You can still do press fit magnets in a drilled hole, but you can't use pressure. It's counterintuitive but if you chill the metal the hole will get larger not smaller. If you chill the metal sufficiently it's effectively and largening the whole big enough to accept the magnet without any serious force and then the magnet gets clamped when the metal warms back up to room temperature. Chilling also doesn't damage the magnets that way. You might be able to use dry ice to chill it, but I'm more effective mix might be liquid propane.
Ah, I love that feeling when I find a new niche channel to follow. This is delightful, I want to try my hand at a version of this when I have the time for in-person games again. This would be great for a “Box of Doom” style roller for the table!
The explanation of the mathematical principle is because the number of independent rolls is increased. Basically, if you wanted an equal distribution of any number between 1 and 12 when using 2 dice then you have to use a d12.
I love, love, _love_ these display modules. I kind of wish I could just throw them into every project I do with numeric output that doesn't need to be readable in the dark. Pity they're a bit much for that... I wonder if it's an economy of scale problem?
I kind of imagine the spinning cylinder to look like the spinning pillars scene in The Golden Child movie. "I-I-uh-I-uh-I-uh-I-uh want a random numberrrrrr"
OK that's pretty cool no wonder it kept getting recommended to me for the past two weeks. At first I thought it was gonna be some sort of digital dice didn't realize completely analog both would've been cool.
You could make the cylinder design by having a bunch of disks with 45 degree mirrors to redirect the displays. Redirecting the light instead of the magnetism.
You could implement percentile dice the same way geometric dice do: with multiple independent randomisers. Just use more than one spinning disc/drum. You could send the dials spinning at different speeds with the same mechanism as an old-fashioned slot machine.
i don't understand any of this, i hate physics but it's been so lovely watching someone do these things passionately ,, it's really granting as it is special to you, seeing people do the things they love is precious
I suppose making it a hex die can give you 0-15, or 0-F. Just my computer engineering experience figuring a way to expand beyond just a 10-digit die. I actually have trouble reading dice while playing D&D with friends. So the ease of seeing the results appeals to me. But I'm looking into getting some rechargable backlit dice.
I have a set of red D6's that were very consistent at low rolling, so consistent in fact that in DND I always had low stat characters to the point i even had an illegal character rolled up I can't remember the exact rules I think you needed at least 1 stat 14 or higher and an overall positive if you added up the modifiers
Is there anyway you could do an expanded examination of the mechanics for the 7 segment display? I did a bit of a deep dive on flip dots so I sort of get it, but most I see use electrical charge around a winding to flip the polarity on the display and get the flap to flip. Based just on what I see here I don't see how putting a second magnet up to the permanent magnets would change their polarity, but maybe I've been out of physics class too long. Are they permanent magnets in the chassis, or a lump of iron that just holds the magnetic charge (or is that the same thing?)
Nevermind, I watched again and I'm seeing my mistake. The iron pieces are the core that take the charge the magnets you refer to are actually the ones on the 7 segment display. I got mixed up because the dialog is a little opposite to the video; we see the front when discussing the iron, and the back when discussing the magnets.
Technically if you use something like hex, you could theoretically go from 0-15 (0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A - 10, B - 11, C - 12, D - 13, E - 14, F - 15)... though, unless you got really creative B and 8 would be confusing... I suppose you could use b instead, but then you'd get confused with 6? D has a similar problem with 0... Playing card denominations might be more intuitive... then you have 1, 2, 3, 4, 5, 6, 7, 8, 9, J, Q, K = 1D12... add an ace, and you have a D13.
The whole idea with dice is that it doesn't take 4 minutes to roll a random number. Literally isn't broken so you can't fix it. Truly a case of FTD here.
how about a big wheel all the dice working out from the center, as a column. so the dice that need more room get more room as the concentric circles of magnets get bigger. oh just realised how that wouldn't work with the magnetic detents at set intervals.... I can see a solution with a sliding display that can be moved to the desired ring and the magnetic detent is attached to that.
The final build looks like something out of a Cyan Worlds game that would be a little in universe flavor to learn the numbers of a fictional written language.
Actually, if you cover the disk with thin aluminum sheet, so individual magnets are no longer visible, it will be even more satisfying and "mysterious".
@@m1geo Probably just straight up burning. It's a cermet, there isn't any aluminum metal in the magnets. That wouldn't work since they're sintered! It's just straight up burning, for a thermite reaction you need an oxidized metal and a metal that wants to pull the oxygen. Rust (iron oxide) + aluminum = pure iron + aluminum oxide + heat. There is no oxidized metal in a neo magnet, so it can't be a thermite reaction. That means it's just straight up metal fire :) I mean there is a tiny amount of aluminum used probably as a flow modulator and plasticity enhancer during the pressing, that just kinda does nothing after sintering, but you aren't getting a thermite reaction with .2% aluminum. It's just normal combustion oxidizing all the metals
@@dogsarebest7107 "Probably just straight up burning." Well, it's glowing. The drill bit would melt long before reaching the temperatures required to ignite the metals in the magnet.
