3d Printed Cycloidal Drive for Robot Arms

[@PaulGouldRobotics]

Nice design. Adding bearings around the inner pins should increase efficiency too.

[@roTechnic]

Thank you. The inner pins (and the holes around them) are the parts that I do worry about wearing out - and, you're right, changing them out for bearings would help. Maybe I should... I think at some point I'll set one of these gearboxes running non-stop for a couple of weeks straight and see what wears out.

[@TDOBrandano]

@@roTechnic or maybe putting the pins themselves on bearings, if the diameter is an issue.

[@LawrenceKincheloe]

I've been looking at adding a spacer to support the inner pin and a slip ring like a bushing for it to ride against. I've also been looking for a FEA study on this gearbox design, because my understanding is that all contact surfaces and forces are normal to the point of contact, which means that most of the gear wearing would be caused by compression distorting the shape of the gears, as opposed to slipping, causing wear. If that's the case, then an impact resistant plastic, surrounded by a compression coupling(steel ring/belt/ect), might make for the ideal gearbox design.

[@roTechnic]

@TDO Brandano that's a good idea. @Lawrence Kincheloe I think all contact surfaces are normal to the point of contact around the edge of the cycloidal disk, but I'm not sure about the inner pins and the holes they ride against - if you have any info about them I'd be fascinated to hear it.

[@Mrshotshell]

@@roTechnic rctestflight has a few videos about wear testing 3d printed cycloidal gearboxes

[@p3dee]

I'm starting to fall in love with mechatronics after seeing how this works. I'm only used to CAD/CAM/WebGL programming.

[@DiffEQ]

4:01 Why wouldn't you put the set screw on the flat of the shaft? No only is there more material there, but you have much more contact to prevent slippage since the flat become a fulcrum point.

[@roTechnic]

yeah. I only noticed I'd done that after I reviewed the video! for the testing I did have the set screw in the right place :D

[@username4294967296]

Amazing how much force this can generate without breaking any plastic! Looking forward to the continuation videos of you building the robot arm.

[@roTechnic]

Thank you. Yeah - I'm always amazed at how strong PLA 3d prints are!

[@JulianMakes]

Just wonderful! I’d also love a robotic camera person! I’ll be having a go at the add on later today, thanks loads!

[@TheRealStructurer]

Like your build, how you explained it (right level and speed) and the fact you engage with viewers 👍🏼 Subscribed and looking forward to the next video

[@roTechnic]

Thanks for the sub!

[@adsalesmanguy2251]

You can make a waving arm with on of these. You’d put it at the arm base near the torso. You would have to make a hand and fit it in on the metal beam thing you had to carry the weight and you could use the same program to make a waving arm.

[@PhG1961]

Waw, impressive. A bigger one... well of course ! Size does matter !

[@rondlh20]

You can clamp some metal part, like a pulley, to the stepper axis to create more surface area. The stepper force should be the limitation, not the plastic...

[@roTechnic]

That is a great idea. Watch this space...

[@gregarmstrong4653]

how about bearings for the pins to roll around in, oversized by the overall eccentricity. Alternatively, low friction sleeves on the guide pins.

[@roTechnic]

yeah, I did think about that. I've run this gearbox for about 10 hours and i can't see any wear yet. that may well change under load, so I've always got the option of adding these later - I think the sleeves would be better than bearings - they should be more compact

[@dekutree64]

Fatten up that ring and you'll get a lot more torque before it flexes out of shape. Usually cycloidals have the ring outer diameter around 1.3x the ring pin circle diameter.

[@roTechnic]

Oh, good to know! thank you. is there anywhere in particular you got that info from?

[@dekutree64]

@@roTechnic Just do an image search for cycloidal drive and look for ones that are professionally designed and measure them. I'm not sure how they arrive at that proportion, though. How much deflection is acceptable at the design torque? The only equations I've been able to find for calculating deflection are for thin walled pipes, but should get you a ballpark estimate. Pipe squashed from above will deflect by .149*F*r³/EI, and bulge out at the sides by .137*F*r³/EI. E=Young's modulus of material (3.5GPa for PLA), and I is second moment of area, which in this case is (OuterRadius⁴ - PinCircleRadius⁴) * pi/4. The force can be calculated by assuming pressure angle is about 45 degrees between the disc lobes and the fully engaged pins, so radial force is equal to tangential force. For example if output torque is 10Nm and ring pin circle radius is 5cm, then total radial force on the ring is 10Nm/.05m=200N. You can also use that to see if your eccentric bearings are strong enough (in this example 100N each if you have two discs).

