Remora - ethernet NVEM / EC300 / EC500 cnc board

kwarup wrote:

Cloned  github.com/scottalford75/Remora-Novusun/tree/feature/DMAstepgen with.
git clone github.com/scottalford75/Remora-Novusun.git --branch feature/DMAstepper
ping 10.10.10.10 and is ok.
Uploaded Remora-NVEM.bin to my novusum-nvem ..
sudo halcompile --install remora-nv.c
sudo halcompile --install nvmpg.c
linuxcnc Remora-Novusun/LinuxCNC/ConfigSamples/remora-nvem-basic/remora-nvem.ini
and the result is:
Unexpected realtime delay on task 0 with period 1000000
This Message will only display once per session.
Run the Latency Test and resolve before continuing.
joint 1 following error
emc/task/taskintf.cc 975: Error on joint 1, command number 109

joint 2 following erroremc/task/taskintf.cc 975:
Error on joint 2, command number 169

Thinking on this a little more, I've seen a similar behaviour when the config file is corrupted and does not get passed correctly. Resulting in no step generators being loaded. Please try a full chip erase and re-flash so that the default 3 axis config is used or alternatively re-upload a config file to the board. Having a serial monitor attached would help to see what is going on, but this would be my first approach.

I've just downloaded the firmware in the main branch and  done a upload again to my novusum-NVEM without recompiling the driver remora-nv and everything is working..  

Bari wrote:

This might not be obvious to everyone but all this beta testing of Remora over Ethernet should be done on the simplest of networks without any shared traffic. One host LCNC computer with Ethernet to one switch port and the NVEM on the same switch on another port and nothing else for now until this project is solid.


I'm using dual ethernet card, one for Internet another to Remora .. 
 

I've spent the last two weeks developing an alternative to the DDS accumulator approach to fill a DMA stepgen buffer with the hope of being able to achieve higher step rates. Unfortunately with no success. The alternative approach is still computationally heavy.

Looks like the only way to go for crazy high step rates is to use hardware PWM in conjunction with DMA, which is the approach that MX_master shared.

Time to throw the towel in on the DMA stuff for now and focus on stability and additional features as well as work on the STM32H743.

Would 200 - 250khz be a reasonable step rate? I discoverd that clearing of the buffers (memset) was consuming more time than the step calculations for the 6 axes. So if we can use DMA to clear the buffers the step calculations take less than half a servo period, so still time to do Ethernet communications etc.

I would think yes, most good quality stepper drives have 200kHz as the maximum rate.
At some point/speed the cabling becomes an issue, but since the drives are on the same board, not an issue in this use case.