@@vinny142 No, definitely not. That's a tungsten carbide endmill. They happily turn steel into burning shards, happily turn titanium into raging fires, what makes you think it won't do it for this? It has to do with PHYSICS. Small particle size means LOW THERMAL MASS and HIGH SURFACE AREA. That means it burns REAL EASY. Why do you think microscopic powder generated is HARDER to ignite, than a solid chunk of tungsten that weighs several ounces? Even ignoring the fact that one particle igniting, generates heat that then raises the temperature of adjacent particles to either very near, or above the autoignition temperature. That's how exothermic reactions are defined. Have you seen any machining videos? Never seen a spark happen? Never see ribbons of metal burn? Touch a 9v battery to steel wool and think about it Static electricity can happily ignite aluminum powder or titanium powder, but the spark doesn't vaporize your finger. Because.. physics.
Since you just want random numbers and don't care about them being arranged sequentially in the spinner, you can overlap the positions and reuse magnets from the previous position for the next one. EG 2,3,5,6 and 8 all have three horizontal segments so you can get away with just one set of magnets to flip them on for half the wheel and then another set to flip them (as needed) in the other half of the wheel. You can also take advantage of the fact that the top half of the vertical segments are the same in 2/3 and in 5/6 whereas the bottoms are the same for 3/5 and 6/8. Meaning you just need to encode which bits to flip, not the whole state of the display per position. And, by just encoding the flips not the full data, you can cut down drastically on the space needed in the wheel.
I like this line of thinking, but I can't help but think that there would be some inevitable failure in the magnets flipping every node every time, at least when the disc is spinning fast. The one nice thing about parking the display over a full array for every number is that every node lands on a magnet that ensures it reads correctly. And having the same number of magnets on every number also helps balance the weight and spinning speed to reduce bias. I would think since magnets determine where the wheel stops, having more or fewer on different numbers could make them more or less likely to appear. I could be completely wrong though.
@@shotguntornado if you have a reset field (all on/all of) maybe it could be enough. Because in the end you only care about the correct flips once it slows down. And you should already have the bias in the wheel as it is. The one that has only "repellors" should be more unlikely than one that has only attractors or do the displays not work like that?
Won't that mean it only works in one direction, though? Since a magnet doesn't "flip" a bit, it's specifically writing a 0 or 1. If you set it so going from 6 to 8 just entails activating the top-right pair, then going from 8 to 6 will only keep them on.
@@SollowP I didn't even consider that, yeah you can't just choose to flip a left or right side. You can't flip a 3 into a 5 without it becoming a 9 for example.
this got me thinking about how cool it would be to have a revolver where the cylinder has the magnets around the outer circumference, and have the display mounted on the top. would be fun to play a western style game, pull out your revolver, give it a spin and have the 'roll' shown to you. add a little flair to your roleplay. especially so if your rolling D6s for damage since many revolvers are 6 bullet capacity so it would really be fitting
Nice build! Regarding the cylinder + magnet channels idea, at 04:25 you could tilt the display about 18° clockwise (or 12° CCW) to have the Y positions of each magnet be somewhat equally spaced, and it might be possible to design a cylinder where the 7 magnets for a given digit are not in a straight line, but in a weird pattern that matches the exact position of the display. It's a bit difficult to explain but I can sketch something if you'd like :) smaller magnets in the cylinder would probably be required (and a ton of magnets if you want a d100)
This is an incredible project. I have a deep affection for those mechanical flip displays, machined metal, and random number generation. Really a perfect storm for me. Thank you for your fantastic content
Just a minor nitpick: chi-squared is usually pronounced by mathematicians with a hard "k" sound (for the greek letter chi), not a "ch" sound. Great video!
More importantly, p = 91% doesn't mean "91% it's fair". It means "91% chance to see at least this much variability with a fair spinner". You can't assign a probability to the likelihood the null hypothesis is correct without a prior belief about it's correctness
Instead of bending steel bars to get your magnets into a linear configuration, I recommend trying lengths of coaxial cable, which should be shielded well enough to prevent that crosstalk you ran into. By the way, first-time viewer here - great video!
Coaxial cable will not transmit magnetic flux... Not static flux anyway. Your idea about shielding is correct but coaxial cannot shield against a magnetic field. Coax only shields against induced magnetic fields and that's a dynamic situation only.
packs of thin steel wire, with a wrap of kapton followed by mu metal tape and another layer of kapton, and maybe bismuth or pyrolytic graphite filling the space between them. That ought to work quite well.
Ack I'm sorry I missed your booth at Open Sauce. Hope you had a great time. I think this project is very understated in terms of all the cool design decisions involved in creating the spinner. Keep making cool stuff!!
to be fair I started at lapidary, cutting open geodes and agate, a wee touch on jewellery making and then the algorithm decided it was dice so I had people carving dice on diamond wheels casting them, all sorts. even just some showing off dice. now the algorithm has brought me back to machining just with more dice. cool video dude.