[@mrteemug5329]

I'm getting big Atomic shrimp vibes and I like it. Subbed and liked, I want to see that robotic arm.

[@roTechnic]

Yeah, you've gotta love Atomic Shrimp - I don't think I'll be branching out into cooking videos as well as tech though :D

[@mrteemug5329]

@@roTechnic :D I guess a collaboration with him will have to suffice

[@810mb3Rg]

Nice video! It bugs me how you didn't put the set screw on the flat side of the shaft haha

[@roTechnic]

thanks. I realised that just after I’d shot the cad sequence where I added the screw. the model you see me put together has the screw over the flat, but I thought no-one would notice the cad was different. looks like I was wrong :D

[@brunobock372]

I had the same problem with the motor attachment on a different gearbox. Whenever we get into heavier loads with 30 printed parts this can happen very quickly. Especially with PLA

[@swannschilling474]

Super nice, I was actually also planning to mount a camera to my arm, but the vibration that you see on the first axis translates down the arm and amplifies! My arm moves quiet smooth, but for your project you might also look into belted or pulley transmission as they are a lot smoother than cycloidal gearboxes....although I am really in love with cycloidal drives! 😇

[@roTechnic]

Thank you! Your arm looks awesome by the way! I've played with belt drives in a previous video, and I'm about to try another design using eccentric cycloid gears to see how that works. Maybe I should do some vibration tests before committing to a design...

[@chehinegargouri9671]

​@@roTechnic Hello, is this kind of transmission non backdrivable (Irreversible system) ? Thanks.

[@peterhenson823]

Innovative. Awesome job! and great video. no fluff about your channel just straight to the heart of robotics. Great stuff man! The cycloidal drive was something I had never seen before. Thanks for sharing!

[@roTechnic]

thank you

[@BeeRich33]

Perhaps a bearing for the inner rotation? Not sure if that would help.

[@roTechnic]

If you mean on the 5mm rods that transfer the drive, then yes, that's a great idea. I might add some in a future revision

[@renanchagas7052]

Congratulations, awesome design!

[@roTechnic]

Thank you very much!

[@JimMaz]

Nice channel. It's gonna get big.

[@gerry5134]

Amazing design 👍🏼

[@chrisBruner]

As it's back drivable, you might need an encoder or angle sensor to know its current position.

[@roTechnic]

yeah, that would also make up for any backlash that develops over time too. I’ll definitely put one on the arm.

[@chrisBruner]

@@roTechnic I was thinking also that even if you don't need the second cycloidial gear, you could print it as part of the first gear. IE two gears 1 print.

[@2testtest2]

@@chrisBruner It's possible I'm misunderstanding, but I don't think printing a second cycloidal gear as part of the first one would have much benefit. As I see it at least, the point of having more of them would be to have them move opposite of each other (180 degrees out of phase), in order to cancel out some of the unbalance in this design.

[@roTechnic]

Yeah, unfortunately the centre of mass of the rotating cycloid stays the same, but the point it revolves around moves due to the eccentric cam, So it would need another disk on a cam offset in the other direction to counteract it.

[@chrisBruner]

@@roTechnic Yes, I understand that. I'm suggesting that instead of printing 1 gear, you print 2 half as tall, as one print job.

[@qcnck2776]

Excellent progress. What do you mean by multiple cycloids in the gearbox?

[@roTechnic]

Thank you. When you spin an eccentric disk really fast it tends to vibrate. So drives that are meant for faster motors have a second (or third etc) disk stacked on top on the first one, but offset by 180 degrees, so when they spin the eccentricity of one counteracts the eccentricity of the other and kills the vibration. Here, the motor doesn't spin that fast so we don't need to bother.

