I'm not going to have a rant about the current state of motion control electronics, suffice to say its 'messy': split between: - the really dreadful and dangerous - prime 1980's technology - one-size-fits-all board which don't fit all - 'black box' technology which needs a degree in software to figure out It's led me to an idea for reworking the current status quote - I'd appreciate peoples thoughts?... I'm looking to design/build a distributed architecture for motion control which will allow machine electronics to be put together from individual nodes, connected by a common multi-drop bus. For wiring a machine, you run a single common bus around the machine, then drop whatever nodes you need (e.g. motor control, switching, sensors) off the bus when you need them. A single master interface (either passive PC interface or embedded micro) then runs the system by talking to the nodes who do the actual work. This multi-drop bus then gets published as an open standard: for anyone to implement to. The advantages of such an architecture then become: - you can mix-and-match electronics to suit the machine: eg if you need a big power driver and four small ones, do it: they'll all talk the same language. - pay for what you need, don't pay for stuff you don't. - wiring becomes massively simple: a single bus feed around the machine, not hundreds of individual wires going everywhere. I won't go into future design detail here (gets complex very quickly): suffice to say I think it's do-able, and can be cost-competitive. Would anyone be interested in this as an end user? Would anyone be interested in supporting the development of this? Any thoughts - good or bad? Mat
No, RS485 most likely. I2c isn't resilient enough over long runs nor if you put it too close to motor drive wires. Can bus would be perfect but the micros needed are too high end. LIN bus would be electrically good but too slow
This sounds good in theory, though in practice it's rather like what LinuxCNC does in-software. Plug & play modularity in "real" CNC would be amazing, though. Would be interesting to see a proposed example architecture of say, a Super Mini Mill or a Robodrill or something.
I think you should take a look at recent micros. CAN bus is getting cheap. The Teensy 4 has CAN bus support with library support - $20 for a 600 mhz, 1MB RAM micro. Also, there are some devices (VFDs, servo motors iirc, ...) supporting CAN bus as well.