Hey! Great video. I noticed some yellowing around your eyes. It might be worth checking out, as it can sometimes indicate things like jaundice, allergies, or even just fatigue. Hope everything is okay!
the thing I like about your channel is that you just do what you think is intresting and fun without worrying that it may not perform the best on youtube or other stuff like that
Small note, "chi squared" is pronounced like Kai with a hard k, it's the written name of the Greek letter χ which is the symbol used in mathematical notation. Definitely not a huge deal or anything, just a small fyi.
It bugs me so much that you placed the numbers in order (0, 1, 2 and so on) rather than have opposite sides of the disc add up to 11 like they do on a dice. But DAMN that sound is satisfying! :D
@@wyw876 that might be worth working out although you would also have to consider the rest of the set passing by. That would make it considerably more complex to figure out. Assume we’re spinning clockwise. The right side pair of magnets on the disk first approaches the left side segments, then reaches the correct ones on the right side. The left side magnets are at their left side segments but then continue past the right side segments as they leave.
Great video! If you want to ream a hole oversized with a nominal reamer, you can shove a greased-up paper towel wad into the hole. You'll manage to oversize it by about 1.5 thou. If you need a little less slip, if you run it at a higher RPM and feed more slowly, you can shave off a bit less. As with drill bits, you'd be surprised how a reamer can be manipulated!
Thanks for sharing yet another marvelous building endeavor! Should you ever have more time than ideas - let’s recreate and improve Konrad Zuse Z1. That programmable mechanical computer is able to perform float point calculations but is prone to metal stripes stuck and bent irreversibly. The existing replica is close enough to the original to suffer from exactly same problems.
I love projects like this. They start with just a question or curiosity and then morph and materialize out of thin air. No one asked for this specific thing. Not even you. And yet, now it exists, and it's AWESOME. Love it.
Absolutely love the project! And I hate to be the insufferable "that guy" but a p-value of 0.91 doesn't mean "a 91% chance its fair." A p-value of 91% means that if we assume the d10 is fair, there's a 91% chance of obtaining a distribution as reasonable as the one you obtained. Therefore you could say "this isn't a crazy distribution to get from a fair d10, so I can't disprove the possibility that it's a fair d10." If you had a p-value of 1% then the correct statement would be "if this was a fair d10 there's a very low chance that I would've observed this crazy distribution, therefore I doubt this is a fair d10."
I would love to see a miniaturized version of this, maybe of a handheld size. Of course, that would mean that an even smaller flip digit display would need to be made as well.
You solved two problems (percentile is just two of these). You only need a single bit display to extend it to a d20, as many early d20s did double duty before the modern d10 shape took hold. They would have 0-9, and 0-9 again but with a dot. If you were using it as a d20, the dot was worth 10. If you were using it as a d10, you ignored the dot. In your case, perhaps use a second digit that either underlines the number, or doesn't. Then the viewing window could hide most of the extra display and you're only adding one magnet per number (although twice as many numbers). If you build a d20 disc this way, it can double as one digit of your percentile die so you wouldn't need a second d10. You could also use it in conjunction with the d10 to speed up that Eldritch Blasting. Similarly, you could make a dual-purpose d12/d6 spinner, and a dual purpose d8/d4 spinner.
It'd take a bigger disc, but you could do a D20 if you compromised in using A-J (with a weird g) for 11-20. Or, a single flip dot for +10 (eg. dot-3 is 13, while no-dot-4 is just 4). You'd still need a big wheel for it though.
I think when you heated the metal with a gas torch you took it past the curie point which messed with its magnetic affects. You want a nice ordered line of iron atoms, and now there kinda all over the place, not passing the affect down the rod, it just goes out the side or wherever.
The iron rods weren't magnets themselves. They were just supposed to "conduct" or guide the magnetic fields from the neodymium magnets. Once they cooled back below the curie point, they would be able to do that as they'd be attracted to the magnets again. I think the problem was that the magnetic field needs a return path, and without one provided, it would follow the adjacent rods back inadvertently flipping other segments.
Fascinating! that’s a chonky boy. You should look up demley mechanical dice from 1920, they’ve always amazed me but I don’t quite grasp how this works and there’s no quality content explaining the mechanism on youtube. please make one for us, I’ll subscribe to your channel for now.
100 rolls is a tiny drop in the bucket of a "true" test. but no one ever has time for that :p I do wonder what sort of non-random weight the a) arranging the numbers 0-9 in order caused (since there's a non-random distribution of "change polarity" and "keep polarity" interactions. and b) its been shown that repeated human inducted motion (spin the wheel or Roll the dice) has it's own measurable bias. In all likelihood, those (and whatever other environmental and manufacturing factors) probably don't mess with things enough to care about in real use, but it'd be interesting to see how many decimal places you have to go out before you can see the effect they have.
Totally not to weird you out, and maybe it's your lighting, but the green hue under your eyes is medically concerning... I hope your okay and taking care of your health! You always make super cool stuff, and I'd feel bad if I didn't at least say something.
