3D Printed NEMA 17 Precision Gearbox with Position Feedback!
HTML-код
- Опубликовано: 5 окт 2024
- I'm so thrilled to present the final design of my 3d Printed Gearbox for a NEMA 17 stepper motor. The new design is a major improvement on the first prototype in so many aspects, just to name a few things: stronger construction, a lot of mounting points, position feedback with a hall effects sensor, more gear reduction and thus more torque.
The video has been some time on the way now, and has taken quite some effort, so I really hope you enjoy it!
If you haven't watched the first video already, you can find it here:
• NEMA 17 Gearbox with N...
-------------------------------------------------
RESSOURCES
Files for 3D printed parts:
www.thingivers...
Additional parts needed:
NEMA 17 Stepper motor + driver (from Banggood)
Hall Effect Sensor (from Banggood)
3 pcs. 5x11x5mm Ball bearings (from Amazon)
25x37x7mm Ball Bearing (from Banggood)
2x3 mm neodymium magnet (from eBay)
5mm steel/aluminum round bar (from eBay)
2 pcs. 200 mm GT2 belt loops (from Amazon)
Ruthex M3 Threaded inserts (from Amazon)
M3 and M4 nuts and bolts (from Amazon)
NB: If you make it, let me know! Send me a link/photo/video of your build - or post a make on Thingiverse ;)
----------------------------------------
Follow me on Instagram to get projects updates and much more:
/ emilostuff
![Natanael Cano - Amor Eterno [Official Video]](http://i.ytimg.com/vi/EO4prCs6WpM/mqdefault.jpg)
appears out of nowhere - drops a couple of ultra high quality videos - disappears. fu**ing legend.
I wouldn't call it position feedback, but homing or indexing, still interesting design.
@Robo Cop lol, community disagrees ... content creator even refers to it as "zeroing" and "homing" in the actual video - as such title is a bit misleading as most people looking for position feedback are likely after more than a single position that is achievable with an end stop or homing switch on the axis as opposed to in the drive train.
I agree, surely an encoder would have been better.
@@teenflon even better, there are hall effect encoders, which can monitor the angle of the magnetic field. one radial magnet on the shaft center and such sensor near it
This channel is underated. Great content, looking foward to more!
Thanks again very much! I'm glad to hear that!
I think you should try an optical encoder, or use more magnets. I think it will make it respond and get back to the position pretty fast.
Because the EXPENSIVE servo motors or closed loop steppers respond to feed back in less than a millisecond.
Good luck
Had a couple of ideas watching this, though they might require additional compute power compared to your current solution. First is that instead of a single Hall effect sensor, you could use two, with one each on the input and output spindles, hopefully allowing an increase in precision and accuracy and restoring some of the functional flexibility. Second is that with either one or two sensors, you could replace the solitary single-point magnet with a polarized magnetic strip along an inner circumference of the matching gear or pulley, essentially functioning as an absolute encoder, as you mentioned to Omar Aissani in a comment below.
I know I'm a bit late to the party, but - I think that the torque values are directly correlated to the holding torque of the stepper (because failing point was stepper loosing it steps)
Maybe it's an obvious remark, but I think it's worth to mention that this design won't overload itself and cause mechanical failure, which could be handy for some designs.
I admire how good it is described. I just see virtually every advantage and disadvantage of this solution now. The only thing I don't know is how will it behave after being worn off - backlash due to loosen off the pulleys. Actually I cannot name any more information I would strive. Good job.
These are the slickest FDM 3D prints I've ever seen.
really nice design!
having minimal backlash can be critical and is hard to achieve on a 3d printed basis. great job
Thank you! Cheers!
Full positional feedback next? :) Would be cool to have an open source affordable closed loop servo based on nema17 motors
Great content, fantastic ideas, nice video capturing, excellent voice over which is crisp, clear and always on point that explains well.
As Stefan said your channel is underrated. Please continue the brilliant work you will be definetly discovered.
I support you 🤗❤️.
This makes me so glad! I'll be continuing to make these videos and hopefully they will spread to more people. Thank you for the support!
