Very cool! Another tip is that you want to have the walls as thick as possible so that the slicer will not print unneeded lines that could slow the bearing down. For overkill when printing bearings I set the wall thickness to 3.0 mm so that the ball bearing can roll around the grove of the print and not hit cross lines.
I experimented with this design, I made the design in Fusion 360, these were the changes I made. I minimised the clearance between the diagonal 45 degree contact surface of the inner and outer race to zero. You can adjust the pre-tension on the ball by the screws. For a smoother surface I separated the inner and outer race and printed them in vase mode and sequential printing and set the extrusion width to 1 mm, layer height 0.15 mm. I changed the print orientation 180 degrees, so the wide part was down. This minimised the deformation during cooling, 45 degrees angle prints like to curl up during printing. The inner and outer races press fit in the rest of the design, you use a more wear resistant filament for this. You could also machine them is you have a lathe. The results are very promising, there is no play in the bearing. The bearings are silent, you can still feel the balls skipping to the next layer, which are at a slight angle in vase mode. I did this first test with 3mm balls, I didn’t have any 4mm balls. I will try 5mm next.
@@chipcode5538 wow, interesting. My goal was to make a simple design with as few parts as possible to make assembly easier. But it's very cool that you are exploring ways to increase quality. Cheers!
I see to remember that it's a Good Idea to use an odd number of bearing balls. I think it distributes the load better if there are imperfections in balls or races. Much like a three-legged or five-legged chair is more stable than four or six legs on an uneven floor.
three-legged chair don't wobble on uneven floors. 4, 5 or 6 legs will always wobble. If you have an even floor, more legs means more stability. Doesn't have anything to do with odd or even. (-:
Nice work. BTW, I've found that you don't need threaded inserts in situations where you aren't taking the screws out that often. Just thread direct into the plastic.
Besonders bei einem Fahrzeug ist aber mit Verschmutzung der Lager zu rechnen und dann sollte die Reinigung wenig Mühe machen. Meine BB-Pellets tausche ich öfter, weil sie rosten, wenn die Messing-Schicht abgenutzt ist.
this is a really cool idea to save quite a bit of money on prototypes, large bearing can get expensive fast, now i can print them and test if my ideas work before commiting
i really like this approach. When making mine a couple years ago the flat sections were printed diagonally which left a slughtly bumpy surface. to fix it I added a very narrow and shallow groove where the balls made contact. it still fused as a flat surface but the print path was concentric with the bearing. it worked much better after that.
This sounds interesting, but I'm not quite able to picture it. Was the modification you're describing take place entirely in the sketch for the race revolution?
Great content, reall nice to see a custom created bearings rather than being tied to standard offerings, would be great to see more testing at speed or different loads applied
Well, I think I spent about two weeks to figure it out and refine the design. It wasn't the first idea I had. So it was not THAT easy, the video just doesn't show that :)
Very well explained and definitely better suited (and cheaper) for 3D printed applications that do not require the thousands of Newton and tens of thousands of RPM that even the smallest of steel bearings can deliver. A significant advantage is that you can design the bearing as part of your machine. Steel bearings are designed to be fitted in metal seats and on metal axes. I just subscribed. Cya !
I like where this project is going, thanks for sharing the design. As an engineer I was quickly making some calculations and changes in my mind while you were showing the 3D prints: 36 balls with 72 kg of load is 2 kg per ball and with a flat surface, the contact area of ball and surface is a single point of contact which is an infinite small area which results in an infinite high force on the plastic race. As long as you rotate the bearing, this will even out along the full circular contact area but as soon as you stop moving, things change. In time (and temperature) you will see that the balls create small dents at a certain location. Will this be a problem ? Yes, definitely. When will this become a problem ? It all depends and we will only know when we try this. 15 years ago we also stated that using PLA for 3D printing was a bad idea since it is bio degradable but we still do and printed stuff holds a very long time.
Interesting point. Generally I think that these bearings should not be used with high load. Deformation under continuous load will 100% be a problem and AFAIK all regular filaments has this problem, only special stuff like CF-nylon or PC can handle that. But these filaments quite hard to print and bearings should be printed with decent quality.
Deformation is key. Temporary deformation of the ball and the plastic are simply part of how the bearing works. If I were to go about calculating deformation, I would assume that all the deformation happens in the plastic, and go from there. If the balls were also plastic, it would be harder to calculate, but something tells me it ought to work better. Do the vendors of balls provide a recommended force per ball? If so, I might optimize my design by finding a ball material with stiffness close to my filament choice, then go with the ball loading recommendation. Permanent deformation to smooth out the races sounds like a good thing. I wonder if I could get the permanent deformation I want, but avoid the deformation I don't. My first thought is to load the bearing with a weight, then get it spinning fast enough to generate the heat to just oh so slightly slightly soften the races. Then my guess is removing most of the load while keeping the bearing spinning would allow the plastic to cool while keeping the smoothed shape. I don't know if that would work, but if it did, it would take precise control of the recipe. Time and temperature have to be just right, just like cooking. Maybe add some lubricant at the right moment to cool rapidly. Possibly cook with one kind of balls (steel or aluminum maybe), and then switch the balls out for use (plastic). I wish I had the time to do all that experimentation.
If you go with 6mm, you can use plastic BBs instead of bearings. Generally good for light loads outdoors like a lazy susan, so the bearings don't rust.
