High Torque Cycloidal Drive (NEMA 23 Motor)

I design and assemble a cycloidal actuator based around a NEMA 23 stepper motor. Hopefully, this design will eventually be used in a robotic arm.
Gain access to CAD files and support my work - / levijanssen
The motor I used in this project - amzn.to/2GonCgU
Disclaimer: I may earn an affiliate commission when items are bought through this link.

[@safetyinstructor]

Ounce inches? Burger pattys per hot dog... Metric is the answer

[@Mysli0210]

My reaction was pretty similar xD

[@jorgenskyt]

My cursor instantly hovered frantically above the "thumbs down" icon ... imperial "units" gives me the creeps :-/

[@safetyinstructor]

@@jorgenskyt I know, So complicated and frightening.

[@SaHaRaSquad]

Unicorn turds per full moon Also I wonder how this thing turns at all, considering the "wobbly" wheel seems to be directly connected to a fixed output. But maybe I overlooked something.

[@quaraluch]

And the whole project should take about 15,6/31 metric light yards of time to complete :).

[@legomanTJ]

Removing pins may remove overall friction, but having more pins will have more surfaces for load distribution

[@lanechandler9469]

Instablaster

[@lambsauce5445]

doesn't make a difference if it can't move at all ;)

[@gagatube]

Saw another video on Cycloidal drives where small bearings were used on all the contact surfaces - they demonstrated the design _is_ back-driveable.

[@lambsauce5445]

yeah, James Bruton I think ;)

[@klausnielsen1537]

You made your own driver for the stepper, then designed a rugged metal pin and delrin based cycloidal drive and made it work!! What makes you think there is something you can not do? So well done. Keep it up 👍👍👍

[@AndyRRR0791]

I think you drilled out the wrong holes. You need to allow eccentric motion between the output drive pins and the white cycloidal gear.

[@xxportalxx.]

Yeah I don't know how it would even be able to turn without that, interesting

[@ahmadseblini5184]

Right, that is why it doesn't have a contineous and smooth rotation as the acrylic is being smashed by the 4 Pin (from output shaft).

[@justtestingonce]

You need a second disc to offset the vibrations for when the 1 disc hits the pins.

[@owenjones9608]

Yep, with the cams for each cycloid disc set at 180 deg to each other.

[@franciscoferraz6788]

@@owenjones9608 I see... That's beautiful

[@cruzemissile5409]

design a direct drive 3d print head using a very small stepper motor and this concept.

[@neur303]

Second that :) I only found one project using a cycloidal gear for an extruder. Wonder what's the reason we don't see this more often for extruders.

[@Chris-hn4lp]

​@@neur303 There is too much slop in the mechanisms as well as too much friction. Therefore it would have horrible backlash in an extruder, and would extrude inconsistently. Planetary gears are superior in just about every way to this. Of course this does allow a bigger gear ratio in a small form factor, but the negatives aren't worth it.

[@neur303]

@@Chris-hn4lp I read quite the opposite. That they minimze backlash in comparison to other gears.

[@neur303]

@@Chris-hn4lp It seems planetary gears offer more torque per volume. Which would be quite reasonable for a direct drive as lower volume should also equal lower weight. But I still believe there's more to it. Like you wouldn't be able to use a planetary gear with plastic gears. And the cycloidal gear should be a lot easier to manufacture. www.machinedesign.com/news/article/21829580/comparing-cycloidal-and-planetary-gearboxes

[@williamfosterXD]

@@Chris-hn4lp yeah this a one of the few designs without backlash, not that it is an issue for the extruded as it can be correct with extra restart on retract ...

[@IIBLANKII]

3:24 tape the hole up first I know there is a bearing, but metal shavings/dust+ magnets are just something that will give you a head ache.

[@avigetsbored]

would be pretty cool too see you test this baby out, maybe puts some weights on a pulley and see when it cant pull anymore ?

[@machinemaker2248]

When designing for 3D printing, keep in mind that every measurement in Z has to be a multiple of your 3D printer's Z step size. If it isn't, your parts will be off by whatever amount is necessary to bring the height to the nearest z-step. Also, use machinist's feeler gauges to get your bed dead nuts accurate.

