Advanced Search

I have had this gravel sound. Starts by sounding like tick, tick gradually transitioning to gravel and going back to tick,tick and disappear.
Repeats itself after some minutes.
It may not be what causes it for you, but in my case it was a slight difference in the DC cycle time and linuxcnc cycle time.
One is running slightly faster than the other. Both nominally 1 millisec, Combined with some latency of the linuxcnc computer
ethercat will send and receive data not from the current servo cycle, but from the previous or the next cycle.
The gravel sound would appear with the transition over this mixed period to stabilize when the cycles are so much out of phase that they always receive data from the same servo cycle phase.

You can check with halscope of the actual position for example. When the gravel sound happens you will see that
the actual position isn't a smooth line anymore as is expected. Instead you will see that it stays on the same
position for a cycle, to maybe jump double distance another. This is worthwhile checking. 
Two curves in halscope, commanded position and actual position, with and without gravel sound would be interesting to see.

The TXEN pins corresponding GPIO pins would need to be set as outputs 
and into the correct state in the hal file, something like:

setp hm2_7i95.0.gpio.014.is_output true
setp hm2_7i95.0.gpio.014.out false
setp hm2_7i95.0.gpio.017.is_output true
setp hm2_7i95.0.gpio.017.out false
 
 

something to try - switch to relative moves, then G1 Z0 ? Even though you're not moving, it may count as a wait point for M3
Or use M66 instead? You could hook that up to a button near the spindle, or a GUI button.