Check out my favorite 3D printers, the K1 Max: amzn.to/3txQUC6 or if you are on a budget, the Ender-3 V3 SE: amzn.to/3FkCPus I hope you enjoyed this video and learned something new! For more content like this, please consider supporting me on Patreon: www.patreon.com/howtomechatronics P.S. If you are interested in the measuring tools: Force meter: amzn.to/3FEUAnS ; Digital dial indicator: amzn.to/3PDvSco
It wouldn't be an easy comparison, a planetary gearbox would require to be bigger than these (if we also want 1:25) or have gears with way smaller teeth (a 3D printer would have problems with that).
I tried to 3d print small S springs in PLA for a circuit probe and found the PLA would brittle fracture if left under strain over night, and otherwise lose flexibility eventually. I switched to PETG and have yet to break the S springs. Might give PETG a try for the flex spline. It's more sticky than molten PLA, but doesn't require a heated enclosure like ABS. Excellent video presentation. Bravo.
For the Australian Rover Challenge I used a flexspline that was printed in solid TPU of 95A Shore hardness and it performed really well. This choice was landed on after a significant amount of testing.
For the flex spline, it is something that need to flex, so I would definitely go with TPU or Nylon, PLA is nice for prototyping and part that needs strength but not for flexibility. Anyway this is a really interesting test, I'd love to see it with a different material for the flex spline.
I would try PTEG for the flex part. It's way more flexible than PLA but nearly as easy to print. I also worked with TPU and would guess that it is to soft for this application. But maybe some of the harder variants could be worth a try.
@@Daniel-bu4gx I reckon a TPU with Shore Hardness D60 or greater would be good for this. That's similar to a skateboard wheel. It would have excellent resistance to abrasion and creep. But yes, even a regular PETG would be vastly better than PLA in this application.
That outcome really shocked me, in terms of durability over time. When I first saw them both in operation, the cycloidal looked like it had more flex and strain on some parts. But thank you, I’m not sure why this showed up in my feed, but it did and I learned something!
I’ll have to go watch that Cycloidal drive video after this one - I used to work in a factory that made cycloidal drive three phase motors, and I always wondered what was so special about the cycloidal drive
Now that you have created wonderful 3d printed prototypes to prove the concept, I'd love to see this taken to the next stage, and produce metal parts, either from CNC parts, or maybe testing out parts made from metal 3d printers. Great work on all this.
@@HowToMechatronics you should be able to make the cycloidal gears out of aluminium with one of those cheap chinese 2040 CNCs. the parts are ideal for a cnc, just make sure to check the backlash as those machines are known to have some.
I agree. Or also, build the ring without the bottom. Create a separate disc for bottom. Connect ring and bottom with "L" shaped brackets. In this way you can have different materials, plastic for the flexible toothed ring and metal for the torque transmission.
That, and using vertical slots would also greatly improve the longevity. It would also allow for a more shallow cup, reducing the lever and stresses at the corner.
We are using harmonic gearbox with 80:1 ratio as a phase adjustment device between two or more rotary axes on printing presses. There are many different cases where these compact gearboxes are used for other than just speed reduction. You can cover those cases at some point of time. That would broaden your viewers perception of gearbox.
It may be too much to explain in just a comment, but how does the harmonic gearbox work for phase adjustment between axes? It sounds like an interesting application that I’d never heard of before.
@@DaveEtchellsI would assume the stepper that drives the secondary print rollers is mounted on the harmonic drive or attached and spins forward or backward to change the phase. Just like automotive VVT technology that advances and retards the camshafts in relation to each other and the crankshaft, despite all being driven by the same chain. It’s used to optimize valve timing.
To be totally fair, the cycloidal drive has quite some metal parts, which the harmonic drive don't. Also, the size is optimized for the cycloidal drive, not for the harmonic drive. I suppose an harmonic drive with longer deformation, could even have smaller diameter and get better result. Also, the joint between the back and the side of the harmonic inner gear (the one breaking), could be reinforced in many ways.
Dude you're killing it. I need to work with you or for you. I love the delivery and competence and clarity. Nice walk through the progress of logical thought. Ontop of the work you kindly made a video. Thank-you
If you plan on properly (and by that I mean more thoroughly) testing the harmonic drive principle you might take a look at the less known double rigid external gear config (one with 50 and one with 48 teeth, exactly as many as the belt flex spline). For the same compactness of the design you're getting a much higher torque...
@@HowToMechatronics i know, I built a full metal one myself. Seeing your video makes me wanna build a 3d printed pancake one myself just for shits and giggles 🤭
@@ErosNicolauSo you suggest something, the person says “that’s a great idea” and you basically say “duh, that’s why I said it” new project: social skills.
You probably do not need any "Expansion" setting, they are not needed to get dimensionally good prints. Most important setting is (if you're using Cura, I don't know in other software) the "Slicing tolerance" set to "Exclusive". I was struggling like you with prints until I changed "Slicing tolerance" setting from "Middle" to "Exclusive" end now every print is perfect to the 0.05 I would say (I tested some clearances changing by 0.05, and you could tell that there was change). You can find this setting in "Experimental" section in Cura. Even 3D printed threads right from the Fusion 360 are working righty away without any change in clearances. This "Horizontal expansion" setting, as many other settings, was created when 3D printing was at its beginnings. People back then tried to solve problems in not always to say the best ways. And now these settings still are there, but they are not really have to be used nowadays. There was a video explaining that phenomenon on RUclips but unfortunately I Forgot channel and name guy who made it. Maybe someone reading that can comment below the link. EDIT: Of course for this to work printer have to be calibrated (e-steps)
Thank you very much for this great video. Little Tipp: You could try to shrink the length of the teeth on the harmonic drive, so you don't need to stretch the material that much. Also it might be a good idea, to use 1 bigger bearing instead of 3 (or one) smaller ones, to stretch the pressure / load.
