I’ve attempted to install a headless OS on here and then install sandify, but it might not support the necessary versions of node.js. I tried experimental versions, and I made headway, but didn’t get it running last night bashing my head into that wall.
I had trouble with installing npm and nodejs on a pi. Once you have it installed, running sandify is trivial. All the code runs client side. The server is just about building the static site.
You could also use npm to build the site from a full desktop computer and just install the static site on the pi. I am not sure the exact steps. But I am sure that would work.
These are great leads I will investigate! Thank you very much!
@jeffeb3 Good news, built and placed the sandify on my pi, and can access it via my network. Is there a way to export these gcode files to the MCU board? I’ve got them linked with a home made male to male micro usb cable. I think the pi sees it:
pi@raspberrypi:~ $ ls -l /dev/serial/by-id/
total 0
lrwxrwxrwx 1 root root 13 May 14 17:53 usb-STMicroelectronics_MARLIN_STM32G0B1RE _CDC_in_FS_Mode_204930764741-if00 -> ../../ttyACM0
Would I need to install octopi on here instead of the other OS I have on there now?
trying this
You can’t (easily) send gcode to be saved over the serial port. You need to store it somewhere and then play it back. Octoprint is probably the most robust solution, but it doesn’t run that fast on a pi zero. There still won’t be an easy way to copy the files you create directly to the octoprint server. You need to download them in whatever browser you are using (it only copies it locally, nothing is on the server) and then upload it to octoprint.
Other potential options are cncjs, sandypi, or (depending on the co troller you use) webui on an esp32.
I’ve got a BTT SRK Mini E3 V3, it has an SD Slot for flashing, I imagine it can store things too. Another idea I’ve tossed around is grabbing an old laptop and just removing the screen and it should be better until RPi stock stabilizes.
The esp wifi expansion card can add webui to that controller. It is possible to send gcode over serial, but it isn’t quick (you have to break it up and send it over gcode) and you also need a way to control when it gets played back. So you need a few more pieces.
I have some pi alternatives and I haven’t had any time to try them. I am sure several of them will work. I just am not sure how much unknown software I need to accept into the core image to run some of these.
I’ve also got some Arduinos lying around if that’s a better option, or is the Pi my limiting factor?
You’re in uncharted territory. However you can solve it, go for it.
What you’re missing is the gcode sender piece.
The skr can manage to play files off of the sd card, but it still needs something to control it (to tell it which files to play), and the only way to upload it is through the gcode (slow) interface. You would need to write something to stream the files over the usb.
Octoprint or cncjs from the pi can control it too. But the pi zero may limit you. I bet it would work, but it will be a bit groggy. Octoprint is very hard to break.
Adding an esp wifi device to the skr will give you a faster way to upload to the skr sd card, and control the playing of the files. An arduino isn’t going to work. This adds more hardware (but they are <$10) and more network configuration.
Are you trying to generate fresh unique patterns on the fly and run them automatically? That is challenging but theoretically possible.
Typically the patterns are generated offline by human interaction with sandify website via browser (no locally hosted version necessary) and the patterns are streamed to the device from PC or Pi, or uploaded to SD and external connection initiates them.
I think it is also possible to auto-start gcode using Marlin’s built-in feature, so you just apply power and it runs with no OctoPi and no USB PC or anything. (Could come in handy when there is unintended WiFi jamming, but you would have to be running Marlin.)
Maybe you are already aware and you are pushing the limits with live, fresh generation but just checking you’re not making extra work for yourself to run a sequence of pre-generated patterns.
So update, I’ve been pointed towards SandyPi which seems like a good option. Purely to get motion going I’ve opted to use an old laptop and once I have movement I’ll begin building a table. There’s mention of running SandyPi on a pizero and I’d like that if possible. I have no expectations, and I will focus on features after I have a functional motion system and a nice table I’m proud to display.
Update update:
I’m gonna try multiple things. Hopefully with the least resources I can manage.
But I’ve got this BTT SKR Mini E3 V3 which has no precompiled firmware, so I’m wondering what modifications did V1 do to their “Optimized” Firmware on other boards that I can do myself.