Oh how I wish i could find one of those 7-segment displays for myself. 2 weeks searching and no luck. I would love to have made one myself out of some laser cut wood for my D&D kids. I absolutely love this and would love to see how you would tackle a D20.
Man, you have an uneven distribution of blood in your face (yellow eye sockets). Please contact a doctor - there could be problems with your thyroid or liver.
I love it! I have a thought for how to make it easier to remove inverted magnets: first drill through-holes at all the right index positions, between half to 3/4 of the clearance hole diameter. Then have a drill press with the clearance hole bit & a depth stop. The larger bit should naturally center on the pilot holes well enough; start with a depth too shallow & "sneak up to" the depth where the magnets are flush to the surface. Now if a magnet is backwards, you should be able to knock it out with a punch from the back side pretty easily, especially with a glue that takes 24 hours for "full cure." I'm a mechanical engineer with above-average manual machine shop experience 😉
I know the distribution looks good but it's, in a way, false. Dice are not numeric, they are probabilistic. In other words, you cannot use the mean value, you have to use the probability of a specific value occurring. Easiest example is the natural 20. If you are going for a natural 20, it doesn't matter that the 4 dice add up to 20, you want one to be 20! It's a probability not a sum or mean. Similarly if a game mechanic requires a double, then it's the probability of one dice matching another. This becomes more important when micro bias and variation occurs... 2 nearly perfectly balanced dice might have a slightly larger chance of summing a smaller number for example. For more on this Google what ordinal numbers are.
This isn’t anything to do with consistency or skill, this is a DOWNSIDE to using multiple dice for RNG. 10d6 and 1d60 are completely different methods. If you use multiple dice you need to use it SPECIFICALLY for the normal distribution. Otherwise you should use another method
The sound truly is delightful. I want that to be THE sound effect for "some sort of RNG is happening".
It does sound like rrrrrnnnnnnnnnng-g-g--g-----g------g :D
The way your story parallels what became of music, from tubes to flat sheets, quite interesting. Nice design, GG!
You can still do press fit magnets in a drilled hole, but you can't use pressure. It's counterintuitive but if you chill the metal the hole will get larger not smaller. If you chill the metal sufficiently it's effectively and largening the whole big enough to accept the magnet without any serious force and then the magnet gets clamped when the metal warms back up to room temperature. Chilling also doesn't damage the magnets that way. You might be able to use dry ice to chill it, but I'm more effective mix might be liquid propane.
Ah, I love that feeling when I find a new niche channel to follow. This is delightful, I want to try my hand at a version of this when I have the time for in-person games again. This would be great for a “Box of Doom” style roller for the table!
The explanation of the mathematical principle is because the number of independent rolls is increased. Basically, if you wanted an equal distribution of any number between 1 and 12 when using 2 dice then you have to use a d12.
Coherent channel themes, who needs them
I love, love, _love_ these display modules. I kind of wish I could just throw them into every project I do with numeric output that doesn't need to be readable in the dark. Pity they're a bit much for that... I wonder if it's an economy of scale problem?
Tom Scott brought me here. Content kept me here!
I kind of imagine the spinning cylinder to look like the spinning pillars scene in The Golden Child movie. "I-I-uh-I-uh-I-uh-I-uh want a random numberrrrrr"
OK that's pretty cool no wonder it kept getting recommended to me for the past two weeks.
At first I thought it was gonna be some sort of digital dice didn't realize completely analog both would've been cool.
You could make the cylinder design by having a bunch of disks with 45 degree mirrors to redirect the displays. Redirecting the light instead of the magnetism.
Your videos are very enjoyable. I love your curious, hands on character. You share your thought process well!
I really want to make a D20 version now, a few tweaks would be required in the design, but it would definitely be doable
You could implement percentile dice the same way geometric dice do: with multiple independent randomisers. Just use more than one spinning disc/drum. You could send the dials spinning at different speeds with the same mechanism as an old-fashioned slot machine.
i don't understand any of this, i hate physics but it's been so lovely watching someone do these things passionately ,, it's really granting as it is special to you, seeing people do the things they love is precious
I loved spinning this at open sauce. So satisfying! Would you consider making this design open-source so other people could make their own?
I suppose making it a hex die can give you 0-15, or 0-F. Just my computer engineering experience figuring a way to expand beyond just a 10-digit die. I actually have trouble reading dice while playing D&D with friends. So the ease of seeing the results appeals to me. But I'm looking into getting some rechargable backlit dice.
Hey Hey! Glad to meet you at Opensauce, I was the big guy helping with Cody's Suit booth. Hope to see ya next year
I've got no idea why but I'm getting massive ze frank vibes... Cool machine
I can confirm that this was very satisfying to spin. Glad I had a chance to stop by and chat on Friday!
really interesting and really clear and concise naration.
Great project! This is why I watch You tube. Thank you for posting!
Would you mind sharing a link to the display segments? They look amazing!
Man I’m so bummed that I missed this at opensauce. This is awesome!
