Tuning analog Fanuc DC servos?

I´m trying to tune the x axis of a machine with Fanux DC servos and I cant seem to get rid of the overshoot at start
I have used this page as a reference gnipsel.com/linuxcnc/tuning/servo.html
But changing FF2 only makes it worse no matter what I set it to.

I also think the P is very low but setting it higher will produce more and more ocsillation.

Any ideas?

Attached is a screenshot with the settings and halscope

mm or inch?
Are these velocity mode servos?
If so, how do the drives get their velocity feedback?

Normally you would not use D on a velocity mode servo
(requiring it indicates something is wrong in the drives velocity loop tuning)

Hmm I figure it could be that I got the tachometer feedback wrong, I simulate a tachometer from the encoder velocity, you helped me alot with that if you remember.
D helped with oscillation at low speed.

The analog output is quite strange to, on half speed it only outputs 1,2v, maybe that indicates that the the drive get a too high tachometer signal so that the pid loop does not need to send a higher signal, or should be the other way around maybe

Its mm and velocity servos.

I guess I would verify the the velocity mode worked properly before tuning

If you get a runaway in open loop mode (no feedback = all PID terms 0), the
tachometer feedback is backwards

It may also be that your tachometer feedback is in the correct direction but scaled improperly

The analog output is quite strange to, on half speed it only outputs 1,2v, maybe that indicates that the the drive get a too high tachometer signal so that the pid loop does not need to send a higher signal, or should be the other way around maybe

This does sound like the tachometer feedback is backwards (the tachometer feedback is positive so the PID output is being "helped" by the feedback )

I think I found the problem, a pretty long time ago when I connected the motors and tested the basic functions I tested and measured the rpm of the demounted motor and then it spinned 2500RPM at 10v and the tacho voltage was also correct
But now I wrongly asumed that the maximum speed for the old control (12000mm/min) was the same as 10v to the drive and when I checked now the pitch of the screw and that with 2500rpm would be 30000mm/min! no wonder it was over reacting.
I wonder how the machine constructors thougt when they skipped 60% of the servo capacity?

I had very little time left to try it out but imediately the slow movements was much smoother.

Have to write a manual for myself so I dont forget to do all things in the right order its usually quite long between the times that I mess with linuxcnc. love it one day and hate it the next