Engineering student here... I'm hoping to make a 3d printed robotic arm over the summer break. I might just have to bash together a few of these gearboxes to accomplish that! Great video, and fantastic instruction!
Must say bro you've done a precise work on it!
Thanks man! I'm glad you like it 😊
nicely done, just one thing , this is not position feedback but built in homing
position feedback is that it will report at every position where it is. , can be done with hall effect sensors as well
thankx for posting and explaining.
You're right!
Where have you been all my life? Very underrated channel. I give it 5 stars. Would watch again.
This does seem a lot better than the 3D printed gear based gearboxes I’ve seen.
I agree that this is underrated and you deserve a lot more subscribers. Really good design and video - thanks.
I thought you sounded danish, and When i Saw the danskvand, i was reassured! Nice video, will go and watch your other videos, but i can already say you got yourself a new subscriber😊
Loved it, it even feels as if Christoph Waltz is narrating.
Hey Man, awsome work.
I just went to your channel to check out your other content.
I was shocked to see you didn't have 100k subs.
Your content is clearly A++ as im sure you already no.
As everyone is pretty much saying, ill be tuning in for your career👍
Keep up the great work 👍👍
I'm so glad to hear that! Thank you so much!
@@Emilostuff i agree. i think the youtube algorithm is holding you down. :P
Found this channel, immediately subscribed ! That's a great design i want to see more.
Here are some improuvements i can think of:
- Stepper motors usually don't get hot so u can include it in ur enclosure.
- Adding a pcb that contains a small microcontroller and a stepper motor driver.
- Create a hole for a connector, so u can just provide 12V and pins for communication with the master microcontroler.
- Create ur own library, OOP will be the best route to take to get and set data E.g stepper.getPosition();
Hi! Thank you very much! And thanks for your suggestions - I like them a lot. I did briefly consider making it a more 'complete' solution with all the electronics inside. So maybe I will look more into that, If I revisit this design. However I do have another stepper gearbox design in mind that uses an absolute encoder. I think this would be an even better candidate to make fully integrated.
I recommend against including the stepper motor into the enclosure, personally, unless you also want to take the time to print or mount a set of cooling fins. Whether they get hot, how hot they get, etc; depends on a lot of factors, and plenty of people have had their stepper motors get very hot over the years, especially on longer or high-speed prints.
Adding a dedicated microcontroller and driver is fine for those whose projects require only one motor, but it tends to become a balancing act for people who need several motors and already have a monolithic control board (ex. RAMPS, etc). A reasonable compromise might be to include a simple coupling board so that your connectors are secure and all in the same location, and a list of recommendations for drivers and microcontrollers. This would also help to keep the production cost down in the end.
Stepper motors definitely get hot under load
"We will be watching your career with great interest"
Very well thought out Project and Video. Nice clear Voiceover and easy to follow along. Good Camera shots and editing as well as a good choice of background Music.
The only thing left for you is the "and a special thanks goes to all of my Patreons and my sponsor Raid Shadow Legends" :D just kidding
Hi Beredro! This made me smile so much! Thank you for your kind words - I'm very glad you like the video :) Hopefully your prediction will come true some day!
I know I'm kinda late to the party but there's an easy upgrade for your system, you could place a potentiometer (variable resistance) without stopper attached to the gears, by doing that you would be able to know the exact position of the arm simply by reading the voltage of the potentiometer and utilizing the magnet to "zero" the system or reducing complexity, just home it yourself
But using potentio means using analog signal which usually means *noise* . I'd rather get myself an encoder.
But if you want to keep the price cheap, yea i'll go with the potentio as well
@@rizalardiansyah4486 I've made pure analog control systems, if you know what you are doing noise it's not an issue, first of all, get good potentiometers, not cheap ones, second as long as you have a stable input voltage and your cables are good the result will always be the same, encoders are only needed in hostile environments, like a factory with a high degree of temperature variation or high amount of vibrations
Congratulations, great engineering....it's the IBM process "Plan your work, work your plan."
