@@3DprintedLife you are promoting a company whose cloud logistics division directly supports ICE kidnapping brown children, among other human rights abuses.
For interfacing shafts with 3D printed parts in a VORON, we use 20T pulleys with the top flange popped off, and with screws driven in to fully stabilize it. It gives you a solid and cheap interface between a steel shaft and a printed part.
On yeah I built a voron extruder a while back and remember that feature. I'm not sure I have room but I'll definitely play around with this and see if I can make it work. Thanks for the tip!
@@3DprintedLife if you’re tight on space, use the 16T gear. They are much smaller, but still provide better interface than the nut + set screw approach.
Thank you! Yeah gearboxes are a lot of fun. The efficiency is definitely not great, maybe 60% or so right now. I intentionally went for a very tight tolerance between the flex spline and housing teeth to keep backlash and deflection to a minimum. I'll be doing a lot more testing with this gearbox once I get my control board made and start building the arm, so I can throw in some efficiency tests too! Perhaps even comparing performance and efficiency of different materials used for the flex spline.
@@3DprintedLife I'd love to see a video(s) with the mechanism printed out of different filament and tests of efficiency and longevity. Would be cool to see the arm casing printed out of Carbon Fiber PLA. Thanks for the awesome content!!
yeah use a planetary gear box instead, actually you can buy a nema 17 with a planetary gear box with more torque than this off the shelf it has a 20:1 ratio, It might cost a little more but its metal and wont brake. if you had one on a nema 23 well you would have close to 2x the torque of what this one you made or you could have similar toruqe with a 10:1 ratio.
Thanks for watching! If you enjoyed, check out some of my other videos like a pneumatic 3d printed liquid piston engine! ruclips.net/video/QdH_su27DYg/видео.html
@@jbarnhart2653 well labor for me yes, it's way more. But I'm doing all the hard work and releasing as open sourced so others don't have to. Materials and assembly time will be significantly under even the lower priced arm I showed (which btw that is the price of raw parts, you still have to build that arm yourself too)
@@3DprintedLife Yes but open source for people who have all the same equipment that you do. More expense. I've done a ton of 3D printing and fabrication and even the best looking, most popular and in my opinion well thought out projects and stl's require "tweaking", in other words a lot of the work you did, in this video. How much do you "pay yourself"? When I'm asked to look at stuff like this, it's at least $40 an hour. Price goes up the more tech and skill needed...only way down is by making a bunch at once...(no) but even then, I don't think that except as a "challenge" project that it would be economic. "Build your own..." usual involves love of the work or better quality, rather than saving money or time.
Brilliant design! Also, I really like that you do everything by yourself. I believe the experience you gain and all those things you learn when doing this is super valuable and the best thing is that noone can take it from you. I also applaud that you share with the rest of the world here. A true passionate inventor/tinkerer, enjoying the process.
I learned a while back that I'm gonna end up spending more and taking longer by doing everything myself, but the experience and knowledge gained more than makes up for it. Usually. Sometimes I go too far down the rabbit hole and never finish projects
If you're doing a lot of structural prints I'd recommend upgrading to a direct drive extruder. Even after calibrating my ESTEPS my bowden setups always have a tiny amount of underextrusion at the start of a tool path. This introduces a bunch of weak spots for a failures to occur, which will be even more significant under cyclic loading due to the way fatigue cracks form. I like the animated CAD visuals keep up the good work!
Yeah, good point. I've spent a lot of time tuning to correct for this, but going direct drive would make everything a lot easier and stronger. Maybe one day :) And thank you!
I have zero idea about the math, I would not know where to even begin when I had to make something like this and then I would fail again when it comes to the software controlling the arm. And yet here I am, watching and marvelling at all this expertise and at the fact that people are actually making things like this at home. I mean, I have a 3D printer, I am sooo proud about the first thing I've made in Tinkercad myself, a soap holder, and then I see this guy... :D
Very Nice! Most will never realize the amount of work and talent it took to do this. Hopefully some sponsors will see this and drown you with new toys.
Great video. Still appreciate seeing the progression, rather than skipping to the finished version. I'm sure by the time you have printed 6 you will have refined the design much more and need to reprint the first one! Don't forget that while the single design is attractive from a scalability perspective, your robot arm will thank you for making lighter gears near the effector and beefier ones near the shoulder. And with your parts being printed tooling = free so CAD time is your only real excuse not to have a couple or few sizes.
Yeah all good points! I'm only planning on using this gearbox on the two main hinge joints for exactly that reason. I'm still figuring out how I'll handle wrist motion and the rotary platform, but I expect I'll go with a simpler reduction technique like just using belts.
Cool project, great video quality, and most importantly, you were clear and concise about your thought process, you really brought us along for the build. You got my sub fam. Keep this up
I'm making a robot arm for a school project and will probably also go with the steppers. This might come very useful. Thank you for making it open source.
In my experience Taulman Bridge is an absolute nightmare to print with, you need very high temperature enclosure to prevent warping. I switched to Polymaker CoPA and PA6-CF, both of which prints almost as easy as PLA, sticks perfectly to glue stick, doesn't warp, etc. As long as you keep the filament dry and your printer hotend can handle the temperature.
I just got a harmonic drive from harmonic drive. Without measuring (they are tiny), the teeth do look involute. The drawing and 3d model they gave me doesn't include any info on them, I suspect its their secret. Nice work on this, very impressive to come up with your own new design on something this complicated.