I think they are both great if made of right types of steel, and not plastic. They are very compact and expensive, but zero backlash is what they are also known for.
I use them for work, they really have zero backlash. They also are efficient only on specific range of torque. If wrongly dimensioned for the work, they became inefficient really quickly.
you could also try with an inverted timing belt. a hefty 8 mm pitch would work well I think in your case. I have a design with 3mm pitch and it works really well. durability would be quite good theoritically.
Test the harmonic planetary drive! I think it's better than both. It's a harmonic drive that can be made with completely rigid parts. Two ring gears of the same diameter with tooth counts differing by one, one of them fixed and the other connected to output. A rotating input carrying (at least) three spur gears around in contact with both ring gears. Two of the three spur gears have teeth out of phase by one-third of a tooth, but all have the same number of teeth in contact with both rings. Reduction ratio is equal to the tooth count of the output ring.
the teeth on your flex spline are pretty big. You can probably get them down to MOD 0.8 which will reduce the amount of deflection the flex spline will need to undergo AND increase the gear ratio per the same gearbox diameter. Also, try making a elliptical bearing with BBs like the youtube ZeroBacklash does with his strainwave drive gearbox.
Thanks for the input! Yeah, a smaller gear module would help as we would need less deflection of the flex spline. Though I'm not sure how low we could and get a accurate one with the 3D printers. We can definitely achieve better results with improving the design.
Very nice job on the video and the explanations. Cycloidal drives are used in small tracked vehicles like excavators. As I recall High Dash was a manufacturer of them. They are used to drive the tracks. They last a long time if the fluids are changed at the manufacturer's recommendations. When they fail they are generally replaced completely. They are generally just as expensive or more so to purchase than a planetary drive. I suspect they are used because they may be a touch smaller. I always preferred a final drive but that was likely because that is what I started with. I've never worked on a machine that I know of that had a harmonic drive.
I was surprised to see that much backlash in the harmonic drive; commercial metal ones are zero-backlash. Looking at the video, it seems like the flex spline teeth could be a little bigger without binding, that’d probably reduce the backlash a good bit. I still wouldn’t expect it to be as strong as the cycloidal mechanism, but a tighter-fitting nylon spline should be both more durable and lower backlash. (Nylon is also “slippery-er” than PLA, so might be more efficient.) Great series, BTW!
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 seem to be the right one to ask. Theoretically you should be able to construct a 3:2 ratio Strain Wave/Harmonic and 2:1 Cycloidal drive. I am certain that reality is somewhat different, at the very least in case of the strain wave/harmonic drive, but how low could you potentially go, in reality. How would wear and tear, efficiency and such be affected and would there be any advantages to go for these drive types for lower gearing? I think i can think of a couple, but I am very much out of my comfort zone and dare not draw any conclusions. I suppose though that it is really only the cycloidal drive that is relevant when it comes to lower ratios, but still, I am curious. Thanks
Just bought the parts to try your Cycloidal design. Pretty excited! I'll be 3D printing for now but I will try milling the parts on our Tormach mill if the backlash isn't too bad
I think even if you made the flex drive from thin spring steel for maximum flexibility and decent durability, You'd still break them! You can't have parts flexing all the time without it stressing them in some way. Probably with spring steel, you'd stress the areas between the teeth. Something that will probably improve durability, but will increase back lash is to have the flex ring float in dove tail slots in the back plate. This will let them move independently of each other a little bit. The flexible section will have to NOT be constrained from flexing anywhere so that there are no high stress zones.
Hats off to you! That was a heck of a lot of work by the looks of it and a very thorough look into this topic. Well done! I've increased my understanding as well.
for the harmonic drive I had the same problems you indicated but basically you made a few mistakes that I had made too. the first you have left too much space between the teeth this causes a kickback. the teeth of the widest part of the wave generator must come practically in contact with the backlash in this way and equal to 0. also such a large teeth is not ideal. in 90mm I made a 180-teeth flex spline in high tenacity module 0.4 resin. another big mistake and in the design of the flex spline and its material. The flex spline must have a fillet radius at the base and must have a thicker flange reduced from its diameter where the hubs attach. this means that it does not break during deformation. it can also help you to do it in a harder material like abs or petg with full infill. I repeat I made it in resin but I still have to test its strength. another big mistake is in the wave generator which, done in that way, does not guarantee correct deformation and tends to block. you could make a flexible bearing with simple 6mm airsoft pellets I solved it like this and now the movement is practically perfect with 0 backlash the couple and to be tested.
with printing one could try a few things in geometry on the flex spline to reduce strain energy while mitigating loses in torque strength since your stuck with thick walls due to plastic. When you handle the metal harmonic drive splines it becomes clear that the thickness is very important. The way it necks down just outside the mounting flange is interesting, probably ground, and the teeth are way shallower than your model. They have a more stubby trapezoidal tooth profile. They also have pretty tight tols when it comes to housing alignment and offsets, getting a plastic printed one to work as proof of concept is pretty cool.
A more durable way to build a 3D printed harmonic drive is instead of printing the flex spline cup, invert a cogged belt. Print an internally toothed ring with a plug or plate or a groove inside into which one edge of the belt will fit. The part the belt goes into has the same diameter and number of teeth as the belt. There's a cup with a rigid connection to drive and bearings, with a supremely flexible spline that is not going to fatigue and break along layer lines.
First of all: Thanks for the video, it was a great comparison and very informative. I liked It very much and also gave it a like. Second, a little negativity: Yes, I know, A 3D Printer is a great tool. But It's a great tool for prototyping! I really don't like, that everything has to be 3D printed nowadays. If you really NEED sucha gearbox - just go and buy it. Or make it yourself on your several thousands, metal slashing CNC mill. But never ever build something like a gearbox out of plastic - if you want to rely on it. These drives are a great way to build 3D-printed rotots, for example. But everything about that should be considered a prototype. You simple cannot expect something in the ballpark of a "real" robot, when it's made of plastic - there's a reason, toys are made from Plastic and Machines are made out of metal. Take all this with a grain of salt. Design you gadget to make it work as best as it could - and then change your material. Because 3D-Printed plastic is just for Prototyping, when it comes to machines, that need to have actual torque. If the gearbox doesn't break, you can be pretty sure, anything else will brake.