Very neat video, has me thinking about all sorts of designs. I would imagine the greatest drawback of your current design would be contact forces of the ball bearings deforming/wearing down along their contact paths. I wonder about adding an internal raceway's, perhaps just an O.D. and I.D. to correspond with the balls rolling contact plane, or more to further constrain them. Perhaps with a strip of something with high hardness, and small tolerance, mechanical tubing for instance. Could cut out thin slivers (thick enough to minimally deform from contact stresses, and wide enough to help distribute the normal force along the plastic face) that fit into notches. Basically use 3D printed plastic everywhere except the balls themselves, and small metal or high hardness plastic rings positioned where the balls would contact.
Nice to see someone else doing this. I printed a thrust bearing to hold my 5kg spools horizontal for my printer. Basically just a bicycle shaped cage to hold 6 metal balls in a ring. I used some spare Plexiglas as the top and bottom to get a smoother and harder raceway, but it would probably work fine if I printed those too.
I really like the design, it gave me a place to start. The only issue I am working on now is that I need to press fit it in an axel through the hole, so will need to figure a different way to attach the parts. I will likely make the center part a single body instead of the outer ring and figure it out from there (as there is more room in the center of the ring
Preload! Your "radial load" bearings are similar to angular contact ball bearings and those perform best with and because of preload. Simplest modification would be to slightly reduce the height of inner race so there's a small amount of (axial) preload when the screws are fully tightened. However 1) this will mean that the balls will start to mainly ride on the 45 degree secondary races and that's not ideal if loads are mostly radial and 2) there will still be some play from the balls sliding laterally on the secondary race. One 'ideal' solution would be to have the balls ride on both primary and secondary races. To achieve that, first make the primary race not vertical but slighly 'open' by -5 to -15 degrees (visual context at 5:10) and then preload can ensure the dual contact on both inner and outer races (though admittedly only if the print is to very close to dimension). More load bearing contact more better. The end result will be very similar to angular contact ball bearings with a contact angle of 25 to 30 degrees (if on dimension and preload leads to dual contact or 5 to 15 otherwise) and the play will be reduced to close to zero (plastic not being rigid). The other 'ideal' option is to simply copy the angular contact bearing design so instead of main+auxiliary race you have a round race with a slightly bigger radius than the bearing balls you're using. The best way to achieve that is to just fillet the above design since that ensures no overhangs past 45. Also use some lubricant (reduces noise and cage friction).
Wow, thanks for taking your time to write such a comment! Yeah, I read about 4-points angular contact bearing, but didn't know that it is usually preloaded, I learned something :) It's probably the best option if very little play reqiered. Maybe will experiment with this. I also realized that I can just do a deep groove ball bearing with Conrad-style assembly :) I did not talk about it in the video because I realized this while making the video, but I show it at the very last seconds. I think it's the most simple design. I just was not aware about the Conrad assembly and thought it will be impossible to make :)
Thanks for the added info! I’d assume you want the bare minimum amount of preload to take up the slack without brinelling the soft plastic races. That seems like a narrow margin to hit? For the second idea, is your suggested as simple as filleting the inner 45° corner with a slightly larger radius with no other design changes? Or does it including adding preload as well?
@@williammartin9751 It's narrow but not that narrow of a margin because plastic has higher elasticity. But good to bring up brinelling because it is a concern since plastic creeps. Therefore all plastic bearings should be 1) run-in till the races are fully burnished (i'd say preload-only at first then under constant load) 2) not kept under significant static load for long periods. Second question: Yes, the design of the races is as simple as that. And with that design, adding a low preload becomes as simple as slightly reducing the height of the inner bearing races. I keep talking about preload because that's what results in the largest performance improvement.
This is a very interesting video. Thanks for sharing. Seeing how it's designed helps a lot when trying to understand it. But there is something I don't understand. Let's talk in 2D which I think makes things simpler. A circle can contact two straight lines and roll between them without sliding. However, each circle (ball) is in contact with 4 lines in your bearing. Doesn't that imply sliding no matter what? For example, see cross sections at 4:35 explaining the auxiliary races.
Hmm. I think it will still mostly roll even with 4 points of contact. It will just roll in a way that it's rotation axis oriented somewhere in between. Hard to put it into words :) Anyway, bearing design with 4 points of contact is very common. Look for "4-point angular contact ball bearing" for example. So it is a bit less efficient, but mostly because of some extra angle between load and contact point. I don't think that there will be significant sliding. In fact even if you roll a ball between two parallel flat surfaces, the rolling axis doesn't have to be parallel to these surfaces, it could be tilted and the ball will still roll just fine. I think it creates some extra friction, but it is not sliding. This extra friction will be similar to friction that "spinning top" has, just a little bit of that. Hope it makes sense, it's so hard to explain without showing/drawing. Thanks for your comment :)
Hi, I'm having a bit of an issue fitting the ball bearings. I miss heard you in the video, I thought you said 4.5 mm balls but you were talking about the size of the holes in the cage. So.. my question is, What size balls did you use? Thanks in advance.
I refuse to press like because brass inserts look cool in white plastic (even though they do). I pressed like because this is a cool video that taught me a lot of interesting things.
Cool work. I wonder if a design that doesn't use a bearing cage but instead closely packed bearings would get rid of the play in the bear you show at ~11 minutes? This is what's used in bicycle hubs and headsets, works very well.