Oh boy this video was looong overdue! I'm so tired of seeing all the "creative" minds of RUclips just skimming the HD subject and going for the pot design, while what you're using (called the "pancake") is so much more compact and nicer! I know because I built a full metal one some years ago and have been waiting to see somebody do the 3d-printed pancake too ever since... So good one, bud! 😍
Working with gear boxes daily i never thought to see them made out of plastic. I've never heard of harmonic drives in my field of gearboxes. The design principle is very similar to cycloidal drive gearbox. neat work!
Found in your discord, this was all printed in PLA! Not sure if you mentioned that in the video, I looked for it but couldn't find it. You should probably put that in the description.
Thank you for this video. I appreciate seeing how it takes time and many iterations to develop something. It helps me to see this because you know a lot more than me and are way smarter than me and, while my projects are far simpler, I have to keep revising my designs over and over. They just keep coming out not-quite-right. Or, just wrong. I keep thinking smarter people just make things perfect right away and I'm doomed to fail because I keep making mistakes. It's good to see that nobody is perfect, but you keep doing awesome things and making awesome progress. Thank you again!
I enjoyed this video more than I planned on. Kudos for the cinematography. I enjoyed it all. You actually finally gave me an excuse to start working on a robot arm. Although I had been trying to stay away from the trials and tribulations that is a 3D printer, I'm at least thinking about it now. I love the narration, the troubleshooting, the production. I'm so impressed with all your skills. Keep it up.
Built a robotic hand when I was like thirteen, back in like 2008 maybe. It was all wood, with programmable tumblers with a live control override set up like a piano. It's glue didn't hold up to humidity long, so literally just fell apart over a wet summer. It used pull strings and springs, and later bands, very close to how a real human hand works. Can just copy the human body and only run a single moter to bring it to life with this method. To add a computerized element, the tumbler spokes can just be raised and lowered, still powered of the single moter. I wouldn't be surprised if someone else is already building something like it.
Well it's open sourced! Play around with it, make it better, or make your own from scratch. Don't let your dreams be dreams! github.com/DDeGonge/OS-ARM/tree/main/cad/HarmonicDrive75mm
I create 3d printed torque sockets for capping bottles at work and using a 3/8 inch drive, we are regularly able to place 60 in'lb of force on them without any problems. They obviously wear after a while but we were really surprised at how much they could take for as long as they can take it.
For the gear tooth profile, you should use a stub profile (addendum = 0,8, dedendum =1) and 30° pressure angle as it seems to be used by commercial strain wave gearboxes.
I've printed with bridge.. make sure to put it and a heated dryer for ideal printing. I put the heating unit of a food dehydrator in a box with my filament overnight. I would use a more aggressive tooth pattern as well. The bridge nylon can either smooth the surface enough to become a bushing or with the friction weld moving parts into stationary parts.
I recently picked up a few 3 phase BLDC motors that are used in camera gimbals to provide pan and tilt. They have 12 windings and 14 magnets so there are 42 steps to go the full 360 degrees. They were less than $5 apiece. They are quite compact and might be able to be mounted internally in a harmonic drive such as yours. I may have to give it a try. Just to be clear, the way your design works is that the belt moves inside the two cups, but the moving cup has more teeth than the fixed cup.
Use either a filled nylon like nylonX (carbon fiber) or nylonG (glass fiber) to reduce warping when you do decide to print the new flex spline. It will also add rigidity
A thought about the design: you have two big gears with teeth facing inwards. The static one is fixed to the motor and has x teeth. The rotating one has x-2 teeth so you have two “matching points” of the big gears. The number of teeth of the “belt” doesn’t matter that much. If you would replace the belt with a planetary gear with 2 planets, the ratio of the big gears as well as the whole working principle would stay the same. But you could skip the ball bearings and instead of the rotor and the belt you would have gears, that give you a far higher material thickness.
That's called compound planetary gear and it works great. More efficient than this, although it does have backash like all rigid gears which is not good for a robot arm.
full disclosure: i'm not into printing and modeling or any of the stuff, but i do enjoy watching these type of videos. but in the video the thing where the 2 bearing are attached to, the thing that broke? around 4:47, i think you were using a nut and bolt to tighten it to attach to the motor. can't you like shave the top of the rod and flatten it? at like between 4:44 - 4:45. and the piece to attach to it a little flat at the top too, like a semi circle, so that you wouldn't need to use a nut and bolt anymore. sort of like how the blades of an electric fan are mounted to the motor. sorry if i can't explain it very well.
Yeah I did try to use that flat on the motor shaft if that's what you mean, but the plastic is just too soft and easily deforms. I have some ideas on how to improve this though!
These vids make you appreciate how much work went into prototyping before the advent of 3d printers. Doesn't work? print another slightly different version. Old school, spend another 3 days redesigning and machining the part.
I made a wooden robotic camera arm for a school project using DC motors from old power tools (18V and 12V I think). Very powerful, but loud and not very precise. For the gearing I just used a nut and a threaded rod at the end of the tool's gearbox. The threaded rod was good because it also stopped the arm from moving up or down by itself. The arm was pretty bad as the wooden structure was very unstable. It was also very unprecise, but I think that could easily be fixed with a few sensors and software. It cost me about 300$ (including the Arduino and everything else).
That's pretty interesting. Whiteline's polyurethane blend called includes PTFE (Teflon) so their bushings theoretically become self lubricating in the same way you're using tape here.
