Delta ASDA-E3 + IgH – occasional one-cycle position update skip

Hi everyone,

I'm developing my own CNC controller in C using the IgH EtherCAT Master on real-time Linux. I'm not using LinuxCNC itself, but I know there are several people here with experience using Delta ASDA drives with IgH.

My setup is:
- 4x Delta ASDA-E3
- CSP mode
- 1 ms cycle / SYNC0
- Distributed Clocks enabled
- Intel I210
- Real-time Linux

I have a strange intermittent problem.
During constant motion, occasionally one drive's Actual Position (0x6064) does not update for exactly one cycle. On the next cycle it advances approximately twice the normal amount.

For example:
Cycle N-1:  +39000
Cycle N:         0   <-- one drive does not update
Cycle N+1:  +78000   <-- approximately 2x

At the same moment, another Delta drive in the same EtherCAT cycle continues updating normally.
WKC remains normal, DC stays synchronized, there are no drive alarms, and 0x1C32:0B / 0x1C32:0C remain zero.
The same drives, motors and EtherCAT wiring work smoothly with a Bochu CNC controller.
I have already spent a lot of time checking Linux RT jitter, SYNC0 shift, DC phase and packet timing without finding the cause.

Has anyone using Delta ASDA with IgH ever seen this exact one-cycle hold followed by a double update?
Is there any Delta-specific EtherCAT, DC, synchronization or servo parameter that I may be missing?
Any ideas would be greatly appreciated.

Thanks,
Fahreddin

UPDATE / SOLVED

Hello everyone;
After many tests, i found any solutions for this issue

Three İmportant Parts
1. Using Xenomai/Cobalt clock as the master cyclic timing base
2. Waiting for the ethercat slave dc clocks to converge before starting motion.(<1 microseconds)
3. Explicity phase-locking EtherCAT SEND to a safe position relative to SYNC0

I hope this saves someone else a lot of debugging time.

Fahreddin