I have a set of red D6's that were very consistent at low rolling, so consistent in fact that in DND I always had low stat characters to the point i even had an illegal character rolled up I can't remember the exact rules I think you needed at least 1 stat 14 or higher and an overall positive if you added up the modifiers
This was cool I just would of put super thin layers of wood over the disk & instead of going 1, 2, 3, 4,… I would of randomly picked an order
Is there anyway you could do an expanded examination of the mechanics for the 7 segment display? I did a bit of a deep dive on flip dots so I sort of get it, but most I see use electrical charge around a winding to flip the polarity on the display and get the flap to flip. Based just on what I see here I don't see how putting a second magnet up to the permanent magnets would change their polarity, but maybe I've been out of physics class too long. Are they permanent magnets in the chassis, or a lump of iron that just holds the magnetic charge (or is that the same thing?)
Nevermind, I watched again and I'm seeing my mistake. The iron pieces are the core that take the charge the magnets you refer to are actually the ones on the 7 segment display. I got mixed up because the dialog is a little opposite to the video; we see the front when discussing the iron, and the back when discussing the magnets.
Bro out here brain washing us with that intro lol
it would be tough to make, but i could imagine a spiral, or a long tape 🤔
Technically if you use something like hex, you could theoretically go from 0-15 (0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A - 10, B - 11, C - 12, D - 13, E - 14, F - 15)... though, unless you got really creative B and 8 would be confusing... I suppose you could use b instead, but then you'd get confused with 6? D has a similar problem with 0...
Playing card denominations might be more intuitive... then you have 1, 2, 3, 4, 5, 6, 7, 8, 9, J, Q, K = 1D12... add an ace, and you have a D13.
I love it, love it, love it! But I really, really want to polish the hell out of it!
Who wouldve expected that problem when drilling magnets o.o
A little bit of tumbling of cleaning up the tool marks, and you could sell the kick out of those.
Wow, haven't seen something this orriginal in a while. Great job 👏
The whole idea with dice is that it doesn't take 4 minutes to roll a random number. Literally isn't broken so you can't fix it. Truly a case of FTD here.
how about a big wheel all the dice working out from the center, as a column. so the dice that need more room get more room as the concentric circles of magnets get bigger. oh just realised how that wouldn't work with the magnetic detents at set intervals.... I can see a solution with a sliding display that can be moved to the desired ring and the magnetic detent is attached to that.
Wait, since when did dice have a zero?
The final build looks like something out of a Cyan Worlds game that would be a little in universe flavor to learn the numbers of a fictional written language.
Good timing too, the Riven remake is out today
@@HubrisIncbased vr mode
I really want to try those games
I like that it shows non-numbers between valid outputs. That’s a feature, not a bug.
Just as you said "Happy RNG day" I thought to myself "I hope the date is different every year" and you didn't disappoint.
9:39 you just released the little magnetism demon trapped in the magnet
So that's how magnets work!
This video really reminds me of a classic 2010 UA-cam video. And I mean that as a most sincere compliment! Thank you for showing up in my feed!
Same here. I love it's not overly produced.
@@ddegn Indeed, but it's structured enough to keep you engaged!
Actually, if you cover the disk with thin aluminum sheet, so individual magnets are no longer visible, it will be even more satisfying and "mysterious".
The aluminum may act like a brake though. While it isn’t ferrous it is metal and responds to moving magnetic fields.
@@NandRIt wouldn't be a breaking effect though, as no eddy currents would be generated as the magnets would be spinning with the aluminum.
@@electrifyingvids3545 There are magnets on the display though...
@@joey_f4ke238 Perhaps they could switch to silver plated plastic? The silver would provide negligible eddy currents.
We tried that, and a small portal opened up, so we shut it down, like CERN in 2027, fyi, the future sucks!
I'm glad I randomly landed on this video, particularly on this day.
I randomly missed a day
i love dice and dice adjacent peripherals but my favorite part was drilling into that magnet. would not have thought it would look like that
I wonder if there was a termite reaction going on? Iron and aluminium burning?
@@m1geo Probably just straight up burning. It's a cermet, there isn't any aluminum metal in the magnets. That wouldn't work since they're sintered! It's just straight up burning, for a thermite reaction you need an oxidized metal and a metal that wants to pull the oxygen. Rust (iron oxide) + aluminum = pure iron + aluminum oxide + heat. There is no oxidized metal in a neo magnet, so it can't be a thermite reaction. That means it's just straight up metal fire :) I mean there is a tiny amount of aluminum used probably as a flow modulator and plasticity enhancer during the pressing, that just kinda does nothing after sintering, but you aren't getting a thermite reaction with .2% aluminum. It's just normal combustion oxidizing all the metals
@@dogsarebest7107 "Probably just straight up burning."
Well, it's glowing. The drill bit would melt long before reaching the temperatures required to ignite the metals in the magnet.
@@vinny142 No, definitely not. That's a tungsten carbide endmill. They happily turn steel into burning shards, happily turn titanium into raging fires, what makes you think it won't do it for this?