I agree with some of the other commenters; the Harmonic gearbox didn't get the most fair representation IMHO. That's not to say that you are shilling for cycloidal or anything, but rather that the way you went about it was sadly not the most optimal science. I would recommend a follow up video where you flip it around and design for the Harmonic gearbox first, then match that with a cycloidal equivalent. Furthermore, I would use a filament that is made to flex for the flexible part, as PLA is about the worst material on Earth for such things. This isn't 2014, most of us can print with TPU now a days.
For the flex spline, what if you mad that separating it in to 3 types of parts. Back plate to couple in to the input, a ring gear for the output, and several "fingers" (basically cantilever springs) that go between the input plate and output gear. This would kinda look like a drawn out clutch pressure plate. The advantage is the printing direction could be down the axis of the flex spline. You could also curve in where it connects to the input plate, reducing the stress concentration and mix up the materials for each part. Great video and lots of fun!!
Or I guess you could get something like a 1/16" thick sheet of HDPE and wrap that in to a cylinder instead of printing the cantilever spring fingers. Either way thanks for sharing the projects!!
Thanks for the video! I think I still stick with the harmonic drive because of the reduced backlash but I can't deny the durability of the cycloidal drive is very appealing. 👍
harmonic gearboxes usually have a flexible bearing as the wave generator, it leads to smoother operation. you are right that the flex spline is the weak link. but that's a failure of the material. I've seen metal ones where the teeth ripped off the flex spline before the wall of the flex spline failed.
There are quite wide 3GT closed belts, like 20-25 mm width, with various lengths. You could glue one side to output shaft and use another as running track for wave generator. With PLA the glued side will still be the weakest point - the belt under load might just rip out PLA it's glued to. So the next sorta improvement would be to glue the belt not to a 3d printed part, but to an aluminium disk of appropriate diameter and thickness about 3mm. With an epoxy and proper surface preparation it should be quite strong. It'd be very interesting to see how much load it can survive:)
Wow, great video! One of the best dyi on this topic. Really awesome job! Thanks for sharing your experiences and very interesting comparative approach.
Hello friend, use a closed timing belt htd8m as a flexible part, you'll get long durability also the outside part could be another timing belt just inside out to the rigid part.
The flex gear for the harmonic drive could be replaced with a vulcanized rubber timing belt for better reliability. As for it's operating limits it did be interesting to find out. But from an operation and maintenance stand point I could see people might go with harmonic drive due to the reduced need to troubleshoot point of failure. As the common saying goes on the shop floor "the more the parts the more points it would fail"
I really enjoyed this thank you. I'm guessing you could use 3 or even 4 colloidal disks to further strengthen, smooth and potentially reduce backlash. I'm curious how much slack comes from minor fit of stainless teeth and collets. Will be following and attempting to emulate your work, though you make CAD look so easy!
Trying different materials would be a good next step, like more flexible materials for the flex spline. Nylon might be a good choice, it is quite flexible whilst being very tough and having very good layer adhesion, it is also low friction.
You need a test to measure torque while in motion so that you get the dynamic efficiency with parts in motion. Something like a disk brake (from a bicycle), similar to an engine dynometer. Or lifting weight with a rope and winch. The brake dynometer works very similar to your torque test, but the lever is coupled to the gearbox output shaft only through the friction of the disk brake.
I've 3d printed several strain wave gears of a very similar design. I got the backlash quite low but had the same delamination problems with the flexible geared part and put the project aside for now. I'd love to try your cycloidal design!
They are so cool be hard to get them functional with 3D printing. I also should give it few more tries, printing them with a bit different design and also with other materials like Nylon or ASA maybe. Cheers!
Yes! I bet a good option would be to have that part MJF printed in nylon12 by like JLCPCB or Xometry. Also, the reason I wanted to build the strain wave gear in the first place was to make a DSLR camera tilt and pivot set up. This is such an awesome choice for that if the backlash can be improved @@HowToMechatronics
That is not how you test for efficiency but for static resistance. Efficiency is describing the power transmission - a static load is by definition 0 power. The 2 gearboxes might perform very different under different load situations. Both have their unique properties but also the "good old" gearboxes like planetary also have their own advantages.
I think you could shorten the length of the flex spline gearbox by adding flexibility to the flex spline walls and shortening the flex spline. If you removed material from the thin wall of the flex spline in an n-tuple helical pattern, it would allow greater displacement of the side of the spline with the teeth. It’s possible removing material would cause too much increase in stress concentrations though.
I never saw that deforming gear ⚙ version. That's so clever. That's not how the metal version you show works, no? I mean, the end result is the same but achieved by a different mechanism. Seems so simple. (Wave generator! I love it.)
Hey. First at all you made two great videos about designing and 3d printing cycloidal gearboxes. As I wanted to build myself something similiar, I searched for DIY 3d printed gearboxes. After watching both videos of you about cycloidal gears, I noticed that to you made them slightly different. One time you have used steel rods in the housing and this time you used ball bearings for the pins. Also the first time you have used brass bushings and now you are using ball bearings for the roller on the output shaft. As I want to design myself also a gearbox, I would like to know which of these two cycloidal gearboxes works better? Which has less friction? Which is more precise? What are the problems of the different designs? If you have already answered these question somewhere else, I am sorry, but I couldn't find a comparison on you webpage. I assume in your first model with the steel pins you could achive a bigger transmission ratio, but with more friction.
Hi, pretty cool video. What about noise? I designed a two stage planetary gear (using herringbone gears) and it's very, very noisy, even with grease. Especially at high RPM. By design the cycloidal drive should be much quieter I think?
