Testing: Cycloidal vs Harmonic Drive 3D Printed Reducers
ฝัง
- เผยแพร่เมื่อ 20 พ.ย. 2024
- Over the last few weeks I've built two versions of Cycloidal Drive Reducers, and one Harmonic Drive / Strain Wave Reducer. This video is about testing both using an ODrive and an encoder on the motor so we can energise the motor coils correctly and apply motor holding power. I also want to make sure that the Cycloidal Reducer is tough so that is subject to some extra resilience testing. My Cycloidal Reducer will get developed further, probably removing the bearings, so we can use it in the next version of openDog.
You can support me on Patreon or buy my Merchandise:
***************************
Patreon: / xrobots
Merchandise: teespring.com/...
***************************
Affiliate links - I will get some money of you use them to sign up or buy something:
***************************
Matterhackers 3D printing supplies: www.matterhacke...?aff=7500
Music for your TH-cam videos: share.epidemics...
***************************
Other socials:
***************************
Instagram: / xrobotsuk
Facebook: / xrobotsuk
Twitter: / xrobotsuk
***************************
CAD and Code for my projects: github.com/XRo...
Huge thanks to my Patrons, without whom my standard of living would drastically decline. Like, inside out-Farm Foods bag decline. Plus a very special shoutout to Lulzbot, Inc who keep me in LulzBot 3D printers and support me via Patreon.
HARDWARE/SOFTWARE
Below you can also find a lot of the typical tools, equipment and supplies used in my projects:
Filament from: www.3dfuel.com/
Lulzbot 3D Printers: bit.ly/2Sj6nil
Lincoln Electric Welder: bit.ly/2Rqhqos
CNC Router: bit.ly/2QdsNjt
Ryobi Tools: bit.ly/2RhArcD
Axminster Micro Lathe: bit.ly/2Sj6eeN
3D Printer Filament: bit.ly/2PdcdUu
Soldering Iron: bit.ly/2DrNWDR
Vectric CNC Software: bit.ly/2zxpZqv
Why not join my community, who are mostly made up of actual geniuses. There’s a Facebook group and everything: / 287089964833488
XROBOTS
Former toy designer, current TH-cam maker and general robotics, electrical and mechanical engineer, I’m a fan of doing it yourself and innovation by trial and error. My channel is where I share some of my useful and not-so-useful inventions, designs and maker advice. Iron Man is my go-to cosplay, and 3D printing can solve most issues - broken bolts, missing parts, world hunger, you name it.
XRobots is the community around my content where you can get in touch, share tips and advice, and more build FAQs, schematics and designs are also available.
Next week is a totally different robotics project - Patrons and TH-cam Channel Members have it already!
bro wtf 13 hours ago?
Which book shall i study to understand these concepts like...harmonic drive etc. There application n all.
Pls help
Hi James I think a abacus drive would be worth looking into as a powerful, compact, and backdriveable actuator
You should install a separate gear reducer on the strain bearings of the motor, that should help with the loss of torque
@@andrewstrauss7166 what is the difference between cyclodial and abacus ? and also, how cool would it look if a reversible inception drive was used in a project, led lit of course
th-cam.com/video/0-uSUrcRsyw/w-d-xo.html
At 25kg per leg looks like James is going to be able to ride this puppy. Can't wait to see the final design.
Only when stationary. Moving (3 legs, each 25 kg) sounds too close for comfort.
That would be so cool, although as the others have pointed out have to consider the added forces/torque from acceleration during motion. Literally a statics vs dynamics problem lol
Next project open pony?
It's 2022, goodboi v3 is out, and can carry a load of 400 lb. Life is good
Moving, you'd need have 2 legs on the ground at a time, so they'd need to support (say) 80kg static, which means 160kg dynamic (you "weight" twice as much at peak when you're moving about). So he'd need 80kg per leg to ride it.
So, if he had a 40:1 reduction, that would do it. No idea if that would still be backdriveable.
-3 videos of people telling him to stop testing the tourque with his fingers
- james: I'll do it again
69th like. Nice.
That thing isn't really able to do much damage not more like a other dude shaking his hand.
