an easier robot arm actuator...
Вставка
- Опубліковано 8 лип 2024
- Experimenting with the cycloidal drive form to try to make a compact actuator for robot arms
The code for the python demo and Fusion plugin as well as the CAD files for both models (in fusion and STEP formats) are available at: github.com/roTechnic/internal...
These are the supplies i used:
NEMA 17 stepper: amzn.to/3gtK3Qd
Arduino Mega: amzn.to/3nx3Oui
5mm steel rod: amzn.to/3HMBGeV
Thin wall bearings: amzn.to/3u0Whrd
small bearing: amzn.to/3K5e9H8
eSun PLA+ filament: amzn.to/3HvzVT5 - Наука та технологія
If the the gear had a herringbone sort of pattern, that would eliminate thrust along the axis.
Solid idea
Seconded. I've used double helical gears in a couple of projects and they're great. All of the force is transmitted in a direct and constant manner.
I wonder if that would reduce the pulsation as it moves between teeth, most visible in the closing frames of the video.
@@keithwins Yup. Helical gears always have at least two teeth engaged.
That, and it also would remove any unintended torque moments from the gearbox, which could become an issue as these gears get (longer, wider?). I wonder how well it would go with a more extreme twist rate, rather than the small drive gear being swept at 360 degrees, what if it was swept at 720, or 1080 degrees? I'd think it would produce more, smaller contact points, evening out the load along the gear teeth.
This is what access to 3d printers does. It brings ideas into reality that we didn't even dared dream about. I am so impressed with where this is going.
Yeah. I also want to use my printer for this kind of things… however, Blender has a physics simulator that in a few years could simulate mechanics without printing them.
@@robertoguerra5375 fusion 3d already has that.
no it just lets idiots play around with idea's they found in a book from the 70's
because this has deffinatly already been thought about. tested. and discounted for various reasons. like motor power required. wear and tear. or production costs.
there is nothing new about this. and almost none of the things like this that you see.
they didn't come up with the idea themselves even if they tell you they did.
@@darkracer1252 Many ideas have been invented multiple times so your comment is effectively trash. Best you can say is that the idea wasn't fully researched before being presented as original. Outside of scientific journals or patent applications, that's perfectly fine. I'd also like to point out that just because an approach isn't particularly useful in a production environment with CNC or injection moulding doesn't mean it won't be effective for low-end FDM.
Not saying it's awesome... haven't tried it. But, I'm interested in seeing where it goes.
@@4Fixerdave
THIS idea was fully researched.
the youtuber is a rtard for even attempting it.
why? because other solutions are easyer to make. cheaper. more efficient. and stronger.
this design really looses on all fronts
and no im deffinatly not one of the people who thinks that failed science from the past should never be tried again with new advancements.
this one in particular is going to have to wait untill we break thermodynamics by removing friction
basicly break the laws of physics
Next step would be to add a planetary gear set to that for more constant load application. Having 3 driving cycloids could smooth things out significantly
3 or 4 driving cycloids could also serve as radial bearings for the driven ring. Something is still needed for axial loads.
3 or 4 driving cycloids could also serve as radial bearings for the driven ring. Something is still needed for axial loads.
@@hanelyp1 HERRINGBONE planetary excentically cycloidal gears should work for axial loads. I kinda just wanted to say the phrase, these big words are adding up.
I ventured into the comments to suggest the same - planetary to drive multiple pinions. Only I would increase the overall diameter and locate the stepper on the interior
Back to the Backlash >>>
Excellent, very clear and detailed explanation on the workings!
More ideas for you to consider:
1. having 2 motors drive it can give you more power, better rigidity (positioned opposite), and control over backlash (just advance one motor further to tighten the backlash). This way you can move a robot arm quickly and not precisely, but have it arrive very exactly and without backlash to target point.
2. have many (herringbone) rollers between an outside housing like now and between an inside off-center rotating-wobbling disk like in cyclodial drives. You get higher gear ratio than this and way more contact points to stress - hence a stronger gearbox.
Excellent ideas! I will see if I can implement them - thank you!
For those figuring out why this is different from a regular big gear with thread on the inside and a smaller gear attached to the motor, one revolution of the input here is results in the outter gear rotating one 'teeth', while with regular gears the big gear moves with the amount of teeth the smaller gear has.
