Can you verify that encoder position or pos-fb doesn't change after a few seconds?
Just so we know?

You may not be able to fire up halscope in time for that,
or if you are unable to catch it in halshow.
I propose you use sampler/halsampler.

Add last in your hal file

Adjust net names to yours.

Within a few (10 max) seconds after you started linuxcnc type the command "halsampler -t data.out"
After the coordinate on the axis screen has happened, hit Control-C on halsampler.
You have now a readable file of recorded values. Open the file in Excel/Calc and plot XY plot.

 [/code]

I have not, I'll give it a look.

I've found the EtherCat experience to be absolutely positive so far. I had this retrofit up and running (with a lot of help from the forum) in a day or so compared to about a month on the previous machine.

While it's certainly possible that it's related to EtherCat, I don't think so. The machine movement is absolutely rock solid and silent and so long as I don't enable the machine until Lcnc has updated the position values.
Lcnc is changing the machine start possition after it loads and it looks to me like it's loading the g54 values from the linuxcnc.var file about 5 seconds after it loads. It seems to correspond with the axis toolchanger message in the bottom right disappearing but that could be purely coincidence.
The 'jump' occurs every time I restart Lcnc but I'm not powering off the drives. If I set the drives to display encoder count the value doesn't change when I close lcnc so if it were an encoder feedback issue I'd expect to see changing values on the drive its self.
It seems more like an issue of incorrectly saving machine coordinates on close or load.

wrote:

Well a slightly newer version could work too.

Try whatever came right before "Bookworm", see if it's old enough.
 

Thats the thing. The LinuxCNC version before Bookworm was on a Wheezy ISO. There was no linuxcnc version released on Bullseye (5.10 kernel) which precedes Bookworm. We don't want to hear Ubuntu being suggested. That's bad advice (as is using Wheezy which is akin to suggesting people drive a Model T Ford)

Can you tell about your computer hardware and how you installed linuxcnc?  What Linux version?  What does a latency-histogram --nobase --sbins 1000 display? There could be network latency issues on newer Debian versions, but that only affects Mesa ethernet cards. I would be looking more at secure boot, isolcpus and BIOS settings.

Hello everyone.

I've always been very curious about how LinuxCNC code works, and I participated in translating the documentation into Spanish when it was at version 2.8. Now I'm looking for opinions in a forum about a need some of us might have (I suppose): documentation of the source code itself.

I love images. A flowchart is a treasure for me. Yes, I know: code is the best documentation... but sometimes it's a real headache (we all know that, right?).

That said, let's get down to business!

I've prepared the PDF attached to this post. First of all, apologies: some of it is in Spanish (mainly the descriptions), but that doesn't obscure the general idea.

The document relates to the halcmd_completion.c file and is just an example. It only took me half a day to prepare, and the flowcharts haven't been proofread. I insist: I'm asking for your opinions on my vision, not for technical corrections.

This isn't the time to discuss the file itself, but rather to share my idea for documentation. I believe any "novice LinuxCNC coder" can find guidance with this type of "document".

I'm also aware that this is a user forum, not a developer meeting place, but... what about "high-end" users?image missing /media/kunena/emoticons/w00t.png 

Thanks in advance for your comments.
J.M. Garcia a.k.a. Abu

I think I have seen those 83 errors as well, probably some issue with how it starts/stops but I agree that should not be an issue. I might try that with the "built in" drivers to see if that behaves any better.
Errors that don't really cause any known issues on my end but I did fix is
<pdo idx="0x1702"> to 0x1600
<pdo idx="0x1b04"> to 0x1A00
That maps to the truly variable PDO mapping, the ones you have set are fixed PDO mappings, and when lcec tries to write to them it gets errors and writes them to the dmesg logs. In addition if you use the correct ones you can map whatever PDO's you want.
I am pretty sure with the xml file you used it does not configure DC sync. When I use the built in "driver" type="OmrG5_KN04H" I see a message about the DC sync, when I use the example xml it does not, which makes sense because its not enabled in master.

refClockSyncCycles = "-1" for older versions of linuxcnc_ethercat ( i.e. lcec ) syncToRefClock="true" works for newer versions.

For anyone out there trying to use variable PDO, for some reason my drive does not support 0x60f4. It took me a long time to figure that out.

Apologies in advance, I'll probably describe this wrong but hopefully the video makes it clear. 
I'm in the early phase of a Weeke router retrofit. I've got my Delta Ethercat drives up and running with LCNC thanks to a lot of help over in the Ethercat section of the forum.
I've got an issue at the moment that I think is probably just basic Lcnc behavior stuff. After I launch Lcnc and Axis loads the axis positions are updating and the machine is trying to move to that new position (if it's enabled at the time). It seems to happen just as the "axis" dialog box in the bottom right corner disappears If it's not enabled then the position for the axis update and everything is good from then on. It looks like Axis is importing the stored values from linuxcnc.var about 5 seconds after startup, the g54 values I think.
If the machine is enabled at the time when the update happens the drives are seeing this position update and trying to teleport to the new position at what seems to be max speed and acceleration making for some very noisy and unpleasant sounds. 
What do I need to do to get rid of this behavior?

Video shows it all pretty clearly. Even though you'll hear me saying that it doesn't do it the first time it clearly does. You can see the Z axis (I think) move about 50mm right at the start of the video.

Ooops, I did not set  pos_cmd_out but its not needed and could be removed from the component

Try the attached component

Untested. Probably has some typos but the logic should be sound.

hi,
I was hoping to enable and disable adaptive-feed from halui but could not find a way to do that. I was hoping for some help on modifying halui.cc so adaptive feed can be directly changed from the hal layer via a panel button. It would be good to have an indication also of the current status of adaptive-feed from halui. 

Currently adaptive-feed can be modified via Gcode M52 P1 and P0. It can also be changed from the linuxcnc python module via the command channel which makes me believe that it should be changeable from halui.cc too.

I mean one easy way  to do this would be to make a python user component that imports the linuxcnc module and will check the halui status from the stat channel and update on a pin and set it from the command channel via monitoring a couple hal pins for changes.

However, I feel halui is the appropriate place to put pins for operating the machine from the machine panel.

regards,
-automata