It has to do with PHYSICS. Small particle size means LOW THERMAL MASS and HIGH SURFACE AREA. That means it burns REAL EASY. Why do you think microscopic powder generated is HARDER to ignite, than a solid chunk of tungsten that weighs several ounces? Even ignoring the fact that one particle igniting, generates heat that then raises the temperature of adjacent particles to either very near, or above the autoignition temperature. That's how exothermic reactions are defined. Have you seen any machining videos? Never seen a spark happen? Never see ribbons of metal burn? Touch a 9v battery to steel wool and think about it
Static electricity can happily ignite aluminum powder or titanium powder, but the spark doesn't vaporize your finger. Because.. physics.
Since you just want random numbers and don't care about them being arranged sequentially in the spinner, you can overlap the positions and reuse magnets from the previous position for the next one. EG 2,3,5,6 and 8 all have three horizontal segments so you can get away with just one set of magnets to flip them on for half the wheel and then another set to flip them (as needed) in the other half of the wheel. You can also take advantage of the fact that the top half of the vertical segments are the same in 2/3 and in 5/6 whereas the bottoms are the same for 3/5 and 6/8. Meaning you just need to encode which bits to flip, not the whole state of the display per position. And, by just encoding the flips not the full data, you can cut down drastically on the space needed in the wheel.
I like this line of thinking, but I can't help but think that there would be some inevitable failure in the magnets flipping every node every time, at least when the disc is spinning fast. The one nice thing about parking the display over a full array for every number is that every node lands on a magnet that ensures it reads correctly. And having the same number of magnets on every number also helps balance the weight and spinning speed to reduce bias. I would think since magnets determine where the wheel stops, having more or fewer on different numbers could make them more or less likely to appear. I could be completely wrong though.
@@shotguntornado if you have a reset field (all on/all of) maybe it could be enough. Because in the end you only care about the correct flips once it slows down. And you should already have the bias in the wheel as it is. The one that has only "repellors" should be more unlikely than one that has only attractors or do the displays not work like that?
Won't that mean it only works in one direction, though? Since a magnet doesn't "flip" a bit, it's specifically writing a 0 or 1. If you set it so going from 6 to 8 just entails activating the top-right pair, then going from 8 to 6 will only keep them on.
Neat idea but the end result would just be the display showing an 8 since the magnets would pass over and flip things which weren't supposed to flip.
@@SollowP I didn't even consider that, yeah you can't just choose to flip a left or right side. You can't flip a 3 into a 5 without it becoming a 9 for example.
Thank you for the quality closed captions. They are greatly appreciated.
this got me thinking about how cool it would be to have a revolver where the cylinder has the magnets around the outer circumference, and have the display mounted on the top.
would be fun to play a western style game, pull out your revolver, give it a spin and have the 'roll' shown to you. add a little flair to your roleplay. especially so if your rolling D6s for damage since many revolvers are 6 bullet capacity so it would really be fitting
Fascinating idea and of course the numbers arranged appropriately will count down remaining ammo
I have a revolver cylinder with a full set of bullet dice just for this purpose
Now I'm imagining 3d printing a toy revolver in your design for use in a Russian roulette "game".
Nice build! Regarding the cylinder + magnet channels idea, at 04:25 you could tilt the display about 18° clockwise (or 12° CCW) to have the Y positions of each magnet be somewhat equally spaced, and it might be possible to design a cylinder where the 7 magnets for a given digit are not in a straight line, but in a weird pattern that matches the exact position of the display. It's a bit difficult to explain but I can sketch something if you'd like :) smaller magnets in the cylinder would probably be required (and a ton of magnets if you want a d100)
This is an incredible project. I have a deep affection for those mechanical flip displays, machined metal, and random number generation. Really a perfect storm for me. Thank you for your fantastic content
Even watching this thing spin is satisfying, can't imagine how extremely delightful it is to spin it in real life.
This was a ton of fun to play with at OpenSauce. I will not admit how much time I spent spinning the dial or just admiring the simplicity.
Just a minor nitpick: chi-squared is usually pronounced by mathematicians with a hard "k" sound (for the greek letter chi), not a "ch" sound. Great video!
More importantly, p = 91% doesn't mean "91% it's fair". It means "91% chance to see at least this much variability with a fair spinner". You can't assign a probability to the likelihood the null hypothesis is correct without a prior belief about it's correctness
latek?
The Kessler Syndrome being perhaps not as unfun as you'd hoped is a pretty great turn of phrase. xD
Where are more details of the game? I checked Mastedon and only saw a few screenshots.
@@bernardandrys2397 Would be cool to see gameplay of this and even better if it were released on Steam.
Did you mess with the color correction of the area under your eyes? It looks greenish.
Very cool!
By the way "Chi" is pronounced like the first half of the word "kite".
Instead of bending steel bars to get your magnets into a linear configuration, I recommend trying lengths of coaxial cable, which should be shielded well enough to prevent that crosstalk you ran into. By the way, first-time viewer here - great video!
