Retrofit Leadwell MCV 600XL

At last got some time to connect everything

image missing http://cnc.rs/wp-content/uploads/2020/02/IMG_20200204_165733-Medium.jpg

Ofcourse there is a problem , It seems it has something with the drives, because I can not eliminate the effect with Linuxcnc tuning.
first drive on X tunes fairly well (about 0.015mm ferrror on start and stop and 0.002 while moving), but has some oscillation while moving. Second drive on Y needs 0.1 I so the ferror centers on line and can not be tuned as good as X. Third drive on Z (counterbalanced) has constant oscillation and needs I = 3 to lower the static ferror down to centerline. It wont tune right , moving up is smooth but down is bumpy and in halscope it is seen as oscillation.

image missing http://cnc.rs/wp-content/uploads/2020/02/IMG_20200206_152248-Medium.jpg
image missing http://cnc.rs/wp-content/uploads/2020/02/IMG_20200206_175816-Medium.jpg

I have the manual for the drives , they are set for velocity mode P0-1 = 4 . There are two AGND pins 35 and 36, if analog gnd is connected on pin 35 it wont work, so I tried on 36 also and connecting both together.
Servo drive manual

Tried P and I parameters in drives Cotrol parameter section P1-0 and P1-1 , cant seem to help , lowering P1-0 leads to poor responce, Tried P1-1 up to 3000 it just slows down the response.
Also tried P3-08 Speed feedback filter time parameter ... from default to 0 ... no help its always the same.
Anyone has any suggestions ?

None of the motors seems sto stand still and hunting .

Search for
tuning granite devices argon drive
There are a lot of info so might find something useful there.
Tuning the drives itself is paramount to having a working system, and any tuning of Linuxcnc with improperly tuned drives will only cause headaches and lots of lost time.

Switching to torque mode, it seems much smoother. Will post my results soon

To my surprise these servos are easier to tune in torque mode. What I am missing still is that I cant quite get out the start and stop overshot , its not much 0.02mm at any reasonable feedrate, but its pretty violent , acceleration is set to 800mm/s
Also there is still present some sort of oscillation , and Z tends to oscillate while going down, its P is lowest like 15 and XY is 60, I was set to 1500 for all axis. FF1 and FF2 have strange effects never seen as in velocity tuning
Thats it for now

I don't think you'll ever be completely rid of overshoot using torque mode with a single PID loop. All you can do is minimize it. You didn't mention D, it will be nessisary to use some D in torque tuning. I've also found that setting a limit for I can also be helpful, especially when very large I values are needed.

Yes FF1 and FF2 do behave very differently with a torque command.

I've had the best luck running two nested PIDs on my torque mode servos. Position command to first loop, that sends a velocity command to a second PID loop that then creates the torque command that goes to the drive. Tuning the position loop then behaves and tunes like an ordinary velocity command servo system.

Yes there is some D but in comparison to P and I its a small value around 0.6 , above 1 and it begins to vibrate at 2 its veri violent so I settled at 0.6.
I dont mind the overshoot but would like to soften the response somehow, might be its because the machine is on box ways and has lots of stick .

For now I will leave it as is , after I finish the ATC and other stuff I may get another round of tuning


This is the best I managed for X but its not same obviously in the middle of travel its like this but it a bit worse on start and end where the ways are tighter

image missing http://cnc.rs/wp-content/uploads/2020/02/Screenshot_2020-02-17_13-22-15.png