Integrating My Harmonic Drive - 3D Printed 6-Axis Robot Arm
Вставка
- Опубліковано 11 лип 2021
- I took my 3D printed harmonic drive from a previous video and a newly designed motherboard to build up the base of my new 6-axis robot arm.
A huge thank you to PCBWay for sponsoring this video! See more here: pcbway.com/
Eventually this arm will be filming my B-Roll for me, so there will be plenty more updates as the development continues! This arm uses my harmonic drive from a previous video, and a lot of work has gone in to ensure this arm is not only cheap and powerful, but looks great too. It should end up costing around $500 total, have a working range of 500+mm, and be capable of carrying 1kg with ease. It wont be super fast, and it wont be super accurate, but it will be an excellent camera minion.
The project is open sourced! Find the WIP files here. I do not recommend attempting to recreate this project yet, I still have a lot of work to do and I will document the parts needed in better detail once the design is ready for others to make.
github.com/DDeGonge/OS-ARM
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
If you'd like to support plus receive some exclusive perks, you can do so here!
/ 3dprintedlife
Join the discussion on Discord! / discord
Follow me on Instagram and Twitter! / 3dprintedlife_official
/ 3dprintedlife
Buy my filming gear! (Affiliate links)
Sony a6400 amzn.to/3d0SMee
Tamron 17-70 lens amzn.to/3vLqYkx
Oben Tripod amzn.to/3dd4S40
Light Panel amzn.to/3Qa1dTs
A huge thanks to all my Patreon Supporters!
- Mom & Dad - Nullun - Jake
- Trent Waddington - Barb - Amy
- Ayydan - Eric - TrueCarry
- Justin Gort - Pavlo - Will Hofmann
- Mikhail - David Jones - Robert W
- Michael H - Tyler Zlatkus - Michael S
- Austin R - Chris Santos - Mark Whittington
- Dominick K - Martin Wilson
#3dprintedlife #robot #harmonicdrive - Наука та технологія
![I Made a Robot Arm to Hold My Camera [$500]](http://i.ytimg.com/vi/e3TNaIyTAnY/mqdefault.jpg)
![I Made a Robot Arm to Hold My Camera [$500]](/img/tr.png)
If you enjoyed the video, consider checking out some of my projects like a 3D printed wankel engine, card shuffler, nerf aimbot, and more!
Any reason to not use something like a big tree 6axis controller?
Any reason you didn’t use deep groove ball thrust races?
That would work, but I wanted to build my own partially for the challenge but also to get it in a small enough form factor and have the power I want and all the other IO and features.. It definitely isn't necessary though! 😁
Deep grooves perform better but the ideal solution here would actually be two angular contact bearings. But getting those bearings in sizes this big are not cheap
@@3DprintedLife
Excellent, to build your own is always a rewarding feat.
I was looking at one of the stepper driver modules that has 20a fets but the module can’t handle it.
Mechanical, electrical, firmware, magnets, IKEA "hacking", bending the test jig out of shape - this video has it all!
And don't forget the cat conent at the end, perfectly balanced content.
@@rafaeldullnig1320 As all things should be
This channel has the quality of a million subscribers channel! I'm shocked there are only 32k subscribers.
@@nicobugs
Agree and subscribed accordingly.
That board debugging sounded painfull...
Just a little, hopefully I'm almost done with it though!
@@OneDeuxTriSeiGo simulation makes only sense if you primarily design analog or simple digital circuits (by your own). What he does here is more like a integration job, the discrete parts around the ICs are mostly recommended by the vendor and the ICs are connected via bus. There is nothing to simulate here. Also getting these chips into spice is not simple and the outcome low.
Best thing to avoid these kinds of failures is to build it on a breadboard before or have experience with pcb design.
Greets from a hardware developer.
@@OneDeuxTriSeiGo oh cool! I am developing with fpgas also since a few weeks. Interesting stuff!
What you described is basically what a professional routing software does. It checks whether nets are connected properly and (if the setup is good) also compatible with each other.
But, as I said it is mostly connecting working chips with each other, like building a PC. If there are routing specific things, you get that information out of the datasheet.
The problem with spice is that you have to build models for the chips or at least some dummies which can be time consuming. Don't know if it is worth that. I have done that during my engineering studies and I it was just frustrating. I would rather build it on breadboard, if the circuit isn't that complicated.