Coaxial cable will not transmit magnetic flux... Not static flux anyway. Your idea about shielding is correct but coaxial cannot shield against a magnetic field. Coax only shields against induced magnetic fields and that's a dynamic situation only.
For magnetis screening..Mu metal is your thing ( mu pronounced moo )
@@jasonudall8614 that's exactly what I was going to comment
packs of thin steel wire, with a wrap of kapton followed by mu metal tape and another layer of kapton, and maybe bismuth or pyrolytic graphite filling the space between them. That ought to work quite well.
You’ve hit it out of the park with your explanation, no, story telling about statistics. It’s honestly a delight to watch.
0:19 I did not know that.
I would squeal for a mechanical clock with this display mechanism.
Ack I'm sorry I missed your booth at Open Sauce. Hope you had a great time. I think this project is very understated in terms of all the cool design decisions involved in creating the spinner. Keep making cool stuff!!
to be fair I started at lapidary, cutting open geodes and agate, a wee touch on jewellery making and then the algorithm decided it was dice so I had people carving dice on diamond wheels casting them, all sorts. even just some showing off dice. now the algorithm has brought me back to machining just with more dice. cool video dude.
Hey! Great video. I noticed some yellowing around your eyes. It might be worth checking out, as it can sometimes indicate things like jaundice, allergies, or even just fatigue. Hope everything is okay!
the thing I like about your channel is that you just do what you think is intresting and fun without worrying that it may not perform the best on youtube or other stuff like that
That spinner is awesome. It looks like something from a star trek casino!
Small note, "chi squared" is pronounced like Kai with a hard k, it's the written name of the Greek letter χ which is the symbol used in mathematical notation. Definitely not a huge deal or anything, just a small fyi.
I have a bunch of tiny neodymium magnets from an abandoned project. I wonder if I can make my own flipdot modules. To the workbench!
It bugs me so much that you placed the numbers in order (0, 1, 2 and so on) rather than have opposite sides of the disc add up to 11 like they do on a dice.
But DAMN that sound is satisfying! :D
I'm wondering if the total forces involved in the number of segments being flipped per digit transition is affecting the randomness distribution...
@@wyw876 that might be worth working out although you would also have to consider the rest of the set passing by. That would make it considerably more complex to figure out.
Assume we’re spinning clockwise. The right side pair of magnets on the disk first approaches the left side segments, then reaches the correct ones on the right side.
The left side magnets are at their left side segments but then continue past the right side segments as they leave.
@heatshield would magnetic eddy currents cancel that out, or introduce even more randomness? 🤔
Looks and feels like something between d10 and roulette. You know "roll on black to open this door"🌝
Great video! If you want to ream a hole oversized with a nominal reamer, you can shove a greased-up paper towel wad into the hole. You'll manage to oversize it by about 1.5 thou. If you need a little less slip, if you run it at a higher RPM and feed more slowly, you can shave off a bit less. As with drill bits, you'd be surprised how a reamer can be manipulated!
What a wonderful project and video! (despite the pentagonal trapezohedra slander :P)
Thanks for sharing yet another marvelous building endeavor!
Should you ever have more time than ideas - let’s recreate and improve Konrad Zuse Z1. That programmable mechanical computer is able to perform float point calculations but is prone to metal stripes stuck and bent irreversibly. The existing replica is close enough to the original to suffer from exactly same problems.
I love projects like this. They start with just a question or curiosity and then morph and materialize out of thin air. No one asked for this specific thing. Not even you. And yet, now it exists, and it's AWESOME. Love it.
Absolutely love the project! And I hate to be the insufferable "that guy" but a p-value of 0.91 doesn't mean "a 91% chance its fair."
A p-value of 91% means that if we assume the d10 is fair, there's a 91% chance of obtaining a distribution as reasonable as the one you obtained. Therefore you could say "this isn't a crazy distribution to get from a fair d10, so I can't disprove the possibility that it's a fair d10."
If you had a p-value of 1% then the correct statement would be "if this was a fair d10 there's a very low chance that I would've observed this crazy distribution, therefore I doubt this is a fair d10."
Severely underrated comment smh
I would love to see a miniaturized version of this, maybe of a handheld size. Of course, that would mean that an even smaller flip digit display would need to be made as well.
Regular and coherent are so boring, I’m here for that drip-feed of neat passion projects to distract me from my pile of incomplete passion projects 😂
You solved two problems (percentile is just two of these). You only need a single bit display to extend it to a d20, as many early d20s did double duty before the modern d10 shape took hold. They would have 0-9, and 0-9 again but with a dot. If you were using it as a d20, the dot was worth 10. If you were using it as a d10, you ignored the dot. In your case, perhaps use a second digit that either underlines the number, or doesn't. Then the viewing window could hide most of the extra display and you're only adding one magnet per number (although twice as many numbers).
If you build a d20 disc this way, it can double as one digit of your percentile die so you wouldn't need a second d10. You could also use it in conjunction with the d10 to speed up that Eldritch Blasting.
