Make with Ada: ARM Cortex-M CNC controller
blog.adacore.comAuthor should check out http://camotics.org/ It's an Opens-Source 3-axis CNC simulator that runs on Mac, Windows and Linux. It not only shows you a visualization of the tool paths but also the resulting cut workpiece.
Disclaimer, I'm the author of CAMotics.
Author is aware that there are many Gcode simulators.The specificity of this one, thanks to Ada portability, is that the simulator code is 100% the same as the code running in the embedded controller.
No other Open-Source GCode "simulators", as far as I know, simulate the actual cut.
You just made me realize I never looked up the word 'Disclaimer' before. Now that I have, I realize that I've been using it wrong.
Some questions come to mind:
- Can it not accelerate smoothly? The graph on the page seems to imply that.
- Also on that graph, the relation between distance and speed on accelerated parts appears to be linear, but, if this is supposed to approximate constant acceleration, the relation between distance and speed squared would be linear (v^2=v0^2+2ad).
- Why did they need to develop that GUI, when there's existing software available? E.g. Pronterface. edit I see it's also a simulator so that makes sense.
On the other hand, I really like such proof-of-concept projects, small enough to understand but large enough that you can see the patterns. One can learn a lot from them, and I'll be sure to check out the code some day :)
On a related note, some shameless self-promotion: my APrinter[1] project is an open-source firmware for CNC devices, is highly portable running on many chips including STM32F4.
Thanks for your comment.
- The software architecture is inspired by GRBL. The split of motion blocks into segments is more manageable for the software, and if you choose a segments size small enough, the acceleration will be very smooth.
- I do not try to approximate constant acceleration.
- The GUI is a way for me to test the Gcode and stepper algorithms (the code is the same in the GUI and in the embedded controller).
Please also comment directly on the post.
Funny, I wanted to do the same stuff on the same board by converting my current C implementation to Ada, it ended up in a disaster, the compiler wouldn't run on my mac (so I used a linux simulator), the real-time feature of Ada were actually clocked on a fixed SYSTICK, and most of the Ada tools were randomly crashing.
the plugin I started writing at the time: https://github.com/nraynaud/ada-intellij
The code I wanted to convert: https://github.com/nraynaud/webgcode/tree/gh-pages/interpola...
where I threw the towel: https://github.com/nraynaud/bldc_ada
When was this? The tools working for me (either Linux or Windows). In fact we released a new version today: http://libre.adacore.com/download/configurations
And we are also working on a bare-metal driver library for ARM Cortex-M. Have a look: https://github.com/AdaCore/Ada_Drivers_Library
Hey, I know you :) I used your web Gcode simulator a lot: https://nraynaud.github.io/webgcode/
funny, I never really used it myself, but I got a few people telling me they use it :)
On the topic of safe languages on embedded systems, I'm very excited for the future of Rust on micros. Having just recently spent probably in excess of 100 hours on an embedded systems project at university, I'm starting to get a bit annoyed with using embedded C.
I'm sure good software engineering practices would eliminate a lot of my problems, but being the only computer scientist in a group of electrical engineer has made my life difficult. The entire project is just a massive mess of cascading makefiles and #defines, and is amazingly poorly documented.
A lot of the code was supplied by our lecturer, and has no documentation whatsoever, and he's gone and made this arcane cascading set of makefiles that goes about 4 levels deep.
Having just recently spent probably in excess of 100 hours on an embedded systems project at university, I'm starting to get a bit annoyed with using embedded C.
You many want to reconsider your career path, then. Because I've been working on embedded systems for a long time now and it's always been embedded C. And it's looking like it's going to be that way for a long time to come. If you're using a kernel like Linux or FreeRTOS, even moreso.
The entire project is just a massive mess of cascading makefiles and #defines, and is amazingly poorly documented.
Again, welcome to embedded systems engineering. Enjoy your stay.
Are CNCs still using stepper motors? If so, why? Hard drives moved on decades ago, becoming much more accurate and fast.