@@OneDeuxTriSeiGo You could do this but it would take a lot longer than just reading the datasheets and properly designing the board in the first place. If you're thorough enough, I expect no obvious mistakes. If you don't know if something will work before you make it, you haven't put enough effort in or don't have adequate knowledge.
This video basically has the same amount of work done in it as my whole year of being a full time R&D engineer. Jaw dropping diligence!
Isn't it sad when our hobbies usually get more love than the place that pays us.
I appreciate that you include all your problems and challenges 😊. It keeps me motivated when I struggle with my own projects.
I REALLY wanted to see it crush the can at the end. Maybe next time?
I did too but just not enough power, hopefully running at full power or using a shorter arm it can!
Would it possible for you to make a video on your considerations when designing the PCB and more importantly, how you debugged it?? It's fascinating how you used the logic analyser to find out what was wrong with the PCB and would love to learn how to do it myself!
If enough people are interested then definitely! At some point I'll start a second channel where I hope to cover topics like this!
@@3DprintedLife i am interested
@@3DprintedLife Please do.
@@3DprintedLife DO it
Im intrested
I find herringbone gears are great for 3D pritning. They're not common anywhere else because they're hard to manufacture any other way, but they're self-aligning along with having the advantages of the angled gears you mention in the video. But feel free to do what you think is best!
Double herringbone gears are superior though, because torque doesn't generate a force perpendicular to the surfaces of the gears
The 3d printing is cool and all, but that cat is pretty darn cute
for this kind of comment is why I still read youtube comments
If your desk is hollow you might want to add some larger washers and/or plywood plate so you don't accidentally torque the screw heads through.
Happy to see that you could test new filaments for the harmonic drive !
Looks amazing !
Dude, I owe you big time. I was so fired up to jump into this. And then you went all mother board on me. Immediately thereafter I saw, with brutal clarity & conviction, the great precipice between your skillsets and mine. It was over before it even began. But, look here! Just knowing that put me ahead of the power curve, so to speak, and that ain't bad. You're a good man!
Such a great job you've done so far, can't wait to see the finished product!!
Thanks! Same here!
The Sebastian Lague of electronics and mechanics. =)
The last few videos are so super well made, I'm happy every time that one comes out!
its insane how you are taking the time to walk through your design process, I am thrilled to see the final result! new sub definitly :)
Thanks and welcome!
Man, this is like engineering that I always wonder about, but never see anyone do. This is great stuff. Thank you.
The 0 stock everything is really speaking to me. The alternative alternatives, it's hell.
Mhmmm not fun at all!
That horizontal motion looks and sound fantastic 👍👏
Excellent content. I think your channel deserves a better name. There is a lot more mechanical and electronics design than 3D printing. My personal experience is that a 3DP audience quickly outgrows a 3DP channel. After building some DIY printers and have you chamber hot enough to print some exotic materials, you are basically done with most 3DP channels. A 3D printer is just a tool for make DIY projects.
Agreed
The Mona Lisa, ohh please no, THAT PCB is a TRUE WORK OF ART. Awesome video I’ll start ordering boards from PCB way they look way better than the ones I get!
Looking forward to the next update!
I just found this channel and have been binging the videos. I was curious how you have learned all of this this and got to this point of being able to develop all of this. Love the vids!
Glad you enjoy it! I learned everything from a combination of university classes, working professionally as an engineer, and self-taught skills for personal projects.
Amazing work, love the enthusiasm, keep it up. waiting for the next video
Thanks I appreciate that!
Remarkable content. You have inspired me to make my own take on a robot arm, and you have mass fame and following on the way!
You're an amazing maker AND UA-camr I only wish I had your skills. Keep up the excellent work👍
the completeness of your explanation, and the style of editing... must be a lot of work (at least the editing part) and I love it! New sub! whoop whoop! :D
Thanks! And yeah you're right, it takes a lot more time than most people realize. But it's well worth it!
@@3DprintedLife happy to hear that! I'ts by far the most important part ;)
Great work man! I know these videos hardly capture all the work you actually do to make that hunk of filament move around a little.
Thanks I appreciate it. Yeah definitely so much more that went into it, it's difficult cramming hundreds of hours of work into a 10-15 minute video! 😁
Omg so eager to see more about this!!
Dude, watching this I just want to give you moneyz to buy some treats to compensate for all that pain endured by failures. Moreover, you persisted and finally made it work. That’s an outstanding and a rare trait.
