=FIXED= Z-axis appears to slip, and other things

I’ve had my 800x800 X-carve for over two years and gotten it to work ok, but I have always had problems with the Z axis, it has always been difficult/impossible to get consistent/accurate depth.

Today I was cutting handles out of beech with a brand new 6mm spiral cutter, all went ok except for the depth, each sucessive handle was several mm less deep than the previous. So I stopped and removed the Z pulleys, made sure the one of the stepper motor was aligned with the flat side of the shaft, and decided to grind a flat side on the threaded rod, I’m not sure if this has been changed since but my rod doesn’t have a flat side for the set screws to but against, so they could slip.

Then I stated carving again, the first handle had the right depth but with the last pass this ¤¤¤¤ decided to offset the cutter 2mm in the X axis, which obviously snapped the brand new Festool €60 cutter for which I travelled 2 hours yesterday to go and buy it… I had cut 12 handles previously with no X slipping at all…

Then I installed another straight 6mm cutter and there again the Z axis was way off and it slipped again in the X axis but less than a mm, I tried with a slower feed rate but with no change.

It seems that when you try to fix one problem, two others show up.

Looks like I was able to fix it, first I reduced the plunge speed by half, which improved but not fixed the problem.
Then I loosened the belt for the Z axis a little but what really solved it was to play around with the power of the stepper motors.

My first thought was that the Z-axis wasn’t getting enough power, so I upped the power and noticed that blocked the axis entirely. So I played around with all the axis settings and noticed I had to set all the settings lower from 90% to about 70%.

I have since cut 18 drawer handles and they were all at the exact same depth with no apparent X or Y slipping. This is quite amazing since I don’t recall it ever being so accurate.

I don’t know what caused this gradually increasing problem though, it worked ok 2 years ago and then gradually this Z axis problem started becoming more and more apparent.

1 Like

Your observation is correct, having more accuracy at lower amp settings matches the explanation found on the Pololu website for the DRV8825 stepper drivers:
For the microstep modes to function correctly, the current limit must be set low enough (see below) so that current limiting gets engaged. Otherwise, the intermediate current levels will not be correctly maintained, and the motor will skip microsteps.

1 Like

Could be an interesting experiment, full step on Z and 4 steps on X and Y, will be the cheapest solution to get more torque on the motors. Interesting what the effect on accuracy and surface quality is.

1 Like

Full step will give you the maximum torque available, but (depending on system) it may be louder in noise and perceived smoothness. Belts will smooth things out in that regard. I use all rods on mine and at 1/1 step is behaves quite ragged, this is tamed down a bit at 1/2 step that I currently use.

Full step and harder acceleration may be quite a bit harder for belt systems, causing them to wear out/snap more easily I would think.

But in our hobby world half step or even full step would be sufficient for most in terms of resolution :slight_smile:

1 Like

Thanks for all the replies, when I initially tested my X-carve after assembly I had lots of offset problems. I read that the stepper motors needed more power, so I recall setting everything to the max setting at which it would still work. This seemed to fix it, but after reading these replies it seems like I should not have set it so high. I don’t think this was explained anywhere.
It certainly isn’t in the troubleshooting guide “no something went wrong” , this probable cause and fix should be added.
Such a shame for that spiral cutter, it went through beech like it was butter, and in relative silence. My previous spiral cutter was burned out after 2 years of use, just replaced it yesterday…

Yea it’s more than just resolution as pointed out. It deals with acceleration and deceleration.

If you look at a stepper driver data sheet, they generally show microstepping as a sine wave and how the steps create that wave. Full steps create a jagged waveform that could be more triangle than sine. The more microsteps, the smoother the wave gets which means better transitions. Overshoot may come into the picture with lower microstepping

It’s analogous to modeling resolution, the higher resolution of the model, the smoother the surfaces are and such.

I’d also be curious how GRBLs accel limits could be tweaked to compensate for lower microstepping and if that would actually do enough to smooth it.