 11:45 mark he shows what he did. You could have one motor that is fixed to the base and have your card holders move side to side from it. here's the link! ruclips.net/video/Gu_1S77XkiM/видео.html
Thanks for the input and the link! I'm a big fan of Stuff made here, and the multiplexer option is a great one that I hadn't considered! A few others have suggested similar ideas so I'll definitely have to see how that would fit in!
So is this project done?
Still in the works!
What about using a tension spring instead of a compression spring?
I had thought about that but didn't see any real benefit, and it would take up more space. Would you see a benefit to a tension spring over a compression style?
@@jackbaumgartel I just thought of it when you shared the 4 spring design. Might be better reliability (less likely to hang up maybe). Would be able to utilize the entire receiving tray capacity with a tension over a compression spring as well.
However not an expert, so I would use my judgement sparingly
Great project Jack, Bravo. I am eager to see additional progress! Are you interested in a sponsor of sorts? I could financially help grease the gears sort to speak...I don't want to speak out of turn, but I am in the midst of sorting my own 20k collection and this would be a lot of help. Let me know...
Hello and thank you! I'm glad to hear you're enjoying the project and I really appreciate the offer! Right now, the best way to support the build is by using the link in the video description. You can make a general donation or help fund specific parts. And 20k is a lot of cards! hopefully I can get you something to help :)
Any updates?
Soon!
Just checking in. Any updates?
Any updates on this?
Cool. I know I talked to you in another video. I have my python code identifying most normal cards in under 1 second but the code is not optimized yet. The game has been out for 30 years. There's oddball stuff I am working on now. I have figured out how to detect foils. I cannot build my machine right now as my chop saw is dead :(
I am thinking 6 bins doesn't seem like enough. How do you plan to sort the cards? I would think 8 is a minimum needed for most common sorts for MTG. 5 bins for each color, 1 for colorless, 1 for multicolored, and 1 for unidentified/junk cards (like those rules reminder cards). I suppose you could sort by color, everything else in the unidentified bin. Then sort that bin by colorless and multicolored. If you could add a couple more bins though you could be a little more efficient with sorting time. I don't think short output stacks is a bad thing. Think about this. Let's say your input bin holds 2 boxes of MTG cards. If you sort by color on average there will be equal amounts of each color. There will be less colorless. Multicolors can be all over the board though depending on set. I think your output bins don't really need to me more than half the height of your input bin. Also, what do you think about pushing unidentified cards backwards or in some dedicated direction? That bin could be tall then and not match your regular output bins.
@@SirPoonga Awesome to hear from you again and impressive that your code is so quick! I talked about different sorting strategies a while back and what I learned is that everyone wants to sort their cards differently, so the ideal number of bins really is personal preference. This machine will have as many as it can fit without being too large that it's no longer a desktop device.
@@jackbaumgartel Right now I am doing a perpetual hash search to narrow down the card match. Look up ball tree image search, combined with phash, can narrow down the search quicker. From there I take the top matches and do a ssim comparison. I think mse works also. These are more traditional computer vision techniques. However, with a robot the position of the card in 2d/3d space will be near identical each time which allows for some shortcuts. You could go Tenserflow and neural networking but I think that is better for a phone app.
For source/training material I have taken the default cards list from scryfall's bulk data api. I filtered out unplayable cards like the rules reminders you find in boosters, oversized cards, artwork cards, etc. I then download each card image, check if it has the identification text in the lower left. If so throw the image away and mark the card in my db that I can find it by the id text. Otherwise, keep the image and phash it. There's some more details to it than that as over the 30 years of MTG they have done some strange stuff that I have to account for. So when searching I first check if there is the id text in the lower left corner. if so I identify it by that. This requires some OCR cleaning of the data as it may think an I is a 1 or a 1 is a 7 because it thinks the flag is long enough. If the card does not have id text I am then doing the phash/ssim comparison.
Right now I am finding the patterns of how the OCR fails. I am randomly picking a card newer than 2013 and not in brother's war. Grabbing the image. Giving it a slight contrast and noise filter so it isn't a perfect scan. Then run it through the identification and see what the OCR comes back with. Anything that isn't a match for which card was picked I note in a log. Once I get enough data I will start looking for patterns of how the OCR fails.
Note, if you also do the OCR thing they changed the format of the id text with MOM and newer. ChatGPT helped me out with making a regex that can identify the parts of the id text in either format.
@@jackbaumgartel ruclips.net/video/ZAWQQYgePP4/видео.html
„Euler‘s buckling cases“ should be something you understand thoroughly if you plan to go with a spring based design.
I‘d advise against it though. If the part isnt readily available and has to be custom manufactored then you‘ve lost one of the original goals for the project: for people to built it (easily) at home. Unless you wanna become a spring distributor and have ppl but them off you.
Thanks for the input, and yeah that's a fair concern. Avoiding custom parts is a major design goal, but so is cost & complexity! I've got some great feedback here so this is something I'll be working on more
Why use springs at all? You already have an actuator poking down each well every time a card is added: why not have the base of the wells start near the top and slide/ratchet down?
If you do actually need them to be automatically reset, then suction the bases back up. If the depth is an issue, add a spring underneath with a clasp that releases the ratchet from up top.
It's not strictly passive but it adds 0 active components and doesn't rely on specific and delicate spring force
This machine doesn't use any suction or an actuator to drop the cards! Sorry for the confusion, I showed a few examples of other machines which used such systems, but this design does not. A few videos back on my channel I did an overview of the mechanical design decisions and whatnot if you're interested!
@@jackbaumgartel aah! I think the z axis was just a huge assumption and the suction was stuck in my head from one of the other examples! I was recommended the video without having seen the design of this one. I'll definitely go watch
Sorry for the confused suggestions ;p
Have you made any progress on your design?
Yes! Sorry for the delay, I've been doing mostly software work so no fun updates yet!
can we get a up date
yes im working on one! thanks for the patience
Update?
OK, but what is the point of these machines?
Simply curious. Clearly there's a niche here for machines and MTG cards, but what do they do? Is it for like building a randomized deck, or sorting cards by color, or what?
Easy of sorting. Sorting can take quite a bit of time and can be tedious so a machine to do it for you is pretty dope
I also sure if the software is open sourced after the machine is created there could very well be someone that could run with it to create random decks.
Great question! The main use case for these machines is card shops that will often buy and sell cards by the thousands. It's expensive and tedious to pay employees to sort and organize the 100,000 cards you just bought and now need to sell. Some serious collectors and hobbyists with large collections might want one too just to help organize their collections.
Could you just use stepper motors and crank it down by the thickness of the cards?
The video talks about why that isn't an option
Yes, definitely could! But because the stacks will fill up at different rates, I’d need a separate motor for each one, which would be pretty expensive compared to the spring strategy
dead project or?
not dead! sleeping maybe but not dead
Time to revive? Got some ideas. Maybe a elasic band contraption. Or a variable resistance pulley like a compound bow.
Any updates?
yes i'm sorry it's been forever! working on getting an update out shortly
@@jackbaumgartel I can't wait!!!
Any updates?
Yes sorry i've been MIA for a while, working on an update video now
@@jackbaumgartel awesome can't wait for the update