Help to find problem in servo

Hi!
I attach several snapshots, please help me to "decode" it.
I have problem with slow moving on my servo (2-3 mm/min)
Hear is the pictire without any mooving, only digital noise.

This is the same noise but in other time scale

Then I make little shot (external HAL button via jog) on 0,25 sec with the JOG scale 10mm/min
The picture show X and Z out and ferror signals

This is the same but in othe time scale

And the I did one shot both axis with the speed 1mm/min.
Help me to understand my problem.

the HAL connected such way

net X.pos-fb <= hm2_[HOSTMOT2](BOARD).0.encoder.00.position
net X.pos-fb => pid.0.feedback
net X.pos-fb => axis.0.motor-pos-fb

net X.pos-cmd <= axis.0.motor-pos-cmd
net X.pos-cmd => pid.0.command

net X.pos-out <= pid.0.output
net X.pos-out => hm2_[HOSTMOT2](BOARD).0.stepgen.00.velocity-cmd

setp pid.0.Pgain [AXIS_0]P
setp pid.0.Igain [AXIS_0]I
setp pid.0.Dgain [AXIS_0]D
setp pid.0.bias [AXIS_0]BIAS
setp pid.0.FF0 [AXIS_0]FF0
setp pid.0.FF1 [AXIS_0]FF1
setp pid.0.FF2 [AXIS_0]FF2
setp pid.0.deadband [AXIS_0]DEADBAND
setp pid.0.maxoutput [AXIS_0]MAX_OUTPUT

### PID SETTINGS START ###
P = 50.0
I = 0.0
D = 0.0
FF0 = 0.0
FF1 = 1.0
FF2 = 0.0
BIAS = 0.0
DEADBAND = 0.001
OUTPUT_SCALE = 250.0
OUTPUT_OFFSET = 0.0
MAX_OUTPUT = 140.0
#the sign may be "-" or "+", it depends of encoder wires connect
#if the servo drive has fatal error [ferror] then change the sign to opposit
INPUT_SCALE = -1000.0
### PID SETTINGS END ###

I agree with Andy I think 'P' is way too high.

If P is too high, it will magnify thread timing jitter and cause large completely
bogus velocity modulation in the stepgen rate (what you are seeing)

Other things:
For final tuning during motion, you should use pid.N.error rather than ferror
because ferror is calculated incorrectly for PID controlled loops when in motion
and will have an error= velocity*servothread period.

To see pid.N.error you need to set the PID comps debug flag

FF1 can be fine tuned when P is small (too much FF1 makes position lead profile, too little FF1 makes position lag profile) changes in FF1 should be very small (it will still be very close to one)

To eliminate the 1 or 2 encoder count dither when stopped, you can add some more deadband

Thank you all, I will test it tomorrow.
I will create pyvcp bar to adjust P. I think the start will be from 10 till 50 (as now) with the step = 1.

And the last question.
In horizontal is time scale in mSec.
But what about vertical scale?
In Volts? In Meters or in what measure units?