[@qcnck2776]

@@roTechnic Thanks

[@alexon2010]

Como eu gostaria de ver como fazer uma cremalheira e uma engrenagem com 3 pinos modelos cycloidal.... gostaria de aplicar esta ideia da cremalheira + engrenagem cycloidal em uma router cnc para como se comporta... mas não sei como seria para fazer... seria uma boa sugestão de video

[@CoolerQ]

I... really want to go and try to build one of these myself. Instant like + subscribe so I can follow along! I hope the design gets open-sourced at some point.

[@roTechnic]

Thank you. And, yes the design will be open source - all the cad and code for my previous videos is on my GitHub github.com/roTechnic/ I haven't added this one yet as there are a couple of changes I'd like to make to the CAD before releasing it.

[@fortoday04]

Can you show us a tutorial on how to write the plug-in?

[@roTechnic]

I certainly can! see here: kzbin.info/www/bejne/r2rZfYmOn99lmZI

[@zwurltech9047]

Great Design, thank you for sharing!

[@smokeyloen8126]

are you going to change the supply list for the update you think, I want to build one

[@roTechnic]

I’ll publish the cad in the next week - it should use the same supplies

[@GWAStech]

Wow mam! This is so insane! Amazing! Great job! 😁👍

[@roTechnic]

Thanks a lot 😊

[@waltermundt]

What about the design of the gearbox determines whether it can be backdriven?

[@janibizjak9966]

it is pretty fu*king impresive, kudos to you! thank you for explanation!!!

[@MrNoipe]

Great video, excited to try making this. Do you have a recommendation for a stepper driver?

[@roTechnic]

Thanks - I tend to use either TMC2208 or DM542 drivers, but I've recently been playing around with the SimpleFOC system - I've got a feeling that will soon become my goto driver

[@koyulab2561]

I have a question, how does the 3D printed part attached to the shaft can slip? What is the scientific explanation behind it, it seems like the part is fitting perfectly fine to the shaft. And how does the screw that you added later on changes the situation?

[@scottb8423]

Plastic is soft! (especially a 3D printed part that might not be solid inside) Even if it fits 'perfectly fine' to the shaft, it is too soft to resist a higher torque; it deforms enough to let the shaft rotate. A hard, metal set screw pushes against the flat face of the d-shaft, and spreads that force through a larger volume of plastic that it is threaded through; if the shaft wanted to turn relative to the plastic, it would now have to push the entire setscrew away. As an analogy, imagine trying to stop a high torque motor by gripping the d-shaft with your hands (don't try this at home); just like the plastic, your hand is not nearly rigid enough to stop the shaft from flexing your hand open enough to turn.

[@koyulab2561]

@@scottb8423 Very nice and through explanation, thanks a lot dude. Are you a mechanical engineer by any chance?

[@roTechnic]

Wow, I couldn't have put it better myself! Thank you!

[@scottb8423]

@@koyulab2561 mechanical engineering technologist, and general tinkerer. You are welcome!

[@scottb8423]

@@roTechnic and thank you for the excellent videos, I'm working through them now. Keep it up!

[@crocobroco4654]

Nice build. Have you seen James Brutons design?

[@roTechnic]

Thank you. Yes, I've seen those - they look really good - if I end up using brushless motors instead of steppers I may have to steal some of his ideas for them 😎

[@ivprojects8143]

Excellent design! Subscribed.

[@roTechnic]

Awesome, thank you!

[@chrisandersen977]

Any place to get the stl?

[@roTechnic]

No STLs I'm afraid, but you can download the CAD from github.com/roTechnic/cycloidalGearbox and print them from there

[@rakshithbaskar6014]

I tried making the drive but the gears are larger and they don't move as freely as one in the video, could you suggest me a way to reduce the size of the gears.

[@roTechnic]

could be that you are printing slightly oversize. you could try scaling it down slightly or following the tips in my video on tuning a budget 3d printer to make it print to exact size

[@rikidrabsch902]

Nice simple design! What was the backlash like?

[@roTechnic]

Thanks. there is a tiny bit of backlash, but in this video: kzbin.info/www/bejne/q2Xcn355ZbmKgpo I added another cycloidal disk which gave me virtually no backlash at all!