Your videos are very well done, you got a new subscriber! Here is my idea for the version three, add a magnetic rotary encoder to make this a pretty nice hobby grade servomotor. Part four could be a video about designing a driver for it. Look up AS5600 for example, you would have to tinker a bit but maybe you can put this on the output shaft to reduce backlash even more, active backlash compensation. If that results in too much oscillation just put it on the servo itself.
Have fun with this! AS5600 you can very fine position feedback with this one!
I am very impressed with the outcome. You persevered in fighting back the results from the first attempt and building a better mouse-trap! I accept that there is a practically unbeatable issue with the flexibility of something that long and its ability to fight or resist movement. I personally don't call the movement backlash - I use that word to explain the exactness or sloppiness of gears and their mesh-ability! I know that the object you built, if made from large block of steel or aluminum and no plastics or belts involved, would not have that amount of movement. But you have made a strong servo. Haven't I heard of strong servos before? Made like a normal plastic servo but in metal? But you did have a self imposed problem to solve and I believe you did it.
I have a need for a stronger mechanism than I have now. I developed and built a vehicle platform that allows me to ride on it. It is basically something that allows a wheelchair to ride on top of and it uses 24vdc motors to move with and a car window motor to steer the front wheels. I use a DPDT switch to steer the wheels from left to right, but the centering is not self managed. I have to do it and it is difficult to get it in the middle. As well as the steering is not proportional. There are a couple ways to fix the overall problem but I want to keep it as a fly-by-wire system (isolated) so that I can implement other features down the road - like RC control of the whole vehicle. This is my pet project - just to see if I can do it. Don't we all have something like that we want to do?
Plus I liked the fact that you talked us through the process with explanations and discussion. That is becoming more of an issue with me - I'm not going to watch an entire video of hand movements!
This video definitely deserves more views!!!
Thanks man! 🙌
I love the camera application at the end! I could see a few of these being used to make a Timelaps rig
What an awesome idea, I definitely know applications where the homing with a hall effect sensor would make all the difference. You should make a video about that (and most importantly the code!!) Thank you!
Great work, good video and sound. Nice to see a Danish guy, making quality engineering content on RUclips, looking forward to more content from you.
Thank you so much! Much appreciated :D
Damm...I expected you to have way more subs.....great video quality. Keep up the good work.
Thanks!
Awesome project!
Concerning the M3 flat nuts I always print the square holes without any supports on my prusa Mk3.
If I can reach the holes afterwardsI clean up the drooped overhang with a knife but generally it isn’t necessary. Especially if precision of the nut position in the “z” axis is not critical I just oversize the z dimension slightly. For a good fit I use 2.2mm height for my square and it 98% fits without cleaning anything up. If I want to play safe I up the dimensions to 2.4 or 2.8mm that for me always fits the nut without any work. This way you don’t need to use expensive and difficult to implement threaded inserts. The nuts can clamp on the layer lines with minimal material in the x and y directions the inserts need more material all around taking up mich more space which you generally don’t want when building things with M3 bolts :)
Good content and excellent video editing.
Thank you very much! Glad to hear that 😊
One solution would be to give a precise encoder at the output. You can make sth similar to digital caliper around the circumference of this output holder.
And the ender3 pro is a great machine, mine printed great from the start, and 6 months later, still printing awesomely. In addition, Angus Devrron of MakerMuse, has a great little tolerance gauge that will help you dial in your CAD tolerances for your printer.
Thanks for the tips!
Looks neat. Looking forward to next version with absolute encoding 👍🏼 Thanks for sharing! Subscribed
It would be interesting to see this turned into an actual servo with an encoder or an analog hall effect sensor.
Next step is a rotary encoder (those looking like a pot or those Will Cogley uses) on the driven gear!
Use selfcentering bearings for the belt tensioner axis, or use dedicated tensioner rollera and have the gear axles solidly mounted 😉
This is awsome working ı very thank you for your hard working and sharing with us
Excellent video and design process explanation - thanks!
Very impressive!
Good presentation of information. I wish you a speedy promotion of the channel and millions of subscribers.
Thank you so much 🙂
Nice video! Subscribed immediately :)
For my taste the background music is a bit too loud, but thats neglectable ;)
Keep up the great work!
Noted!
could you make a tutorial on how is the electronic parts done?
