Ethercat random jitter fix

Try with synctorefclock false, and see if your reset counter also counts up or not 

I did, unfortunately with SyncToRefclock set to false, the dc-phased pin starts flickering.

the reset counter doesnt go up either.
with it set to true, the rest counter goes to 1 or 2 per lcnc launch.
what ive also found, without every driver being ready, the dc-phased pin is already HIGH. Is that supposed to happen?

Can you plot pll-err? Sounds like you have quite a bit of "normal" jitter.
That can be seen in pll-err.
dc-phased is just the condition that pll-err is below 10% of your servo cycle time.

I can yea. what bothers me, is that the DRO was "filled" with the encoder position right away previously. Now, it rarely does but most of the time, takes 5-10 seconds to fully fill in with machine positions. If you enable it before that, you get amplifier errors or a loud axis clunk.

I dont know of anything that changed other than weather

The phasing is done before all-op, it is about phasing with first drive DC epoch, the epoch arrives ages before all-op

Your problem is CPU and or network jitter, the symptoms your are reporting as Hakan correctly deduced, are telling us your latency is too high. 

refClockSyncCycles="1" syncToRefClock="false"

I think is your best bet.

if you still get jitter, I'd say look at your latency, cpu? kernel version? network harware? latency histogram results with a few glx gears?
isolcpus, nohz_full ?

im reinstalling the whole unit today.
i was trying trixie as i had to redo my host for the new cnc. But apparently is this very experimental with debian 13.

Im installing the linuxcnc 2.9.4 iso now, which should still be bookworm. then updating to 2.9.8 and installing the patch again

re-did the pc. even used one with a 10th gen i5 now instead of the old 5th gen. Jitter is at around 10-20k. should be fine.

DRO is still not always filled on start, still takes a few moments for it to do that.
Will test further

@grandixximo 
Do you have any idea what might be causing the delay in the DRO? 
It seems like the drives arent going into OP right away. The delay is always different. Sometimes the DRO is instantly filled with the encoder read (but converted to axis position) values, sometimes it takes up to 10 seconds. 

Could this also be part of the whole issue? The drives PRE-OP to OP delay. 

TAKUYA wrote:

It seems like the drives arent going into OP right away. The delay is always different. Sometimes the DRO is instantly filled with the encoder read (but converted to axis position) values, sometimes it takes up to 10 seconds. 

 

Seems a bit unexpected to have up to 10 second delay before OP on ethercat. Is this a new config or was it previously working?

Does anything relavant show up in dmesg?
Once it finally populates the DRO, is everything working after that?

The config was previously working just fine. I dont know what changed that caused this.

Ill check dmesg again when im home.
As for the working state after the DRO populates... Its hard to say. I can definately move every axis just fine, but theres a 50% chance that clunks occur. Sometimes more clunks, sometimes less clunks. Sometimes they even appear mid-cycle which is a no-go.

ive swapped out all the ethernet cables now, re-imaged my lcnc host, even used a new host without any success.