In another way, you can say that with this system the driving gear only has one teeth.
No it doesnt. If that inner gear had teeth it could easily reproduce the 12:1 ratio seen here AND be a better design. Planetary beats this in a number of ways.
@@wizrom3046
with planetary you also have more gears. multiple reductions.
this is a single transfer of force. less parts to break fail.
but also way more friction and thus less wear resistant and in the end an absolutely stupid idea that got thought up and discounted decades ago
Ah, thank you! I was wondering about that. The other difference here, I see, is that the gears are slipping on each other quite a lot, whereas with "regular" gearing they don't slip very much.
@@BrooksMoses a rounded gear as seen here is a poor choice because of the angle of power transmission from one gear to another. The closer that angle is to 90 degrees (ie true radial) the more integrity the power transmission has, and is likely to have less friction too.
Fantastic work, I love how clearly you explained the creation of the cycloid :D I'm tempted to try and make multiple stage internal eccentrically cycloid reducer. Looking forward to your next ideas.
Thank you! Let me know if you do it!
Very glad youtube waved your video at me! Extremely handy to see python code in Fusion 360 doing the sort of thing I need, and your presentation is spot on. Thank you. I'm about to dive in and explore all your other videos.
As I continue to find more of your videos I'm more and more impressed with your concise explanations and great ideas. Keep it up mate!
Super unique design! I really like these gearbox experiments.
Glad you like them!
Very nifty. Good job. The video didn't waste any time which is good. We can all freeze or replay these so there is no need to go slower (given it mostly shot over my head and thus got a rewatch).
Very fluid, clever and clear demo, thank you!
Thank you. You are faster than I expected. Looks really good.
no worries! please keep giving me ideas!
This is exactly where I am now. Thanks for the insight.
So cool that you keep working on this project!! Nice update, makes me want to build one!! 😇
thank you, I sometimes feel that I’m just procrastinating here and putting off building the arm this is supposed to be for!
Your videos are absolutly perfect! thank you a lot for the work you do.
Wow, thank you!
@@roTechnic in which way it is better than "normal" gear? I think it is exactly the same but with less teeth, and also the teeth are slipping and suffering more wear.
Amazing work and the narration is even better. Very impressive
Interesting.. thanks for trying the inside out design... Might not be as good, but now we know for sure.. every data point is useful!
Oh, absolutely! And it was a fascinating project!
Gosh, you int 'alf clever! And so refreshing to hear a British chap doing interesting stuff on UA-cam. I hope I haven't broken any rules. 😀
I need to try one of these
First vid of yours I’ve seen, subscribed!
awesome videos.... delighted I came across your channel!! thanks for sharing your knowledge
If you're putting the drive gear into an elbow like you have here, you don't need gearing teeth to go all the way around the circle, the elbow stops the rotation at three or so teeth on either side of centre.
So there's some space saving available in this specific use-case.
You have a very soothing voice.
Just rewatched this. One advantage of thei gearbox, is that it can have a hole through the gear, which is perfect for wires.
Amazing video!
I need to learn that snap your fingers trick to speed up my projects 🤔
Wonderful video.
It looks like you have room for a second motor and input gear. 2x the torque, and the second motor could be clocked to eliminate backlash!
6:00 - In fact I do have a crazy idea 🤪. - It could do no less than do away with all metallic bearings and screws. - And perform even better. - No joke!
Namely: Drive a single planet of a tapered roller cycloid gear-bearing containing at least three planets.
Best thing a single fat 3D printed screw can pretension the whole assembly together to zero backlash!
The tooth trajectories need to be of a (soft sinusoidal) herringbone shape in order to prevent slide-out of cone roller planet gears.
In fact I made a prototype of such a gear-bearing. Just not driven at a planet but passive.
Guess I need to make a video showing it off such that it gets more clear what I mean.
Is it like what you said with the last video? A planetary gear system using the cycloidal ring and sun gear, and the cycloidal worm gears?
Now put 3 (or 4) of the internal gears 120 (or 90) degrees apart and put large planetary gears on the end with a small sun gear. Probably not too practical since another set of gears will introduce backlash, but there's still internal space that could be used to increase the gear ratio.
