Ethercat random jitter fix

Just to complete the information. The xml file contains drives that aren't actually present, so you can manage machines with different configurations. The real machine only has the x-y-z axes and an analog spindle.
So address 0 (I/O card), address 1 x-axis, address 2 y-axis, address 3 z-axis. The subsequent addresses aren't actually present, but this is done so as not to change much in the .hal file.
But I repeat, on Debian 10 the same xml files are used, the function syncToRefClock="true"

sync0Shift should not be 0

sync0Shift needs to be at least a third of your cycle

Set to 500000 not 0

Also please post files as attachments, or link to paste bin or Google drive, don't paste the entire text in the forum.

Also run ethercat master command, share the output here, preferably in a text file.

I would suggest the first device to be DC enabled, which it is, and the xml devices should match what is actually connected to the machine.

Have different versions of the xml and Hal if you have different machines with different configurations.

To rule out issues with anything outside the programs, did you try to swap SSD with the debian 10 you know is working on other hardware?

I would suggest taking an analytical approach to this.
As I understand it, 4 registers are important:
607A target-position
60FF target-velocity
and
6064 actual-position
606C actual-velocity
I simply enter a value in 607A and 60FF, and the axis moves to that position. LinuxCNC evaluates 6064 and 606C. If the target and actual values match, LinuxCNC is satisfied.
That’s my understanding.
Now the question arises: Does the axis vibrate when stationary or while moving?
If it vibrates when stationary, which shouldn’t happen, it’s constantly receiving new positions and oscillating.
Control loop?
Do the axes oscillate while moving?
Unstable, fluctuating setpoints?

At this point, I would use Wireshark to compare the traffic between the two systems.
I think the findings will be very revealing.
 

The problem only occurs during movement and is a noise transmitted to the mechanics, the kind you usually hear about on the LinuxCNC forum.
It seems to have solved the problem. First, I restored the original lcec_conf, Lcec_coffigen, and Lcec.so files. I changed the parameters refClockSyncCycles="-1" and syncToRefClock="false" in the XML file.
For now, I've left the sync0Shift = "0" parameter.
I'll test the machine and update you on how it works.