Love it, can’t wait to see the final price of the robot arm and the footage it will make with a camera mounted on it! I would say you earned a new subscriber but I was already subscribed! Haha I guess it was meant to be.
Did you try any alternative belt tooth profiles? If you were using standard 3D printer belts a lot use GT2, which is curvilinear profile. The label on the belt you were showing off around 2:45 unfortunately didn't say the type. Curvilinear is quieter and higher torque, but at the cost of backlash. I'm curious if using a trapezoidal profile belt like XL/MXL would be helpful, although I could also see inverting the belt messing with the potential benefits.
If you got delamination problems you can bake prints on the printed. Just cover the print with something (a box/aluminium foil) and set the bed to 80°C then leave it for 30 minutes. I do not know how much this increases strength, but on a basic make 2 parts and try to break both basis it makes a noticeable difference
Just want to mention, it makes sense you were getting too much slop with the rubber belt. They're designed to be run in tension and have a certain amount of expansion expected. So without that tension it would flex too well.
You could use an ordinary timing belt turned inside out, as a flexible gear ring. You could match two timing belts with slightly different number of teeth to match your gear ratio.
Don't be afraid of printing Nylon. Some "help": - Make sure that the filament is dry (seriously, 50-80°C for around 10h and print directly from dry storage) - PVA gluestick on a Prusa Satin sheet works really solid for me (93mm high part on a 2mm wide ring measuring 18mm outside holds like glue) - Enclosures help a lot (even cheap ones) - Nylon oozes a lot, factor in some time to play with retraction settings to reduce blobs and zits - Using a bigger nozzle (using 0.6 right now) can help with layer adhesion - Manual post processing might be required - Try to design/print the part so mechanical stress goes against the layers, Nylon with bad layer adhesion tends to break along the layers
Nice, been wanting to try a design like this myself. I think the only 'downside' ive seen mentioned for this type of design is that the gearing ratios tend to be 'too high', limiting your top speed. But for most applications I can think of, doing it a little slower is an acceptable compromise. The DIY gearbox community definitely needs a well-worked out design like this. Too bad the belt was not cooperating. A printed one is never going to be ideal I think; youve got all the shear concentrated in two spots along the layer lines. Perhaps a belt with a somewhat stiffer backing added to it would work; or otherwise a printed version (kinda annoying to be stuck with existing belt teeth profiles not optimized for this purpose), but hot melted onto a thin metal ring like a slice of a thin walled pipe, or some other stiff material capable of transferring tons of shear. Not easy to come up with off the shelf solutions there really... could consider doing something with composites but not im seeing any easy ways of attaining a nice uniform pattern with the right diagonal fiber orientation without making some kind of custom winding jig... sounds complicated. DMLS titanium would be a lot less labor intensive; but yeah you mentioned cheap.... But if youve got a good and stiff backing, you can crank up the with the width of the belt and attain some really crazy torques (without a stiff backing all the load is probably going to concentrate where the teeth meet and mess them up there).
Yeah I was considering adhering some spring steel to the backside of a belt to get the best of both worlds. It would still have more backlash and deflection than the plastic flex spline since the teeth of the belt will have some give, but it would probably run quieter, smoother, and longer. But finding the right sized closed loop belt was really tricky, especially finding extra wide ones that are this small. I'm pretty sure printing nylon will be the way to go, since nylon has closer to isotropic properties when 3D printed which should improve the shear strength of the part, and I can make the spline any size I want. Plus, the better wear resistance and more naturally lubricious properties should be a good compromise between rigidity and lifetime. But I won't know for sure if that's enough until I build up the arm and start testing it for real!
@@3DprintedLife Agreed, nylon will be a pretty decent option. Armadillo might be another good one if the nylon turns out to be to warpy; its got properties quite comparable to nylons but is super easy to print.
That's seriously impressive. The cost of the final product is a bit misleading though; sure, the parts themselves might cost a fraction, but the work & labor that went into the R&D is surely somewhere at 300 manhours (wild guess), not counting the time spent learning about this at school/uni :) But again, it is super cool to see something like this happen, and the printer's resolution is also great, with there being no need for surface treatment
Yeah that's why I'm specifying the cost is just for materials. But the goal isn't just to make this for myself, but to share the design with all so they can replicate my results for the actual price of materials!
hey ! great work. Nylon isn't really hard to print. I recently printed Taulman's alloy 910 Nylon on Ender3 v2 using stock glass bed and elmer's purple glue stick. It stuck pretty well without warping. The sample was flat on its long side, a 1 cm by 1cm profile that was extruded for 15cm.
I was making planetary gearboxes for my plastic extruder. It kept breaking and eventually reached the limit of the plastic regardless of my design. I only had one stepper driver as well so what I did was just make two them and wired them in parallel. They were nema 17s and the stepper driver was only a 2A but it never popped the driver and never broke the gearbox anymore. Other things broke 😅. I feel your pain in this video with reiterating though. You tend to learn so much from it though. Great video thanks.
Commercial boxes often use an eliptical bearing instead of rollers, which increases the number of theeth in contact. But, where do you get an eliptical bearing? You take a thin section bearing and squash it a bit in a vice! 3D Print an eliptical boss to fit tightly in the centre of the bearing to stop it returning to circular. With this approach, you can go back to your original toothed belt design, as the belt is stretched over the bearing, which removes (most of) the backlash. The combination of the two significantly increase the overall strength.
