Experimental Harmonic Drive Reducer - 3D Printed

last time i made a cycloidal drive reduction which uses two cycloidal discs to make a ten to one gear reduction this is all 3d printed in pla although it could be upgraded with more advanced materials i did quite a bit of testing including lifting heavy weights and making a crazy jumping robot leg which eventually destroyed itself but left the cycloidal drive reducer intact my aim was to make something more compact and which will give us a better than five to one belt reduction ratio i'm currently using in open dog version two however we really want something around ten to one reduction so that the drivetrain can be back driven and give us natural spring in the joints to make dynamic robots using the motor driver holding power as a virtual spring i'm pretty happy with the cycloidal drive i developed although there are a few improvements that could be made which i detailed at the end of the last video however i've been looking at other types of innovative drive reductions and this time i'm going to attempt to make a 3d printable strain weave reduction or harmonic drive a strain weave reduction uses a flexible spline which is deformed by a wave generator attached to the input of the drivetrain this runs in a ring gear with teeth facing inward and slowly rotates as the wave generator deforms it there are already a few videos on youtube about 3d printed strain weave reductions simon merit did some development a few years ago using an inside out htd profile belt as a flexible spline this is pretty successful and there are a few good videos in simon's channel to check out however development seems to have gone as far as it could using a drive belt and as far as i can tell from the videos this is because the tooth profile of an htt belt is round but simon comments that ideally we need more of a triangle shape for efficiency and smooth movement in general it's possible to 3d print flexible materials though which i've done a lot of in the past mainly to make tyres for vehicles and robots so what about if we could 3d print a custom flexible spline with a bit of trial and error i designed and printed a ring gear and a flexible spline with triangular shaped teeth and a wave generator with two bearings on it this seems to run pretty well the reduction ratio of a strain weave reduction is the inner teeth minus the outer teeth divided by the inner teeth which in our case is 40 minus 44 all divided by 40 which is 0.1 or a 10 to 1 reduction this is the same ratio as the cycloidal reduction i built so we can compare the two so let's print some better test pieces the flexible spline is printed in ninjaflex cheetah which is around a 95a shore hardness the rest is pretty standard pla thanks to 3d fuel for the filament for this project and lots of other projects in my channel this time we want to harness the output power of the flexible spline so it's made like a cup shape and it's bolted into a rigid piece with the teeth locked into a recess both pieces are also bolted together with some m4 bolts and lock nuts my wave generator is a handle with some bearings on and some bolts holding them and that fits neatly into the flexible spline to deform it as you rotate the handle that all fits into a casing and now you can see that strain weave motion as i move the wave generator and if you look closely you should be able to see those nuts moving in the base it's pretty tricky to hold it all together so i've made another wave generator handle which is round and fits perfectly on top of the main outer casing and that makes it easier to turn and demonstrate how this works so i do have to hold it together at the moment because there's nothing holding everything together and it will ride apart otherwise but you can see as i turn the handle that we get the output moving in the opposite direction at a 10 to 1 ratio with either of the wave generators inserted though it's almost impossible to back drive unless we actually flex the casing or the flex spline so that it skips this is a complete contrast to the 10 to 1 cycloidal drive which easily rotates even with the motor attached now it's time to make a new version which we can mount everything on bearings and somehow hold all the sections together so it doesn't ride apart and also power the wave generator with a motor and harness the output for testing my flex spline is again printed in the same tpu material which is flexible and this is black this time instead of red because i've run out of red the rest of the casing and all the other components are going to be printed in 3d fuel standard pla if it works in pla then it can always be upgraded to another material in the future but using standard 3d printable materials makes it really accessible to anyone else who wants to build one and all of this cad is going to be published as open source and you can find that in the link in the description to this video this time my outer casing which has the inner facing teeth has a large bearing fitted into the top and that is of course where the output mounts which plugs neatly in there and that output is going to hold the flex spline the flex blind is of course flexible and that again mounts into the output and that fits in with all of those teeth locked in all the way around that plugs in neatly to the outer casing and fits into that big main output bearing and it rotates fairly freely without the wave generator fitted again the flex spline and the rigid output piece are bolted together but this time i've put the bolts facing outwards so we can attach something to it i'm using the same motor that i used with this cycloidal drive because i can't buy any more at the moment and this holds the wave generator the base of that is a 3d print with three captive m8 nuts in it so we can screw that onto the motor and screw everything on top all assembled it looks like some bearings spaced with washers and another piece on the top to brace those apart in the middle there's another captive nut and that allows us to put a further bolt in which will go all the way through to the output the motor mount just screws onto the bottom of the main body with the inner facing teeth and of course the last step is just to put in the flexible spline with the output now to hold everything together there's the extra bolt which goes in the middle and that's again mounted on a bearing which is in the middle of the output disc obviously this is going to rotate in the opposite direction and rotate at a different speed and hence it has that separate bearing but putting that bolt in is going to hold everything down so the whole mechanism doesn't ride apart in use now it's time for a quick ad from the video sponsor which is phantomwallet phantomwallet is available in a number of sizes and finishes there are two lineups phantom r and phantom s phantom wallets include a special mechanism that fans out your cards for easy access there are lots of accessories for customization available which attach to phantom wallets such as a coin holder money clip key holder a multi-purpose silicon band id holder and tracker phantom wallets are very stylish and also compact so they reduce pocket bulge considerably check out phantom wallet at the link below in the video's description