@@fischX You know that now, yes. But when TESTING, i.e. not knowing how much torque it can output, you're risking damage to your hand
It's also not cycloidal as some have said. But good criticism is not always picked up unfortunately, probably due to the amount of stupid youtube comments so they don't read it. If they do and if the criticism is correct, but not acting on that, I'd find that weird.
So this video was available last week for Patreon supporters I believe based on one of his community posts. So it could just be that since he's working approximately 2 weeks in advance he just didn't see those comments from the first video by the time this video was filmed.
"Mummy, who's that funny man on the skateboard with the strange stick?"
"Oh dearie, that's just James, the village robotics genius."
"But Mummy Mummy, are the three-dimensional models for that gearbox available for download online?"
"Yes child, he's a very good man and he publishes his designs as open source."
How do I get them? Looking at the V2.stl with an online viewer I cannot see the differnt pieces by themselves...
@@GregorShapiro It's not an STL, it's an STP (or STEP) file. There's a difference. Any 3D CAD program or viewer should be able to open it. You may have to enable options in your viewer to have parts of the assembly be displayed and adressable as separate pieces.
STEP is quite practical as a file exchange format but It's emphatically NOT a source file format. It's hard to change anything about a part in a STEP file. You lose all sketches, extrudes and other operation history, assembly constraints, etc. If you want to, say, change the diameter of something, you're better off just designing the whole part new.
It would be great if James could consider publishing his F3D files along with the STP files. Those are the true source for his design.
@@olavl8827 Even in an STL viewer it isn't much better. The 'parts tree' has but two 'leaves': one is the 22 ball bearings, the other is the rest of the construction which cannot be broken down any further.
I can color or hide one or the other or both of these 'leaves' but can't isolate the cycloidal discs or any other feature.
I would like to take sections so I can trace the shapes in the 3D modeling tool iam familiar with so I can extrude appropriate heights of the various parts and send them to a slicer for the 3D printer I hope to get access to when Covid restrictions are lifted.
@@GregorShapiro I'm not even sure if you read my previous comment, so I'll repeat: James' file is not an STL but a STEP file. You need a STEP/CAD viewer, not an STL viewer. STL is not a CAD file format.
If I open the file in FreeCAD I can select and manipulate all the individual parts of the assembly. But this does depend on the import settings that I configured. The situation will be different but similar in other programs.
@@olavl8827 Yes I did read your previous post. I did install a STP viewer that is what I found 2 leaves in the hierarchy.
3:10, C'mon man. I'm not going to tell you to not test it with by hand but at least do it from the other side of the table so that it doesn't have a chance to pinch your arm between the table and the lever.
Overall, like the idea. I think the Cycloidal Drive will work out well in this project. I agree that at the very least the outside bearings need to go, if I understand correctly, the disc is rolling across the outside pins anyways and thus friction there should be very small.
If you modify the geometry of the cycloidal disks, they won't need a full pin to roll against. This would let you print roller ring section in nylon and integrate the pins into the ring itself and possibly cut the outside diameter a bit, further reducing weight.
Upvote this. Every time he haphazardly grabs that lever, I feel like my heart is about to jump out of my chest.
I know it shouldn't be strong enough to seriously hurt him but I did cringe a little every time he grabbed it, all he to do was clamp it so he could easily grab it on the upward stroke and it would also eliminate it pulling the table up as well
Yeah, I'm fully expecting a video titled "Well it seems I've lost my fingers" at some point in the future.
@@deflepperdrocks12954 Looks strong enough to break an arm or at least a finger if it hits you unlucky. I'd call that serious and I agree with the other commenters who want James to be more careful.
@@olavl8827 his palm/arm should easily be stout enough to stall the motor at the distances he was grabbing and at the same distance even a single finger would likely survive although badly bruised, the only really dangerous spot in my opinion is the radius just above the tabs that connect it to the board as the rotating orange mass has very little clearance and you have substantially less leverage working in your favor
I would love to see him test the danger with something like a wooden dowel but I doubt he'll even see these
1:54 massive respect for the fact that it has 3 massive red stop buttons on it
It should have one that does everything really!
