Hi JBear78, thanks for your comment. This set up brings room for more improvements, such us collision detection, auto teaching… Since the encoders are absolute, home position is also improved and more accurate.
Excellent. I look forward to publishing your complete project. Can you already give a prediction? I have everything stopped here, just waiting for this.
Thanks Valmor, no predictions yet but I am wondering about release first the CAD, mechanical and electrical component list, without controller details. Then people can start ordering the small components, printing the parts and assembly the robot without the controller, the controller PCB will take more time..
This is great work. I want to retrofit this for my Nema 23 models and do the same thing as you. Is it possible for you to share the two files please? I'm just being a bit lazy is all.
Hi Lukxy, thank you! Main knowledge about CAD, mechanical components, electrical etc from university studies, the rest reading a lot on the internet xD
Hi Fubatchi, when the magnet is not present you get that noise. If you check the last video you will see that the noise is gone, since the three motors has a magnet present.
Hi @bschwand, thanks for your comment. I used stepper motors for two reasons, the first is the price. The motor itself is cheaper and also the electronics to control it. The second is the size, stepper motors with planetary gearboxes are quite compact. To find similar size within brushless motors including gearbox is not easy, they tend to be wider in diameter, which does not fit too much in the current design.
@@ZeroBacklashRobotics I understand AC servomotors are more expensive, but I was thinking of bldc motors like used in RC vehicles, those are cheap. There is also an open source driver to drive them like steppers, even microstepping ( because a bldc is essentially a stepper motor). I imagine standard planetary reducers like used in steppers could easily be adapted too.
Hi, even those bldc motors are pretty big. If you can find a bldc motor reference as small as Nema11, with precision planetary gearbox, at a decent price, would be an option. But I think this does not exist. The idea of the project is to keep the cost at the minimum level, while functionality is not lost.
@The101Superman I choose them because they are accurate 14bit (~0,05deg), fast running over SPI interface, and cheap. You can find them on Aliexpress under 10€. You cannot find in the market encoders under that price with such accuracy. Remember one of my main targets of this project, is to keep the cost as low as possible, while keeping the same functionality of a real industrial robot in scale format. ;)
Hi Prajesh, it need to be tracked counting the number of turns of the stepper motor, based on that info and the gear reduction you can calculate the current output angle of the joint.
@@ZeroBacklashRobotics would it not be advantageous to put these encoders on the joint output side in order to count for all the backlash and slack in the mechanism?
Hi @Prajesh, thats indeed a very good question. And I asked myself why not do it in that way, because it gives nice features, such backlash control and also better homing position since your absolute position is unique. But I decided not to do it in that way. For four main reasons. First you loose your gear resolution, second the design of the wrist is based on a differential drive, which means its difficult to get the real axis 5 & 6 since they are dependent from each other, third one the encoders should be placed near the wrist which increase the volume giving a weird looking, and the last one, J4,5&6 are pure mechanics J4 & J6 can rotate endless, adding your encoders in the wrist will limit that rotation, since you will have to do wire management. Hope this answer your question. Thanks for your good comment! ;)
I work with Yaskawa robot and the encoder is on the servomotor too =) Better accuracy with the gear ratio and exactly for the same reason of endless rotation on the T axis (the axis number 6 on Yaskawa robot). Backlash is reduced with some features. - Epicycloidal gear box used zero-backlash gear to eliminate the problem. - Harmonic drive is naturally zero-backlash reducer - And for basic gear association they used spring to constantly maintain position. Also helicoidal gear is much better in this situation.
Some filters can be applied for false readings, but what I experienced is not required, since is absolute encoder running over SPI as long as your communication is not interrupted you dont miss values.