LinuxCNC + Orange Pi (allwincnc)

mwidlok wrote:

By the way is the orangePi H3 the best board that is usable for this project? Are any others working?
 

I am voting for OPi PC.
Look for an H3 with a long pinheader and at least 1 MB of RAM.

I also tested EtherCAT on it. It worked.
But it would be better to create a native EtherCAT IRQ-free driver for it.
It's in my long-long-term plan.

I think you meant 1GB of RAM but yeah!

Of course. Sorry.
Long pinheader and at least 1 GB of RAM.

OPi H5 also works!

Wzor, could You elaborate more about H5 support & installation process? What versions of system/LCNC You use?

I found board with H5, that is available now:
linux-sunxi.org/Libre_Computer_Board_ALL-H3-CC
I really like it because it is open hardware / open software design with 2GB of RAM and nice components. I'm thinking about getting one for myself and then prepare drivers/arisc code tailored for me. What do You think abut this board? Any experience?
Thanks,
Michael W.

Libre_Computer did not meet. You can try. Everything is the same as on H3.

www.loverpi.com/products/libre-computer-board-all-h3-cc

It is a bit confusing when you order since the product description is for a H3 but the H5+2GB option has the H5 for $45usd. The H3+1GB option is sold out at $35usd.

www.loverpi.com/products/libre-computer-...ariant=3133794385933

wiki.loverpi.com/sbc:libre-computer-all-h3-cc-h5

wiki.loverpi.com/sbc:libre-computer-all-h3-cc-h3

Amazon $40 www.amazon.com/Libre-Computer-Board-ALL-...ritium/dp/B07D4KTRWR )))

If you read though all the comments for the board on Amazon you'll see the problems with u-boot and Armbian. MX_Master had to battle with these and it's really frustrating when you just want to focus on writing your application and the bootloader and Linux distro is so poorly slapped together. Once we sort this out we should have some pretty stable results.

I'm back from vacation and starting back on my project to set up an Orange Pi for my milling machine. I have LinuxCNC installed, a WHB04B-6 pendant installed and all seems to be working. I also programmed up a ESP8266 microcontroller to act as a step counter. What I am using the step counter for is to see if the OPi/LinuxCNC is returning to zero steps after a run of g-code. The ESP8266 program uses 7 pins as follows:

1. 3 pins for ISRs (These are the "step" signals for x,y and z
2. 3 input pins for x,y,z direction
3. 1 pin for resetting the step counter.

When an interrupt is triggered the associated ISR checks the appropriate direction pin state and either adds or subtracts a step from the counter for that axis. The main loop writes the step counts to the serial stream back to my laptop at one second intervals. The reset pin can be used to force the counters to zero. It is a very short program and I don't see any issues with it. However, when I run it with the OPi setup I occasionally get step counts that are not zero when I return to the starting point. How I run this is as follows:

1. start LinuxCNC, home, and insure "machine" is at 0,0,0
2. reset the step counters on the ESP8266 to 0,0,0
3. run the g-code program, which to this point has been the "LinuxCNC" default program that loads with the AXIS interface.
4. MDI G0 X0 Y0 Z0
5. read the step counter.

Most times the step counters return to 0,0,0 as expected, but occasionally I will have an axis that is one step off. It is random which axis and which direction it is off. Probably happens in one of ten runs . I suspect this may be an anomaly of my step counter setup, but am not 100% sure of this. The ESP8266 should be plenty fast enough considering that I have the LinuxCNC set up for my machine parameters which is about 30 kHz maximum step rate. Being an amateur programmer I think I have a little investigation to do. Does anyone see anything obvious in my description above that raises concern?