[@fabientuizat1129]

tres belle réalisation merci de partager !!!!

[@roTechnic]

Merci beaucoup!

[@greendsnow]

hey! how about the encoder? where do fit it in?

[@anasqai]

Why the middle moves like that? If a normal gearing = lesser voltage needed right? Can i subject-buy and have ez job making these?

[@roTechnic]

Less voltage also means less torque - here we keep the voltage the same and gear down the RPM, increasing the torque

[@waynesardullo2482]

very nice design

[@roTechnic]

Thanks a lot

[@andrewmalaty8]

Can you share the plugin you wrote for fusion?

[@roTechnic]

in this video: kzbin.info/www/bejne/r2rZfYmOn99lmZI I describe how it works, the code is linked in the description.

[@Joshua-kr5fq]

Amazing love your videos!

[@roTechnic]

Thanks so much!

[@RenatoYamamoto481]

nice keep the good work

[@roTechnic]

Thanks, will do!

[@theobserver9849]

Awesome video.

[@IngCristian]

Wow...

[@TheTeditor]

Impressive!

[@MicrobyteAlan]

Awesome thanks 👍

[@joshuavincent7884]

Nice!

[@roTechnic]

Thanks!

[@pieterpennings9371]

You're an F'ing genius lol.

[@Puner54]

That is someone Cool!!!

[@roTechnic]

Thank you

[@kewintaylor7056]

Good idea....

[@roTechnic]

Thank you!

[@nou5440]

nice

[@roTechnic]

Thanks

[@SP-yw2tl]

Instead of 3d print cut the pieces out of aluminium plate then it will be strong.

[@roTechnic]

I'm hoping that the 3d printed pieces will be strong enough for my purposes, but if not I may have to do this. Thanks for the comment :)

[@vijayjayaram606]

Wow

[@PhG1961]

This is really an awesome video. What about a planetary gearbox ? Or this cycloidal drive with a much faster brushless motor ? I'm looking into a robotarm build myself and so far I'm planning the use of a brushless motor comined with a planetary gearbox or drive belts like over here: kzbin.info/www/bejne/iaDRnYydl8Z-l80

[@roTechnic]

Thank you very much! I'm debating whether to use a planetary box for the base of the arm - I worry about the accuracy of 3d printing for really small gears, but I've got lots of space in the base to scale it up. I've got a feeling I may have to move to brushless motors instead of steppers, but I'm going to try steppers first and swap them out if I don't have the power

[@PhG1961]

@@roTechnic Great ! This is an interesting point of view. I have some really powerfull steppermotors but it's a compromise between cheap and accurate and fast. I'm in te process of building a humanoïd robot, based on a mannequin. The movement of the arms (upper arms) has to be smooth and powerfull. So far, I didn't find any stepper capable of doing so. Anyway, I'm looking forward to see your next improvement.

[@roTechnic]

​ @Philip Gevaert Thanks, I think this is where brushless motors could really shine - at the limb lengths you're talking about I think that you'd find it hard to get smaller stepper to move fast enough and with enough torque, whereas a brushless motor with a double reduction would keep it small and fast, but would draw more current.

[@PhG1961]

@@roTechnic Indeed, high currents are an issue. I'm still in the study fase. If I manage to get anything working I'll let you know ;-)

[@SHHeo-vp2uq]

와우!!!!!!!!! 천재

[@lucasrem1870]

U use the wrong motor, you only loose energy when you compensate that with gears! Find the right PWM motor you need, or build it, gearing makes it too complex, you need it to be less complex, so you have less failures!

[@roTechnic]

What motor do you suggest I use?

[@lucasrem1870]

PWM with enough torque that fits the need, less parts is less failure!

[@ekstrapolatoraproksymujacy412]

this is not how set screws work

[@roTechnic]

yes, I know! I only noticed after filming the cad part. when you see me assemble it the screw goes on to the flat. I hoped no one would notice. sorry 😳

[@TOTALLY.REAL_FBI.AGENT2]

Cheese

[@shamsulchowdhury8240]

Prosthetic ELBOW Joint please, thats the best use of your design. Please send me your contact so that i can join in your effort..i need a prostehtic eblow joint design.

[@Naxt366]

nice