(I definitely can relate to that. I’ve been consistently and randomly bricking atmega chips in a custom RC receiver and transmitter design for almost a year - hand solder the board all night then brick it next evening. Rinse and repeat. Until I read on some obscure forum a tiny note that the problem might be caused by a defective USB port on my laptop. Fix - connect the programmer via a USB hub. Works like charm since then. I made multiple RC cars running with it. And created more customized boards after.)
This channel has the quality of a million subscribers channel! I'm shocked there are only 32k subscribers.
Thanks I appreciate that, maybe one day! :D
Whenever I want to turn green with jealousy, I tune into your channel. Amazing work.
Your cat does exactly what my cats do. Always when i work on somethin technical they ere curious and try to help me.
What a great project for 3d printing!
Of course the IKEA desk is hollow. :-)
Brilliant project. Thank you for documenting it.
Thanks, glad you liked it!
Damn! Anyone can build something, it takes real skill to troubleshoot something complex like this!
More Ampsss!! What an absolute unit
awesome stuff, I liked when the arm went whooooow.
You've got my respect for sure man. This is a great project! I program a 6 axis welding robot at my job, so I can't wait to see what you've got planned for it. Good luck and good job with presentation too!
Great video as always
Very well presented. Good job!
Thank you!
You should see if it's possible to create a continuously variable harmonic drive.
Bonus points if you can use compliant mechanisms for all of it.
perfect robotics
I just watched as an art vídeo, it's so complex and high level job ..
Dude you are absolutely brilliant
Thank you! But no I just have wayy too much free time 😀
nice work! i'm looking at doing something similar with two of these assemblies forming an 'actuated universal joint'.
Man, i Love your videos and all the expertise you out into your projects.
Keep it up. "I'm loving it" - McDonald's
I appreciate that!
We have the same desk. Very epic. Good desk
Nice! And yeah it is great for the price, probably should have expected it to be mostly hollow though 😁
Dude
Wow
This it incredible..
Very cool! I can't tell very well from the video, but hopefully you left enough space between the wall and where you mounted it so the arm can fold without hitting the wall. Also, I can relate with it looking smaller in CAD.
Yeah there's about a foot so it should be ok!
Awesome video! Keep it up!
This is amazing!
I think you could eliminate the need for the flexible gear by using two gears with the same tooth profile in place of where the bearings are on the elliptical drive part
Nice, you are so patient, incredible. You could consider to do an incremental approach, first use TMC5160 stepsticks to prove the concept
Yeah if I continue to have trouble getting these mosfets to behave I may do that. I was also considering just using the stepsticks for the 5160's on the motherboard by default since the price is about the same as integrating it directly, it just won't look as pretty :)
you're my hero!
Probably been mentioned before but double helical as opposed to single helical gears is better due to canceling out the thrust load the gears would otherwise produce, and when 3d printed don't have the downside of being any harder to produce. Second, for the sake of part count and simplicity I'd personally use two angular contact bearings or tapered roller bearings as opposed to the 3 bearing setup you've got. Slew bearing if you had the money.
For this particular application though there isn't much of a radial load. The main loads are axial and moment loads perpendicular to the axle (not sure if this is the best way to explain it) from the arm being cantilevered out. Something like a lazy susan would probably work just fine.
seems like you are reinventing a lot of stuff
Yessir, it's more fun that way!
Hey how is the arm video going? Really excited to how it turns out!!
Hey! Awesome project and I really look forward to see more of this.
But when I saw you putting screws in under the table I was worried they would rip out when you finally have your camera on it and out for max torque. Maybe put in a kind of backplate like on motherboards to distribute the load and prevent ripout. Should be an easy job to design for you and cheap to order online with the dxf drawing.
Good call, I'll definitely be making a back plate before putting any crazy loads on this arm. Should help with stability as well
Going through the same growing pains on my own TMC2130, 32u4 board. PCB design is hard. Using JLCPCB and assembly to save time.
If you mess up paste you can just put more flux on and drag solder the pins with a drag solder tip
Yep simply incredible as always. Are you doing this full time?
Nope! Still have a full time engineering job
Seperating the function each bearing must fulfill by only letting one take radial and one axial loads is a good way to keep complexity down while making sure you know how to dimension your bearings accurately, however, you can use 'overconstrained' bearing pairs and still determine which bearing takes which loads. Think of Tapered Roller Bearings for example who are often used to decrease the effective distance between Forces when used in an X configuration when having to bear high loads! Cheers.
You're right but you aren't accounting for cost. Angular contact bearings are expensive and you can't easily find cheap versions since they are way less common than radial and axial bearings. The bearings I found are very cheap from Amazon or eBay. Admittedly these bearings are terrible, but when the arm is primarily plastic and loads are way way under the bearings rated capacity, it does not matter.