[@BrosBrothersLP]

Yeah but you have to think about the first layer too. Because for adhesion you squish it most of the time

[@BrosBrothersLP]

For simplicity i stick to all dimension being 1/10th of a mm tolerance and use that as layer size

[@FreeOfFantasy]

@@BrosBrothersLP There is the elephants foot setting for that to compensate the overextrusion in the first layer.

[@techko0240]

@@FreeOfFantasy the elephant's foot setting just adjusts the xy expansion of the first layer I think @BrosBrothers is talking about the not perfectly defined hight of the first layer ;)

[@dhruvbose8294]

Adaptive layers solves this issue to a great degree.

[@sennithblacky]

Your output shaft can only do circular motion, however your gear moves in a wave form around your motor axis. Pining these two together creates a lot of friction. Using long holes towards the center of your output shaft would allow for the wave motion of the pins and no sticking.

[@maxhouseman3129]

Oz/Inch? Please use Nm 😎 If you need reliable and good stepper drives, I recommend Trinamic.

[@nou4898]

14:33 the motor has laggy internet connection

[@othoapproto9603]

nice work, but I think pins on shafts greatly weaken the shaft by subtracting up to 40% at a crosssection. Grinding a flat spot is much better IMO.

[@PraxZimmerman]

The clamping on the cnc router gave me chest pains. Great video though 👍

[@brocksdaddy081910]

Was looking for this comment.

[@TesserId]

...though some hammers are more hammer-like than others. Love the Adam Savage quote. And, I too have found that particular style of pliers to make a really good hammer on many occasions.

[@baconsnot]

I think most cycloidal reduction drives have two sets of gears with 180 degree offset lobes. That may reduce your backlash somewhat

[@S0ulinth3machin3]

The offset lobes won’t reduce backlash. The purpose of the 2nd lobe is to eliminate vibration. Any mechanism which includes an eccentric rotator will have vibration. The 180 degree out of phase counter-rotator cancels it out. As for backlash, if a cycloid is manufactured with the correct tolerances, there will be zero backlash. All teeth are simultaneously engaged. The lack of backlash is why cycloids and harmonic drives are used in robotic arms. Without the lack of backlash, repeatable, precise positioning is not possible.

[@Coltgov191145]

I have noticed that you never put any grease or lubricants in your gear systems you would greatly reduce friction and increase produced torque values if you used some grease...

[@JorgeMarioManuelOrtega]

u need to make it double staged to cancel the wobbling. been there. done it.

[@MM-24]

what does double staged mean?

[@JorgeMarioManuelOrtega]

@@MM-24 one way is to make the output shaft ot the motor moves the first stage and the output of the first stage moves the second stage but that does not cancel the wobbling ... its like puting two of this devices in line . u get a lot of reduction. the way wobble got canceled for me was to put two of the center parts of this design ( the one that moves) 180 degrees from each other one above the other. so you dont have the wobbling , its cheaper than the first way i told you , it has less parts but ... you dont get the added reduction of a proper double stage. chose your destiny...

[@NicholasRehm]

Completely off-topic from the video, but what camera are you running to record?

[@Sigmatechnica]

Try to get 5000, 6000 or 7000 series aluminum. You will find it machines *much* better without bending out of the way and going all gummy and pushing out of the way. most of the aluminum sheet you find is 1000 grade which is pretty much pure aluminum, great for bending into curved shapes and anodizing, awefull for anything else.

[@zhangyichao4612]

Oh, this is a very streamlined design, I like it.

[@victorv175]

Good job! I tried bearings instead of pins for the roller and got smooth running.

[@LeviJanssen]

If your system is similar to mine, how did you ensure the straightness of the output pins? I think this is my primary issue.

[@franktkalcevic5342]

@@LeviJanssen I had the same problem so I used a second disk/plate on the other side of the lobe. My gearboxes, though, are pulley driven so I can assemble them from both sides - I don't have to stack parts on top of a motor - www.franksworkshop.com.au/wordpress/2018/03/04/3d-printed-hypocycloidic-gearboxes-iii/

[@Chris-hn4lp]

As other people mentioned, you drilled out the wrong holes. The holes in the white piece needed to be bigger

[@A.Lifecraft]

Those ridges that are supposed to keep things precisely in place should be deburred and cleaned up.