4:58 you were so close to actually finding out numbers, just vary the 8kg distance from 0.25m to 0.50m
He was even closer, he could have just measured the angle at which the motor stalled at 4:24 and calculated the arm from that.
@@BigHorse4200 Looks like it _almost_ stalled around 40° to me. That would be about 30Nm.
Imagine one day all of the TH-cam engineers come together....not for money...but to get the human race to a better place in the Cosmos...we would overcome anything in onur path...including the impossible...nice project James Burton...All love support from Africa...🔥🔥🔥🌊💯
Ahh, so the next step is the dog flipping onto its back and rolling away on wheels attached to the shoulders!
That's a "Haters gonna hate" GIF waiting to happen :P
Nice
New torque unit - 1 finger per screwdriver.
Love these gearbox series.
Wouldn't it be 1 finger screwdriver, as finger is the force and screwdriver is the length?
@@KentoCommenT Yea, right
finger * scredriver
I really appreciate the amount of r&d he puts into all of his robots
The redesign of the cycloidal disc looks like it would give it a lot more torque.
Really enjoying this build, you're approaching everything like a NASA engineer.
Build. Test. Improve. Test.
@@glebmarkovski was gonna say exactly that. All true engineers follow that approach within cost and production constraints. It's those evil beancounters that demand production before testing.
You on a skateboard increased your coolness level considerably.
There are quite a few other skateboard videos in my channel - mostly 3D printed ones!
Your videos are truly great! It is very nice to see you test your designs, improving them with some goal in mind, and retest them. Very much looking forward to see the "improved disc version" and "bushings version", and hopefully as different tests, so that we can make out what changes gives which effect!
This was posted at 1 am where I’m at in America so I’m glad that my bad sleep schedule allowed me to see this sooner
2 AM in Chile! But I can't miss any of your videos
One AM in Wisconsin lol so same
@@conkillious.-.9462 hey! I'm in Wisconsin too!
@@doofenschmalphys__4376 such nice wether recently right 😂😂😂
The same here, almost 4 am in Brazil
2 AM in ontario canada (5:20 AM now)
I can't believe I just found this channel a week ago.
Same. This guy is amazing.
I think the biggest weakness of this particular strain wave gear is that the flex spline is a very soft material, meaning that when the output has enough resistance the wave generator can just compress the spline and roll past it instead of driving the whole mechanism forward.
Yeah. With the two point ellipse and the triangle gears he is also not getting anywhere near the contribution from the geometry (%10-20). The flexspline cup is also WAY too deep given the material properties (Metal flexsplines are that long to allow for the needed flex). The cup is going to twist right out of alignment. Harmonic shows that the backdriving torque (www.harmonicdrive.net/_hd/content/documents/FB_Component.pdf) is commensurate with the input torque (slightly higher as might be expected) so there is a design flaw
James: * attempts to stop cycloidal drive *
Cycloidal drive: "Hey! I'm workin' here!" * flips table *
I'm glad you're sticking with the cycloidal drive, and looking to remove some of the bearings. if you use a proper cycloidal tooth profile the gear teeth have pure rolling contact where they mesh, so simple pins or a bigger cycloidal gear will do just fine. No need for bearings or even bushings.
Also, I think nylon is a bad idea for the cycloidal gears. Soft, squishy and spongey materials soak up energy and reduce efficiency/backdrivability. I think this is what is causing problems with your harmonic drive. Think of it like trying to walk through mud vs on pavement. Personally every time I've used a squishy material in a drive system it's come back to bite me.
Loving your work!
5:25 "So we now need to do some comparisons with an actual dog leg" I'm calling the SPCA
I am glad I'm not the only one with the wobbly wheel problem. I bought some wheels that were not seated properly when they were manufactured and i experienced the wobly with 4 wheels. LOL Thanks for sharing!
Your videos are both informative and fun! One of the best TH-cam channels.
This really has me hopeful for my dream project. The hope is to make a "mech" that "eats" recyclables and churns out stuff made from the plastic or metal picked up. Mostly focusing on the out product being parts for our "guerilla solarpunk" art projects. So like ride this thing around, feed it litter/stuff from bins, it pops out parts for like a plastic bench with artwork embedded in the design so you could use a paper/pencil to it.