Beautiful.
Thank you! Cheers!
@02:47 Maybe consider 3d printing a bearing race along the outermost edge of the plastic housing and fill it up with a bunch of bearings and a bearing spacer for additional support? 🤔
Fantastic
man i enjoy so much your content, im a lazy engineer that doesnt have his math down, make some courses to go full engineering on calculating n stuff i would be so down, also a course on pcb design my man! where have you been my whole life lmao, i feel like stefan is the testing god, but you are the design engineer god
Hahah thanks glad you enjoy it!
I do not know has anybody mention about this before but, the involute profile shown in gear design accelerator in Inventor is not precise in user interface.
You need to right click and export tooth profile manually. When you do that, gear turns out a cylinder with a sketch on it.
Sketch will include the splines for the single cut (in real life profile of the hob). You need to either extrude cut or sweep cut (depending if the gear helical or not) then you need to pattern this cut according to teeth number with a circular pattern.
After this you will get a proper gear geometry for CNC-CAM or 3D print purposes.
The geometry in design accelerator UI is rough that you may see interference between spurs.
FYI.
Very interesting and very annoying that it does that, I didn't know! thanks for letting me know!
Excellent video (as expected). Now I really feel the urge to continue my rover-with-an-arm-project. May I ask you where you have acquired your knowledge about mechanical, electrical and software engineering? Just from doing it as a hobby or does it come from your major? I'm asking because I have to choose a specific major and I'm aiming for something interdisciplinary.
I majored in computer engineering with a minor in mechanical. Computer is basically a mix between EE and embedded fw so between that and mechanical I got a very diverse education. Then I did lots of projects on my own which I have learned a lot from, mostly designing 3D printers or projects that used 3D printers of course!
@@3DprintedLife Thank you for the answer. I really appreciate it!
Congrats!!!!! You got sponsored
Thanks! :D
man....really interested in one day building one of these for time lapses with my mirrorless camera.
Excellent project! I am super cheering for your success 😁👍 I think you overthought the axial issue: deep groove ball bearings with big diameter like this one usually can take 50% of the nominal static pressure as axial force. Man… I bought a similar bearing: it is rated for 10k Newtons… this is 1000kg, so, it would handle 500kg of axial load man! The plastic won’t handle half of it 🤣
Well yes that is true, but the price for 2x deep groove bearings this large would almost definitely be more than my 1x radial and 2x thrust bearing setup. Plus i enjoy over engineering things like that, it's good practice as long as I know that I am over-doing it!
@@3DprintedLife when it comes to personal projects, if its worth engineering its worth over-engineering in my books!
Informative great job
I just found this channel and love your content. Been looking for more maker channels. Sub'd!
Thanks, and welcome!
How accurate is it? Do you think its possible to make a 6 axis cnc so you can machine some parts for your projects?
In axis 2, you could have a much longer motor for higher torque
Kudos
so coool!
Awesome work ! wouldn't resin print the harmonic driver reduce the backlash ? Seeing all the pain you went through designing your own stepper drivers, you could do the same for BLDCs. Price wouldn't be that much worse, a lot less noise, smoother rotation. By the way, have you seen skyentific's design ? lots of interesting ideas, especially the differential drive.
Have fun and keep going !
Oh, I've tried the diy bldc route for a long time. It's way way more difficult than integrating these stepper drivers haha. I think what I will do is use the normal stepper modules for the 5160 drives since those don't really save me any cost and it will make integration/repair easier, but keep the 2041s on the main board since those are super cheap and capable chips.
@@3DprintedLife I don't understand how BLDC are more difficult ? it's like exactly the same hardware (+ 1 pole) and almost the same software ( even if you do SVM or FOC). Anyhow, I understand your thinking, i'm just curious as if you're not refusing BLDC a bit too fast :)
What about using a hypocycloid gear they are also easy to print
have you considered using 3d printed baerings with airsoft bullets or steel balls?
Where did you learn all those skills?👏🏻
A combination of school, job, and working on personal projects!
Beautiful Project! I want to know which software you used to do the motion simulation (at 0:30). Thanks!
Autodesk inventory!
this is super cool, and I'll be following this closely and maybe become a patreon. but in all honestly, I dare questioning the choice of the harmonic drive: as cool as it is, in a robot arm a 102 reduction ration is the realm of high-speed brushless motors, not steppers. which obviously would require a totally different type of board to drive them... looking forward to the next instalment!