[@Unmannedair]

Not the cycloidal great design I was looking for, but still very interesting. I guess I was looking for harmonic gears

[@haenselundgretel654]

An easier setting for your nema23 might be an arduino with a cncshield, a drv8825 or better stepper driver and estlcam as control software. Much easier to set up and vary.

[@LeviJanssen]

I only made a custom drive because I didn't have one lying around. A proper shield/circuit would have been desirable, yes.

[@haenselundgretel654]

@@LeviJanssen it's just the arduino with a "cncshield" for about 10 bucks at aliexpress and a stepper driver like a drv8825 or a tmc2208.

[@Narwaro]

What in the name of god are ounce inches?!?

[@malakiblunt]

I have no experience of cnc routing (but i do with manual milling & turning) your cutter looks very long. which is just asking for chatter added to the already less than rigid structure of the machine, the shorter you can make the cutter (shorter cutter, or cut the shank and pushed all the way into the collet) would improve the finnish and accuracy your getting on your aluminum parts.

[@TesserId]

The wobbly vid was funny at first but got to be an eye strain. Being a person working with mechanical devices, might want to find a camera mounting with some isolation from the work table. As a side note, having recently been to the dentist, I was a bit fascinated with the ceiling mount for the work light, with it being totally free within it's range of motion and total isolation from everything else (except maybe the floor above). This won't cover all needs though, which is why it's just a side note.

[@illusivec]

I recently decided to design a 3D printed cycloidal gearbox and throw it on top of a 2300kv brushless motor... After a lot of fiddling, I actually got it to work. I think the ratio was 29:1 It ran for 10 glorious seconds. Then the motor windings burned and the magic smoke came out of the ESC. I think my problem was that the main gear was sitting right on top of the housing. Even though I put some grease in between them the friction was still too high. I was planning on putting some ball bearing between them but I just don't have the time. They are a bitch to design and manufacture but it's wild how such a compact gearbox can produce 30:1 ratio.

[@sevdev9844]

In the future, please mention the non-US metrics as well, not only once-inches. Thanks.

[@OsmioGonzalez]

It is a realy good proyect. Actualy, it look prety profesional and in the same line of the clasical way stepper motors look.

[@EEtransmit]

Wauw! Cool project. Little tip: don’t mess up clearance and tolerance... 😉

[@TheNefastor]

It's not friction that prevents you from backdriving it, it's mechanical advantage (or lack thereof). 20:1 ratio means this is like trying to use a level to lift something, but the fulcrum is 20 times closer to you than to the thing you've trying to lift.

[@DJRockinRob]

Wow! That was fun to watch! I would ABSOLUTELY want to see pulling a car with this! What an amazing device! :-)

[@LetsPrintYT]

Wow, really great video! Love it 😍

[@kennethkustren]

Do NEMA swim well ?? I know a guy ... tracked submersible vehicle ... has issues with suspension, and buoyancy... has limited underwater vision capabilities ... no airtight compartments for batteries ... and wants to do it DIY,... himself . Weird, eh ?! .

[@techytools5668]

can you make all parts with cnc i have cnc but not 3d printer. and can u make detailed video for all parts. please ur design is compact and elegant i want to build this. please can u provide me design dawings if u can.

[@MrAwawe]

Come on, ounce-inches? I've heard of foot-pounds, and even inch-pounds, but ounce inches? You imperial types can't be accused of not being creative, that's for sure. Has anyone ever said "You know, I really like the unit foot-pound, but it would be much better if it was 3/4 as big"?

[@goguadevaratu7185]

where's the oil?

[@arned432]

Yea good luck stop 6000g/cm3 of torque with fingers. But like most cycloidal drives on youtube you don't make ofset disc.

