3D Printed Cycloidal Drive V2 - Much Better!

in part one of developing a cycloidal drive i made a cycloidal drive reducer and there's lots of explanation in that video but essentially it uses this cycloidal disc which wobbles and that's a bit like a ring gear with a smaller gear inside doing a reduction and my five pegs you can see there are the output and i'm really looking for something that can be back driven so that we can use it for dynamic robots like robot dogs that rely on a low reduction ratio of less than ten to one so the motor can be back driven and that gives natural spring in the leg holding it in place with a really high power brushless motor and so far in this robot dog which is open dog version two i'm using only five to one reduction belt ratios so what we really need is to get nearer to 10 to 1 make sure it's back drivable and make something light and cheap and easy to manufacture hopefully with 3d printing to make it really accessible however in testing the first version which looked like it was working pretty well we found there were a number of issues and in fact the gearbox broke on one of the first tests to lift some mass up and in fact what happened was the cam on the motor which makes that cycloidal disc spin off center completely snapped through that gearbox seemed to run okay but it only had one cycloidal disc so with the elf center cam with its bearings on and the off center cycloidal disc it made the whole gearbox wobble quite a lot and vibrate the plan therefore for version two is to have two cycloidal discs which are 180 degrees out of phase so that all together they spin on center i used bearings before to make this run really smooth and i'm going to do that again but we're also going to brace the end of the cam shaft in the end of the motor so we have a bearing at the top and the motor bearing at the bottom i'm intending this to be open source and 3d printable and all the parts to be accessible such as the bearings and nuts and bolts therefore i'm doing all the prototyping in pretty much standard pla with a half millimeter nozzle and four perimeters and about 20 infill and if that works then we can also upgrade to other materials such as 3d printable nylon or pet g or something else to make the parts tougher if we need to but i'm hoping that pla should just be fine for this prototyping thanks to 3d fuel for the filament for this project and lots of other projects so check out my channel for more 3d printing projects and check out 3dfuel.com as with the first version the main body for my cycloidal reduction has a bunch of bearings around the outside that that cycloidal disc can run on each one only turns a tenth of a turn each time that cycloidal disc comes round so they're not going round and round really fast but it just saves plastic grinding on plastic which is where some cycloidal gear reductions tend to fail i've got spacers which fit on top of that so that we can put another layer of bearings on top of that for the next cycloidal disk so there's my second disc and that seems to run just as well as the first if we look at those discs side by side you'll notice that they're actually slightly out of phase with each other by 180 degrees the motor i'm using which is the same one as last time did have a center pin but i've now removed that so that i can get some extra space and that's for the new camshaft which has a captive nut in the bottom which will be blocked by that pin otherwise it does have four bolts though which will hold it on perfectly well there's a bolt in the middle which should run perfectly on center or thereabouts and the off-center cam you can see in the lower section which runs off center to move the cycloidal disc as before i have five output pegs which we'll have bearings on to actually harness the output of the cycloidal discs that this time fits on a larger bearing in the base and there'll be a larger bearing in the top of the motor as well and that whole assembly fits on top of the motor mount so that the five output shafts turn independently from the cam on the motor it looks like we've got the right clearance there between the output shaft and the motor cam so now we can continue to stack up the bearings on the center part i'm using bearings the same size for this which are 12 millimeters on the outside and a four millimeter internal diameter we also have a bigger bearing on the camshaft this time which is 20 millimeters internal and 30 millimeters external diameter and that fits nicely with our first cycloidal disc in order to get the next stage to run 180 degrees out of phase we need to build this in stages otherwise we wouldn't be to get the first bearing and the first cycloidal disc on so i've made a little keyway and an off-center piece which attaches with two screws and remember that m4 bolt is going to run through the middle of everything to make it nice and strong so it's time for another set of bearings and another set of spacers for the output shaft and then we can add the second cycloidal disc so now we should be to see both discs are running but they're running 180 degrees out of phase with each other and of course that cam is also offset by 180 degrees so altogether it should run on center minimizing vibrations the last part of the camshaft assembly is a cap which fits into the bearing on the top of the motor and that m4 bolt which screws all the way through the camshaft and into the captive nut right next to the motor