LinuxCNC compatabillity

[ShadeTechnik]

Not sure if this is even a reasonable request since I don't use grbl or grblHAL in anything and I don't know much about programming either. I do use LinuxCNC alot, and I use Teensy 4.x's alot with custom PCB designs. I know there's sort of an overlap in what LinuxCNC and grbl do, but this seems to have most of the blocks in place to slip some kind of LinuxCNC HAL interface into, I'm curious what it would take to get something like that going. People have worked on Teensy's with LinuxCNC but it always seems to die since it's usually just 1 guy doing a random project.

I'm talking about something similar to what mesa does with their HM2 FPGA software. There are cards with ethernet and SPI interfaces with the stepgens and encoder modules, it'd be great to see something like that on a Teensy since it's pretty accessable, though I'm not sure if grblHAL is a means to get there. It's been discussed here:
https://www.forum.linuxcnc.org/18-computer/43212-teensy-based-cnc-board
https://forum.linuxcnc.org/24-hal-components/39813-teensy-4-1-linuxcnc?start=20

[terjeio]

You are suggesting using a Teensy instead of a Mesa card? If so then the driver may contain some usable code.

[ShadeTechnik]

Yeah more or less, it's been done before. Like I said those projects just didn't seem to hold much steam. I'm just curious if there's any interest from the developers of grblHAL here to support such a thing.

[karoria]

There are several more gcode commands that are supported by LinuxCNC like cutter radius compensation (G41/G42) which is super useful for some applications. grbl doesn't support it yet. So, I will be interested if this functionality is added to grblHAL by any means.

[ShadeTechnik]

Personally I am more interested in adding support for the LinuxCNC API to the grblHAL core so that cards using that can be made available. And perhaps the API specs can be used for dual core processors where a core is dedicated to step generation?

I can't think of any MCU boards that LinuxCNC supports that are beyond anything of the MCU's you already support in grblHAL. MCU's are just starting to get attention rather than just being tinkered with in LinuxCNC. Pretty hard to justify an MCU over an FPGA for anything but a smaller machine, but it's starting to get there. I think people running LinuxCNC would be alot more likely to use the stuff you already support if it were available to LinuxCNC rather than the other way around

[terjeio]

Pretty hard to justify an MCU over an FPGA for anything but a smaller machine, but it's starting to get there.

I do not get this - bigger machines need more processing power? At what size does this become evident?

I think people running LinuxCNC would be alot more likely to use the stuff you already support if it were available to LinuxCNC rather than the other way around

I have spent a lot of time developing grblHAL from Grbl, and also on ioSender, leaving my CNC machines largely unused. Adding yet another programming project that will keep my machines idle for way longer is out of the question.
However, I may consider it if I am paid as a professional programmer - at my current rate it will not be cheap.

[ShadeTechnik]

Pretty hard to justify an MCU over an FPGA for anything but a smaller machine, but it's starting to get there.

I do not get this - bigger machines need more processing power? At what size does this become evident?

I didn't say "processing power" it's more about cost. A $200 FPGA card (+PC cost since LinuxCNC does not put the interpreter on the controller like grbl) is alot easier to justify in a $5000+ mill than it is in a $600 3D printer. I can't imagine a scenario where an MCU is better than an FPGA since cores run in parallel but I can think of plenty of scenarios where an FPGA is overkill. Personally I have a lot of stuff where a Mesa card is completely unnecessary and I just like to do custom stuff.

I have spent a lot of time developing grblHAL from Grbl, and also on ioSender, leaving my CNC machines largely unused. Adding yet another programming project that will keep my machines idle for way longer is out of the question. However, I may consider it if I am paid as a professional programmer - at my current rate it will not be cheap.

Well like I said I didn't realize you were the only programmer on this. It's a good open source project, I'm sure a lot of people are grateful you did this. Like I said I'm personally only interested in the Teensy 4 stuff so if it were something I was looking to pay I'd skip grblHAL and get someone to port this https://github.com/SebKuzminsky/linuxcnc-hal-teensy to a Teensy 4, maybe I'll go that road if necessary. This just seemed like it would be a really cool sort of merger.

[kfmut]

I do not get this - bigger machines need more processing power

It's not about processing power per se, it's about available inputs/outputs, encoders, analog outputs for servo drives and spindle. Modular and highly configurable solutions like Mesa cards(clones included) is clear win here and all those dev boards which are basically done after 4 axes and bunch of inputs are definetly not. LinuxCNC is mainly used to retrofit older commercial equipment and some entry-level new commercial equipment.