[@tuberroot1112]

You have a cnc mill, why are you making junk plastic parts in plastic? That is why you pins are not straight. You are also drilling stuff by hand when you could be finishing with pocket operation on a mill to get it spot on. You eccentric looks very rough and fragile, why isn't a milled part? I see people desperately trying to find a use for 3D printers because they bought one. Once you've made the key fob and realise everything else breaks, you can forget them. I see none of these parts which would not be better made on the mill. Instead of a plastic gear with a metal insert and a metal plate on each side you could have a metal gear with a shoulder and one plate, no insert, no screws ... and pins would be straight and solid.

[@Moronicsmurf]

Could you... use a smaller amount of "loose" pins, and leave space between, so that you could put each pin on a small bearing supporting it at top and bottom, which "should" reduce friction, but also increase load bearing capability as it would need to rub up against an outer wall.

[@villarsrl]

how about some grease inside ? that should help

[@TimSheehan]

You americans use foot-pounds... and also OUNCE INCHES? WHYYYYYYYYYY

[@eppicshotgun1281]

Milk of magnesia works well as lube for working with stainless steel.

[@pierdolio]

Don't know if this is a joke or not but what does the MOM do? Prevent galling, (which is the biggest problem when working with stainless)?

[@eppicshotgun1281]

@@pierdolio idk what it does but a welding instructor of mine said he used it all the time in the nukes.

[@chattha3272]

I was surprised to see you had just 4k subs. I was expecting around 2 million subs. For such amazing and original content, you are underra ted. Good job lad.

[@kishor_VU2MZT]

NICE MODIFICATION ,, I HAVE A QUESTION , ON REST WITHOUT POWER SUPLLY HOW MUCH IS RESTING TORQUE OF THAT MOTOR , WILL YOU ABLE TO MOVE MANUALLY ON RESTING SRATE & WITHOUT POWER SUPLY

[@rodstartube]

You can't trust your precision mechanism to a 3D printed part... or worst, several 3D printed parts.-

[@RickBaconsAdventures]

I feel like I may have seen one on your bench but I didn't see you use it. You need a "Noga Deburring Tool" It makes getting rid of those machining/cutting burrs SO freaking easy. Very cheap too. Check out ABOM79, This Old Tony, Keith Fenner, etc for tips on making really clean metal parts. Machining is fun!

[@cedric69graf]

Review shit! It looks like magic - beautiful, but how and why is not clear.

[@danielmorrison7579]

Your cycloidal disc is eccentric, and mates with your output disc using the four pins. How is it that the output disc does not follow the same path?

[@TheKapitanMajtas]

One thing that always bothers me while I see milling on those ploters. Why in hell every one is using such a long tools for milling 5~ mm plates? Makes no sense

[@rodryk5605]

Cool project. Honestly its always a success when your part doesn't end up smoking, or melting, or breaking.. You get my meaning.

[@chrisalbertson5838]

At the start of the video, the output of the drive is not smooth. Is this a video artifact, due to friction or manufacturing defect or just inherent in this kind of reduction drive? If the un-even motion is inherent in the design it is a show-stopper.

[@darkwinter7395]

"Every tool is a hammer" - well, except my sister's ex... he's just a tool.

[@ryzenryne8747]

USE IT AS A PART OF A PULLEY SYSTEM

[@rustycobalt5072]

Just found your channel and am not only impressed but inspired. And I've seen you're going to college soon. You have more than the capability to thrive at any ivy league college. *Please don't* they will exhaust you with irreverent "busy-work" and you won't gain as much from their curriculums (as a self-learner). So much so you might lose your special spark. Especially with the "cultures" and horrendous bureaucracy that enshroud them Personally, I'm at Montana state in Bozeman and they have a great pace for hw, lectures and exams with enough time and freedom to do projects both through the school and independently (atleast for me) Wherever you choose, please don't try to get the most prestigious, as that doesn't really matter. Just one that you can afford and that allows your mind to be free to learn as you have done already

[@landonhillyard]

Second this. My research opportunities paid out dividends, and probably wouldn't have existed at an ivy league.

[@JohanDegraeveAanscharius]

Genius!!! Curious for the lifetime ... Can you run it loaded for a few weeks?

[@timramich]

The hole in the square black piece that sits on the motor should have the exact same size as the pilot sticking out of the motor. It's going to be a standard size for any motor of that NEMA 23 size.