Excellent content, yet so humble :)
Thank you - that's a really nice thing to say :)
what you could do, to give it more strength, is have the outer ring, the eccentric gear and then an inner gear that helps stabalize, the gears.
Now looking forward to seeing you try a planetary cycloid version of this :D
The goal is to have high gear ratios, but also reduce space as for a dynamically balanced cycloid gearbox, we need minimum 2 and even 3 cycloids for fully balanced dynamics.. That increases space and multiplies the number of bearings or rollers needed for real gears that aren't just 3D printed. So if a planetary cycloid hybrid gearbox can achieve the same or higher gear ratios than a regular eccentric cycloid in less space or the same space, then it would be an interesting and highly practical innovation.
You're absolutely right! I'm definitely going to try this out in the future! Thanks for the ideas!
you could improve this by adding more macaroni pinions like a hybrid with a planetary gear set to distribute the force. Then mirror the mechanism like a herringbone gear to get rid of any axial forces. I might have to try this!
friction is always the first hurdle , complemetary forces are next . the trouble starts with wear of parts due softness and over-use . but very nice transparent developing , inspirational
wow it looks so easy
A second gear along the inside, so the driver sits flush on both ends would create more torque relative to its volume and you'd really see the magic of this design. You're only grabbing on the outside: You also need to grab on the inside.
Herringbone arrangement with multiple driving gears synchronized via a timing belt (chosen for anti-backlash, but perhaps gearing would suffice?) could increase the torque capacity and eliminate the need for a large outer bearing (at the cost of several smaller ones for the driving gears).
you could split the outer piece into several plate-like layers with straight edges that are stacked together and then fit bearings along the little pinion. It would then not be a continuous mating surface anymore, but rather about 6 bearings that sequentially push on the outer part. Backlash would probably become unbearable and you'd have trouble mounting the bearings on the little part, but friction would be eliminated.
Impressive....
Maybe you can try to make one a little bigger, so the motor fits inside side by side with the worm. Then you connect the motor to the worm with simple gears or a belt.
The advantage is that it would have a neat form factor with no protruding appendages.
You can also use small bearings if you place them to the sides, and make the cycloidal cage as two facing cups with a central axle.
Very good video. Just a recommendation wile editing use a plug-in that removes the strong "ss" from the audio
Now THIS is a worm drive.
I think there's an interesting trade off to find between the height in mm of one sweep of one tooth, and multiple sweeps. In this example, 3 sweeps at 10mm or 1 sweep at 30mm.
are you thinking that the multiple sweeps could transmit more torque due to the increased helix angle?
@@roTechnic Perhaps! And maybe there would be interesting torque / efficiency trade offs the other way, half a sweep, two pinons and a short gt2 closed loop keeping them in sync?
And of course, herringbone configuration, two sweeps back to back!
"so let's do that" amazing haha
Excellent design. What are your thoughts on a herringbone design, will the off-tangent forces be too high? Also is it possible to put 4 internal gears to spread the load around? I've replicated your last design with four bearings at 90deg on the inputs cam and four stepped cycloidal gears at 90deg phase shift. I will try and do the same with your current design. I will use your python code to generate the internal cycloidal profile. Many thanks.
I can’t see why a herringbone design wouldn’t work - one of my viewers made a herringbone version of my last design that seemed to work nicely.
what benefits are you wanting from the herringbone? if you’re looking to reduce side-thrust, I’d consider stepping the gears instead of sweeping them. if you want to double the nodes in mesh then sweeping the gears through 720 degrees instead of 360 would achieve the same thing.
yes, you should be able to add as many pinions as you can fit in! in fact another subscriber has suggested a planetary version, which could be interesting.
I’m not sure what you mean by “4 bearings at 90 degrees”, any chance you could share a picture or video?
one other quick thing: on my design I used 5mm contraction - I’d reduce that now - I can’t see much benefit in contraction for EC gears unlike normal cycloidal drives
I'm thinking planetary internal cycloidal gearbox......should be able to make it smaller, and between the extra contact area and those areas operating in opposition to one another the torque capacity aught to be pretty good for the size.
Another thought is a roller design. It looks like a planetary gearset but all parts are smooth. The balance between load based friction and the properties of the lube help determine torque transfer qualities. A benefit to this design is that slippage is accounted for when overtorque is applied. The rollers simply slip without the mechanical deflection and breakage of a gear or other positive engagement system.
