This is not a question, this is me bragging I did it! A working proof of concept! :-D Below is a video of it running 3 steppers with acceleration profiles and independent speed. [www.youtube.com] A little while ago I realized an I2S digital sound output looks a lot like a SPI output, and SPI outputs can drive shift registers. back then I was focused on the ESP8266, so I build a small 32 bit shifter board based on 4x 74H595 and wrote a realtime bitbanger demonstrating it as a GPIO expansion (see [forums.reprap.org]). Then, a couple of days ago I found a few post of people using I2S DACs to improve sound quality on single-board-computers, more specifically Raspberry Pi and Orange Pi. Then it hit me: I2S is a realtime digital output present on most SBCs, and we may use may shifter board to control a printer with it! I just happened to have an idle Orange Pi One at hand and a few days to spare, so decided to try it out. Code is here: [github.com] Shifter PCB was previously released here: [github.com] What was done: - Replaced /boot/bin/orangepione.bin with the included fex file, so as to enable and configure I2S output. - A C++ program renders motion for 4 steppers (only 3 connected on the video). - The C++ program outputs 32 bit words, a bitmask for step/direction of several steppers. - alsa player (aplay) feeds the 32bit words directly to the "sound card" (not really a card, part of the SOC). - The "soundcard" outputs raw I2S, and feeds a 32bit shift register. Sample rate is 192k words/s. - Just 5 wires from OrangePi to the shfter: GND, 3.3V, I2S-Data, I2S-BitClock, I2S-WordClock. - Shifter outputs drive the stepper drivers. Bits 0-3 are STEP, bits 16-20 are DIR. - I let out a tear of happiness. :'-D What could we do with this: - Discard the arduino-like controllers used on pretty much every single motion control board out there. - Remove distinction of software and firmware, as they would run on the same processor. (Eases development). - Make printer software mostly hardware independent, as long as there is I2S and alsa on the backend. - Modify reprap-firmware to be more OS-friendly, and run as a backend on the *Pi. - Have a quad-core computer running accurate motion control for the crazyest of mechanics, at high sample rates. - Theoretical limit of 46 steppers (2 channels, 24 bit per channel, using 2 bits for a secondary shifter for DIR bits). - Have a vision feedback system for calibration? OpenCV? - Have a full-blown linux system control your printer: - python, nodejs, octoprint, and whatnot providing web/touch interfaces. - Fast uploads, as there is no serial link anywhere. - A really nice interface to the printer: OpenGL and HDMI displays. - Real easy updates, apt-get style. Required for a working printer, but not tested: - I2S input via alsa-record could allow parallel-in serial-out shift registers to read 32 bit data. - Temperature reading via RC charge time? - Endstops, and a few buttons. https://reprap.org/forum/read.php?2,685993,685993#msg-685993 Mon, 11 May 2026 16:42:21 -0400 Phorum 5.2.23 https://reprap.org/forum/read.php?2,685993,860135#msg-860135 https://reprap.org/forum/read.php?2,685993,860135#msg-860135
What happened to this project, any progress ?

Seems to me that the way to go with this is to bring up LinuCNC/MachineKit on a cheap arm based board like this project did [github.com] , then write a hal driver using the I2S out and input of the board for linuxcnc.

Then you get a full, extremely flexible CNC controller. (I admit I am biased to LinuxCNC which I use with my CNC)]]> bschwand Developers Mon, 30 Sep 2019 15:46:19 -0400 https://reprap.org/forum/read.php?2,685993,687022#msg-687022 https://reprap.org/forum/read.php?2,685993,687022#msg-687022 But I am looking forward to it!
Excellent job, @lhartmann!]]> rklauco Developers Fri, 15 Jul 2016 04:55:31 -0400 https://reprap.org/forum/read.php?2,685993,686970#msg-686970 https://reprap.org/forum/read.php?2,685993,686970#msg-686970
For now I just tested wiring I2S data output pin back to I2S data input. I also tried pulling the wire and it stopped working, so this is not an internal loopback. Since I2S-IN and I2S-OUT share the same clocks it will probably work fine with a 74HC165 shifter. I'll try making a pcb for these soon.

For the test I used 2 processes: reader and writer. See [github.com]
- The writer.cpp just writes a tag (0x11223344) then 127 words (32 bit) of a continuous counter.
- The reader.cpp looks for the tag, then checks if the following 127 words are a correct counting.

Issues:
- I could not yet get these working on the same process (loopback.cpp), loosing data. A thread or proper use asynchronous IO with select(...) should probably do the trick.]]> lhartmann Developers Thu, 14 Jul 2016 23:35:38 -0400 https://reprap.org/forum/read.php?2,685993,686368#msg-686368 https://reprap.org/forum/read.php?2,685993,686368#msg-686368
Another option would be to use 555 analog timers in astable configuration, but with the NTC as one of the resistors. If I2S inputs work as I hope they will then I may be able to find out the NTC resistance by reading the timer period.