I tried with no cage. Too heavy, rattle a lot. And it doesn't fix the play. That play actually is by design. The ball diameter in the CAD was 4.1mm instead of 4mm, so 0.2mm play is expected. It's not too much I think, I am fine with that in my rover. Actually it is possible to make the gap smaller, but it requires a well-adjusted printer that prints really well. Or there is an other option. Just print with no gap at all and assemble. It will not spin freely first, but if you forcefully spin it with some load for some time it will wear-off all roughness and it will be perfect bearing with no play.
In my 3d printed bearings, I use plastic balls for toy guns. I make them so I can put as many balls as is bossible in the outer race. I then whack the inner race with a hammer to suprise the plastic only 1 time. They fall together and run quite smoothly. I've used them in a few projects and they seem to be working as none of them have failed.. I use no bearing cage.
How is the preload in the end? How much play (or on the other hand, how much stiffness) do your bearings end up with? Can you recommend some interference dimensions?
Very cool video, definetely going to try and make a bearing with some bearing balls I have left from from broken drawer rails. Question - why not insert the balls mid-print? Kinda the same way people embed nuts and magnets into prints. Reduces the likelyhood of damaging the cage while filing it with balls.
I was thinking about it, but did not figure out how it could be done. Every time I was thinking about a way to do that I realized that I need to print unsupported surfaces and also the extruder is not able to print really close to the ball due to it's shape. It works for nuts because they are basicly flat objects so you can insert them at the point where there is enough depth to put it completely inside. You can not do it with the ball I think.
@@Positive_Altitude what about just making the clearance between the ball and the frame as minimal as possible (so like 4.05 instead of 4.1 for a 4mm ball) and then just leave the unsupported overhangs as is? Basically just let the overhangs print over the balls and then break them off after the print is cooled off. But now that I'm typing this all out I'm starting to think that maybe this approach won't decrease the likelihood of damaging the part at all.
The smallest I made has inner/outer races diameter = 20mm/28mm . Works absolutely fine, I would say maybe 10mm/18mm is the smallest you can reasonably have with 4mm balls.
Cool idea, but how long do they last? and what loads can you apply? and what material is best? but it can always hold for small stuff.. okay i have to do some testing :-) and can you create a pressure greased bushing for longer lifespans and load??
I show a simple load test in the video. I think 4kg per ball for thrust bearing and maybe around 1kg per ball for radial bearing (because most balls are not loaded in this type of bearing) I will do the proper load and wear test in the future. Materials... I personally don't like PLA for mechanical parts. And everything else except PETG could be challenging to print with enough quality and precision. Ofc nylon will be better than PETG, but it's much harder to print. So personally I am planning to work only with PETG right now. Don't know anything about "pressure greased bushing" so can't tell. However, I want to try to make a flexible TPU seal to protect the bearing from water and dust. I have high hopes that this will work.
@@Positive_Altitude Pressure greased bushing is a channel through which you can inject grease into the bearing every so often to lubricate it. Not sure how you would want to lubricate a plastic bearing, though, or which benefits to expect from adding grease.
@@Positive_AltitudeYeah, the low softening temperature of PLA probably means they would deform under lighter load conditions than PETG would... I think. Really more of a hypothesis. As for Nylon, I wonder if its flexibility would cause problems with bearing performance. Might be worth a test. I might hypothesize that PC might be a higher performing material... I suppose I would have to have a good definition of "performance" before I make statements like that 😅
Is there any way to completely remove backlash on a 3d printed bearings? Also your bearings looks bigger then typical manufactured bearings, anyway your way to design a bearing is awesome, I'll try to make my own based on your design, thanks!
To get the lowest backlash you can try to print it without any added gaps. It will be quite tight when assembled, but you can "work it out". Just apply some load in altering directions and forcefully rotate. It will become much nicer in a few minutes. I tried it once, and it worked for me. But make sure that your printer is well aligned. XY-skew should be very very low. Also the smallest bearing of similar design that I made was 20mm/28mm inner/outer races diameter it was fine, maybe it could be even smaller
Видел у какого то ютубера, он делал вставки из металлической проволоки под шарики. Вставлял металлические кольца изготовленные из ровного прутка, делал что то на подобии рельс для шариков сверху и снизу. Это позволяло выдерживать бОльшие нагрузки без деформации пластика стальными шариками.
I tried printing without margin at all. It starts too tight because of printing imperfections but works well after some time. I think you can absolutely do that and it will be the way to achieve minimal possible play 👍
Have you considered embedding the bearings, by pausing the printhead at the correct layer and then placing the carriage inside, before continuing the print?
What is the goal? To reduce the need for the screws? You can split the part and superglue them together instead. Or you can make them snap/lock together. Serviceability has lot so benefits too, especially when making prototypes like the rover.
I was thinking about it but I don't think it's possible. The extruder will collide with balls and I can't think of the design or method to avoid it. Also in real applications I use these bearings as an attachment point. It's like outer races printed as a part of component A and inner races are printed as a part of component B. So you assemble A and B and the bearing at the same time.
have you considered using plastic bbs to have similar hardness materials contacting instead of metal and plastic which will probably wear down the plastic very fast?
This is cool but I imagine there's very few applications where it makes sense. Sure quality industrial bearing can get expensive but common grade bearing are pretty cheap, even with diameters around 60 or 70mm, we're talking like $10 - and even cheaper Chinese bearings will have better tolerances and radial play than your 3D printed bearings. This is useful if you need something within in a few hours, or maybe for very specific weird geometry, but I don't see much other value.