Oh! Smart! I like the belt idea... but what if we used TPU for the flex spline? This stuff is super resilient and still pretty bendy. I wanna say I'd try tpu for the flex, and resin printed abs-like anycubic resin for the hard parts. Its so tough, but unlike some of the resins I've tried it literally will flex in half without snapping if you make it thin enough to bend. Nutty stuff...
@@3DprintedLife cool! The only thing they taught me in high school is... Nothing. Professors in electronics and computer science in Italy sucks sheet... How i wish to have actually le as red something useful instead of systems history and... C... Not C++, not C# just C and assembly. In 2015. Useless as hell.
This is great, and no I want to build one. I will note though that while I'm far from an expert, iirc nylon is actually quite soft compared to petg or pla. The reason it's used a lot is that while it's soft and will flex it won't break or crease as easily and is better at certain temps. If you want it to be rigid, I think you'll need the fiber reinforced stuff.
Yeah that is true, though I think the increased durability and wear resistance would justify the small increase in gearbox deflection under loads. Fiber reinforced is very abrasive so that would work well at first but I think the performance would fade quick!
To improve the durability of your prints, you could enable some "elephant's foot compensation" (no, not _that_ Elephant's Foot), and pack your printed part with some sand so that you can temper and anneal it. Don't forget to freeze-dry your nylon filament, immediately store it in a vacuum bag with some desiccant, and when you print with it, feed the filament directly from your dehydrator. Also make sure you have a dehumidifier running in your room during all printing operations, and coat your printed Nylon parts with a hydrophobic substance like Neverwet.
This is actually the most impressive 3D printed strain wave drive I've seen on RUclips. Hopefully the video gets the recognition it deserves.
Thanks I really appreciate that! :)
cool but pretty useless functionally when you could just use a direct drive motor
and no, they aren't "100 dollars"
@@3DprintedLife you are promoting a company whose cloud logistics division directly supports ICE kidnapping brown children, among other human rights abuses.
@@Blox117um what?
I knew I recognized that voice. Hah. Sweet design dude. Using the belt is very clever.
Thanks Shane, hope you've been doing well and congrats on your insane success! If you ever find yourself with one too many CNCs just let me know 😁
@@3DprintedLife😂
For interfacing shafts with 3D printed parts in a VORON, we use 20T pulleys with the top flange popped off, and with screws driven in to fully stabilize it. It gives you a solid and cheap interface between a steel shaft and a printed part.
On yeah I built a voron extruder a while back and remember that feature. I'm not sure I have room but I'll definitely play around with this and see if I can make it work. Thanks for the tip!
@@3DprintedLife if you’re tight on space, use the 16T gear. They are much smaller, but still provide better interface than the nut + set screw approach.
@@mr_voron yeah true, great suggestion I'm definitely going to try and make that work, thanks!
This is what I was thinking immediately after seeing the screw/nut combo break!
do you have a link to this? I can't picture it in my head
Great work! Makes me want to work on gears and gearboxes again. Any idea on the efficiency of such a design?
Thank you! Yeah gearboxes are a lot of fun. The efficiency is definitely not great, maybe 60% or so right now. I intentionally went for a very tight tolerance between the flex spline and housing teeth to keep backlash and deflection to a minimum. I'll be doing a lot more testing with this gearbox once I get my control board made and start building the arm, so I can throw in some efficiency tests too! Perhaps even comparing performance and efficiency of different materials used for the flex spline.
@@3DprintedLife I'd love to see a video(s) with the mechanism printed out of different filament and tests of efficiency and longevity. Would be cool to see the arm casing printed out of Carbon Fiber PLA. Thanks for the awesome content!!
yeah use a planetary gear box instead, actually you can buy a nema 17 with a planetary gear box with more torque than this off the shelf it has a 20:1 ratio, It might cost a little more but its metal and wont brake. if you had one on a nema 23 well you would have close to 2x the torque of what this one you made or you could have similar toruqe with a 10:1 ratio.
@@3DprintedLife You can replace the flexible spline with a planetary gear and reduce the output teeth by the number of planet gears.
@@Side85Winder But what about the backlash? That's the most important thing on a robot arm.
Thanks for watching! If you enjoyed, check out some of my other videos like a pneumatic 3d printed liquid piston engine! ruclips.net/video/QdH_su27DYg/видео.html
amazing
try vapor bath of carburetor cleaner on the flexspline (pla) instead of the nylon ?
Nice, but your labour materials and equipment use seem to add up to a LOT more than the price of the lower cost arm you showed.
@@jbarnhart2653 well labor for me yes, it's way more. But I'm doing all the hard work and releasing as open sourced so others don't have to. Materials and assembly time will be significantly under even the lower priced arm I showed (which btw that is the price of raw parts, you still have to build that arm yourself too)
@@3DprintedLife Yes but open source for people who have all the same equipment that you do.
More expense.
I've done a ton of 3D printing and fabrication and even the best looking, most popular and in my opinion well thought out projects and stl's require "tweaking", in other words a lot of the work you did, in this video.
How much do you "pay yourself"? When I'm asked to look at stuff like this, it's at least $40 an hour. Price goes up the more tech and skill needed...only way down is by making a bunch at once...(no) but even then, I don't think that except as a "challenge" project that it would be economic.
"Build your own..." usual involves love of the work or better quality, rather than saving money or time.