I may have some other I2C or SPI ADCs lying around too... Will give it a try once I get some spare time again. I'm pretty sure those are expensive models, though...]]> lhartmann Developers Wed, 13 Jul 2016 07:21:07 -0400 https://reprap.org/forum/read.php?2,685993,686296#msg-686296 https://reprap.org/forum/read.php?2,685993,686296#msg-686296 ADS1115]]> rklauco Developers Wed, 13 Jul 2016 01:13:47 -0400 https://reprap.org/forum/read.php?2,685993,686294#msg-686294 https://reprap.org/forum/read.php?2,685993,686294#msg-686294 4channel 12bit I2C adc]]> rklauco Developers Wed, 13 Jul 2016 01:00:46 -0400 https://reprap.org/forum/read.php?2,685993,686293#msg-686293 https://reprap.org/forum/read.php?2,685993,686293#msg-686293 But I see your point.
I have to google a bit, but I remember having i2c chip with 8ADC and 1 DAC, but only with 8bit resolution. So I have to search some more. That chip was, however, under $2.]]> rklauco Developers Wed, 13 Jul 2016 00:57:26 -0400 https://reprap.org/forum/read.php?2,685993,686290#msg-686290 https://reprap.org/forum/read.php?2,685993,686290#msg-686290
I am already regretting getting the one. It is nice, but I am missing the 3 USB ports and the wifi of the Orange Pi PC+... In order to use octoprint+wifi+webcam I already need an USB hub. :-/

Next step: Finding out how to run RepRapFirmware on Linux.

P.S.: Updated the video, a commented version. I did not realize that mobile platforms and chromecast do not show video comments...]]> lhartmann Developers Tue, 12 Jul 2016 23:53:29 -0400 https://reprap.org/forum/read.php?2,685993,686005#msg-686005 https://reprap.org/forum/read.php?2,685993,686005#msg-686005 Btw, for analog in you'd be better with i2c, wouldn't you?]]> rklauco Developers Tue, 12 Jul 2016 02:34:30 -0400 https://reprap.org/forum/read.php?2,685993,685993#msg-685993 https://reprap.org/forum/read.php?2,685993,685993#msg-685993
[www.youtube.com]

A little while ago I realized an I2S digital sound output looks a lot like a SPI output, and SPI outputs can drive shift registers. back then I was focused on the ESP8266, so I build a small 32 bit shifter board based on 4x 74H595 and wrote a realtime bitbanger demonstrating it as a GPIO expansion (see [forums.reprap.org]). Then, a couple of days ago I found a few post of people using I2S DACs to improve sound quality on single-board-computers, more specifically Raspberry Pi and Orange Pi. Then it hit me:

I2S is a realtime digital output present on most SBCs, and we may use may shifter board to control a printer with it!

I just happened to have an idle Orange Pi One at hand and a few days to spare, so decided to try it out.

Code is here: [github.com]
Shifter PCB was previously released here: [github.com]

What was done:
- Replaced /boot/bin/orangepione.bin with the included fex file, so as to enable and configure I2S output.
- A C++ program renders motion for 4 steppers (only 3 connected on the video).
- The C++ program outputs 32 bit words, a bitmask for step/direction of several steppers.
- alsa player (aplay) feeds the 32bit words directly to the "sound card" (not really a card, part of the SOC).
- The "soundcard" outputs raw I2S, and feeds a 32bit shift register. Sample rate is 192k words/s.
- Just 5 wires from OrangePi to the shfter: GND, 3.3V, I2S-Data, I2S-BitClock, I2S-WordClock.
- Shifter outputs drive the stepper drivers. Bits 0-3 are STEP, bits 16-20 are DIR.
- I let out a tear of happiness. :'-D

What could we do with this:
- Discard the arduino-like controllers used on pretty much every single motion control board out there.
- Remove distinction of software and firmware, as they would run on the same processor. (Eases development).
- Make printer software mostly hardware independent, as long as there is I2S and alsa on the backend.
- Modify reprap-firmware to be more OS-friendly, and run as a backend on the *Pi.
- Have a quad-core computer running accurate motion control for the crazyest of mechanics, at high sample rates.
- Theoretical limit of 46 steppers (2 channels, 24 bit per channel, using 2 bits for a secondary shifter for DIR bits).
- Have a vision feedback system for calibration? OpenCV?
- Have a full-blown linux system control your printer:
- python, nodejs, octoprint, and whatnot providing web/touch interfaces.
- Fast uploads, as there is no serial link anywhere.
- A really nice interface to the printer: OpenGL and HDMI displays.
- Real easy updates, apt-get style.

Required for a working printer, but not tested:
- I2S input via alsa-record could allow parallel-in serial-out shift registers to read 32 bit data.
- Temperature reading via RC charge time?
- Endstops, and a few buttons.]]> lhartmann Developers Tue, 12 Jul 2016 01:19:53 -0400