The best part about Tuesdays are these videos.
I've been working on some cycloidal gear drives as well! Your video definitely gave me some insight for durability & lash back. Happy building!
Priceless James! I've just started looking at drive types as I decide the style of robotic arm to print. I'd ideally like to get maximum leverage and torque of course. Even more so as I'm wanting to design interchangeable tool heads for the arm to use it to drill, mill, cut, etc... I'd like to make it strong enough to work with wood, metal, plastic. The cycloidal drive looks at first glance to be more than capable of handling that! You're the bomb Bruton, THE BOMB!
This seriously makes me think about putting some cycloidal gears into practice. Very inspiring.
For the torque test, instead of using a lever,
uses a pulley attached to the output of the reducer, and a cable with a weight that is wrapped around the pulley when the motor turns. This way you have a constant resistant torque!
Looking forward to the next iteration of the cycloidal drive with bushings.
I think the downfall of the harmonic drive was simply due to the flexible material of the core deforming at the tooth level to allow the gears to slip. Commercial versions use clever metal rings that flex at the ring level but are almost completely inflexible at the tooth level.
I think that 25 kilogrammes was more of your limit rather than that of the leg's lol. Great work as always
Yeah I'd think you'd want to brace the leg against something that won't move and let it press on the scale to get an accurate measurement.
Yes it'll do more, although the motor is starting to back-drive, which is what I want to absorb impact, but not under normal operation for standing still. It's more than enough overall though.
Skateboard freestyle/Guitar solo montage. You absolute mad lad.
I'm surprised you didn't attach a rubber ball to the end of the arm and have the motor press down on the scales you have to test the torque that way. That'll give you a much more accurate result than hanging arbitrary weights off the end.
Cycloidal drive definitely looks like a winner :D Excellent work!
have tried using PTFE filament instead of bearings. I have quite god experience with that.
I would recomand to use it with dualextruding and just for the surfaces where the friction takes place.
Igus makes filament of their oil infused linear bearing material too
PTFE is really quite cheap to buy and mills like butter. You can probably just mill out the whole outter race and use PTFE sleeves as inner bushing.
7:10 Wow that's fantastic! The cycloidal motor also look really good!
Great walkthrough as always 👍
Thanks for sharing your experience with all of us 👍😀
Incredible! Loved the testing on the skateboard there =D This channel is one of my favourites for sure!
thanks!
The VESC can track rotor position @ 0 speed if HFI is enabled :)
Yeah, his motor doesn't have a sensor but he says it's the vesc's fault. I don't think he knows how to use it
I have 4 of that vesc in my skateboard and it can make a sinus curve with a rotor input. To me it seems he has only glansed at the settings or haven't bought the program Vesc-tool.
i mean most motor controllers are designed to do that
@@Blox117 not really
This is coming along nicely, you're one of a kind James!
Nice design and well put together video James. I'm glad you're going to try bushings - it was what I thought when you produced the first variant. Not that it would save on cost, but there are also needle roller bearings available (I am sure you are aware of these, of course) which could work in reducing weight and allow the cycloidal gear to have fewer and smaller holes cut into them as well as perhaps reducing the overall size of the entire gearbox. But then bushings would probably also be perfectly fine considering the RPM you are asking of them. Perhaps not quite as smooth but much easier to work with. If you dont have a lathe or access to one and need some made up, let me know and I'd be happy to make them for you.
I think that its really cool of you to post all of your cad files online for free
Have you considered printing the cycloidal mechanism in a stiff TPU? It's extremely impact resistant and can still hold it's shape if printed solid. It could help absorb some of the impact shock.
I love this, gives me ideas to modify my wheelchair to make my own power drive. Just wish I had a 3D printer and/or someone to help me
Can't wait to see a new dog with that cycloïdal reduction system !
Awesome, I got to try building this for my snowboard.
Love those encoders. really cheap & accurate.
When you consider the mass and materials that your cycloidal drive is made from, it's pretty impressive. Light, torquey and durable under load. You rarely get all those features together when plastic parts are involved. LOL
Thanks, yes that was the plan, and we can still lose those 22 bearings!