Great work - nice design!
Nice work! I tried designing a gearbox like this, but ended up abandoning it. Mine was going to be driven by BLDC motor, and have a gear stage before the belts. That would give very high power to weight ratio, and still much lower backlash than a servo since there is only one gear stage, and the backlash from it is reduced by the total belt ratio.
subscribed. really nice project. i like how you designed the output plate and the inside equivalent around the bearing, its quite elegant and i assume strong
If you make nut holes on the sides rotate them 45 degrees, so that you don't have to print them with overhangs. ;)
Nice Solution, but if you want to make it less dynamic dependent the encoder should be attached to the second shaft to measure the real position, otherwise you will always measure angle filtered by some second order system. off course a reduction will be necessary to keep the encoder precise enough. Nice job!
I was looking for repurposable gearboxes for different types of motion. This should meet that need nicely.
If I'm not mistaken the "Servo42B" does the same thing you want. Its a closed loop stepper motor with up to 256x microstepping :-)
You're right! but I want the gearing as well for greater torque :)
Towards the end it didn't look like you were losing steps. It seemed to rest back in its "stretched" position where there's 500g worth of leveraged tension on the belts, but the motors seemed to take the same amnt of steps.
Very satisfying to watch. Thanks 🙏
Glad you enjoyed it
Great job men congratulations, just watched it and subscribed really well
try using an AS5600 breakout board (hall effect encoder) from Aliexpress to add servo capability. That would turn your project to a whole new level
Well, under "positional feedback" I understand "continuous positional feedback". Not just feedback at one specific point. A rotary encoder would probably be more useful (preferable on the output gear, to compensate for the springiness of the belts, at least a little). Apart from that, it's a neat design. I don't think the threaded inserts are the best solution though. If designed correctly you can easily print the slots of the nuts etc. without any support. And the nuts are a lot quicker and easier to insert, compared to threaded inserts. Not to mention they can probably take more force.
Very very nice, looking forward to more videos!
Nice work. I think you should use an adjustable idler to tension the belt rather than the output pulley. This will allow the output pulley position to remain static so that it's location is absolute should this assembly become coupled to something else as a driving unit.
Do you think the hall sensor would detect the magnetic domains in the material used to make flexible refrigerator magnets? IIRC in one direction the material has narrow stripes of alternating north and south oriented megnetism. It might be possible to glue it to a gear to increase the resolution. Although, I think an optical sensor from an old mouse wheel might be better. If you were feeling really ambitious you might be able to use the principle that makes digital calipers work. It is a cool project as it is but I think you need more resolution to call it a servo, so that if perturbed a little it can move back to where it was.
this thing is beautiful
Great work!
Thank you!
I might scale this up for a Nema34, for a project.
Have you throught of using Worm gear these can move high and stay in one position
Worm gears can be very hard to print. If you print them lying down, the down side will be squashed, or filled with support material, and the up side will have layer lines, making it weak. If you print them standing up, the shear stresses can rip them apart in-situ. I've had some luck with printing them vertically, then powdered-salt remelting.
@@voron_design to make matters worse, the less backlash you have in a worm drive, the higher is the friction, grinding away material all the time
I’d love to make a star tracker with something like this
Why does this channel not have a million followers
awesome design, I learned a few things :) thank you for sharing!
Great to hear!
The torque is more because you didnt include the weight of the leverarm, if you make the lever symetrical on both sides of the axle you can exlude it from the calculation
Cool design but be aware of potential overheating problem and significant decrease of the stepper motor performance under high speed.
A thing that a sugest you to do is try to do a controller inside the motor itself and make itself Control the necessary movements to keep the arm into the exact same angle , not by strength or system resistence,but eletronicaly.
other thing i sugest is to use a potenciometer,like those from the mouse Wells,conected to a gear to get How many gear teeths has passed by for each motor movement,equationaly you can get How many degree you moved by the amount of theeth haver been passed by the potentiometer,as the mouses itself do
And If possible,you could gently put the libs or algorithims tou hás used to your tests ir development?
a place like github would love such a amazing project as yours
Nice project. Your video is well done also. Going to download the files now to see the details. Can you tell me the ratio of the gear reduction that you achieved?
you should build a person 3d scanner camera system with this
Another great video! Would you consider to publish the Fusion360 files, to facilitate modifications/adaptations? Additionally, I'd like to ask what the advantages and disadvantages would be to use a BLDC and encoder, driven by something like ODrive, as compared to a stepper motor. Many thanks again for a great video! Hope that more people find their way here soon!