I JUST watched Tenet last night! Your intro was awesome!
Thanks glad you liked it! I was definitely inspired by Tenet for that one
🤣🤣🤣❤👍...watched already twice!
And thought exactly the same!
@@3DprintedLife that movie was fantastic
This is very well done! The testing with the load cell is genius.
Thank you! Btw I've been following you for a few months now, really enjoying your content and can't wait for more! You deserve way more subs!
@@3DprintedLife mentioning that you subbed to this seemly random cool guy earned you AND him a new sub
excited to see the full robot arm in action! i always wanted one of those automotive car building robo arms!
I am too! :D
Brilliant design! Also, I really like that you do everything by yourself. I believe the experience you gain and all those things you learn when doing this is super valuable and the best thing is that noone can take it from you. I also applaud that you share with the rest of the world here. A true passionate inventor/tinkerer, enjoying the process.
I learned a while back that I'm gonna end up spending more and taking longer by doing everything myself, but the experience and knowledge gained more than makes up for it. Usually. Sometimes I go too far down the rabbit hole and never finish projects
@@3DprintedLife I know. Have been down that rabbit hole way too many times.
If you're doing a lot of structural prints I'd recommend upgrading to a direct drive extruder.
Even after calibrating my ESTEPS my bowden setups always have a tiny amount of underextrusion at the start of a tool path. This introduces a bunch of weak spots for a failures to occur, which will be even more significant under cyclic loading due to the way fatigue cracks form.
I like the animated CAD visuals keep up the good work!
Yeah, good point. I've spent a lot of time tuning to correct for this, but going direct drive would make everything a lot easier and stronger. Maybe one day :) And thank you!
For anyone else with the same problem, esteps won't fix it, but linear advance will. You can also reduce speed and acceleration.
You should design in stabilizing rollers at 90 degrees to the main bearings. That would function as a limiter for oscillations at high rpm.
Ahh good tip, I was wondering why some other designs had those bearings. thanks!
This shows a great amount of dedication!
I have zero idea about the math, I would not know where to even begin when I had to make something like this and then I would fail again when it comes to the software controlling the arm. And yet here I am, watching and marvelling at all this expertise and at the fact that people are actually making things like this at home. I mean, I have a 3D printer, I am sooo proud about the first thing I've made in Tinkercad myself, a soap holder, and then I see this guy... :D
Very Nice! Most will never realize the amount of work and talent it took to do this. Hopefully some sponsors will see this and drown you with new toys.
Great video. Still appreciate seeing the progression, rather than skipping to the finished version. I'm sure by the time you have printed 6 you will have refined the design much more and need to reprint the first one! Don't forget that while the single design is attractive from a scalability perspective, your robot arm will thank you for making lighter gears near the effector and beefier ones near the shoulder. And with your parts being printed tooling = free so CAD time is your only real excuse not to have a couple or few sizes.
Yeah all good points! I'm only planning on using this gearbox on the two main hinge joints for exactly that reason. I'm still figuring out how I'll handle wrist motion and the rotary platform, but I expect I'll go with a simpler reduction technique like just using belts.
This guy: designs, iterates, builds and programs complete, strong robot arm.
Me: happy my vase turned out.
You go, guy, kudos, cheers and congrats!!
Thank you!
Awesome video once again! I can’t wait to see the next one!
Thank you random internet person!
Never even heard of strain wave motors until this video, great stuff and just subscribed!
Cool project, great video quality, and most importantly, you were clear and concise about your thought process, you really brought us along for the build. You got my sub fam. Keep this up
That's pretty epic 😍. Didn't even know that existed until now.
I'm making a robot arm for a school project and will probably also go with the steppers. This might come very useful. Thank you for making it open source.
Outstanding performance from your harmonic drive design! I would be proud of it!
Thank you!
I like that you show your own failures and go for a lower budget range ,most channels like this try and do it by whatever means possible
In my experience Taulman Bridge is an absolute nightmare to print with, you need very high temperature enclosure to prevent warping.
I switched to Polymaker CoPA and PA6-CF, both of which prints almost as easy as PLA, sticks perfectly to glue stick, doesn't warp, etc.
As long as you keep the filament dry and your printer hotend can handle the temperature.
I just got a harmonic drive from harmonic drive. Without measuring (they are tiny), the teeth do look involute. The drawing and 3d model they gave me doesn't include any info on them, I suspect its their secret.
Nice work on this, very impressive to come up with your own new design on something this complicated.
Ah awesome, good to have some insider info haha thanks for sharing that!
Oh boy this video was looong overdue! I'm so tired of seeing all the "creative" minds of RUclips just skimming the HD subject and going for the pot design, while what you're using (called the "pancake") is so much more compact and nicer! I know because I built a full metal one some years ago and have been waiting to see somebody do the 3d-printed pancake too ever since... So good one, bud! 😍
Thanks! And oh cool I was wondering if this variation had a specific name haha :D Glad you enjoyed!
@@3DprintedLife www.harmonicdrive.net/products/component-sets/pancake/fr-2
Working with gear boxes daily i never thought to see them made out of plastic. I've never heard of harmonic drives in my field of gearboxes. The design principle is very similar to cycloidal drive gearbox. neat work!
Thank you!
Looks very promising, incredible how much 3d printed parts actually endure before failure
It really is, the tech has come a long way!