[@ignacykarp3720]

Hwhat the heil are ounce inches?! (I really don't get american units)

[@proto_hexagon5649]

10:31 what if put bearing instead of pins? that reduce friction but i dont know if still add torque.

[@dobrom]

Называется "Волновой редуктор". Я видел такие на спутниках, когда работал в космическом производстве в 1991 году. На советских спутниках...

[@Rob_65]

There is a lot we cannot see without the design but it looks like the holes that hold the pins from the output disc do not match the holes in the internal gear. The holes in the internal gear should match the pins plus the eccentric motion in order for the output pins to move freely without (much) play. With the pins being crooked in the output disc I am sure that is not the case anymore. Drilling out the plastic part of the output disc by hand and then forcing the pins in location with the aluminum disc will not help - if anything it will exaggerate the problem. A good solution would be to use two discs, one on each side of the internal gear, to make sure the pins are mounted straight.

[@ThomasShue]

Why not grind a fly spot on the shaft then drill the hole

[@nrdesign1991]

This would be nice for 3D printer extruders, mercilessly shoving the filament

[@DgtalBreakz]

Looks so neat man... Im being mesmerized by all the different implementations of your idea!

[@JesseSchoch]

I can see why you chose to use pins but wear isn't really going to be your main concern with the plastic gear and the printed base is going to always have flexing. Have you considered milling the pin profiles into the base so it is one solid piece of aluminum? This would solve your bending issues if you can get your CNC feeds and speeds dialed in (those parts seem a bit crufty!). You may be better off with something like Jeff Kerr's "Anti-Backlash Compound Planetary Gearhead" which is designed to flex and uses the flex to reduce backlash with a similar reduction ratio and it can be easily printed.

[@portlyoldman]

Ounce Inches ‼️‼️‼️‼️‼️‼️⁉️⁉️⁉️⁉️WTF.? This isn’t the Middle Ages any more mate 🧐 Great video though 😎😎

[@jackwhitestripe7342]

sir, hayırlı olsun birader, hindistandan selamlar

[@mp-xt2rg]

Use the shortest bit possible when doing steel. Or just sink it as deep as possible in the collet.

[@blairconner8670]

Have you tried a full 3d print version of this build? I suspect it'd be more challenging to get the right tolerances.

[@Ryzler13]

Tools.. Beautiful work with beautiful tools. Now you can make robots to keep making parts and assemble them and make more robots and unleash AI and it will do it by itself. I bet somewhere out there there is an entire manufacturing chain that is doing that. I bet. Robots wont take your job, says everyone who isnt doing the job that robots are now doing. 200 years from now people will be saying robots wont take your jobs. It will be AI saying it. "We are AI, We wont take your jobs now here is Andro 5000 with the weather"

[@deltapixels]

what length were the Ø4 pins and what size bearing did you use?

[@АдилДжусупов]

Great job! I think not continuous (not smooth) rotation of the output shaft is caused by step rotation of the step-motor multiplied to gear ratio (x20).

[@fabricioleinat]

divided, not multiplied.....

[@АдилДжусупов]

@@fabricioleinat Right. My mistake

[@varunahlawat169]

everything was good until you used "ounce inches"

[@whatevernamegoeshere3644]

Wait. Ounce inches is an actual unit? o_o 3:27 that dab of wd40 probably screwed up the greasing of the bearing. Be careful.

[@amos9274]

If you ever do such a project again think about using the tmc 2209 drivers they will remove all noise and are controllable over a single wire. Cool mechanism tho

[@robmckennie4203]

How did you think this was going to work if you have an eccentric element rigidly coupled to a concentric element? Maybe I'm a moron and don't understand how it works, but it seems like the pins should be running in slots, not just put into sized holes

[@JohnBaldiga]

Based on your wasteboard, I see someone's been making some ear savers.