When the overtorque condition is over the transmission simply goes back to work.
This is ised in a supercharger design from a few years ago but I'm sure it has other applications.
it`s good to see all these ideas played with they can all lead to new and ingenious ways of doing things... I suppose an improvement for strength at least would be to have multiple Rotors inside that internal cycloid... When looking at it though it`s all sliding surfaces.
he took a hit at the comment section a more advanced version of so many experts joke well done mate
Amazing design and implementation. I do wonder about the Fusion 360 script though - all those individual line segments will eventually cause performance issues, and make the breps very complex. Doing a best-fit spline, with a tight fit and lots of control points, may be an option. Or using arc segments. This is similar to how the "Smoothing" function works in F360's CAM workspace.
you don't NEED to use big bearings. gears on shafts can be mounted with bearings just bigger than the shaft, you press the bearing into the center of the gear. You can also have nested shafts like in a motorcycle clutch basket
The reason i needed large bearing on this is to allow the motor and pinion to be mounted close to the edge of the internal gear. Can you see a way I could do that with smaller gears?
@@roTechnic you could use multiple rollers with their own smaller bearings. and you can use a smaller bearing on a shaft like the OP said. and then mount the motor on the front side the side that's open to the camera in the video. or a combination of both.
seeing as a large bearing like that can actually get quite expensive.
Nice work! The python code and visualization is especially helpful to show exactly how the curves are generated.
I wonder if there would be any way to make a planetary version of this. There's an awful lot of empty space inside that could be filled with more pinions to multiply the torque capacity, but I'm not sure how to go about driving them, unless just putting another of the same shape pinion in the center would do it.
dammit! that’s a great idea. i’ll see if I can work out how to do that.
a pinion driving another pinion wouldn’t be very efficient as there isn’t much surface area contacting, but the sun gear could be a single tooth, the planets could be larger, and the ring would be larger still. not sure what the reduction ratio would be, but it could work
nice
The next step would be an in- AND out runner, wouldn't it? Two opposing pins, one on the inside, one on the outside. Is that even possible due to the different diameters of inside and outside? 🤔
use this drive as a cable spooler so you can utilizee smalller bearings and slim down your arm joiunts by making them cable actuated.
Genius!
We’ve been using worm gears for articulation for… ever.
You could experiment by placing the bearings on the gears inside
How does this compare to helical gears? Also, you might want to try a herringbone style because then the small gear won't move around.
please make a planetary version
Hey, wouldn't this desing allow for several inputs? Since its hollow, it could be possible to just add more motors, maybe even add another gear on the center to keep all of them synchronized and keep the input gears in place to avoid skipping.
Improvement: If you rotate the shape of the pin through a multiple of 360° it can transfer more force to the outer ring. Changing the geometry of the pin, making the shape a little bit more complex, the 'relevant' contact surface (the part applying force) increases and you get more torque for the same part count. If the material itself holds up, you can double or triple the torque.
But wouldn't you still end up with the same total contact-surface area? You get more individual contacting areas, but they all get smaller.
@@roidroid The harder the material the less your insight is true. Think of difference btwn. laying on top of soft balls vs. on a bed of nails. In the first case the balls compress and the surface area increases. In the latter case your skin gets destroyed by a single nail but not by hundreds. The load of the body can only be supported by multiplying the force arrows. Preventing piercing of skin is analogous to skipping of gear dure to torque. In practice material science and physics combine.
@@discoveringthegardenofeden7882 oh coz none of them are flat surfaces, but are rounded? Interesting
At the end, it looks like the bearing is slipping. Is it possible to have the cycloidal gear deeper? Seems like it's almost just rolling the outside gear. Love your content. This is the type of channel that I go to youtube for.
good spot! during the torque tests the pinion bent slightly - it doesn’t effect the normal running, but as soon as you take the housing off it wobbles like crazy!
you can make it deeper - increase the size of the pin or decrease the contraction.
Супер!
Could you do an internal and external gear, and herringbone booth?
can you tell me the procces to get the equation of the offsetted hypocycloid to use in Solidworks
because a normal hypocycloid is discontinuous and cannot to drawn by equation driven curve feature.