@@3DprintedLife Bro this is such a sick project love your work
Found in your discord, this was all printed in PLA! Not sure if you mentioned that in the video, I looked for it but couldn't find it. You should probably put that in the description.
Awesome, the video with the finished arm is going to be insane!
Thanks, I'm very excited for it! And thanks so much for the support on Patreon!
@@3DprintedLife Glad to give what little support I can :)
Awesome video and thorougher explanation of the thoughts behind design process.
Glad you liked it!
Satisfying the design manufacture and testing !! keep up the good work.
Good job! Enjoyed your tongue-in-cheek style. Fits your fellow engineers very well. 😉
Thank you for this video. I appreciate seeing how it takes time and many iterations to develop something. It helps me to see this because you know a lot more than me and are way smarter than me and, while my projects are far simpler, I have to keep revising my designs over and over. They just keep coming out not-quite-right. Or, just wrong. I keep thinking smarter people just make things perfect right away and I'm doomed to fail because I keep making mistakes. It's good to see that nobody is perfect, but you keep doing awesome things and making awesome progress. Thank you again!
No one is perfect, even though I try! Just wait until my next video, there is going to be a lot of pain and a lot of failure in it :D
@@3DprintedLife I'm going to be in it??? LOL **weeping**
I enjoyed this video more than I planned on. Kudos for the cinematography. I enjoyed it all. You actually finally gave me an excuse to start working on a robot arm. Although I had been trying to stay away from the trials and tribulations that is a 3D printer, I'm at least thinking about it now.
I love the narration, the troubleshooting, the production. I'm so impressed with all your skills. Keep it up.
Thanks so much I really appreciate it, glad you enjoyed! :)
random reminder that CNC mills also exist
Haven't the slightest clue about any of that BUT really well made and explained video. New subscriber UnLocked.
Glad you enjoyed it, welcome! :)
Subbed! Your design, demo, and open source mindset is amazing!
Thanks I appreciate that!
9:50
Almost did the good old knuckle slam there
worthy of a subscription! hope you finish it, eventually! ty
Thank you, and welcome!
Built a robotic hand when I was like thirteen, back in like 2008 maybe. It was all wood, with programmable tumblers with a live control override set up like a piano. It's glue didn't hold up to humidity long, so literally just fell apart over a wet summer. It used pull strings and springs, and later bands, very close to how a real human hand works. Can just copy the human body and only run a single moter to bring it to life with this method. To add a computerized element, the tumbler spokes can just be raised and lowered, still powered of the single moter. I wouldn't be surprised if someone else is already building something like it.
This has been my dream project for so long!
Well it's open sourced! Play around with it, make it better, or make your own from scratch. Don't let your dreams be dreams! github.com/DDeGonge/OS-ARM/tree/main/cad/HarmonicDrive75mm
@@3DprintedLife I know! Thank you so much!
I am an old cannon cobbler and pocket watch tinkerer, so this is outside my interests but as an engineer and novice 3d printer, this is wonderful!
I create 3d printed torque sockets for capping bottles at work and using a 3/8 inch drive, we are regularly able to place 60 in'lb of force on them without any problems. They obviously wear after a while but we were really surprised at how much they could take for as long as they can take it.
Very good video. I'm looking forward to seeing updates on the arm :)
Thanks!
Wow! That was super fun and enjoyable to watch
Glad you enjoyed it!
Really enjoyed the video, thank you. You are a real engineer.
Thanks!
Good job! You have good production and good content, the work will payout. Keep it up!
Thanks, will do!
Looks like some of the pricing on the arms in the beginning was legit. Rd in a bigger company plus higher quality parts. Really solid work man.
So glad to find your channel.
Love your work. Keep up!
For the gear tooth profile, you should use a stub profile (addendum = 0,8, dedendum =1) and 30° pressure angle as it seems to be used by commercial strain wave gearboxes.
This is the first time I've seen your channel great video subscribed!
Awesome, thank you and welcome to the channel!
hi bro, that´s a lot of hard work!! thank for sharing your knowledge
Thanks glad you enjoyed!
Great design and development. Well done.
Thank you very much!
I've printed with bridge.. make sure to put it and a heated dryer for ideal printing. I put the heating unit of a food dehydrator in a box with my filament overnight. I would use a more aggressive tooth pattern as well. The bridge nylon can either smooth the surface enough to become a bushing or with the friction weld moving parts into stationary parts.
your willingness to embrace so many failures is what makes you a goat
I recently picked up a few 3 phase BLDC motors that are used in camera gimbals to provide pan and tilt. They have 12 windings and 14 magnets so there are 42 steps to go the full 360 degrees. They were less than $5 apiece. They are quite compact and might be able to be mounted internally in a harmonic drive such as yours. I may have to give it a try.
Just to be clear, the way your design works is that the belt moves inside the two cups, but the moving cup has more teeth than the fixed cup.
Oooh cool idea!
Use either a filled nylon like nylonX (carbon fiber) or nylonG (glass fiber) to reduce warping when you do decide to print the new flex spline. It will also add rigidity
Hugely impressive young man, you give far too much credit to others.
I appreciate it, but I really couldn't have done this without building off of what other's have done!
I didn't even know these drives existed, fascinating
Fantastic video, all the best for your projects; Keep up the great work!
Thank you!
Patterns
Mom and Dad.
Priceless!
