I got my AR3 up and running and have made a few upgrades, but I'd like to get them dialed in before sharing. So yes, still very interested...
I just started building my AR3 last week and have a ton of experience in Solidworks. I hope to have it running by the end of next week. I'm game to contribute where I can...
These are the things I want to look at after finishing my build. I'm doing few hours evey day now...
- test encoders on joints vs on the motor
- 48v behavior on j1-j5
- increase length of arms. j2 arm and j6 bearing arm. will likely create a new arm assembly I can switch back and forth with the original. nothing fancy. just open the cad model and increase the length + do some mass reduction cutouts.
- try out larger nema23 stepper motors on j2 and j3.
I know big changes are coming on software side so I'll wait and see for those.
I'm wanting a welding type robot so payload can be small.
I'm might be talking to myself here... but thinking about taking the j0 plate and printed base and replacing it with a mobile battery powered aluminum chasis with mechanum wheels. The control box would be incorporated into the chasis and the internal mains power supply would go away. j0 could also be direct drive. The base would likely need an optical floor tracking sensor but otherwise only has the minimal electronics to power the wheels. Overall cost could be less than current robot with a 7th axis.
You are not talking to yourself. Its great to hear everyone's ideas and see all the creativity that goes into these projects.
2. remove too much wiring, use CAN bus (or etc) to reduce wire into minimal (come with encoder)
I've done some research on this topic as I kind of find CAN usage elegant in this case and found surprising fact that whole industry is kind of putting some tax on CAN based solutions. I simply cannot grasp why can based drivers are so expensive in comparison with not-integrated solutions.
Using DC/servo motors with gearboxes and CAN is completely out of the price range. Speaking about My Actuator[6] and CubeMars[7] actuators here.
Using Nema motors with 3rd party CAN driver/controller is just a bit cheaper, but still expensive. CAN driver/controller itself is in $100-200 range[3]. The most cheap here is something called AnanasStepper which started on KickStarter and you can see price development on [1][2]. E.g. from $19 to $50-70. Still even for this price the cheapest I've been able to find.
Anyway, the question is if there is no any other open-source CAN/driver/controller project available, like in the past was Mechaduino project (just controller/driver but no CAN)[4]. Looks like academia is still searching for cheap controllers/drivers as they are still thesis done on the topic, like this one from Czech Republic's VUT[5].
Have I missed anything on my research on the topic?
Thanks,
Karel
[1]: https://www.kickstarter.com/projects/249504031/ananasstepper-30-a-servo-stepper-for-multi-axis-control?ref=discovery&term=ananas%20stepper
[2]: https://www.ananasstepper.com/store
[3]: http://www.stepping-motor.cn/uim242.html
[4]: https://tropical-labs.com/shop/mechaduino-0-2-pcb/
[5]: https://www.vut.cz/www_base/zav_prace_soubor_verejne.php?file_id=223888
[6]: https://www.myactuator.com/dowload
[7]: https://store.cubemars.com/
I hope planned AR4 design will also be open-sourced. If so, and if community should be involved in design, is it possible to kind of use less pricey 3d modeling options? I've had a look and it looks like so far preferred solid works is in the price range of $4k which may be basically a price range of arm itself and is basically 1/2-1/3 or so of chinese-based arms (ufactory[1 xarm]).
Anybody got experience with open-source or more cheap 3d modeling software? What about FreeCAD[2] if I may ask directly?
Thanks,
Karel
[2]: https://www.freecadweb.org/
I'm not sure where you're at with inexpensive CAD programs, but I really enjoy using OnShape. It has a free and paid version. The major difference between the two is privacy. If you opt for the free version, your designs can be seen by anyone with an OnShape account if they search for it. Paid subscription is costly, but then allows for your designs to be private.
I'll check that out. Im still using Solid works at the moment mostly because that's what most of the universities are using and have asked for. I do include the step files with the models so hopefully that helps some but I agree SW is a bit spendy.