I think that some filet or chamfer on the area where the break in harmonic drive occurs would be a good improvement. Also, PLA is not suitable for parts which are under constant deformation, probably PET-G, nylon or ABS would be better for this application. Thank you for interesting content.
In the real life the Harmonic Drive is made from special alloy steel, the strength of the flex spline is not a problem at all. The number of teeth is much-much higher on both circular spline and flex spline so the backlash is really zero, at least unmeasurable. That makes it perfect choice for robotics, where the zero backlash is fundamental. Although the Cycloidal Drive's backlash is also very low, this gives an advantage to HD. Another problem you didn't meet, that is the vibration. You know the real machines may rotate at much higher revolution and that might add vibrations to the Cycloidal Drive due to it's eccentric nature. That can be a serious problem, vibrations kill machines in long term. Your 3D printed elements are not strong, not heat-resistant, dimensionally not accurate things, very hard to make good power transmission and driving elements with these properties.
Great video! Looks like that was a lot of work. I'm surprised PLA performed that well. I'm curious how much of a difference switching to PETG would make. Would the extra elasticity help the harmonic drive enough to make a measurable difference in reliability and performance? How much of a difference? Also would that elasticity hinder the cycloidal drive max force?
Thank you! Yeah, I would like to know what difference PETG would make, but I didn't have the chance to test it this time. I will try to make some tests with different materials in a future video. Cheers!
When I do harmonics I make the "Cup" out of a good Nylon Also the discs of a Cycloidal - never usa pla for gears especially if you have - like me - regular (non stepper) motors running at 4000 to 12000 rpm. Also if you put another set of teeth on the inside of the "Cup" (flexor) then instead of bearings you can use gears and you add the reduction from that as the inside is now a 1 stage planetary and you can drive the sun
Harmonic drive is kinda cool. Not entirely sure of its practical applications but the way it works, I'd not seen that before. With these tools and a driver board like arduino or r-pi you could set up a nifty Rube Goldberg machine or with many of them, do something like what Tim Hunkin does with janky but artful animatronics.
the cup-shaped design for the flexible gear seems very weak, since there are forces pulling the layers apart as it deforms, even when running without any load
The main downside of the harmonic drive is the amount of strain put on the inner workings because materials don't like bending or stretching whilst moving in opposition to the driving force
It feels like making the wall of the flexible spline gear a plain circle was a mistake. Making it a wave below where the bearings contacted it on the inside *should* give you a lot more durability without it either cracking at the base plate or right below the teeth. You'd probably never see this in a machined harmonic drive because its not the point of failure in a solid metal spline and machining a variable geometry wave for the wall would be very difficult.
Regarding the cycloidal drive, I was wondering if a much durable version can be made with bicycle components. Rollers in the chain wil act as the rollers, the sprocket with act as the cycloidal disc.
Instead of generating the equation to under 2pi then imperfectly filling in the gap, why not generate to pi (or more), and double/triple it and cut overlapping segments?
Check out my favorite 3D printers, the K1 Max: amzn.to/3txQUC6 or if you are on a budget, the Ender-3 V3 SE: amzn.to/3FkCPus
I hope you enjoyed this video and learned something new! For more content like this, please consider supporting me on Patreon: www.patreon.com/howtomechatronics
P.S. If you are interested in the measuring tools: Force meter: amzn.to/3FEUAnS ; Digital dial indicator: amzn.to/3PDvSco
You should throw a planetary gearbox in there as well and use it as a baseline.
Yes please.
It wouldn't be an easy comparison, a planetary gearbox would require to be bigger than these (if we also want 1:25) or have gears with way smaller teeth (a 3D printer would have problems with that).
If the intended use is robotics, isn't planetary gearbox excluded on account backlash?
@@ihydf that depends on the use-case: not all scenarios require much precision.
Yep, I guess I should try a planetary gearbox as well.
I tried to 3d print small S springs in PLA for a circuit probe and found the PLA would brittle fracture if left under strain over night, and otherwise lose flexibility eventually. I switched to PETG and have yet to break the S springs. Might give PETG a try for the flex spline. It's more sticky than molten PLA, but doesn't require a heated enclosure like ABS.
Excellent video presentation. Bravo.
For the Australian Rover Challenge I used a flexspline that was printed in solid TPU of 95A Shore hardness and it performed really well. This choice was landed on after a significant amount of testing.
TPU is amazing stuff.
For the flex spline, it is something that need to flex, so I would definitely go with TPU or Nylon, PLA is nice for prototyping and part that needs strength but not for flexibility.
Anyway this is a really interesting test, I'd love to see it with a different material for the flex spline.
Yep, thanks for the input! I should try different material as well as different design.
Yep, Nylon would work great.
I would try PTEG for the flex part. It's way more flexible than PLA but nearly as easy to print.
I also worked with TPU and would guess that it is to soft for this application. But maybe some of the harder variants could be worth a try.
@@Daniel-bu4gx I reckon a TPU with Shore Hardness D60 or greater would be good for this. That's similar to a skateboard wheel. It would have excellent resistance to abrasion and creep.
But yes, even a regular PETG would be vastly better than PLA in this application.
Adding a fillet at the fallow edge on the flex spline might also solve a lot of issues!
That outcome really shocked me, in terms of durability over time. When I first saw them both in operation, the cycloidal looked like it had more flex and strain on some parts. But thank you, I’m not sure why this showed up in my feed, but it did and I learned something!
I’ll have to go watch that Cycloidal drive video after this one - I used to work in a factory that made cycloidal drive three phase motors, and I always wondered what was so special about the cycloidal drive
Now that you have created wonderful 3d printed prototypes to prove the concept, I'd love to see this taken to the next stage, and produce metal parts, either from CNC parts, or maybe testing out parts made from metal 3d printers.
Great work on all this.