Looking forward to this build
A thought about the design: you have two big gears with teeth facing inwards. The static one is fixed to the motor and has x teeth. The rotating one has x-2 teeth so you have two “matching points” of the big gears. The number of teeth of the “belt” doesn’t matter that much. If you would replace the belt with a planetary gear with 2 planets, the ratio of the big gears as well as the whole working principle would stay the same. But you could skip the ball bearings and instead of the rotor and the belt you would have gears, that give you a far higher material thickness.
That's called compound planetary gear and it works great. More efficient than this, although it does have backash like all rigid gears which is not good for a robot arm.
I think we found the new 3D printing RUclips super genius
Incredible! Congratulations! And thank you!
Impressive work my dude!
Appreciate it!
Excellent work. very helpful to see so many destructive test also
Breaking things is my favorite!
full disclosure: i'm not into printing and modeling or any of the stuff, but i do enjoy watching these type of videos. but in the video the thing where the 2 bearing are attached to, the thing that broke? around 4:47, i think you were using a nut and bolt to tighten it to attach to the motor. can't you like shave the top of the rod and flatten it? at like between 4:44 - 4:45. and the piece to attach to it a little flat at the top too, like a semi circle, so that you wouldn't need to use a nut and bolt anymore. sort of like how the blades of an electric fan are mounted to the motor. sorry if i can't explain it very well.
Yeah I did try to use that flat on the motor shaft if that's what you mean, but the plastic is just too soft and easily deforms. I have some ideas on how to improve this though!
These vids make you appreciate how much work went into prototyping before the advent of 3d printers. Doesn't work? print another slightly different version. Old school, spend another 3 days redesigning and machining the part.
I made a wooden robotic camera arm for a school project using DC motors from old power tools (18V and 12V I think). Very powerful, but loud and not very precise. For the gearing I just used a nut and a threaded rod at the end of the tool's gearbox. The threaded rod was good because it also stopped the arm from moving up or down by itself.
The arm was pretty bad as the wooden structure was very unstable. It was also very unprecise, but I think that could easily be fixed with a few sensors and software. It cost me about 300$ (including the Arduino and everything else).
That's pretty interesting. Whiteline's polyurethane blend called includes PTFE (Teflon) so their bushings theoretically become self lubricating in the same way you're using tape here.
great, want to see further progress.
Excellent video, very entertaining and also very well explained ! Really cool stuff !
Thank you!
First time here! Really cool video man! I scrolled down and expected more subs than 11.8k! lol ...Here's another one! 👍
Oh don't worry, it's at 11.801k now! :D welcome!
You are EXTREMELY patient!
Oh, you'd be surprised haha
@@3DprintedLife Well in that case you're a fabulous video editor :) Well done, excellent piece of work, great video.
Love it, can’t wait to see the final price of the robot arm and the footage it will make with a camera mounted on it!
I would say you earned a new subscriber but I was already subscribed! Haha I guess it was meant to be.
Subscribed! Very helpful information here buddy. Especially if you are on a very tight budget and want to build a robot.
Thanks for the sub, and that's the goal!
This would make a great gate opener!
Did you try any alternative belt tooth profiles? If you were using standard 3D printer belts a lot use GT2, which is curvilinear profile. The label on the belt you were showing off around 2:45 unfortunately didn't say the type. Curvilinear is quieter and higher torque, but at the cost of backlash. I'm curious if using a trapezoidal profile belt like XL/MXL would be helpful, although I could also see inverting the belt messing with the potential benefits.
If you got delamination problems you can bake prints on the printed. Just cover the print with something (a box/aluminium foil) and set the bed to 80°C then leave it for 30 minutes.
I do not know how much this increases strength, but on a basic make 2 parts and try to break both basis it makes a noticeable difference
Just want to mention, it makes sense you were getting too much slop with the rubber belt. They're designed to be run in tension and have a certain amount of expansion expected. So without that tension it would flex too well.
the Greatest video i ever seen please keep doing this project i want to build the same thing
Thank you, don't worry I'm not stopping!
You could use an ordinary timing belt turned inside out, as a flexible gear ring. You could match two timing belts with slightly different number of teeth to match your gear ratio.
Don't be afraid of printing Nylon.
Some "help":
- Make sure that the filament is dry (seriously, 50-80°C for around 10h and print directly from dry storage)
- PVA gluestick on a Prusa Satin sheet works really solid for me (93mm high part on a 2mm wide ring measuring 18mm outside holds like glue)
- Enclosures help a lot (even cheap ones)
- Nylon oozes a lot, factor in some time to play with retraction settings to reduce blobs and zits
- Using a bigger nozzle (using 0.6 right now) can help with layer adhesion
- Manual post processing might be required
- Try to design/print the part so mechanical stress goes against the layers, Nylon with bad layer adhesion tends to break along the layers
I really appreciate all the tips, thanks!
Nice, been wanting to try a design like this myself. I think the only 'downside' ive seen mentioned for this type of design is that the gearing ratios tend to be 'too high', limiting your top speed. But for most applications I can think of, doing it a little slower is an acceptable compromise. The DIY gearbox community definitely needs a well-worked out design like this.
Too bad the belt was not cooperating. A printed one is never going to be ideal I think; youve got all the shear concentrated in two spots along the layer lines. Perhaps a belt with a somewhat stiffer backing added to it would work; or otherwise a printed version (kinda annoying to be stuck with existing belt teeth profiles not optimized for this purpose), but hot melted onto a thin metal ring like a slice of a thin walled pipe, or some other stiff material capable of transferring tons of shear. Not easy to come up with off the shelf solutions there really... could consider doing something with composites but not im seeing any easy ways of attaining a nice uniform pattern with the right diagonal fiber orientation without making some kind of custom winding jig... sounds complicated. DMLS titanium would be a lot less labor intensive; but yeah you mentioned cheap....