Thank you! Yes, great idea. I'll see to it, that those will be uploaded soon.
I think there are many advantages of using BLDC's over steppers, both in terms of power, speed and noise. The only downside in my perspective is the increased cost and complexity of using eg. an ODrive. But BLDC systems are only getting cheaper and cheaper, so it's definitely the way to go, also for me at some point ;)
You can pull in any design to fusion 360 and convert it to editable
Hey, just joining in with other comments - I subscribed after the very first video I saw back in April.
Thank you so much! Glad to have you here :D
Next version use magnetic rotary encoder, like the AS5600... super cheap and will give better results than a hall sensor. :)
Why not use an absolute encoder e.g. AS5047 for getting rid of homing at all?
Agreed! Definitely next step!
Interesting way to have feedback from a stepper motor . I would also like to what is the name of the aluminum extrusions? I think those will be very handy for DIY projects.
12.5 Kg cm 1.22 Nm
1.23 Nm
Nicely done.
You can easily get rid of the issues with the press fits of the nuts by designing the upper surfaces with bridging in mind. I quite like the design, so I might attempt to make the modifications myself. I recommend you check out gregsaun's maker cheat sheet on GitHub for a codified versions of some of the modifications I would make.
Cool!!! If you had 4 magnets in the hall effect sensor would it's accuracy be better?
I'm sensing a future diy ultra cheap direct drive wheel for sim racing
Good project! But the video would be muuuch better without this background"music" and any chance to hear the device working!
Very nice implemetation! ty
that hall effect sensor is just for homing, we want a proper encoder on the out put shaft. If you have a proper feed back controller, it will fight back any torque you apply to it and not just lock your stepper motor. it will improve your precision drastically. heck you dont even need a stepper motor, you can just use a DC motor, I think there are very accurate hobbies stepper motor were it uses optical encoder instead of potentiometer as feed back.
Did you think about building the gearbox with a worm drive?
Theoretically there should be no backlash left because its self stoping.
Correct me if I am wrong.
Great content ! Hope to see more.
A gearbox based on a worm drive is a great idea, but it won't remove the backlash between the gears unfortunately. There will always be some play between the gears, otherwise they wouldn't be able to turn. a worm drive will make remove the backdrivability though, which means that it can't lose steps even under too heavy load. That can be a good thing depending on the context :)
The deflection I'm going to guess is in part due to the tensile elastic strain of the belt or the bending elastic strain of the aluminum bar. I think you could create a closed loop system to correct the angle from the displacement on static positions using a rotary encoder over a Hall sensor. That would be more difficult to incorporate though.
I have a question though. I've seen that a lot of people usually go between three different approaches: a belt, a planetary gearbox, and a strain wave (harmonic) gearbox. What was the reason you chose the belt system over the other two?
Edit: My apologies for any inconvenience in necroing a two-year old video.
Great project, but could you please explain to me how you fit the small gear to the 5mm NEMA17 motor axle
Hi this is awsome thank you for the design I wish to make a few changes to fit my stepper motor can you please upload the cad files instead of the STL
u da man! it would be really great to see the code! may be next video?
Could make some kind of brake system to hold it at desired location.
Good Job!
wow! great work! instant subscribed.
Thanks a lot! 😊
Great work man. Very well made video. New subscriber here. As some already mentioned this is not true positional feedback but more of a homing feature. You should have a look at implementing a proper encoder in there 😃
Hi, great design thanks. Have you a link for the aluminium extrusions you talked about please?
great man..............looking more from you
Great video, the parts look very crisp and clean. What 3D printer are you using?? I am thinking of buying one for making some small parts for prototyping.. what design software you use??
RUclips recommendations are getting better later
well done!
thanks!