Backlash causes erratic servo movements

[mrtned]

I have used Mach3 and Mach4 for years and decided to switch to UCCNC. When the backlash is enabled, the servo motors lose their position and become erratic when jogging. I have read the other posts regarding others complaints and have adjusted comp accel and accel with very little improvements. I never had this problem with MACH 3 or 4. Is there a fix to this issue? I have also downloaded the latest version of UCCNC and it behaves the same.

[cncdrive]

There is no fix for this issue, because this issue is yet unknown.
Can you please check if the same issue happens with the same tests you do (e.g. jogging what you mentioned) when you disable the backlash?
Also can you post your profile file and tell me which axis or axes has this issue?

And yes, comp.accel is what sets the acceleration for the backlash compensation and it has to be higher than the normal acceleration parameter to work properly,
however the UCCNC automatically adjusts the comp accel higher than the normal acceleration parameter, so it takes care that it can't be a problem that the comp.accel is too low.

Mach3 backlash compensation works different, it stops the axes when doing the compensation while the UCCNC doing it without stopping the axes.

I'm wondering though how can the servomotors loose their positions? I mean even if the UCCNC would do the compensation incorrectly servos are closed loop, so they should not loose position except if the frequency goes too high. So, one thing to check is the kernel frequency, maybe it is set too high and the signal can't go through which would explain jerky and erratic movements and also lost steps.
Please note that if you used mach3 previously with LPT port for example that could only do 100kHz as maximum while the UCCNC can do 400kHz with the ethernet controllers, while they are adjustable to even as low as 25kHz. The default is 100kHz...

[mrtned]

I'll try posting the profile tomorrow. All Axis exibit this issue. I have the same results whether I use 25KHz or 100KHz, I normally run 25KHz. The condition can be readily seen when jogging, for example, in step mode .01 resolution. I tested it using a backlash of .03 When I increased the comp accel from 240 to 300 it did seem a little less erratic in a sense. What I mean is, instead of the axis just randomly jerking each time the axis + was it, it would advance in the correct direction for several times before randomly reversing direction. Keep in mind, this all happens each time the axis is jogged in the SAME direction, not just when changing directions. Meanwhile, the DRO readout advances in the correct direction.

The backlash adjustment should only occur when reversing. This jerking effect happens VERY frequently even after the axis initally reverses. In other words, when backlash is enabled, the selected axis will jerk and go in random directions even when the same jog button is hit sequentially over and over..... ex: if I jog y axis in the + direction and keep hitting the + direction button the axis will randomly go in the - and the + direction. So there's actually 2 problems that I can initially see, first is It's like the program is trying to apply faulty backlash moves each and every time the axis moves in the same direction instead of only applying it when the axis initially reverses direction. Second problem is the backlash isnt always moving the axis in the correct direction. I hope this helps.

[wgpeters]

I also have this problem. My system is Mach4, uc100, Sherline lathe, Vistacnc pendant, HP Laptop. Backlash is set to .003 in Mach4 (backlash measured with dial indicator)

When the jogging distance is set to less than what the backlash is set to, I get the following behavior:
Jogging in one direction .001 inch seems ok, but jogging in opposite direction does not. Each jog of .001 causes the stepper to go correctly, but immediately back up by the amount of the backlash. Success jogs of .001 produce a move of .001, then the next jog does not move, then the next moves .002. This happens both with the on-screen jog button, and with the pendant. Jogs in .001 increments with the pendant causes loss of position verified with a dial indicator.

When the jogging increment is greater than my backlash setting, the motor reversal seems to happen only at the end of the jog, but the jog distance seems correct as measured by my dial indicator.

Now I also tested moving .001 increment using on g0 codes going in successive .001 increments (g0 z .001, g0 z.002, ... g0 z .010) and this works fine.
So the problem seems to be only in the jogging functions. Motor acceleration rate does not seem to have any effect on the problem.

Bill

[wgpeters]

I also have this problem using Mach4 and the UC100. Has there been any resolution?

[wgpeters]

My setup is Win10 HP Laptop, Mach4 Hobby, UC100, Vista Pendant,Sherline Lathe, Sherline Nexgen Mill

System does not jog correctly when backlash setting is non-zero in Mach. This affects all axis.

Backlash adjustment is applied after each jog even though direction has not changed. Sometimes, the stepper goes the wrong direction, and often the jog amount is wrong. For instance a .01 jog may go .01, or may go something less. This happens whether the jog was done from the pc jog button, or the pendant, so I believe the pendant is not the problem.

Smaller jog increments result in more erratic behavior, especially when the jog amount is less than the backlash setting (which in my case is set to .003".

When making successive jogs of .001, the movement is often .001, 0, .002, 0, .002, etc.
When stepping by multiple gcodes, like g0 z .001, g0 z .002, g0 z .003 ..... the system seems to work correctly as shown by test indicator and visual observation.

UC100 is set to 25khz, but testing with 100khz had the same result.

Backlash should only be applied when changing directions.

System works correctly when driven by LinuxCNC so problem is not stepper malfunctions or the Sherline motor drivers.

Do I need new firmware or drivers or something else?

[dammogreen]

I have the same problem I won't re-describe it as my problem is very well described by the previous posts of: "mrtned » Wed Feb 28, 2018 8:50 pm" and "wgpeters "...

Anyone resolve this yet?

Steppers,W10, UC300ETH UB1, 1.2047

[cncdrive]

Check the comp.acceleration parameter of the axis, because the software uses that for the backlash compensation. It might be too high when your steppers can't accelerate the axis that fast and so they loosing steps.
Backlash on the other hand used by lots of Stepcraft users, because that hobby machine mostly have a little backlash and they are 10000+ users already and they using the backlash without any issues, so there is sure no algorithmical or code issue.

[wgpeters]

Where do I find this parameter? I have Mach4.

[dammogreen]

accelerations are on the Configurations page for the specific axis.....I have no luck with different, and even very low acceleration settings so far, with various multipliers of 1.2 to 2x ofthe normal accel parameter. I may do a video later after more tests.