That would be cool, but I need to make some CNC machines myself first. :)
I'm so glad metal 3d printers are becoming a thing now!:D
@@HowToMechatronics you should be able to make the cycloidal gears out of aluminium with one of those cheap chinese 2040 CNCs. the parts are ideal for a cnc, just make sure to check the backlash as those machines are known to have some.
I think that you can increase the durability of the flex spline if you add a radius of 2 or 3 mm in the bottom of the cup around the wall.
I agree. Or also, build the ring without the bottom. Create a separate disc for bottom. Connect ring and bottom with "L" shaped brackets. In this way you can have different materials, plastic for the flexible toothed ring and metal for the torque transmission.
Agreed, it even has extra space for additional reinforcement:)
That, and using vertical slots would also greatly improve the longevity. It would also allow for a more shallow cup, reducing the lever and stresses at the corner.
We are using harmonic gearbox with 80:1 ratio as a phase adjustment device between two or more rotary axes on printing presses. There are many different cases where these compact gearboxes are used for other than just speed reduction. You can cover those cases at some point of time. That would broaden your viewers perception of gearbox.
It may be too much to explain in just a comment, but how does the harmonic gearbox work for phase adjustment between axes? It sounds like an interesting application that I’d never heard of before.
@@DaveEtchellsI would assume the stepper that drives the secondary print rollers is mounted on the harmonic drive or attached and spins forward or backward to change the phase. Just like automotive VVT technology that advances and retards the camshafts in relation to each other and the crankshaft, despite all being driven by the same chain. It’s used to optimize valve timing.
To be totally fair, the cycloidal drive has quite some metal parts, which the harmonic drive don't. Also, the size is optimized for the cycloidal drive, not for the harmonic drive. I suppose an harmonic drive with longer deformation, could even have smaller diameter and get better result. Also, the joint between the back and the side of the harmonic inner gear (the one breaking), could be reinforced in many ways.
Yep, you are totally right.
Dude you're killing it. I need to work with you or for you. I love the delivery and competence and clarity. Nice walk through the progress of logical thought. Ontop of the work you kindly made a video. Thank-you
Seconded: this video, and others I've seen in the series, are of exceptionally high quality.
The amount of deep work to achieve this result is crazy
respect
If you plan on properly (and by that I mean more thoroughly) testing the harmonic drive principle you might take a look at the less known double rigid external gear config (one with 50 and one with 48 teeth, exactly as many as the belt flex spline). For the same compactness of the design you're getting a much higher torque...
Any links for this? I can't seem to google it
@@samuelkleiner6143 I posted a link twice already, my replies keep vanishing. Just look for "FR pancake gear" and you'll see it
Hey, thanks for the input. Yes, that kind of setup could work better and we could get better results with materials like this.
@@HowToMechatronics i know, I built a full metal one myself. Seeing your video makes me wanna build a 3d printed pancake one myself just for shits and giggles 🤭
@@ErosNicolauSo you suggest something, the person says “that’s a great idea” and you basically say “duh, that’s why I said it” new project: social skills.
You probably do not need any "Expansion" setting, they are not needed to get dimensionally good prints. Most important setting is (if you're using Cura, I don't know in other software) the "Slicing tolerance" set to "Exclusive". I was struggling like you with prints until I changed "Slicing tolerance" setting from "Middle" to "Exclusive" end now every print is perfect to the 0.05 I would say (I tested some clearances changing by 0.05, and you could tell that there was change). You can find this setting in "Experimental" section in Cura. Even 3D printed threads right from the Fusion 360 are working righty away without any change in clearances.
This "Horizontal expansion" setting, as many other settings, was created when 3D printing was at its beginnings. People back then tried to solve problems in not always to say the best ways. And now these settings still are there, but they are not really have to be used nowadays.
There was a video explaining that phenomenon on RUclips but unfortunately I Forgot channel and name guy who made it. Maybe someone reading that can comment below the link.
EDIT: Of course for this to work printer have to be calibrated (e-steps)
Thank you very much for this great video.
Little Tipp: You could try to shrink the length of the teeth on the harmonic drive, so you don't need to stretch the material that much.
Also it might be a good idea, to use 1 bigger bearing instead of 3 (or one) smaller ones, to stretch the pressure / load.
I think they are both great if made of right types of steel, and not plastic.
They are very compact and expensive, but zero backlash is what they are also known for.
I use them for work, they really have zero backlash.
They also are efficient only on specific range of torque.
If wrongly dimensioned for the work, they became inefficient really quickly.
I believe that's right. Thanks for the input!
This is awesome. Thank you for making this video. It is also interesting to note that the harmonic drive seems to have a less smooth output.
you could also try with an inverted timing belt. a hefty 8 mm pitch would work well I think in your case. I have a design with 3mm pitch and it works really well. durability would be quite good theoritically.
Yeah, I believe that could work well. Thanks for the input!
Test the harmonic planetary drive! I think it's better than both. It's a harmonic drive that can be made with completely rigid parts. Two ring gears of the same diameter with tooth counts differing by one, one of them fixed and the other connected to output. A rotating input carrying (at least) three spur gears around in contact with both ring gears. Two of the three spur gears have teeth out of phase by one-third of a tooth, but all have the same number of teeth in contact with both rings. Reduction ratio is equal to the tooth count of the output ring.
This sounds *really* interesting!
Wow, amazing idea.
the teeth on your flex spline are pretty big. You can probably get them down to MOD 0.8 which will reduce the amount of deflection the flex spline will need to undergo AND increase the gear ratio per the same gearbox diameter. Also, try making a elliptical bearing with BBs like the youtube ZeroBacklash does with his strainwave drive gearbox.
Thanks for the input! Yeah, a smaller gear module would help as we would need less deflection of the flex spline. Though I'm not sure how low we could and get a accurate one with the 3D printers. We can definitely achieve better results with improving the design.