I don't understand why the output was not a smooth rotation, but instead slightly jumpy. Could somone explain?
the pinion bent slightly in my torque tests (when you heard the "Thunk") - you can actually see it rotating off center in the last shot in the video - this caused the speed to change in part of the rotation.
@@roTechnic Ahh ok thank you very much. The design was very interesting but I couldn't enderstand what was causing it.
I like this idea!Do you think that having more pinions driven by seperate motors would increase the torque? What if you had three pinions ganged together by a belt and driven by one motor?
yes, definitely.
Steeper cycloids and or less of them in the orbiting pattern as well as a larger pinion for more strength?
What if you made it more like a worm drive? That should make it thinner and still allow for back-drive-ability with that style of teeth.
I hereby declare this as the "noodle drive".
A "smooth gears" with Zero teeth would be two wheels rubbing each other, the most simple form.
Gearbox will take two stepper motors and that will be doubling torque, right?
Could be posible put an inside cylinder with the pattern with a free axe in front of the motor power axe. and maybe this way the axe can no "jump" as easily?
The cycloid gear looks to have a lot of wear surface making a plastic version have a short lifespan. I wonder if somehow you could use ball bearings in there, so the balls rotate as the friction surface, like how a ball screw works.
What about herringbone gear?
Would it make sense to double the pinion at the opposite side, coupled with gears to the motorized pinion?
yes it would, although maybe belts over gears for backlash reasons
Would it be possible to combine this with a planetary gear? Thus having three rolling pins that are driven from the center with the motor? It might only get a 1:1 between the sun and planet gears, but it might increase the rigidity as loads are distributed.
I became lost at some point.
How is the pinion gear diameter determined?
How about the offsets?
I do not follow code all that well.
Plan to try Levi Janssen's method of manually plotting out the cycloid.
Only do it on the inside like this one.
Need some help. :)
With so much empty space inside the big gear, could you fit the motor in there?
I had to idea to adjust your script to create a planetary gearset where we move the small gear to the centre and create 3 planets that mesh with this small gear and the ring. So far, not getting the geometry right for the planets!
what would happen to the movement if the gears ar longer/deeper
How about 90 degrees version? Small (worm) pinion between two discs?
GENIUS!
Everyone wants to do magic, but no one wants to learn the math to do magic
Wo dering where can get those bearings?
How about to add more drive gears?
how did you smooth out the cycloid curve and inverte it @2.17 to 2.18 ?!
Do you mean in Python or in Fusion?
@@roTechnic fusion
Actually am trying the equation in Solidworks but it doesn’t turn right
What's the gear ratio? Did i miss it?
Could this be arranged in a planetary fashion?
There is continuous gear engagement, so why does the assembly rotate at an inconsistent rate?
May not be smaller but significantly more dust resistant.
Anyone have a link to a tutorial that can help me learn how to do the rolling circles in python?
Where are you getting the large bearing from?
What if you have three internal pinions that support it, you would not need that external housing with bearings.
I am having a lot of trouble increasing the number of "pins" and decreasing the pin radius without getting any odd overlaps, sometimes it looks pretty good but when I try to put a circle around the ring gear in the sketch fusion360 crashes every time.
let me know the number of pins and radius etc that you're using and I'll have a look
Same problem here:
pin_radius = 5 / 10
pin_circle_radius = 30 / 10
number_of_pins = 12
contraction = 0.5 / 10
I found a solutin:
Let the script plugin draw the segment for you, but do not make the circular pattern yet. First, edit the sketch and use the Fit Point Spline tool to trace around the curve drawn by the script. Only drop a handful of points - enough to accurately recreate the shape, but not so many it crashes your program again.
Then, delete the original line, and continue the original process now using the spline curve instead.
Works pretty well for me.
Hope this helps someone out :)
Can you make it as a plantary gearbox?😊
Couldn't you move the motor closer to the center to make it smaller?
Wait how did you create a sketch in fusion360 from script?! I‘ve kinda been looking for this exact feature!
Where can I find 3D parts models?
I wonder this drive was invented in my university... At peroid of my education!
So... its like a worm gear but colinear with with driven gear?))))
Any chance we will see these fusion files on get hub?
yep. github.com/roTechnic/internalECGears