But if youve got a good and stiff backing, you can crank up the with the width of the belt and attain some really crazy torques (without a stiff backing all the load is probably going to concentrate where the teeth meet and mess them up there).
Yeah I was considering adhering some spring steel to the backside of a belt to get the best of both worlds. It would still have more backlash and deflection than the plastic flex spline since the teeth of the belt will have some give, but it would probably run quieter, smoother, and longer. But finding the right sized closed loop belt was really tricky, especially finding extra wide ones that are this small. I'm pretty sure printing nylon will be the way to go, since nylon has closer to isotropic properties when 3D printed which should improve the shear strength of the part, and I can make the spline any size I want. Plus, the better wear resistance and more naturally lubricious properties should be a good compromise between rigidity and lifetime. But I won't know for sure if that's enough until I build up the arm and start testing it for real!
@@3DprintedLife Agreed, nylon will be a pretty decent option. Armadillo might be another good one if the nylon turns out to be to warpy; its got properties quite comparable to nylons but is super easy to print.
Oh cool I hadn't heard of that filament before, thanks!
That's seriously impressive. The cost of the final product is a bit misleading though; sure, the parts themselves might cost a fraction, but the work & labor that went into the R&D is surely somewhere at 300 manhours (wild guess), not counting the time spent learning about this at school/uni :)
But again, it is super cool to see something like this happen, and the printer's resolution is also great, with there being no need for surface treatment
Yeah that's why I'm specifying the cost is just for materials. But the goal isn't just to make this for myself, but to share the design with all so they can replicate my results for the actual price of materials!
hey ! great work. Nylon isn't really hard to print. I recently printed Taulman's alloy 910 Nylon on Ender3 v2 using stock glass bed and elmer's purple glue stick. It stuck pretty well without warping. The sample was flat on its long side, a 1 cm by 1cm profile that was extruded for 15cm.
Exactly what I needed thank you!
Nylon isn't too bad! Just dry it and it'll print quite nice. Loves to warp though, I found printing draft shield helps, an enclosure would be better.
You earned a sub sir ! Looks like a channel ill be lingering in for the night ♡
Why thank you! Lucky for you I don't have enough content to keep you up all night...yet :D
@@3DprintedLife i came back to watch this again and see if you have made any progress in ur robot arm ? if yes lets us know
I was making planetary gearboxes for my plastic extruder. It kept breaking and eventually reached the limit of the plastic regardless of my design. I only had one stepper driver as well so what I did was just make two them and wired them in parallel. They were nema 17s and the stepper driver was only a 2A but it never popped the driver and never broke the gearbox anymore. Other things broke 😅. I feel your pain in this video with reiterating though. You tend to learn so much from it though. Great video thanks.
Commercial boxes often use an eliptical bearing instead of rollers, which increases the number of theeth in contact. But, where do you get an eliptical bearing?
You take a thin section bearing and squash it a bit in a vice! 3D Print an eliptical boss to fit tightly in the centre of the bearing to stop it returning to circular.
With this approach, you can go back to your original toothed belt design, as the belt is stretched over the bearing, which removes (most of) the backlash. The combination of the two significantly increase the overall strength.
Clear Explanation and detailing
Thank you!
Oh! Smart! I like the belt idea... but what if we used TPU for the flex spline? This stuff is super resilient and still pretty bendy.
I wanna say I'd try tpu for the flex, and resin printed abs-like anycubic resin for the hard parts. Its so tough, but unlike some of the resins I've tried it literally will flex in half without snapping if you make it thin enough to bend. Nutty stuff...
Wowowowowowow. Awesome engineering work
the calm with which you say "i've decided to design my own motherboard"
REALLY? FOR REAL? Where did you learn that? D:
I learned basic circuit design in college but self-taught pcb design. I find it a lot of fun so I take any excuse to design a custom board!
@@3DprintedLife cool! The only thing they taught me in high school is... Nothing.
Professors in electronics and computer science in Italy sucks sheet... How i wish to have actually le as red something useful instead of systems history and... C... Not C++, not C# just C and assembly. In 2015. Useless as hell.
Excellent work!!!
Thank you! Cheers!
This is great, and no I want to build one.
I will note though that while I'm far from an expert, iirc nylon is actually quite soft compared to petg or pla. The reason it's used a lot is that while it's soft and will flex it won't break or crease as easily and is better at certain temps.
If you want it to be rigid, I think you'll need the fiber reinforced stuff.
Yeah that is true, though I think the increased durability and wear resistance would justify the small increase in gearbox deflection under loads. Fiber reinforced is very abrasive so that would work well at first but I think the performance would fade quick!
Great video man
To improve the durability of your prints, you could enable some "elephant's foot compensation" (no, not _that_ Elephant's Foot), and pack your printed part with some sand so that you can temper and anneal it.
Don't forget to freeze-dry your nylon filament, immediately store it in a vacuum bag with some desiccant, and when you print with it, feed the filament directly from your dehydrator. Also make sure you have a dehumidifier running in your room during all printing operations, and coat your printed Nylon parts with a hydrophobic substance like Neverwet.