Dear Builders and Makers, since a while I'm thinking about building a robotic arm for mostly educational purpose to get into this crazy world of stunning robotic motion control. I did plenty of reasearch during the last month about what kind of concept I'm looking for and I found a lot of different concepts for robotic arms. But none of them really was what I was looking for. What I have in my mind is a robotic arm which is made out of less parts, so it should be a minimalistic design but on the other hand it should have a precise movement to make use of it. None of the designs I found online really fitted to that idea so far. What I mean with that is for example the well known "EEzybot ARM" is very basic and uses tiny servo motors for motion control which are not really precise when it comes to movement. Then there are robotic arms like the "Thor" which are made out of hundreds of parts where every joint and gear is 3D printed and self made. (see 10 Best DIY / 3D Printed Robot Arms in 2020 | All3DP for further information) I want a robotic arm which uses precise but easy to control motors (e.g. NEMA17) but they dont have enough torque to run without a gear system. Still I want to keep it simple without making my own gears - therefore I ran into planetary geared NEMA17 (e.g. Nema 17 Stepper Motor Bipolar L=48mm w/ Gear Ratio 27:1 Planetary Gearbox). Those motors have an impressing high amount of torque with a very simple and small built-in gear box. What do you guys think about using those kind of motors for every joint on a 6-DOF robotic arm? I'm thinking about using like a larger one with a high gear ratio of like 50:1 on the bottom axis and then going with smaller ones and less torque for every further joint. When it comes to the joints I am not very sure of how to put them together with a good axial stiffness without any kind of play between them. I thougt about using double row angular contact ball bearings and then clamp the previous joint to the inner ring and the upcoming joint (axis) to the outer ring, while using the torque of the NEMA17 shaft which sticks through the bearing. I hope you guys get my idea without a sketch haha. (see Angular Ball Bearing 3804-2RS-TN > Easy Ordering!, 14,82 € to get an idea of what bearing I'm talking about... they can take axial and radial forces - so they should be good for this usecase) Do you guys have any good idea of how to create play-free joints between the robots axis using geared NEMA17 motors? I would really appreciate your thoughts and ideas about my idea. Thanks! Best regards Julius
Go single-row deep groove or roller, use them in a pair- they have to be tightened against each other to work. Look into bearing pairs, spindle design, that kind of thing, it should provide plenty of ideas for designs. In one theory you'd have, at the end of each segment, a fork (it's call a "fork joint" for a reason), each "tine" containing one bearing (the cup, in the case of a roller bearing). The "gap" would hold the end of the next segment being driven like a clevis eye, riding on lateral screws in a circular pattern that go through the bearing IDs, which are providing the clamping force between two circular flanges. On one side, you drive that swivel with the motor and planetary mounted on the "lower order" segment. Not sure if that made any sense, but it should once you do the bearing research. These types of arms get very heavy very fast- consider higher-force/lower-backdriving forms of motion lower down the order like screw-driven elbows. Small motors will be required nearer the end actuator due to weight and leverage, like pancake filament extruder motors.
Thanks for your fast response! Yea your totally right with the idea of the fork joint and using standard deep grove ball bearing for this case. What I had in my mind when talking about the angular contact ball bearings is to use this kind of bearing for the first joint. I thougt about a design with a motor standing up vertically so the motor shaft sticking up into the sky. Therefore the lowest bearing combination have to take a lot of axial forces compared to other joints with a fork joint design. As far as I know normal deep grove ball bearings are not made to take axial forces - even with a pair design. Your Totally right - for the last two axis which will be something like the robots "hands" ungeared / pancake steppers should work just fine.
Deep groove bearings can actually take axial loads in both directions of about ~40% of their static load rating, depending on the size. You could easily be looking at ~200kgf on a single average-size (2" bore) bearing for a lazy susan base. That said, what I was actually thinking of for the joints *were* angular contact bearings, as you said, because deep grooves don't always need preloading in pairs. A-Cs generally come in three standard angles- the steeper the angle the more force it can take but the less speed it can support without overheating. "Larger" (ish) A-C bearings like 7007s can take hundreds, maybe thousands of pounds of axial force- they're used double-paired in auto-toolchanging spindles with a pneumatic ram on top without any support. They're also not super cheap even at the lower grades. No reason to spend $70 on a high speed bearing when a $15 roller bearing from a lawn tractor would do the job just fine. Downside there is lubrication. Roller bearings don't come sealed and lubed for life, which could complicate matters. Shame, because big crossed-roller bearings are cool.
oh.. your totally right mate. I have mistaken something here. Appreciate your support very much! I'm probably going to make a concept design in Autodesk Inventor and share it here. Thanks!