Your work is awesome you are a genius person thanks for sharing 👍 🙏
The cycloidal drive looks great! Just wonder, if you replace the bearings with bushings do you think it could generate too much heat?
I don't think they actually turn that much, and if the cycloidal disc shape is correct they shouldn't turn at all apparently ;-)
@@jamesbruton Nice!
@@jamesbruton If they won't turn at all ideally, then what is gained by having them, once you have the correct disk shape? Should you just get rid of them altogether and widen the post they were sitting around to compensate?
@@delphicdescant well its still fdm printed pla so if it isnt perfect(wich is pretty common in fdm printing lol) then you would have plastic rubbing on plastic
James - if you want to cheat slightly on the Cycloidal profile, Camnetix Gear-Trans software has a Cycloidal option. The software itself is fairly expensive - but there's a fully functional trial. I've found the results are very good.
take a look at Skyentific's block and tackle joint. seems just perfect for you
Cycloidal drives by nature handle loads better than similarly sized or similar ratioed strain wave drives because of the higher surface contact and rolling features of the cycloidal drive. If you reprint the spline in Nylon rather than TPU, you should see an increase in performance.
Take a look at the Enviolo transmission, might work out for you! :)
Have you played around with dual stage compact belt drives? I've seen some Paul Gould videos demonstrating that and might be a nice option. Love these videos!
I've seen the videos, I'm just a bit worried the belts will skip so I'd like to leave belts behind ;-)
One day i would like to see these torque houses in an electric skateboard. I live in an area with next to no flat areas & terrible road conditions, good old Devon.
I would want to be able to free wheel as well though.
at 3:20 and here, we see the invention of the scissors, a stationary blade and a moving blade, doesnt matter that they arent sharp
i love your videos, but i think losing your fingers might slow you down, you dont want that.
Try a differential Cycloidal arrangement, this saves you the middle output bearings and brings the output shaft to the outside for better momentum attachement
Every time you do that finger stop test at 3:10 I imagine it clamping your finger down against the board and snapping it like a twig. Maybe try stopping it on the up side?
Can't wait to see the nylon or aluminum versions of this!
hes gonna need some funding for aluminum. might be castable but damn thats a lot of work
Very nice work!
Don't you think that going to bushings instead of bearings will increase friction, resulting in less output torque? It would be interesting to see you do a side by side of the two. I suspect the bushing drive will have about 50-70% of the output torque of the bearing version.
The harmonic drive stalls out earlier because the strain wave gear can't mesh with the other gear under load. Oiling the strain gear should improve that. TPU is just too sticky.
I look forward to your video every week
And this is a perfect example of why
Reminds me of one of the small Vietnam fishing boats when you were on the skateboard
I think most 6.6 VESC's can take encoders. I know the Trampa VESC's can, and they are based in the UK and very friendly. They have 100v VESCs that can pump out 250 amps continuously that can take encoders, and you get to use the nifty VESC Tool.
Open source king! Gotta love that
Just want to say: Thank you for sharing all your work. I would like to build the dog, I need a 3D-printer first :)
A robot dog skateboard would be super cool.
I love your projects - big inspiration.
Which book shall i study to understand these concepts like...harmonic drive etc. There application n all.
Pls tell.
That's getting strong, you should scale up a bit and make a steampunk horse and ride into town on it!
industrial harmonic drives are totally backdriveable and also have an acceptable efficiency. With the flexing however, it's probably hard to design it properly. The cycloidal drives makes more sense in this case, but it's not the fault of the concept itself.
I can see someone making an anime about the power stick skating that involves fighting with them as well.
You might want to consider using ceramic bearings which are lighter than steel/metal bearings, if the bushings end up being a bit of a drag.... so punny.
Ceramic bearings would end up quite costly, I hope bushings with geometry improvements end up being enough.
Might investigate a small 2:1 planetary gear reduction, or some other ratio and then change your belt reduction, you can use with the existing belt drive. Put it on the larger pulley so that the belts aren't having to handle as much torque. You might even be able to find or make one that is small enough to fit inside the larger pulleys.
this is comment which will promote James channel by 0.001 %
I've got big respect of what you are doing there! And I was wondering what the Cycloidal drive would cost to build right now. Do you know that?
