Mike's SCARA Robot

[mcardoso]

Hi All,

I don't post here often but thought some might enjoy the project I've been working on since the beginning of COVID. I am an engineer, amateur machinist, and robotics enthusiast. I was fortunate enough to happen across a '99 Seiko D-Tran TT8800 SCARA robot at an industrial surplus supply store. It did not have a controller or have any guarantee to not be broken, but at $200 I couldn't say no.

Over the last year and a half, I've been researching about the robot, figuring out wiring, proprietary AC servo feedback, homing sensors, and building a custom controls system for the robot. Through work, I have access to some Allen Bradley PLC and servo drive equipment that I can mess around with for home projects.




At 130 lbs and a 35" reach, this robot is one of the largest SCARA robots I've seen available. I don't yet have plans for it other than messing around, but it could become a tool changer for my CNC mill, or maybe a 3D printer.

The most challenging part of this project has been developing a custom interface board to sit between the Yaskawa AC servo motors on the robot and my Allen Bradley servo drives. The motors use a Yaskawa proprietary protocol which they refer to as incremental "multiplexed" feedback forcing you to use Yaskawa drives. My custom PCB (based on clocked sequential logic in 74AC series CMOS chips) decodes their protocol and presents a standard incremental TTL with hall effect commutation feedback signal to the servo drives. I am very proud of this design as I was never taught digital logic and just figured it out as I went.




I am fairly finished with the mechanical and electrical side of this project and I was working hard at developing an interface to the robot using G-Code through a totally custom designed interface programmed on an Allen Bradley PLC. There are certainly some G-Code programs out there that could do what I want, but the interface to the PLC allows for some very powerful implementations of kinematics and real-time control rather than just pre-programmed motion.

The full project log can be found here: https://www.hobby-machinist.com/thre...a-robot.84001/

And here are a few videos of it running:
https://www.youtube.com/watch?v=o10d...MACEngineering
https://www.youtube.com/watch?v=-7sY...MACEngineering

I'm happy to answer any questions you have or share more photos and videos!

-Mike

[jwatte]

My custom PCB (based on clocked sequential logic in 74AC series CMOS chips)

This is at the same time totally awesome, and insanely scary :-)

Modern microcontrollers, with interrupt triggered pins and flexible I/O, are almost always able to "take a thing in" on one side, and "put another thing out" on the other side, and you can do the entire thing in software. And if you need something even faster than that, the crown kind-of just goes to FPGAs, and there are several hobbyist-friendly FPGA boards out there. I haven't seen an actual TTL logic decoder/board in a long time. Must have been fun to do it in just plain hardware!

[mcardoso]

Thanks jwatte,

I was problem solving down multiple avenues to get the motors on this robot working, and this ended up being the path of least resistance for me. The motors at full speed generate a quadrature pulse stream at nearly 1MHz. I would have needed a moderately fast microcontroller to sample the pulses, perform logic and memory operations, and finally set outputs. DMA would have likely been necessary, and even then the whole program would need to interrupt, perform operations, return, and re-arm the interrupt within a couple hundred clock cycles. Likely doable, but not easy. I've also never designed a PCB around a microcontroller (other than buying arduinos and adding headers to my board).

I've also never done anything with FPGAs. A couple people on another forum started helping me figure out CPLDs for this project, but I had the discrete logic boards complete by that point. I will agree 100% that FPGA/CPLD is the right way to go after this problem.

In hindsight though, for someone like me who had no formal background in digital logic design (my minor in EE was completely controls focused), designing a circuit like this was extremely insightful. Almost a bit like playing with legos. I used a web based simulator to design and test my logic and then just read datasheets to make sure my logic levels and clock speeds were appropriate. It was certainly more complicated than that, but online forums kept me pointed in the right direction and I never truly got hung up on anything. A very satisfying part of the project.

Mike

[jwatte]

Yeah, logic is fun!

FWIW, the Teensy 4.0 and 4.1 microcontrollers run at 600 MHz these days, and do DMA for you if coddled just right, and their development environment plugs into the Arduino IDE (or PlatformIO in VSCode, which I prefer,) but there's also something to be said for jamming chips on a board :-) Very nice!