Yep agreed! Though the harmonic on its own is only about 30:1, the additional reduction comes from the belt drive going from the motor to that driveshaft and I plan on playing with the reduction there to achieve the best results.
@@3DprintedLife sounds great! looking forward to!
Where does one go for this kind of education? Like how much schooling does this man do or what kind of job does he have that enables him to build such things from scratch? Is this like MIT level education?
Nice!
Where did you gain all your experience with circuits and circuit board design? Want to create some of my own projects but don't know where to start
I learned electronics basics in school and then self-taught pcb design. KiCad is great free software and there's tons of tutorials on UA-cam!
You are really amazing 😍
Thank you so much 😀
Try Cone bearings, they do radial and axial loading in one
That’s true, but those along with angular contact bearings are extremely expensive at this scale which is why I went with a stack up of thrust and radial bearings
@@3DprintedLife True, I didnt actually look at the costs, the diameter bearing you need, would be expensive and way overkill, what about printing a polymer bearing surface from a suitable Igus filament? One roll would have you covered, and they have some for sub 80° C applications that do not require an enclosed printer.
"totally unrelated note" that tooth profile on the side is for the encoder
What pcb and schematic design software are you using? It’s looking like eagle but I’m not sure. Im using circuit maker 2 by alto I’m since I can’t afford designer.
Kicad, it's free, open-source, and awesome!
Skimmed the comments and didn't see mention; please consider using a larger diameter slewing bearing for the main base support. Even printed races with airsoft bbs or purchased bearing balls would have less deflection. Keep in mind that joint is taking overhung moment load so one side of the bearing would be in compression while the other would be in tension. The needle bearing you're using now would need more preload compression to keep from separating on the tension side but is still so small that it will cause excess defection in the base components.
Cheers,
Great points! I'm definitely going to do something like this for the next iteration, I'm just trying to figure out the best way to do it while keeping it cheap and smooth!
@@3DprintedLife I'm a huge fan of air soft bbs with printed races. Add a bit of race interface produces a surprisingly well performing printed component. Depending on your location a bottle of thousands of bbs can be purchased for less than 10$.
Keep in mind the overhung moment though. Bearing race orientation above/below would need compression preload while inner/outer would be more typical to slewing bearing design handling a range of dynamic loading; axial, radial, and moment. I'd be glad to share concept designs if you'd like.
Do you have a write up on the motherbaord? I'd pay for that.
Yea the size translation between CAD and real life sometimes is unexpected lol
would like to know the precision on it aswell.
How is backlash on your rotation base? I have a need for that. Thanks.
I haven't tested it but the gear meshing is very tight so it should be extremely low, there will likely be more positioning error from the plastic twisting and distorting.
Mom & Dad awwwww :D
The desk is not fully hollow actually, there's cardboard inside.... Great design in any case, effective and it looks beautiful!
Haha yeah I sort of figured, but surprisingly I didn't hit any so it must be pretty sparse 😁
@@3DprintedLife Yeah, it's in a triangle grid pattern with vertical slats so you usually won't hit it. I had the same when I was drilling holes to feed cables through it for my 3D printer.
What I have always done when I need stepper control for a project is just buy a cheap 3d printer board. It's an Arduino with 4-5 stepper outputs, plus a handful of random IO. Perfect for prototyping.
Yeah true that could work! My issue with that was there aren't many that can support 6 steppers, and originally I wanted to be able to drive motors with more current than the tmc5160 pololu style drivers could handle. But due to all the fun issues in the video I am revisiting my plan!
Why the use of so many Allen screws? That seems really annoying.
Very cool project though and I'm definitely invested
It does seem excessive, but it's actually helpful in 3d printed designs to keep assemblies rigid since plastic is of course way less rigid than metal
Where did you get the documentation to write firmware for the trinamics? Been looking all day and couldn't find some to hook them up with a esp32 via SPI
It's all in the datasheet, just search "tmc5160 datasheet" (or whichever trinamic driver you have) and it should come up. They have a very detailed section regarding all the SPI commands as well as a setup workflow a bit later down with recommended settings to get things up and running.
@@3DprintedLife oh, okay, thanks! I didn't go all the way down the datasheet as I am quite new to both the tmcs and esp32, and got discouraged. Should have looked all the way. I had also previously found a library for the tmcs on the esp32 but it reported some compatibility problems, and discouraged me even further to dive deeper. Thank you for the advice!