Showroom Low-cost 6dof build

Hi everyone,

Time to start sharing something I have been thinking about and working on in private for quite a while now...

The goal is to reduce the cost of a 6-dof build by eliminating all unnecessary parts, especially moving parts. Think of it like the scene in the Martian where they start ripping all the parts off of the MAV. Carbon tubes? Don't need them. Linear bushings? Throw them out. There are only three rotating parts, the motor, the ball nut, and a big honking bearing. And there are only two linear motion parts, the ball screw and an anti-rotation slider. Check out the video.

Of course, just because I'm simplifying the mechanical system doesn't mean giving up on performance. I'm using Turnigy 6374 149kv motors rated for somewhere north of 500W. The drive in the video is a VESC running in open loop mode, my plan is to close the loop with an AS5047P magnetic encoder and either use 6 VESC 6's or the Odrive. Still lots of work to figure out there on the controls side though. The belt drive using GT2 belts and 3D printed pulleys is currently a 2:3 ratio on a 1605 ballscrew but end-game plan is closer to 1:1 on a 1610 screw. There's a lot of work to do yet...

Nice , not trying to be negative, but have you thought about the effects of lateral/side loads on the screw/nut when it is actually driving a sim rig? The screw may turn into a bit of a banana after a while - causing vibration etc, etc. I guess further testing will tell. Good luck.

Also, if you don't mind, what does the torque curve look like on one of those motors?

There shouldn't be any side loads since it is a two force member--pinned with a ball joint at both ends. In terms of column loading back of the envelope calcs put it north of 10 kN (2000 lb).

I don't really have a torque-speed curve for these motors but Oskar Wiegl put together a spec comparison for various sizes of these motors. I think the actual current ratings are inflated by a factor of two or more due to not having a giant propeller providing cooling but it's a start. I'll post a link as soon as I get enough posts to be trusted

Here's the link https://docs.google.com/spreadsheets/d/12vzz7XVEK6YNIOqH0jAz51F5VUpc-lJEs3mmkWP1H4Y/edit#gid=0

Re Side loads: Just checking that you have considered it, as I'm obviously not visualizing the final setup . Re motor: - Just had a concern on those type of motors start-up torque - once again if you have this sorted ignore me .

Hey Scott, happy to see you interested in this kind of motors, I hope you find an easy way to drive them, would be great.

I added an encoder to track the position at zero and low speed. I'm using the AS5048A over a SPI interface to the VESC. After properly shielding the cable, this works very well. It eliminates the hunting at startup or reversing that you can see in the videos. Running in duty cycle control (basic velocity control) it ran very smooth at low and high speeds. I wasn't able to get the full velocity loop or position loop tuned very well so I'll have to spend some time working with it.

I'm using the VESC 4.12 firmware right now, but VESC 6 has been in the works for almost two years now with significant performance and usability enhancements. Rumor has it should be releasing in the next week or two with backwards compatibility for the old hardware so I'm looking forward to trying that out as soon as I can get my hands on it.

I think I'll pick up an Odrive the next time Oskar puts together another run as well. It would be interesting to have both options to compare.

keep posting! looks greate...

i saw you are using 3d printed pulleys, i think you could get cheap aluminum pulleys in different sizes from aliexpress at cheap, just in case the 3d printed one gets weak

best regards and i am following your posts

fer

I probably could but I like being able to easily test with different ratios and sizes, and not have to wait 6 weeks for shipping.

Also, they are both custom. The motor pulley has a socket for the encoder magnet, and the ballscrew pulley is shaped to fit nicely over the ball nut. I used Droftart's OpenSCAD parametric pulley thing to model both of them (https://www.thingiverse.com/thing:16627)

Wow, so many clever people, I can't wait to see what happens next!