I dunno, I’m not sure anyone’s going to be making front-line tools or consumer products with these, but the advantages for small home projects, teaching, and prototyping proofs of concept are pretty neat. Personally I don’t keep a lot of bearings in stock at home, I rarely need them, and when I do it’s always a different type to that which I have. Having a few 6mm ball bearings lying around to help me out is very flexible and very cheap.
Hello, what a great project. I am an architecture student and I would love to be able to print this project. I think it can be used for the filament spool, what do you think? Is it possible for me to have the stl file? because from the onshape program everything is together and I don't know how to modify it
You can create OnShape account and then you will be able to clone this design and get your own editable copy. There you can take a look at all features, modify or export as STL. Let me know if you have issues with that :)
@@Positive_Altitude I was trying but I can't separate it into pieces, it's a lot harder than I thought. There is a way for you to upload a link to download it for each piece. I think it's a super interesting piece.
@@andresescoda4768 I updated the description. I placed STLs in a new GitHub repo. There is a green button that you can use to download all as a .zip archive :)
You can make the play in the angluar contact bearing adjustable by adding a gap in between the two inner races. Use a rubber O-ring or a piece of TPU as a kind of spring allows for easy adjustability
I think if you have well-calibrated filament (flow rate, pressure advance, retracts, etc) even the regular seams could be quite nice. I tried scarf seams -- not a big difference, so I don't do it right now.
Never heard about it. Anyway, the hotend replacement is not such a big deal and it will be required anyway eventually. I just need the rover to be white. I am going to place it outside in sunny days and white will heat up less.
The left and right side are connected through a differential mechanism. Mars rovers usually use differential bars, but I used differential gears instead. So I have a differential inside (like in car transmission) and the body is connected to this differential through a worm gear. Here I am spinning the worm gear by hand and tilt the body relative to the differential. I will make an overview of this suspension in one of my next videos. :)
@@Positive_Altitude At 4:57 the surfaces in contact with balls meet at angle, not rounded. Is it intentional? I would expect the contact surface to follow the shape of the ball, i.e. be round
@@positron96if there will be grooves it will be not possible to assemble this way, because the cage with balls will not fit in the gap any more. Though it is possible to add half-grooves on opposite races. I mean the corner between main and auxiliary races could be rounded.
This timing is violating my privacy lmaooo I was about to dive in myself into topic since I need cusom bearing for diy steering wheel with arduino, need rather uncommon diameter
I did not try, but I doubt that it will make a big difference. Grooves appears when the material can not physically handle such force and starts to deform. I think the bearing just should not be used with such load. Maybe nylon will increase the maximum load a bit, but at the same time when both balls and races deforms that will increase friction. Ideally all parts of bearing should be as hard as possible (hardened steel / ceramic) because it reduces deformation and friction. Our races are quite soft, but at least we can have proper balls ... that's what I think.
Very cool! Another tip is that you want to have the walls as thick as possible so that the slicer will not print unneeded lines that could slow the bearing down. For overkill when printing bearings I set the wall thickness to 3.0 mm so that the ball bearing can roll around the grove of the print and not hit cross lines.
I experimented with this design, I made the design in Fusion 360, these were the changes I made. I minimised the clearance between the diagonal 45 degree contact surface of the inner and outer race to zero. You can adjust the pre-tension on the ball by the screws. For a smoother surface I separated the inner and outer race and printed them in vase mode and sequential printing and set the extrusion width to 1 mm, layer height 0.15 mm. I changed the print orientation 180 degrees, so the wide part was down. This minimised the deformation during cooling, 45 degrees angle prints like to curl up during printing. The inner and outer races press fit in the rest of the design, you use a more wear resistant filament for this. You could also machine them is you have a lathe. The results are very promising, there is no play in the bearing. The bearings are silent, you can still feel the balls skipping to the next layer, which are at a slight angle in vase mode. I did this first test with 3mm balls, I didn’t have any 4mm balls. I will try 5mm next.
@@chipcode5538 wow, interesting. My goal was to make a simple design with as few parts as possible to make assembly easier. But it's very cool that you are exploring ways to increase quality. Cheers!
I see to remember that it's a Good Idea to use an odd number of bearing balls. I think it distributes the load better if there are imperfections in balls or races.
Much like a three-legged or five-legged chair is more stable than four or six legs on an uneven floor.
Great that you mentioned this, my next design will be Odd. Yes, the indentation on the Crown Cork is also odd.
I think odd numbers have more to do with reducing harmonics.
three-legged chair don't wobble on uneven floors. 4, 5 or 6 legs will always wobble. If you have an even floor, more legs means more stability. Doesn't have anything to do with odd or even. (-:
Nice work. BTW, I've found that you don't need threaded inserts in situations where you aren't taking the screws out that often. Just thread direct into the plastic.
True. Not really needed here, I just like how they looks and enjoy working with them :)
Besonders bei einem Fahrzeug ist aber mit Verschmutzung der Lager zu rechnen und dann sollte die Reinigung wenig Mühe machen. Meine BB-Pellets tausche ich öfter, weil sie rosten, wenn die Messing-Schicht abgenutzt ist.
One of the best videos I’ve seen on 3d printed bearing
this is a really cool idea to save quite a bit of money on prototypes, large bearing can get expensive fast, now i can print them and test if my ideas work before commiting
Thanks! Will try to design them myself! Appreciate that you included the dimensions, will save me a ton of time testing things out!
i really like this approach. When making mine a couple years ago the flat sections were printed diagonally which left a slughtly bumpy surface. to fix it I added a very narrow and shallow groove where the balls made contact. it still fused as a flat surface but the print path was concentric with the bearing. it worked much better after that.
