Backlash compensation not working in z-axis, AXXB-E controll

[mholm]

Hi

Backlash compensation does not work properly on z-axis in my setup. It tries to compensate wile I'm jogging in 0.01mm increments in the same direction. Jogging the z-axis work like it should when backlash compensation is disabled in settings.

I have ruled out everything but the controller or software. I have swapped cables to run the z-axis motor plus hardware with the x-axis electronics, that works well with the x-axis backlash compensation enabled. I have run the z-axis drivers + cables from AXXB-E controllers x-axis pins and that works too with backlash compensation enabled in x-axis settings.

I have also tested with different versions of UCCNC.

The only thing I can think of is the software or the controller doing something weird with the z-axis. Attached is my profile file.


Regards

Mikael Holm

[mholm]

I have done some more testing and narrowed it down to the Steps/Unit. In my setup I must use 160, 320, 640, or 1280 depending on microstepping, to get the measurements right for z-axis. However something doesn’t like that ratio. If I adjust the steps/unit by +-20 the erratic backlash compensation behavior disappears.

On my x- and y-axis I use 400, 800, 1600 steps/unit depending on micro steps set at driver. That works perfectly without erratic movement when jogging and backlash compensation activated. However if I change x- or y-axis steps/unit to the same as z-axis (640 in this case) the erratic jogging behavior appears on these axis too.

Please help me solve this annoying behavior. I’m out of ideas what to try next.

Regards
Mikael Holm

[dezsoe]

Hi Mikael,

Thanks for the detailed report. We will investigate it.

[Strategger]

Hi Mikael, dezsoe,
To complement Mikael's report, I confirm the bug is also present in UCCNC v1.2111 + UC100 controller. My setup is a 3-axis mill with inch units and a profile imported from a working Mach3 installation. The issue:
1) Erratic behavior of Z axis when step-jogging at 0.001" per step. One sees a number F of forward steps, then one backwards; exact number of F seems to depend on the selected steps-per-unit (SPU), but it is not clear and it does not seem to be periodic.

2) When SPU is 10000 on all 3 axes (X,Y,Z), the backlash compensation works fine, as expected. But SPUs of X: 10000, Y: 11000, Z: 12000, for instance, and for many other choices, lead to the erratic behavior.

The points below are new findings with respect to Mikael's post:

3) An axis crossover occurs when the backlash produces erratic behavior: Jogging in one axis (X or Y) also produces movement in the Z axis. It is as though the Z axis became enslaved to X or Y, but this does not happen all the time. I haven't figured out exactly when this coupling occurs.

4) The backlash compensation works perfectly when controlling the UC100 with Mach3 (same computer, analogous settings for Mach3 and UCCNC, and same UC100, without even removing the USB connector). Therefore, the problem *appears* to reside in the UCCNC software more than in the controller module (UC100) itself---because at least 2 controller models exhibit the "same" issue.

Hopefully this information will be helpful to solve this rather annoying issue. Happy to provide more data or do more testing if necessary.
Regards,
David

[Strategger]

Hi,
To complement Mikael's report, I confirm the bug is also present in UCCNC v1.2111 + UC100 controller. My setup is a 3-axis mill with inch units and a profile imported from a working Mach3 installation. The issue:

1) Erratic behavior of Z axis when step-jogging at 0.001". A number F of forward steps, then one backwards; exact number of F seems to depend on the selected Steps/Unit (S/U), but it is not clear.

2) When S/U is 10000 on all 3 axes (X,Y,Z), the backlash compensation works fine, as expected. But S/Us of X: 10000, Y: 11000, Z: 12000, for instance, and many other choices, lead to the erratic behavior.

These are new findings with respect to Mikael's post:

3) An axis crosstalk occurs when the backlash produces erratic behavior: Step jogging in one axis (X or Y) also produces movement in the Z axis. It is as though the Z axis became enslaved to X or Y, but this does not happen all the time. I haven't figured out exactly when this coupling occurs.

4) The backlash compensation works fine when controlling the UC100 with Mach3 (same computer, analogous settings for Mach3 and UCCNC, and same UC100 without even removing the USB connector). Therefore, the problem *appears* to reside in the UCCNC software more than in the controller module (UC100) itself---because at least 2 controller models exhibit the "same" issue.

Attached is one of the profiles that caused problems. Hopefully this information will be helpful to solve this rather annoying issue. Happy to provide more data or do more testing if necessary.
Regards,
David
(Apologies if this message appears twice; I'm new to the forum and having trouble posting)

[Strategger]

Hi,
As an update to my previous post (yet to be approved, I suppose), I have quickly tested the backlash compensation with the newest UCCNC version available to me (1.2113), the same UC100 controller and the same settings (where applicable; cf. attached profile). I hereby confirm that the problem is equally present and behaves similarly as I described for version 1.2111. I hope this helps to debug the issue.
Regards,
David