Linear scale correction: homing to index

Hmm if that were the case this would affect everyone that homes to index and that is not the case AFAIK

I would try a minimal home to index case in master

i seem to have this problem as well. For some reason it only seems to affect one of my 4 axis when homing. After the following error alarm is cleared it usually homes fine the second time. It only seems to do it on the first attempt at homing after starting linuxcnc.

What version of linuxcnc do you have? (exact version is important)

A problem on only one axis suggests a hal file error, can you post you hal and ini files here?

Todd wrote:

i seem to have this problem as well. For some reason it only seems to affect one of my 4 axis when homing. After the following error alarm is cleared it usually homes fine the second time. It only seems to do it on the first attempt at homing after starting linuxcnc.

Sounds exactly the same like my situation, although I currently have only one servo axis running. I also have to restart LinuxCNC to be able to catch the following error again. Once the f-error happened I can unhome and rehome as much as I want without following errors.

But of course the position counters don't step-change anymore during the rest of the session since it will be reset when it is already at 0.

So, what LinuxCNC version do you suggest for further testing? Fresh GIT pull from master? Or one of the stable releases?

I am using 2.5.4, and I also had the same problem when I was testing the new trajectory planner version of master (that was a few months ago).
I've just been living with it

I can post my config if it might help.

I would suggest using master (especially if you use error last target)

posting hal an ini files would help

Here it is.

I dont see anything obvious wrong in the hal file

I would try removing all the unused
net N-pos-rawcounts <= hm2_5i25.0.encoder.0N.rawcounts
lines as I have seen issues with this in other systems

What axis has the problem?

Usually just the X, but once in a while Y also.

Commenting out net N-pos-rawcounts <= hm2_5i25.0.encoder.0N.rawcounts didn't make any difference.

I spent a couple of hours testing homing to index yesterday (with master)
and cannot find any issues.

Can you plot

axis.0.motor-pos-cmd
axis.0.motor-pos-fb
axis.0.ferror
axis.0.index-enable
pid.0.out

with halscope?

halscope should be triggered on the falling edge of index enable
while doing a home sequence. This would help isolate the problem.

This needs a fast sweep, say 1 ms/division and the falling edge of index
enable visible on the screen, and normal, not auto mode trigger mode.