Very nice job on the video and the explanations. Cycloidal drives are used in small tracked vehicles like excavators. As I recall High Dash was a manufacturer of them. They are used to drive the tracks. They last a long time if the fluids are changed at the manufacturer's recommendations. When they fail they are generally replaced completely. They are generally just as expensive or more so to purchase than a planetary drive. I suspect they are used because they may be a touch smaller. I always preferred a final drive but that was likely because that is what I started with. I've never worked on a machine that I know of that had a harmonic drive.
I was surprised to see that much backlash in the harmonic drive; commercial metal ones are zero-backlash. Looking at the video, it seems like the flex spline teeth could be a little bigger without binding, that’d probably reduce the backlash a good bit. I still wouldn’t expect it to be as strong as the cycloidal mechanism, but a tighter-fitting nylon spline should be both more durable and lower backlash. (Nylon is also “slippery-er” than PLA, so might be more efficient.)
Great series, BTW!
@NabtescoMotionControldidn’t Nabtesco own Harmonic Drive at some point?
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.
Very impressive. It takes a lot of engineering education, training, experience and testing testing testing. Thank you.
You seem to be the right one to ask. Theoretically you should be able to construct a 3:2 ratio Strain Wave/Harmonic and 2:1 Cycloidal drive. I am certain that reality is somewhat different, at the very least in case of the strain wave/harmonic drive, but how low could you potentially go, in reality. How would wear and tear, efficiency and such be affected and would there be any advantages to go for these drive types for lower gearing? I think i can think of a couple, but I am very much out of my comfort zone and dare not draw any conclusions.
I suppose though that it is really only the cycloidal drive that is relevant when it comes to lower ratios, but still, I am curious.
Thanks
Nice,
and greetings from Germany.
Best for 2023 :)
Just bought the parts to try your Cycloidal design. Pretty excited! I'll be 3D printing for now but I will try milling the parts on our Tormach mill if the backlash isn't too bad
I think even if you made the flex drive from thin spring steel for maximum flexibility and decent durability, You'd still break them! You can't have parts flexing all the time without it stressing them in some way. Probably with spring steel, you'd stress the areas between the teeth. Something that will probably improve durability, but will increase back lash is to have the flex ring float in dove tail slots in the back plate. This will let them move independently of each other a little bit. The flexible section will have to NOT be constrained from flexing anywhere so that there are no high stress zones.
Been a subscriber for a long time. Your stuff gets more and more technically sophisticated. 👍👍
Thank you! :)
Hats off to you! That was a heck of a lot of work by the looks of it and a very thorough look into this topic. Well done! I've increased my understanding as well.
PETG is a much better choice for flex splines. It isn't as strong as PLA but is more durable, flexible, and better at elastic deformation.
Holy moly man you made these things. I'm amazed.
for the harmonic drive I had the same problems you indicated but basically you made a few mistakes that I had made too. the first you have left too much space between the teeth this causes a kickback. the teeth of the widest part of the wave generator must come practically in contact with the backlash in this way and equal to 0. also such a large teeth is not ideal. in 90mm I made a 180-teeth flex spline in high tenacity module 0.4 resin. another big mistake and in the design of the flex spline and its material. The flex spline must have a fillet radius at the base and must have a thicker flange reduced from its diameter where the hubs attach. this means that it does not break during deformation. it can also help you to do it in a harder material like abs or petg with full infill. I repeat I made it in resin but I still have to test its strength.
another big mistake is in the wave generator which, done in that way, does not guarantee correct deformation and tends to block. you could make a flexible bearing with simple 6mm airsoft pellets I solved it like this and now the movement is practically perfect with 0 backlash the couple and to be tested.
p.s. sorry for my english but i write with google translate
Thanks for the input. Everything you said sounds right. Sure, there are some many ways the harmonic drive to be improved.
with printing one could try a few things in geometry on the flex spline to reduce strain energy while mitigating loses in torque strength since your stuck with thick walls due to plastic. When you handle the metal harmonic drive splines it becomes clear that the thickness is very important. The way it necks down just outside the mounting flange is interesting, probably ground, and the teeth are way shallower than your model. They have a more stubby trapezoidal tooth profile. They also have pretty tight tols when it comes to housing alignment and offsets, getting a plastic printed one to work as proof of concept is pretty cool.
You can reduce the backlash of the cycloidal drive by offsetting the two eccentricities by a couple of degrees.
A more durable way to build a 3D printed harmonic drive is instead of printing the flex spline cup, invert a cogged belt. Print an internally toothed ring with a plug or plate or a groove inside into which one edge of the belt will fit. The part the belt goes into has the same diameter and number of teeth as the belt. There's a cup with a rigid connection to drive and bearings, with a supremely flexible spline that is not going to fatigue and break along layer lines.
Good job. Creative use of bearings for wave generator instead of using sliding surfaces
one thought on the flex drive is to take advantage of the 3D printing and make a series of bellows or waves in the plate and tube.
First of all: Thanks for the video, it was a great comparison and very informative. I liked It very much and also gave it a like.
Second, a little negativity: Yes, I know, A 3D Printer is a great tool. But It's a great tool for prototyping! I really don't like, that everything has to be 3D printed nowadays. If you really NEED sucha gearbox - just go and buy it. Or make it yourself on your several thousands, metal slashing CNC mill. But never ever build something like a gearbox out of plastic - if you want to rely on it.
These drives are a great way to build 3D-printed rotots, for example. But everything about that should be considered a prototype. You simple cannot expect something in the ballpark of a "real" robot, when it's made of plastic - there's a reason, toys are made from Plastic and Machines are made out of metal.
Take all this with a grain of salt. Design you gadget to make it work as best as it could - and then change your material. Because 3D-Printed plastic is just for Prototyping, when it comes to machines, that need to have actual torque. If the gearbox doesn't break, you can be pretty sure, anything else will brake.