This sounds interesting, but I'm not quite able to picture it. Was the modification you're describing take place entirely in the sketch for the race revolution?
I really like your visual diagrams made with Onshape. Clever use of the tool.
Great content, reall nice to see a custom created bearings rather than being tied to standard offerings, would be great to see more testing at speed or different loads applied
You all make it look so easy. Been struggling for weeks on this simple concept. Thank you for the video.
Well, I think I spent about two weeks to figure it out and refine the design. It wasn't the first idea I had. So it was not THAT easy, the video just doesn't show that :)
Very well explained and definitely better suited (and cheaper) for 3D printed applications that do not require the thousands of Newton and tens of thousands of RPM that even the smallest of steel bearings can deliver. A significant advantage is that you can design the bearing as part of your machine. Steel bearings are designed to be fitted in metal seats and on metal axes. I just subscribed. Cya !
100%, they will never be better than metal bearings, but sometimes we just need things to be simple and "good enough"
I like where this project is going, thanks for sharing the design.
As an engineer I was quickly making some calculations and changes in my mind while you were showing the 3D prints: 36 balls with 72 kg of load is 2 kg per ball and with a flat surface, the contact area of ball and surface is a single point of contact which is an infinite small area which results in an infinite high force on the plastic race. As long as you rotate the bearing, this will even out along the full circular contact area but as soon as you stop moving, things change. In time (and temperature) you will see that the balls create small dents at a certain location.
Will this be a problem ? Yes, definitely. When will this become a problem ? It all depends and we will only know when we try this.
15 years ago we also stated that using PLA for 3D printing was a bad idea since it is bio degradable but we still do and printed stuff holds a very long time.
Interesting point. Generally I think that these bearings should not be used with high load. Deformation under continuous load will 100% be a problem and AFAIK all regular filaments has this problem, only special stuff like CF-nylon or PC can handle that. But these filaments quite hard to print and bearings should be printed with decent quality.
Deformation is key. Temporary deformation of the ball and the plastic are simply part of how the bearing works. If I were to go about calculating deformation, I would assume that all the deformation happens in the plastic, and go from there.
If the balls were also plastic, it would be harder to calculate, but something tells me it ought to work better. Do the vendors of balls provide a recommended force per ball? If so, I might optimize my design by finding a ball material with stiffness close to my filament choice, then go with the ball loading recommendation.
Permanent deformation to smooth out the races sounds like a good thing. I wonder if I could get the permanent deformation I want, but avoid the deformation I don't.
My first thought is to load the bearing with a weight, then get it spinning fast enough to generate the heat to just oh so slightly slightly soften the races. Then my guess is removing most of the load while keeping the bearing spinning would allow the plastic to cool while keeping the smoothed shape. I don't know if that would work, but if it did, it would take precise control of the recipe. Time and temperature have to be just right, just like cooking. Maybe add some lubricant at the right moment to cool rapidly. Possibly cook with one kind of balls (steel or aluminum maybe), and then switch the balls out for use (plastic).
I wish I had the time to do all that experimentation.
If you go with 6mm, you can use plastic BBs instead of bearings. Generally good for light loads outdoors like a lazy susan, so the bearings don't rust.
I really appreciate this type of content on TH-cam. Properly informative, even for a newbie like me 😂
Very neat video, has me thinking about all sorts of designs. I would imagine the greatest drawback of your current design would be contact forces of the ball bearings deforming/wearing down along their contact paths.
I wonder about adding an internal raceway's, perhaps just an O.D. and I.D. to correspond with the balls rolling contact plane, or more to further constrain them. Perhaps with a strip of something with high hardness, and small tolerance, mechanical tubing for instance. Could cut out thin slivers (thick enough to minimally deform from contact stresses, and wide enough to help distribute the normal force along the plastic face) that fit into notches. Basically use 3D printed plastic everywhere except the balls themselves, and small metal or high hardness plastic rings positioned where the balls would contact.
Nice to see someone else doing this. I printed a thrust bearing to hold my 5kg spools horizontal for my printer. Basically just a bicycle shaped cage to hold 6 metal balls in a ring. I used some spare Plexiglas as the top and bottom to get a smoother and harder raceway, but it would probably work fine if I printed those too.
Thank you. The information about permitted load made this already informative video 2 times better!!
Very interesting! Cool design and great graphics. ❤
This is such a good design and such a good tutorial! Thank you for this!
Nice video, It's really awesome to see the rover coming together!
Good job! Thank you for sharing your experience!
I was also thinking of making my own bearings some time ago, but never done it. Nice job, it looks great!
I really like the design, it gave me a place to start. The only issue I am working on now is that I need to press fit it in an axel through the hole, so will need to figure a different way to attach the parts. I will likely make the center part a single body instead of the outer ring and figure it out from there (as there is more room in the center of the ring
Nvm, I am stupid, just going to make the inner part thicker and the screw holes will need to go all the way
Preload! Your "radial load" bearings are similar to angular contact ball bearings and those perform best with and because of preload. Simplest modification would be to slightly reduce the height of inner race so there's a small amount of (axial) preload when the screws are fully tightened. However 1) this will mean that the balls will start to mainly ride on the 45 degree secondary races and that's not ideal if loads are mostly radial and 2) there will still be some play from the balls sliding laterally on the secondary race. One 'ideal' solution would be to have the balls ride on both primary and secondary races. To achieve that, first make the primary race not vertical but slighly 'open' by -5 to -15 degrees (visual context at 5:10) and then preload can ensure the dual contact on both inner and outer races (though admittedly only if the print is to very close to dimension). More load bearing contact more better. The end result will be very similar to angular contact ball bearings with a contact angle of 25 to 30 degrees (if on dimension and preload leads to dual contact or 5 to 15 otherwise) and the play will be reduced to close to zero (plastic not being rigid).
