B-axis drift

[henroys]

Hi
I'm running grblHAL on the Mach3 redboard, configured for a 5th axis.
After my +17000-lines program the B-axis is repeatably 3.5 deg off.
The lines looks like this typically: X-26.74Y-3.73Z174.45A25.44B-711.872

When moving the B-move first in the string, ie B5X1Y2Z3A4 instead of X1Y2Z3A4B5 the error is smaller. Maybe half.
When removing all other axes and just sending the B-moves the error is gone.

I've tried various stepper resolutions, polling interval to no avail.

What could the cause of this?

[phil-barrett]

Have you calibrated your B axis? Is there any backlash in the mechanism?

[henroys]

There is no backlash in the B-axis. But I run compensation on x, y and a. Might try and turn that off and see.
B is simple to set up, 4 degrees per stepper revolution.

[henroys]

Update on my issue:
I have found that microstepping causes the drift. After reducing the resolution to 200 steps/revolution on the driver (DM556) and 50 steps/deg in the sender, the B-axis returns to zero just fine. So the problem seems to originate somewhere in the mach3 board.
When routing the same steps with microstepping to my B-axis using the A-axis output it also returns to zero fine.

[terjeio]

Then step/dir pulse delay ($29) may be too short for your stepper drivers.

[henroys]

0 delay works fine on the other four axis which has the same drivers.
I tried to increase $29 in steps all the way to maximum (20ms?) without any improvement.

[terjeio]

Have you tried swapping drivers, e.g. Z with B? Same result?

[henroys]

No I haven't. But I will...