Feed rate difference on x axis by factor of 10

As a reasonable starting point, I would set (for all axis)


MIN_FERROR = .100
NOTE: these are not signed , this is the ferror limit at 0 speed

FERROR = 1.00
NOTE: this is the ferror limit at maximum speed

OUTPUT_SCALE = 10
NOTE: this should match the output limits

MAX_VELOCITY = 3.000
NOTE: you had 20 here = 1200 IPM I doubt the Tree can do this

MAX_ACCELERATION = 6
NOTE: a fairly leisurely 0.5 seconds to full speed to start. may be improved later

After this is done i would attempt tuning, there still may be a X axis problem but
we are starting from a more reasonable place

These problems are very unlikely to be PC related

Do you have a list of signal names I should be looking at withHalscope?

Take a look here:

gnipsel.com/linuxcnc/tuning/servo.html

I followed the tutorial and I got the X axis in the when x is traveling in the positive direction correct. Feed error is not visible at 5mv /div as in the tutorial. However when x is traveling negative I get a square wave form on the X vel cmd. I tried correcting this with changing P but without real success. I noticed in the manual concerning the ini file variables, they build a table to linearized the output and product a comp file . There is also a variable "Output Offset" I have neither of these listed in my ini file. Can I just add the variables in my file?

X velocity command comes from LinuxCNC and is unaffected by hardware Are you sure you mean X vel cmd?

Did you fix the things in the ini file as I suggested?

The signal that shows a square wave is axis.0.joint-vel-cmd which is X axis on my machine.

Currently my settings are

P = 5.0
FF1 = 0.93
FF2 =0.21

X+ seems to like this and runs fine
X- now travels the full length but does so being pulsed by the square wave

I can reduce the the frequency of axis.0.joint-vel-cmd by increasing P value but then the axis does not move.

It would seem that X+ is correctly tuned then X- should be also. Hence my question about Output Offset variable.

axis.0.joint-vel-cmd is not affected (Or certainly should no be affected) By PID tuning at all

Did you fix the things in the ini file as I suggested?

Yes I have been using the parameters you suggested. I am currently using those settings and adjusting FF1 and FF2. So far no improvement. I have used the servo tuning method from Tree Milling Machine.It uses an ohm meter to set the Tach signal Signal level
Compensation level and Balance level. There is another method given by Servo Dynamics using an oscilloscope which is used to adjust the on board pots on the driver boards. Maybe I should be using Hardware Tuning method #2 before following John Thortons tutorial. The Tree method assumes that I am Using a Dynapath control so maybe it is not a valid method,