The other 'ideal' option is to simply copy the angular contact bearing design so instead of main+auxiliary race you have a round race with a slightly bigger radius than the bearing balls you're using. The best way to achieve that is to just fillet the above design since that ensures no overhangs past 45.
Also use some lubricant (reduces noise and cage friction).
Wow, thanks for taking your time to write such a comment! Yeah, I read about 4-points angular contact bearing, but didn't know that it is usually preloaded, I learned something :) It's probably the best option if very little play reqiered. Maybe will experiment with this.
I also realized that I can just do a deep groove ball bearing with Conrad-style assembly :) I did not talk about it in the video because I realized this while making the video, but I show it at the very last seconds. I think it's the most simple design. I just was not aware about the Conrad assembly and thought it will be impossible to make :)
Thanks for the added info! I’d assume you want the bare minimum amount of preload to take up the slack without brinelling the soft plastic races. That seems like a narrow margin to hit?
For the second idea, is your suggested as simple as filleting the inner 45° corner with a slightly larger radius with no other design changes? Or does it including adding preload as well?
@@williammartin9751 It's narrow but not that narrow of a margin because plastic has higher elasticity. But good to bring up brinelling because it is a concern since plastic creeps. Therefore all plastic bearings should be 1) run-in till the races are fully burnished (i'd say preload-only at first then under constant load) 2) not kept under significant static load for long periods.
Second question: Yes, the design of the races is as simple as that. And with that design, adding a low preload becomes as simple as slightly reducing the height of the inner bearing races. I keep talking about preload because that's what results in the largest performance improvement.
Super nice design!
Your bearings are beautiful!
This is a very interesting video. Thanks for sharing. Seeing how it's designed helps a lot when trying to understand it.
But there is something I don't understand. Let's talk in 2D which I think makes things simpler. A circle can contact two straight lines and roll between them without sliding.
However, each circle (ball) is in contact with 4 lines in your bearing. Doesn't that imply sliding no matter what?
For example, see cross sections at 4:35 explaining the auxiliary races.
Hmm. I think it will still mostly roll even with 4 points of contact. It will just roll in a way that it's rotation axis oriented somewhere in between. Hard to put it into words :) Anyway, bearing design with 4 points of contact is very common. Look for "4-point angular contact ball bearing" for example. So it is a bit less efficient, but mostly because of some extra angle between load and contact point. I don't think that there will be significant sliding. In fact even if you roll a ball between two parallel flat surfaces, the rolling axis doesn't have to be parallel to these surfaces, it could be tilted and the ball will still roll just fine. I think it creates some extra friction, but it is not sliding. This extra friction will be similar to friction that "spinning top" has, just a little bit of that. Hope it makes sense, it's so hard to explain without showing/drawing. Thanks for your comment :)
Good job. I'm waiting for improvements and testing.
This is really cool and well explained! Please do more!
Good tutorial! These would be cool to make, now I just need a suitable project
Great design idea.
Thank you for sharing your wealth of knowledge. Ill be playing around with this .
😃
Hi, I'm having a bit of an issue fitting the ball bearings. I miss heard you in the video, I thought you said 4.5 mm balls but you were talking about the size of the holes in the cage. So.. my question is, What size balls did you use? Thanks in advance.
I refuse to press like because brass inserts look cool in white plastic (even though they do). I pressed like because this is a cool video that taught me a lot of interesting things.
I am happy that is was useful for you, cheers!
I refuse to press like Lol.. what a guy.
@@kbruin79 ah, but I did press like. I just won't be coerced into it. Still a really cool video and deserves the like.
Cool work.
I wonder if a design that doesn't use a bearing cage but instead closely packed bearings would get rid of the play in the bear you show at ~11 minutes? This is what's used in bicycle hubs and headsets, works very well.
I tried with no cage. Too heavy, rattle a lot. And it doesn't fix the play. That play actually is by design. The ball diameter in the CAD was 4.1mm instead of 4mm, so 0.2mm play is expected. It's not too much I think, I am fine with that in my rover. Actually it is possible to make the gap smaller, but it requires a well-adjusted printer that prints really well. Or there is an other option. Just print with no gap at all and assemble. It will not spin freely first, but if you forcefully spin it with some load for some time it will wear-off all roughness and it will be perfect bearing with no play.
This is EXACTLY what I needed!
This is amazing. Good work and soooo useful
In my 3d printed bearings, I use plastic balls for toy guns. I make them so I can put as many balls as is bossible in the outer race. I then whack the inner race with a hammer to suprise the plastic only 1 time. They fall together and run quite smoothly. I've used them in a few projects and they seem to be working as none of them have failed.. I use no bearing cage.
How is the preload in the end? How much play (or on the other hand, how much stiffness) do your bearings end up with?
Can you recommend some interference dimensions?
Good idea and well produced video, thanks.
superb video. i think i'll make quite a few of these. very interesting ideas. thank you!
What a great video! Really enjoyed.