[@uboldin]

I think two Cycloidal Disc 180-degree offsets will keep the motion smooth

[@uboldin]

Due to the eccentric nature of the drive, if the cycloidal disk is not balanced by a second disk or a counterweight, it will generate vibrations that propagate through the driven shafts and the body. These increases wear on the exterior teeth of the cycloidal disk and the component bearings. The static imbalance is corrected with two discs, but a small dynamic imbalance remains. This is generally considered acceptable for most applications. To reduce vibration, high-speed drives use three or more discs to correct the imbalance; the outer discs move in unison, in opposition to the middle one, which is twice as massive. en.wikipedia.org/wiki/Cycloidal_drive

[@mous24000]

tol super nema17 und darauf win platte wider mit einer welle 5mm oder rausch bar mit 8mm serje und fille pobelem weren nicht

[@AlchemyOfDeathBand]

Hey boss, what about the coilgun ?

[@chrisw1462]

NEMA - National Electrical Manufacturer's Association. Say it with me. _Neema_

[@Bobby11]

I assumed that rotation is how a stepper motor works anyway. But the reduction of speed is showing it more. Thats and possibly exaggerated by the pins

[@gbbeeler]

A little hint for making those full depth cuts through metal, use cyanoacrylate glue to bond to a metal base that acts as a heat sink and cut all the way through your thickness of stock to be cut. If the metal stays cool enough (by keeping your depth of cut small enough) the glue will hold throughout machining for thin parts like you fabricated. When the stock your cutting gets too thick, the side of your cutting tool will grab and toss the part with this technique with full length cutting flutes, but you can cut thicker stock by adding a draft angle to just keep the tip of your tool engaged with the part when cutting deep. To release, heat the metals and the CA glue will fail avoiding warpage and dinging from manually shearing the joint (a little acetone soak works well too, but takes longer). This and some of the other good machining tips I see in the comments will really improve your finished parts.

[@bitrage.]

Just a curiosity...why didnt you just file a flat? It would of been easier and stronger then weakening the shaft by drilling a hole in it

[@brandonberchtold9484]

I would reccomend using a much shorter endmill on your router, like only a couple mm longer than the thickness of your material. That'll improve the cut quality and reduce chatter

[@The_Privateer]

This is better than a planetary gear arrangement how?

[@excitedbox5705]

why not grind a flat into the motor shaft instead of drilling a hole. You probably got metal shavings in that motor while drilling. Blow a ton of air in it to prevent shorts but they may be stuck to the magnets.

[@cho4d]

love the video... wish you used metric or at least put metric text overlay somewhere :) (in reference to inch pounds)

[@fragapple]

Hey, I really like your build. What kind of aluminium is that? It looks like you should go for higher feeds, the chips are way too tiny. This way the end mill produces enormous amounts of heat, because it mostly rubs over the surface. It will definitely die sooner than later :) Also it might make sense to go for holding bridges, which are easier to remove. Just my two cents. Did you make progress with the uneven rotation speed of the gear? How would you consider to fix it? Best

[@4dirt2racer0]

i admit im a picky person n have n eye for problems but this was a pretty dam good video man u dont have a bunch of sophisticated tools, but youve made aLOT work with what u have i respect the hell out of that if u keep makin videos like this youll do good for sure i hope u have more like this minus the mandatory ads :p jp

[@miguel18alvarado]

Esa presión para hacer el hueco pudo dañar el eje

[@odinata]

I have a question: would it be possible to make a functional, useful cycloidal drive on a 28BYJ-48 stepper?

[@derarty4290]

Good Question!

[@maxhouseman3129]

Sure. Its only a matter of construction techniques.

[@odinata]

@@maxhouseman3129 I wonder if the decrease in size would make it bind....

[@maxhouseman3129]

@@odinata If the whole construction is smaller, than it should work.

[@The-Weekend-Warrior]

Try to design for a 1:10 ratio if you plan to use in robotics. Also the pins inside should be able to rotate as the eccentric wheel pushes and leans on them (not rattle though)... maybe have them on tiny bearings? Also, maybe lubricate the inside of the mechanism...

[@dadsfriendlyrobotcompany]

Maybe for a 4th axis for your little CNC?

[@fernandoi8958]

3:30 this is why I have nightmares at night lol

[@orion7353]

Why was punching the metal parts out required...

[@michaelvaughn2287]

You don't have a grinder ?! Really all that equipment?!