it's quite tight though and a bit tricky to screw in because the motor wants to turn and it's really hard to hold the whole mechanism still but once that's fitted everything seems to run pretty smoothly and now it's time to finally put the top of the motor on that has another lodge bearing in to match the one at the bottom and we finally have a cap which holds a bearing in the middle which matches the cap on top of the camshaft and that fits perfectly into the big bearing which will make our bolts to attach whatever we attach to the output shaft of the reduction as before i'm using a large lipo that consoles plenty of current an arduino in a box with a knob we can provide a servo pwm signal to a v esc this is really meant for skateboards and only uses the back emf from each coil to work out what the motor position is of the brush's motor in order to energize it so we'll run the motor fine but it won't be very good at holding a position and it's probably not ideal for torque tests but the motor seems to run fine and i can't stop that with my hands so what we really need is a proper test with a big lever to grab again the mechanism does make some sound as it works and that's largely owing to the bearings being loose because they're mounted on m4 bolts instead of tolerance bar so they rattle slightly but vibration is greatly reduced in the mechanism compared to the version one and that's mostly because those two cycloidal discs and the cam have two parts which are 180 degrees off center so all together it runs on center but yep that seems immensely strong so let's do some proper testing now it's time for a quick ad from the video sponsor which is atlas vpn atlas vpn is a tool that connects you through an encrypted tunnel across the internet to another location where you can access the internet from as if you were there this means that your real location is hidden and you get a new ip address from an alternative isp at the new location atlas vpn is supported on any device including windows android mac os and ios and offers a 30-day money-back guarantee popular streaming services such as netflix and amazon prime have geo-restrictions due to licensing rights this means that lots of shows may not be available in your location some services such as hbo hulu and espn may not be accessible in some regions at all using atlas vpn allows you to browse the internet from another country as if you were there so you can unlock the library of various shows which would otherwise be inaccessible grab atlas vpn for as little as 1.39 a month before the deal expires using the link in the description to this video i thought i should open it up again and put some silicone grease in before we do too much testing because although most of the friction is between the cycloidal discs and the bearings the cyclodal discs are resting on each other and also resting on the top and bottom housing and everything's still 3d printed in pla which is pretty low temperature melting point we don't want it all to melt with any friction and that seems to run pretty free and i'm pretty happy with the results so far i think that's a pretty smooth motion and hopefully it's really strong with that double braced bearing camshaft in terms of mass it's a bit over 1.1 kilograms the previous version was 850 grams both including the motor which weighs about 450 grams out of the total mass so it's time to make a bracket to hold that motor that i can screw down to something and attach it to the bench and also a piece to attach a lever to which we can attach some mass to this time i've made a piece which will take a piece of 2020 extrusion so i can vary the length really easily we still have a 10-1 reduction so no doubt really that that will be fast enough for a robot dog's leg joints both for the hip and at the knee can't really see speed being an issue there compared to the five to one belt reduction if i try and stop that with my hand it feels pretty powerful it'll probably take my fingers off if i wasn't careful i've screwed in a screw 100 millimeters from the center of rotation of the output and first of all we're going to try a four kilogram mass tied on a string yep no problems there let's try eight kilograms yep that still seems to work fine as well what about 12 kilograms well the motor drive is having some issues but if we go at the right speed that lifts it with no problems and now 16 kilograms well this is where the motor driver can't handle it ideally we need an o drive or something that uses an encoder to work out how to energize the motor but my gearbox is still intact and i don't think there's been any mechanical issues i thought i should check inside to check there haven't been any issues and everything just seems to be absolutely perfect and still all held together in one piece so it's much better than the version one i'm pretty surprised how strong that is and this of course isn't the best motor driver though this is just using the back emf that comes off the coils or the brushless motor to work out the motor position and work out how to energize the motor so it's all well when you're using it on a skateboard which is what it's for when you're going really fast and the motor is turning and there's no problems when you stall the motor then it gets really confused and that's why basically the motor driver just won't control the motor anymore even though the gearbox is perfectly strong and i'm pretty sure the motor is perfectly strong