Thank you! You are absolutely right. 3D printers are great for prototyping and having some fun too. :)
I agree with some of the other commenters; the Harmonic gearbox didn't get the most fair representation IMHO. That's not to say that you are shilling for cycloidal or anything, but rather that the way you went about it was sadly not the most optimal science.
I would recommend a follow up video where you flip it around and design for the Harmonic gearbox first, then match that with a cycloidal equivalent. Furthermore, I would use a filament that is made to flex for the flexible part, as PLA is about the worst material on Earth for such things. This isn't 2014, most of us can print with TPU now a days.
The audience _is_ impressed. As usual here. 😊👍
Glad to hear it, thank you! :)
For the flex spline, what if you mad that separating it in to 3 types of parts. Back plate to couple in to the input, a ring gear for the output, and several "fingers" (basically cantilever springs) that go between the input plate and output gear. This would kinda look like a drawn out clutch pressure plate. The advantage is the printing direction could be down the axis of the flex spline. You could also curve in where it connects to the input plate, reducing the stress concentration and mix up the materials for each part. Great video and lots of fun!!
Or I guess you could get something like a 1/16" thick sheet of HDPE and wrap that in to a cylinder instead of printing the cantilever spring fingers. Either way thanks for sharing the projects!!
Thanks for the video!
I think I still stick with the harmonic drive because of the reduced backlash but I can't deny the durability of the cycloidal drive is very appealing. 👍
Fair enough! :)
They have no backlash. If they do, they are broke
harmonic gearboxes usually have a flexible bearing as the wave generator, it leads to smoother operation. you are right that the flex spline is the weak link. but that's a failure of the material. I've seen metal ones where the teeth ripped off the flex spline before the wall of the flex spline failed.
I am not afraid of tomorrow, for I have seen yesterday and I love today.
There are quite wide 3GT closed belts, like 20-25 mm width, with various lengths. You could glue one side to output shaft and use another as running track for wave generator. With PLA the glued side will still be the weakest point - the belt under load might just rip out PLA it's glued to. So the next sorta improvement would be to glue the belt not to a 3d printed part, but to an aluminium disk of appropriate diameter and thickness about 3mm. With an epoxy and proper surface preparation it should be quite strong. It'd be very interesting to see how much load it can survive:)
I would recommend trying to print the flex spline in PETG, which is more flexible than PLA and just about as easy to print.
Yep, it would be better for sure.
Wow, great video! One of the best dyi on this topic. Really awesome job! Thanks for sharing your experiences and very interesting comparative approach.
Hello friend, use a closed timing belt htd8m as a flexible part, you'll get long durability also the outside part could be another timing belt just inside out to the rigid part.
The flex gear for the harmonic drive could be replaced with a vulcanized rubber timing belt for better reliability.
As for it's operating limits it did be interesting to find out.
But from an operation and maintenance stand point I could see people might go with harmonic drive due to the reduced need to troubleshoot point of failure.
As the common saying goes on the shop floor "the more the parts the more points it would fail"
I really enjoyed this thank you. I'm guessing you could use 3 or even 4 colloidal disks to further strengthen, smooth and potentially reduce backlash. I'm curious how much slack comes from minor fit of stainless teeth and collets. Will be following and attempting to emulate your work, though you make CAD look so easy!
The Fanuc robots we have at work use all metal harmonic drives in their joints. I've always been fascinated with them since I first saw them there.
May I ask which type are you using? I remember that the cycloidal gear is also commonly used in their R2000 series.
The bigger the Fanuc robot, the more you'll see a Cycloidal@@TheBananaJoint
For robots, mass is critical. Torque Density would have been nice to compare too. But fantastic work!
Trying different materials would be a good next step, like more flexible materials for the flex spline. Nylon might be a good choice, it is quite flexible whilst being very tough and having very good layer adhesion, it is also low friction.
Also using the 3d-parts as positives for mold building to test more materials that cannot be printed easily but manufactured with molds.
You need a test to measure torque while in motion so that you get the dynamic efficiency with parts in motion. Something like a disk brake (from a bicycle), similar to an engine dynometer. Or lifting weight with a rope and winch.
The brake dynometer works very similar to your torque test, but the lever is coupled to the gearbox output shaft only through the friction of the disk brake.
Very cool. What was the weight difference between the 2 drives?
I've 3d printed several strain wave gears of a very similar design. I got the backlash quite low but had the same delamination problems with the flexible geared part and put the project aside for now. I'd love to try your cycloidal design!
They are so cool be hard to get them functional with 3D printing. I also should give it few more tries, printing them with a bit different design and also with other materials like Nylon or ASA maybe.
Cheers!
Yes! I bet a good option would be to have that part MJF printed in nylon12 by like JLCPCB or Xometry. Also, the reason I wanted to build the strain wave gear in the first place was to make a DSLR camera tilt and pivot set up. This is such an awesome choice for that if the backlash can be improved
@@HowToMechatronics
That is not how you test for efficiency but for static resistance. Efficiency is describing the power transmission - a static load is by definition 0 power.
The 2 gearboxes might perform very different under different load situations.
Both have their unique properties but also the "good old" gearboxes like planetary also have their own advantages.
I think you could shorten the length of the flex spline gearbox by adding flexibility to the flex spline walls and shortening the flex spline. If you removed material from the thin wall of the flex spline in an n-tuple helical pattern, it would allow greater displacement of the side of the spline with the teeth. It’s possible removing material would cause too much increase in stress concentrations though.
Acres of almond trees lined the interstate highway which complimented the crazy driving nuts.
Excellent work and great presentation!
I never saw that deforming gear ⚙ version. That's so clever. That's not how the metal version you show works, no? I mean, the end result is the same but achieved by a different mechanism. Seems so simple. (Wave generator! I love it.)
Great job, excellent presentation,thanks!