I recently could have used a high tourge increase the for a College project but was too lazy to make some 3d printed gears worke for me with such a high reduction. If this Cycloidal drive is cheap (conserning everything needed) I might really get one myself.
And also my biggest respect for that robot. I guess as an Mechatronich student that will be my goal within the future to build such amazing thingies😀
For a second I thought the thumbnail was about comparing funky new-generation pokemon that I hadn't heard about yet
Heii James - Thank you very much for this really nice videos!!
What do you think is the achievable with your drive? Looking for a drive for mounting a telescope a bit more compact✌🏻
This is exciting, well done James :)
The skate esc you were using supports encoders.
James: I can stop the drive just using my hands.
James' hands: Band-aid and a fresh cut.
The skateboard esc has a sensor input, you just need a sensored motor. Not sure if the o-drive sensor is compatable.
awsome Men!! greating from Ensenada Mexico.
Hi James, just watching with my 3years old son show asked me to make the same wheeled stick to go outside make some skateboard in the street. ,😁
You may be try Igus filament for the inside wheel with some "chromed" steal bar( Sorry for chromed, I don't the right english word for that)
Keep going !
This design could revolutionize equatorial mount
James, maybe you know this already but onshape has an Cycloidal gear generator, this makes experimenting with ratio's a lot easier Search for: onshape Cycloidal gear generator
If you want to use bushings instead of bearings, maybe consider using Teflon since it's self-lubricating.
That looks like it would be great as the internals of an one-wheel type skateboard.
So, the VESC (skateboard ESC) can actually read hall sensors or encoders, and run sensored instead of sensorless
James: tries to stop the second drive
Drive: am I a joke to you *moves table instead*
For some reason your 3D printed gearbox design studies, especially with the long lever at the end, make me think that an open source "adult fun machine" is kind of inevitable.
in order to reduce weight and cost you need to test designs without bearing. Have you thought about 3dprinted discs with a brass bush insert???
Great video ! Very helpful. Any scope of developing a BLDC controller with encoder(closed loop) for high power applications ?
Love how that cycloidal drive is working. I couldn't quite work it out from what was in the video - but how does the efficiency compare between the cycloidal and the belt drive? i.e. the torque you can get out of the motor direct vs after the belt drive or after the cycloidal drive. Of course, there is the 5x or 10x increase from the "gear ratio", but that is then reduced by the friction, etc of the mechanism. I'm curious what that reduction is.
It's difficult to say, although the belts are very tight so maybe that causes more friction?
I hope we see more RUR soon!
I wonder if the strain-wave reducer is binding due to the squashy nature of the material used for the flex-spline.
PETG is somewhat flexible without being squashy, I wonder if that would work better...
I am really happy with my Open Source Cycloidal Gearbox too, its backdrivable and has a ton of torque...I also realized that a standart ESC produces a lot of noise compared to the Odrive!
All .stl file are available if you like to check them out! 😊
Nice!
@@jonastjepkema Thanks, glad you like it!! 😊
Hey all, just FYI, the equation as copied and pasted into SW had some issues. I copy and pasted it directly into SW2022 and it came up with something that looked more like a flower than a gear. After watching a video on Stepbystep-robotics channel I was able to figure it out. He had done some intermediate calcs on a couple of the terms and then replaced those terms in the equation and removed the associated parentheses that are no longer needed. This solved the problem. Also, in SW you can use the equation editor to lay out your terms and then put the equation into the sketch using the terms with, with " around each term. This makes your disk editable without having to re-copy/paste the equation in.
You could probably cut weight by reducing the number of interior bearings on the cycloidal disks from 5 to 3 . I doubt you would have integrity issues if were applying 67% more force on each bearing. With 19.5 Nm torque (assuming the bearings are at 2 mm from the center of the disk) each bearing is applying ~ 76N of force. If you reduce to 3 bearings they would each apply 127 N, which is pretty reasonable. I bet the cycloidal disks could manage that pretty well and cut the number of bearings by 4.