Similarly, you could make a dual-purpose d12/d6 spinner, and a dual purpose d8/d4 spinner.
It'd take a bigger disc, but you could do a D20 if you compromised in using A-J (with a weird g) for 11-20. Or, a single flip dot for +10 (eg. dot-3 is 13, while no-dot-4 is just 4). You'd still need a big wheel for it though.
I think when you heated the metal with a gas torch you took it past the curie point which messed with its magnetic affects. You want a nice ordered line of iron atoms, and now there kinda all over the place, not passing the affect down the rod, it just goes out the side or wherever.
is there a way to restore it? like, magnetising the metal while it's red-hot and cooling?
The iron rods weren't magnets themselves. They were just supposed to "conduct" or guide the magnetic fields from the neodymium magnets. Once they cooled back below the curie point, they would be able to do that as they'd be attracted to the magnets again. I think the problem was that the magnetic field needs a return path, and without one provided, it would follow the adjacent rods back inadvertently flipping other segments.
Fascinating! that’s a chonky boy. You should look up demley mechanical dice from 1920, they’ve always amazed me but I don’t quite grasp how this works and there’s no quality content explaining the mechanism on youtube. please make one for us, I’ll subscribe to your channel for now.
If random values is the end goal, the sequence of the numbers on the disc should also be randomized.
100 rolls is a tiny drop in the bucket of a "true" test. but no one ever has time for that :p I do wonder what sort of non-random weight the a) arranging the numbers 0-9 in order caused (since there's a non-random distribution of "change polarity" and "keep polarity" interactions. and b) its been shown that repeated human inducted motion (spin the wheel or Roll the dice) has it's own measurable bias. In all likelihood, those (and whatever other environmental and manufacturing factors) probably don't mess with things enough to care about in real use, but it'd be interesting to see how many decimal places you have to go out before you can see the effect they have.
Totally not to weird you out, and maybe it's your lighting, but the green hue under your eyes is medically concerning... I hope your okay and taking care of your health! You always make super cool stuff, and I'd feel bad if I didn't at least say something.
Oh how I wish i could find one of those 7-segment displays for myself. 2 weeks searching and no luck. I would love to have made one myself out of some laser cut wood for my D&D kids. I absolutely love this and would love to see how you would tackle a D20.
Man, you have an uneven distribution of blood in your face (yellow eye sockets). Please contact a doctor - there could be problems with your thyroid or liver.
Cheerz Bruv 🍻. Got my Sub, That means a lot. This is AwesomeSause.
For removing the misaligned magnets, before drilling them out, did you try using a ca glue remover/debonder/desolver?
That thing sounds like a shaman staff. . . Why do I want one?. Great build. Great idea.
I love it! I have a thought for how to make it easier to remove inverted magnets:
first drill through-holes at all the right index positions, between half to 3/4 of the clearance hole diameter. Then have a drill press with the clearance hole bit & a depth stop. The larger bit should naturally center on the pilot holes well enough; start with a depth too shallow & "sneak up to" the depth where the magnets are flush to the surface.
Now if a magnet is backwards, you should be able to knock it out with a punch from the back side pretty easily, especially with a glue that takes 24 hours for "full cure."
I'm a mechanical engineer with above-average manual machine shop experience 😉
I've had success using a spring center punch to shatter magnets in the holes then a pick to get them out. Also CA glue uncuring agent helps
Is it possible a subtle user might be able to influence the outcome by choosing how hard to spin?
You are very cool. I just wanted to give that affirmation, love your project!! :)
thats crazy man. you have so much cool content and yt has only now recommended me your channel. how many channels like yours are just never promoted?
All that work and still we haven't a "10" on our d10 😂
I saw your booth at Open Sauce! really cool stuff you have. I didn't know how much time and effort went into such a seemingly simple device.
I know the distribution looks good but it's, in a way, false. Dice are not numeric, they are probabilistic. In other words, you cannot use the mean value, you have to use the probability of a specific value occurring.
Easiest example is the natural 20. If you are going for a natural 20, it doesn't matter that the 4 dice add up to 20, you want one to be 20! It's a probability not a sum or mean. Similarly if a game mechanic requires a double, then it's the probability of one dice matching another.
This becomes more important when micro bias and variation occurs... 2 nearly perfectly balanced dice might have a slightly larger chance of summing a smaller number for example.
For more on this Google what ordinal numbers are.
I love the sound of it SO. MUCH.
Glad I found you through your Geochron video!
0:40 immediately outing yourself as having incredible taste, nice
This isn’t anything to do with consistency or skill, this is a DOWNSIDE to using multiple dice for RNG. 10d6 and 1d60 are completely different methods.
If you use multiple dice you need to use it SPECIFICALLY for the normal distribution. Otherwise you should use another method
It sounds like a rattlesnake. The sound of "you're in danger!" Perfect for an RNG scenario 😅
I was thinking it sounds like a rain-stick.