Hey. First at all you made two great videos about designing and 3d printing cycloidal gearboxes. As I wanted to build myself something similiar, I searched for DIY 3d printed gearboxes. After watching both videos of you about cycloidal gears, I noticed that to you made them slightly different. One time you have used steel rods in the housing and this time you used ball bearings for the pins. Also the first time you have used brass bushings and now you are using ball bearings for the roller on the output shaft. As I want to design myself also a gearbox, I would like to know which of these two cycloidal gearboxes works better? Which has less friction? Which is more precise? What are the problems of the different designs? If you have already answered these question somewhere else, I am sorry, but I couldn't find a comparison on you webpage. I assume in your first model with the steel pins you could achive a bigger transmission ratio, but with more friction.
Science is organized knowledge. Wisdom is organized life.
Try increasing the voltage for the motor from 24 volts to 40-48 volts. The driver is protected by current, so nothing will burn.
For future harmonics, i'd try making the flex out of two parts, a rigid backplate and a more flexible ring, perhaps TPU?
Thanks for this comparison! Really interesting!
Glad you liked it!
Hi, pretty cool video.
What about noise? I designed a two stage planetary gear (using herringbone gears) and it's very, very noisy, even with grease. Especially at high RPM.
By design the cycloidal drive should be much quieter I think?
Excellent video. Thank you for doing these.
I think that some filet or chamfer on the area where the break in harmonic drive occurs would be a good improvement. Also, PLA is not suitable for parts which are under constant deformation, probably PET-G, nylon or ABS would be better for this application.
Thank you for interesting content.
Thank you!
In the real life the Harmonic Drive is made from special alloy steel, the strength of the flex spline is not a problem at all. The number of teeth is much-much higher on both circular spline and flex spline so the backlash is really zero, at least unmeasurable. That makes it perfect choice for robotics, where the zero backlash is fundamental. Although the Cycloidal Drive's backlash is also very low, this gives an advantage to HD. Another problem you didn't meet, that is the vibration. You know the real machines may rotate at much higher revolution and that might add vibrations to the Cycloidal Drive due to it's eccentric nature. That can be a serious problem, vibrations kill machines in long term. Your 3D printed elements are not strong, not heat-resistant, dimensionally not accurate things, very hard to make good power transmission and driving elements with these properties.
Hello, taulman Bridge Nylon will be your best friend for the harmonic drive ...
Great video! Looks like that was a lot of work. I'm surprised PLA performed that well. I'm curious how much of a difference switching to PETG would make. Would the extra elasticity help the harmonic drive enough to make a measurable difference in reliability and performance? How much of a difference? Also would that elasticity hinder the cycloidal drive max force?
Thank you! Yeah, I would like to know what difference PETG would make, but I didn't have the chance to test it this time. I will try to make some tests with different materials in a future video.
Cheers!
This is just beautiful. Thank you.
Glad you like it!
Really awesome ! Outstanding work !
Thank you!
Great video, but the tooth profile of a strain wave gear is different from the tooth profile of spur gears.
I believe it would be very interesting if you made a motor for an e-bike.
Great video. It would be interesting to see how good a cycloidal would be with laser cut parts which will be much stronger
Or even just using 100% infill. The cycloidal tooth probably would not have collapsed if there weren't voids in the there.
Thanks! Yeah, it would be interesting to see that.
Merci beaucoup mec, excellent taf !
Did you try a solid harmonic drive? It seems easy to 3D print.
How much lubrication do you need? Are they sealed up well enough to prevent mess?
Thanks a lot for another wonderfull lesson and see you next
My pleasure!
Great video/channel, keep up the good work
Thank you very much!
Instead of PLA, wouldn't TPU be a better choice for the Harmonic Flexible Gear? Do a comparison between the two please.
When I do harmonics I make the "Cup" out of a good Nylon Also the discs of a Cycloidal - never usa pla for gears especially if you have - like me - regular (non stepper) motors running at 4000 to 12000 rpm. Also if you put another set of teeth on the inside of the "Cup" (flexor) then instead of bearings you can use gears and you add the reduction from that as the inside is now a 1 stage planetary and you can drive the sun
Harmonic drive is kinda cool. Not entirely sure of its practical applications but the way it works, I'd not seen that before. With these tools and a driver board like arduino or r-pi you could set up a nifty Rube Goldberg machine or with many of them, do something like what Tim Hunkin does with janky but artful animatronics.
Harmonic Drive Practical Applications - Robotics, Telescope GoTo Drives, Aircraft and Satellite Pointing to name a few...
the cup-shaped design for the flexible gear seems very weak, since there are forces pulling the layers apart as it deforms, even when running without any load
The main downside of the harmonic drive is the amount of strain put on the inner workings because materials don't like bending or stretching whilst moving in opposition to the driving force
What was the mass difference between the two actuators? Seems like the steel pins in the cycloidal would make the mass of the cycloidal much higher.
넌 진짜 천재다.. 너무 고마워
Really good work, thanks
This was a great test build break repeat
It feels like making the wall of the flexible spline gear a plain circle was a mistake. Making it a wave below where the bearings contacted it on the inside *should* give you a lot more durability without it either cracking at the base plate or right below the teeth.
You'd probably never see this in a machined harmonic drive because its not the point of failure in a solid metal spline and machining a variable geometry wave for the wall would be very difficult.
A small fillet at the bottom of the harmonic cup would take the stress of the 90 degree joint to the bottom for longer lifespan :)
Well it's safe to say that due to the presence of the metal pins, the wave strain gearbox should be significantly lighter in comparison
But also, if we print the flex spline with nylon and the cycloidal with perhaps a Lazer cut acrylic it might produce astronomical results
Regarding the cycloidal drive, I was wondering if a much durable version can be made with bicycle components. Rollers in the chain wil act as the rollers, the sprocket with act as the cycloidal disc.
Dreams come true. Without that possibility, nature would not incite us to have them.
Instead of generating the equation to under 2pi then imperfectly filling in the gap, why not generate to pi (or more), and double/triple it and cut overlapping segments?