“It’s all ball bearings nowadays…” -Fletch.😂
Very nice designs. Have you every tried adding a lubricant to the bearings?
Very cool video, definetely going to try and make a bearing with some bearing balls I have left from from broken drawer rails. Question - why not insert the balls mid-print? Kinda the same way people embed nuts and magnets into prints. Reduces the likelyhood of damaging the cage while filing it with balls.
I was thinking about it, but did not figure out how it could be done. Every time I was thinking about a way to do that I realized that I need to print unsupported surfaces and also the extruder is not able to print really close to the ball due to it's shape. It works for nuts because they are basicly flat objects so you can insert them at the point where there is enough depth to put it completely inside. You can not do it with the ball I think.
@@Positive_Altitude what about just making the clearance between the ball and the frame as minimal as possible (so like 4.05 instead of 4.1 for a 4mm ball) and then just leave the unsupported overhangs as is? Basically just let the overhangs print over the balls and then break them off after the print is cooled off. But now that I'm typing this all out I'm starting to think that maybe this approach won't decrease the likelihood of damaging the part at all.
Amazing 🤩 love this keep it up 👍
That's so awesome. Thanks for sharing. :D
Thank you sooooooo much for this. You are AWESOME!!!!!!!!!!
Great video, well done.
This video is excellent for learning TY :)
Very nice. Glad I stumbled onto this. Now we can make bearings of pretty much any size. How small can you go with this design?
The smallest I made has inner/outer races diameter = 20mm/28mm . Works absolutely fine, I would say maybe 10mm/18mm is the smallest you can reasonably have with 4mm balls.
Great work!
GODDAMN BOY IVE JUST SEARCHED THIS KIND OF VIDEO TWO DAYS AGO!
Great video!! Thank you!!!
I've heard that some people use airsoft bbs for bearings. Sounds fun and tbh should work. Tolerances aren't that crucial if it is made of plastic
Yeah, I think it will be good too. Actually it could be the same thing but significantly lighter. I will try it out.
Cool idea, but how long do they last? and what loads can you apply? and what material is best? but it can always hold for small stuff.. okay i have to do some testing :-) and can you create a pressure greased bushing for longer lifespans and load??
I show a simple load test in the video. I think 4kg per ball for thrust bearing and maybe around 1kg per ball for radial bearing (because most balls are not loaded in this type of bearing) I will do the proper load and wear test in the future.
Materials... I personally don't like PLA for mechanical parts. And everything else except PETG could be challenging to print with enough quality and precision. Ofc nylon will be better than PETG, but it's much harder to print. So personally I am planning to work only with PETG right now.
Don't know anything about "pressure greased bushing" so can't tell. However, I want to try to make a flexible TPU seal to protect the bearing from water and dust. I have high hopes that this will work.
@@Positive_Altitude Pressure greased bushing is a channel through which you can inject grease into the bearing every so often to lubricate it. Not sure how you would want to lubricate a plastic bearing, though, or which benefits to expect from adding grease.
@@Positive_AltitudeYeah, the low softening temperature of PLA probably means they would deform under lighter load conditions than PETG would... I think. Really more of a hypothesis.
As for Nylon, I wonder if its flexibility would cause problems with bearing performance. Might be worth a test.
I might hypothesize that PC might be a higher performing material... I suppose I would have to have a good definition of "performance" before I make statements like that 😅
Is there any way to completely remove backlash on a 3d printed bearings? Also your bearings looks bigger then typical manufactured bearings, anyway your way to design a bearing is awesome, I'll try to make my own based on your design, thanks!
To get the lowest backlash you can try to print it without any added gaps. It will be quite tight when assembled, but you can "work it out". Just apply some load in altering directions and forcefully rotate. It will become much nicer in a few minutes. I tried it once, and it worked for me. But make sure that your printer is well aligned. XY-skew should be very very low. Also the smallest bearing of similar design that I made was 20mm/28mm inner/outer races diameter it was fine, maybe it could be even smaller
Thanks, this is helpful!
Видел у какого то ютубера, он делал вставки из металлической проволоки под шарики. Вставлял металлические кольца изготовленные из ровного прутка, делал что то на подобии рельс для шариков сверху и снизу. Это позволяло выдерживать бОльшие нагрузки без деформации пластика стальными шариками.
Интересно, но наверное сделать такое аккуратно довольно сложно.
What do you think about designing your races slightly too tight and letting the balls form smoother, more precise races after a break-in period
I tried printing without margin at all. It starts too tight because of printing imperfections but works well after some time. I think you can absolutely do that and it will be the way to achieve minimal possible play 👍
Отличное видео! А где взять шарики для подшипников?
Спасибо! Можно на амазоне или алиэкспрессе "steel ball bearings" пластиковые тоже можно посмотреть
Have you considered embedding the bearings, by pausing the printhead at the correct layer and then placing the carriage inside, before continuing the print?
What is the goal? To reduce the need for the screws? You can split the part and superglue them together instead. Or you can make them snap/lock together.
Serviceability has lot so benefits too, especially when making prototypes like the rover.
I was thinking about it but I don't think it's possible. The extruder will collide with balls and I can't think of the design or method to avoid it. Also in real applications I use these bearings as an attachment point. It's like outer races printed as a part of component A and inner races are printed as a part of component B. So you assemble A and B and the bearing at the same time.
I love your projects.
Do you work on your projects full time?
How do you earn money?