box and use my coupon code brewton for 10 percent off [Music] for basic testing i'm just using the same old lipo that's got tons of juice a vesc which is actually meant for skateboards and an arduino in a box so i can feed it with a signal and that seems to run the motor perfectly well and everything seems to run really smoothly and it sounds pretty good as well this time however i can actually stall the motor with my hand which i couldn't do with the cycloidal drive although i'd imagine there's already more friction on the motor here due to having to deform that flex spline if i try to back drive everything is extremely hard i can move it a tiny amount in one direction but nothing like the cycloid will drive the mass of the whole assembly including the motor is about 1.2 kilograms the motor weighs about 450 grams this is about 100 grams heavier than the cycloidal drive even though the cycloidal drive had many more bearings in it and that's mainly due to the massive output bearing that i've got in this assembly even though the flex spline is lighter really happy with that so far it sounds amazing like a proper robot actuator so that's pretty good and it seems to run pretty smoothly of course there is more friction in it than there is on the cycloidal drive which is free to move just due to having to deform that flex spline all the way around now the commercial ones of these the harmonic drive company make them are basically made of metal so it's actually a whole metal piece that deforms so i guess there is quite a lot more friction there than other types of drivetrains but it does run really well with pretty much no backlash and you can't back drive it so it's kind of locked like a worm gear now i said before in the last video the vesc is meant for skateboards it's not the most ideal thing for this there's no sensors on this either so it's got no idea where the motor is and in order to energize the coils of the brush's motor in the right order to turn the motor it just uses the back emf coming back off the coil so it's not a very good solution really for torque tests we're gonna do some more tests but then in the future i'm gonna do some proper tests probably with an o drive and an encoder on the back of the motor so we can energize it properly and compare this with the cycloidal drive i thought perhaps where the flex blind isn't flexing near the top where it's attached to the rigid part it might be catching on the outer teeth because it hasn't deformed so i brought that part down so that the teeth are only where the bearings actually push the flex spline into the teeth i also removed one bearing on each side so that it only flexes the flex spline right at the bottom so that seems to run a lot freer it's much happier starting up that motor at low power with that motor driver of course with no encoder so that seems like i've removed quite a lot of the friction however now if i try and stop the whole thing with my hand in fact it doesn't stall the motor we can still see that nut in the middle going round which means the motor is turning but it's just skipping and that means probably the wave generator is just driving over the bumps so i made a new wave generator with the bearings moved very slightly apart by only about 0.4 millimeters however now it barely runs at all probably just due to that minute extra flex in the flex spline and the way the teeth mesh and it's very hard to predict what's going to happen because we don't really know how it's going to flex when we draw it so i tried moving the flex line up so it only covers the bottom bearings and it doesn't have to move it into a curved shape it only flexes right at the bottom so we've got a much shorter flex blind now by about 10 millimeters so this runs a little bit better but at high speed it all stalls and i think that's probably because it's just distorting in a really funny way so we're back to the original wave generator with just one bearing fitted the original long flex spline and the outer casing with the teeth missing at the top and that seems to run pretty much okay again however this one i can still stop with my hands so there's not much point in doing proper torque tests at the moment at least with that motor driver we do need to do some proper torque testing and we need to do that with a proper motor driver probably an o drive where we put an encoder on the back of the motor and it can actually hold its position that said i'm not sure why the strain weave i can just stop with my hands and actually stall the motor using even a skateboard esc whereas the cycloidal drive i couldn't so i think the cycloidal drive it's full of bearings there's hardly any friction in there it might be quite a lot more efficient than this strain weave and they're both ten to one so if you think you know why that is then put some answers in the comments it may be of course when you try and stop this then it flexes that flex blind in a funny way and that just causes the teeth to jam causes some extra friction so i think tolerances need to be a lot tighter on this and probably funny things are going to happen at different velocities as that flex spline flexes all over the place so i think probably the cycloidal drive is my favorite at the moment apart from the fact it can be back driven which we really need for dynamic robots so check out that explanation in the start of part one when i showed all the footage of open dog and explain why those motors need to be back drivable which of course we don't get from the flex spline this could still be useful for things probably at low speed like a the base of a robot arm something that moves around precisely driven by a stepper motor but i'm not quite sure really whether i'm going to develop this further i'm going to publish all the cad for this though if you want to have a go but i don't recommend using it for anything in production but it's a solid model so you can modify it and do what you want with it so feel free to do what you want with it right um yeah all of this stuff is open source so if you want to support me through patreon or youtube channel membership those links are in the description below and patrons and youtube channel members can get access to all the videos up to a week early and sneak peeks and pictures of what's coming up and that really helps to support the channel so next time we are going to do some proper talk tests probably going to do some more resilience tests on the cycloidal drive as well and actually try and destroy it it's all pla at the moment and obviously that could change but i think that's the one going forward for robot dog development and probably lots of other things just due to the fact that i've designed both of these by eye this one runs a lot more efficiently alright that's all for now [Music] you

Channel: James Bruton

Views: 189,779

Keywords: Harmonic drive, 3d printed harmonic drive, 3d printed atrain weave, 3d printed cycloidal drive, 3d printed zero backlash reducer, harmonic drive backdriveable, is a harmonic drive backdriveable, testing harmonic drive, testing cycloidal reducer, experiments with cycloidal drive, strain wave reducer, can you backdrive a strain wave reducer?

Id: QoBgSWkJyM4

Length: 14min 22sec (862 seconds)

Published: Mon Apr 05 2021