[scipione205]

Mach 3 on Win 10 x64 - NUVM V2 Controllore with latest plugin

DM542 stepper drivers, 200°/rev steppers ; 2000 steps per millimeter

UNITS: MILLIMETERS

1 - RANDOM NOT PRECISE MOVEMENTS COMMAND (AND DRO)

It happens quite often.

If I issue the following command:

G1 X10

The controller moves the tool, but on the DRO I read:

9.9995

It also happens when jogging (not just for X axis and not just for the specified coordinate 10).

If I try to re-issue the command again, it doesn't move. The DRO remains still, and so does the machine.

Tried also to visualize the movement with a dial indicator and indeed it doesn't move.

Let me tell that the machine has negligible backlash on the axis (less than 0.005mm) and I didn't dialed in any backlash compensation.

Thought could be that 2000 steps per rev gives a minimum division of 0.5 micron, but trying higher steps per rev doesn't seems to work, and however thought that from 0 to numbers rounded to increents of 5 cents, couldn't be an issue.


2 - STRANGE DWELL BEHAVIOR

.I tried a simple code for testing backlash, moving an axis back and forth 0.5mm several times in order to take note of the detected backlash for every iteration.

The code was like:

DWELL P10. (time for the operator to move to the DTI)
G54
G1 X0 F400
DWELL P2. (time for the operator to read the DTI)
G1 X0.5
DWELL P2. (time for the operator to read the DTI)
G1 X0
DWELL P2. (time for the operator to read the DTI)
G1 X0.5
DWELL P2. (time for the operator to read the DTI)
G1 X0
DWELL P2. (time for the operator to read the DTI)
G1 X0.5
DWELL P2. (time for the operator to read the DTI)
G1 X0

The iterations where 10 at least.
Looking at the DTI, for each iteration it didn't move by 0.5mm, but random portions of that. At the end it returns on 0.
Downfeeding the speed make it look worse.

I tried the same code but just with the first dweel (to let me go to reach the DTI after starting the program), getting rid of each dwell betwenn the moves.
It works ok. Even downfeeded to extra slow speed.

What could it be?

DISCLAIMER:
Never had experience with Mach3, only with HAAS, Siemens, FANUC and UCCNC

[dazp1976]

It's not always an exact science.
It likely just needs fine tuning via auto calibration.
If you go back to 0 by going past it, then come back to 0 passing the lash point, then test the SAME direction everytime with the same result.
It's prob to do with minute screw stretching, bearing movement, nut slop, etc.

Is it 1605 screws? yes?
With 2000 steps per unit you might actually end up with a figure of, 1999.99999999999997 per unit or something stupid like that after auto cal.

[scipione205]

Thank you for your reply, but I didn't understood much, could you tell in other words?
The ballscrew is a rolled ballscrew with preload, not a chineese 1605. But it doesn't matter regarding my question, because the system is open loop.
However when I tried moving several times against a DTI with increment of 0.025mm (resolution of stepper and ballscrew) always got within 5 microns or less, BUT when the strange DRO stuff doesn't happen.