Thanks! No, I have a full time job as a software developer, write some stuff for ad-tech in Scala 🙃
Maybe you could try a 3dhojor filament.😃
Sehr gut - THX
have you considered using plastic bbs to have similar hardness materials contacting instead of metal and plastic which will probably wear down the plastic very fast?
I think that's need to be tested. I wounder if there will be any difference. I will test it eventually :)
This is cool but I imagine there's very few applications where it makes sense. Sure quality industrial bearing can get expensive but common grade bearing are pretty cheap, even with diameters around 60 or 70mm, we're talking like $10 - and even cheaper Chinese bearings will have better tolerances and radial play than your 3D printed bearings. This is useful if you need something within in a few hours, or maybe for very specific weird geometry, but I don't see much other value.
That's true. But I just don't want to order bearings when I need them. And also I can go with any size I want. It's just to make the process easier.
I dunno, I’m not sure anyone’s going to be making front-line tools or consumer products with these, but the advantages for small home projects, teaching, and prototyping proofs of concept are pretty neat. Personally I don’t keep a lot of bearings in stock at home, I rarely need them, and when I do it’s always a different type to that which I have. Having a few 6mm ball bearings lying around to help me out is very flexible and very cheap.
Hello, what a great project. I am an architecture student and I would love to be able to print this project. I think it can be used for the filament spool, what do you think? Is it possible for me to have the stl file? because from the onshape program everything is together and I don't know how to modify it
You can create OnShape account and then you will be able to clone this design and get your own editable copy. There you can take a look at all features, modify or export as STL. Let me know if you have issues with that :)
@@Positive_Altitude I was trying but I can't separate it into pieces, it's a lot harder than I thought. There is a way for you to upload a link to download it for each piece. I think it's a super interesting piece.
@@andresescoda4768 I updated the description. I placed STLs in a new GitHub repo. There is a green button that you can use to download all as a .zip archive :)
You can make the play in the angluar contact bearing adjustable by adding a gap in between the two inner races. Use a rubber O-ring or a piece of TPU as a kind of spring allows for easy adjustability
That's an interesting idea, thanks :)
ze bearing iz nice
Perfeito!
do scraf seams make a difference or is random seam placement enough?
I think if you have well-calibrated filament (flow rate, pressure advance, retracts, etc) even the regular seams could be quite nice. I tried scarf seams -- not a big difference, so I don't do it right now.
not sure how true or significant, but i heard white filament wears your printer faster due to the white dye containing titanium.
Never heard about it. Anyway, the hotend replacement is not such a big deal and it will be required anyway eventually. I just need the rover to be white. I am going to place it outside in sunny days and white will heat up less.
You should add a little gap between the bolting surfaces. Then you could adjust the pre load by printing different size shim spacers.
Or your choice of spring washer.
Well done. Where do you purchase the balls in quantity?
Amazon works for me!
@@captaincognizant73 thank you 🙏
Yeah, Amazon or Aliexpress
@@Positive_Altitude - thank you 🙏
how you are able to tilt the chassis of the rover at the beginning of the video
The left and right side are connected through a differential mechanism. Mars rovers usually use differential bars, but I used differential gears instead. So I have a differential inside (like in car transmission) and the body is connected to this differential through a worm gear. Here I am spinning the worm gear by hand and tilt the body relative to the differential. I will make an overview of this suspension in one of my next videos. :)
👍
zis ball bearinz zat i uzed in ze rover, zere are many of zem hizing in plain zight
:)
Very sick video. Bro even gives away the design files for free. How can you not subscribe? 😮
Why didn't you chamfer the contact surfaces?
Can you explain what do you mean exactly?
@@Positive_Altitude At 4:57 the surfaces in contact with balls meet at angle, not rounded. Is it intentional? I would expect the contact surface to follow the shape of the ball, i.e. be round
@@positron96if there will be grooves it will be not possible to assemble this way, because the cage with balls will not fit in the gap any more. Though it is possible to add half-grooves on opposite races. I mean the corner between main and auxiliary races could be rounded.
@@Positive_Altitude yes, I think I mean rounding the corners
This timing is violating my privacy lmaooo
I was about to dive in myself into topic since I need cusom bearing for diy steering wheel with arduino, need rather uncommon diameter
Lol. This happens to me all the time too ;)
@@Positive_Altitude ehehehe
but thanks a bunch, now i have a really solid guideline on what I'll need to make and how! Much appreciated ❤️
@@Positive_Altitude Do you have an email? I want ask you for help to make a project. want to make a rotary joint for the lab aminal.
@@uestcwcq Sure! I added an email address to the channel info.
Someone is using the correct measuring units are rare these days.
wouldnt using nylon balls be better so it doesnt make grooves?
I did not try, but I doubt that it will make a big difference. Grooves appears when the material can not physically handle such force and starts to deform. I think the bearing just should not be used with such load. Maybe nylon will increase the maximum load a bit, but at the same time when both balls and races deforms that will increase friction. Ideally all parts of bearing should be as hard as possible (hardened steel / ceramic) because it reduces deformation and friction. Our races are quite soft, but at least we can have proper balls ... that's what I think.
Еще шайбу железную подложить и будет огонь.
You mean bearing races
Диме надо послушать больше как принято говорить в country #1, и все будет ok.
Ага :) А знаешь что самое забавное? 95% комментариев про мой акцент написано кириллицей. :) I wonder why...
Может, на русском вещать все-таки?
"th" != "z"
ты русский??
да
Use magnets