as well so what we really need is an o drive that uses an encoder on the motor so it knows accurately where the motor position is and where the phases of the motor are so it can energize it then we can do holding power and everything else and that of course is what's in open dog version two which works perfectly well and it's pretty strong so i'm pretty sure we've got enough torque here with our ten to one reduction as long as the gearbox doesn't break the five to one reduction of course is much weaker in open dog with the belt reduction so i don't think that's going to be any problem but we do need to do some proper testing to see what happens when the dog jumps on the ground really hard to see if the gearbox still holds up so i'm going to put together a crude test leg we'll have it jump on the ground a bunch of times but we probably need to come back and do some proper testing in the future so i've made a crew test leg out of spare bits and pieces and some wood out of the shed with a crank that makes it go around and around and if i get the speed just right we should be able to make it jump on the ground there's probably going to still be some issues with that motor driver because we still don't have an encoder or anything to actually control the motor properly so it's probably going to stall a bit like that if i get the speed just right it seems to work just fine now we can test lots of shocks on that gearbox to see what would happen if it was in a real robot dog let's try the motor in the other direction as well and there is actually quite a bit of mass hitting the ground here due to the gearbox itself and also that massive chunk of wood which is a piece of fence post making up the upper leg and yes that's my foot sticking out because i'm sitting on the table to stop the whole thing falling over but a couple of things seem to have broken yep not the gearbox but actually some thin bits of wood that i just tacked on with a couple of wood screws and also a bolt through some plastic which made up the actual crank lever operating the lower leg but the gearbox still seems to be in one piece and that's running perfectly smoothly or at least as well as it was before so let's take the whole thing to pieces and see what it looks like inside although i'm expecting it to be absolutely fine yep that still seems to run absolutely fine and everything's in one piece even though the parts are only printed in pla but i think those bearings have probably had quite a lot to do with the smooth running and not putting too much load on those cycloidal discs and the mechanism itself well it looks like i've made something pretty good this time i'm going to publish all the cad and code as a solid model so you can modify it and it is open source so you can do what you like with it and that's on github and the links in the description to this video i've published the version one as well which you should probably ignore unless you want to make a much lighter one that doesn't need much load on it so there are several improvements to this we could make we probably do need a spacer between those two discs or we could make the discs out of something more slippery like teflon or delrin although there's no ill effects really so fine having them in pla i think i'd be more comfortable if they weren't rubbing on each other the other thing is obviously there's quite a lot of mass to it and also cost all of the bearings in this cost over 100 pounds in money and one way to reduce mass and cost is get rid of them and instead of actually having all of those bearings of which there's 22 and another 10 in the middle just actually have some things like the nylon spacers i put in between them but actually have fatter ones which are fatter on the outside and that'll probably be fine because they're not spinning very fast each one only gets hit on a tenth of the turn as the disc goes round and then you turn a fraction of a turn as well so maybe we could just use nylon spacers or even 3d printed spacers which are fatter on the outside and just use that instead of bearings then that'd be plastic it'd be much lighter and it'd probably be tough enough but that's something we need to do in a future video and what we really need to do to test the robot dog leg properly is make two of these put one on the knee joint one on the hip joint and then have the thing actually jumping on the ground and seeing how much force we can apply to the top with the holding power with a proper motor driver but that will be coming up in the future when i think about open door version 3. so quick and for my patreon if you want to support me through patreon or youtube channel membership those links are in the description below patrons and youtube channel members can get access to all the videos up to a week early and also sneak peeks and pictures of what's coming up and be involved in all of that discussion alright that's all for now [Music] do [Music] you

Channel: James Bruton

Views: 385,019

Rating: 4.9550714 out of 5

Keywords: cycloidal drive reducer, 3d printed gears for robotics, 3d printed gearbox, 3d printing robot parts, tough 3d printed mechanical parts, experiments with cycloidal drive, strain-weave gear, adding bearings to cycloidal drive, cheap reducer for robotics, strain testing for 3D printed parts, cycloidal drive with bearings in, Experiments with Cycloidal Drives

Id: tgEOpl880KM

Length: 17min 13sec (1033 seconds)

Published: Mon Mar 29 2021