Servo tuning / debugging help - clarification of context….

I would like to ask people who have experience to clarify the impossibility of tuning the servo (s).
I have been struggling with tuning for two days, but still without success. I have read many articles and recommendations, starting with "tommylight - Servo Tuning detailed How To" and ending with "John Thornton - Servo Tuning Tutorial" and many other tuning issues described and addressed.
For the last 3 months, I have been reworking my cutter into a full-fledged 3D cutter diluted via Linuxcnc.
So I summarize what is used
Linuxcnc 2.7.15 (Gmoccapy)
Mesa cards 6i25, 7i77 and 7i84
Encoders - surface Hedenhhain LS803
LTi Drivers (CDE34.005)
LTi Servo Motors (LSK-115-2-20-560)
It can be said that the cutter is 90% complete, so I decided it was time to try to move the machine axes. However, I cannot tune the movement correctly. When tuning, I followed the recommendations of tommylight, I started with a small value of P = 1.15 and I tried FF1 (here I started at 0.001 up to 0.95, but axis.0.f-error has a high level and teeth like on a saw. the movement is choppy and I don't know how to continue, what else to try in.

I also tried the drivers (configured - speed control +/- 10V) by connecting an external control voltage - the motors went completely smoothly.
Also after entering Home (the “index” of the linear ruler and Home position -50 is used), the X-axis is still tightened to this value and it often takes 10 seconds.- video.

I am attaching a .ini and .hal file and a few screens.
Thank you for any advice that will help solve my problem.
I apologize for the spelling.

Thank you very much for your quick reply.
I tried P sinece 0.5 to 5, but at about 3 axis move by itself.
I run to my machineshop to try what heppens, but nothing positive. faling edge is away

thank you tommy.
I read your tutorial many times.
I'll try it step by step one more - probably I'm cycled now.
it is a true, i let go axis for about 3sec. du you mean short press something like 0.2-0.5 sec?
sorry for deadband- i thought this parameter set when get something relevant with f-error.
deadband is possible smooth out tooth on f-error curve?

The sawtooth following error plot looks like standard stick/slip
with a torque mode drive with a low P gain.

If your motor drives are running in torque mode, the tuning will be very different
from the tuning tutorials which assume velocity mode drives

Torque mode tuning typically has no FF1 (but some FF2) , large P, I and
significant D for a single loop system

thanks PCW.
please can you give me some values (or some expiriences ) where is good to start tuning. there is no some tutorials for this type of servo setting.
i don't understand why torque mode? diriver can be set in torque mode, but I select speed control (i think it is the same as velocity mode ??)

today I tried to tune this one axis again.
I proceeded as tommy wrote in his tutorial.

the saw was created by the soft feedback and rigidity of the mill table
however, I cannot tune the marked defects on the f-error curve.
can anyone advise me what to focus on, how to remove the acceleration jump and shift the curve to the zero line?
the picture is at a feed rate of 47mm / min, at higher speeds this initial tooth is not so pronounced there.