Z-axis changes direction mid-print?

[chrzr]

Hello!

I've been using Klipper since the early 2018 and I'm very happy with it. However, in the last two months I had a few problems during printing. For example today, I wanted to print a part, which seemed fine in the Simplify3D preview. After around 30%, I went to the printer and discovered that the nozzle was scraping on the buildplate and that the part had been ripped off. I stopped the print and the z-axis still wanted to move against the plate instead of moving to z = 10mm. I've never encountered such a behavior with my previous printers running Marlin.
I have already looked at the klippy.log, but I could not find anything suspicious.
klippy.log

edit: Ok, so I checked the original stl file and it showed some self-intersecting surfaces. I used netfabb to fix them, but it showed the same behavior - this time even earlier ~1mm z height.

-chrzr

[hg42]

sounds like skipping steps?
The acceleration profile is different with klipper. I am using lower acceleration but higher speeds.
Also junction_deviation behaves different to "jerk" setting of Marlin. I tend to adjust it near to zero.

[StefanMeGit]

What drivers are u using on Z?

[chrzr]

@hg42
Thanks for the hint! I'll try adjusting the acceleration and junction_deviation as soon as I'm back home.

I use the provided config file for the CR-10 with the default parameters. The weird thing is that I've been using Klipper with that config from around January until May without any problems. Then, when I wanted to print a part, which I designed in Tinkercad I ran into this problem for the first time. The .stl also showed self-intersecting surfaces, so I fixed it with Netfabb and it worked just fine, which is why I assumed that this was the problem.
After the said print failed twice yesterday, I started printing another part, which worked fine until about 70%. This time the z axis wouldn't move upwards.

I also don't think that it is related to the slicer, as I ran into this problem with Simplify3D as well as with Slic3r.

@StefanMeGit
At the moment I use stock CR10 (non s) Melzi board with the stock drivers. I also have a MKS gen 1.4 with TMC2130s for x, y, z and the extruder. I might just test it with that board, as I start to think that it the z driver on my board is at fault. I just did not have time to install it yet.

[hg42]

Z axes with spindles often need a very low acceleration...I am not sure why, because they usually have a high gear ratio. I guess it has to do with static friction, which may be very high for spindles.
I found this can change over time because of dirt or oxidation and decreased lubrication (oil drifting away etc.)

[hg42]

my corexy uses 25-50 mm/s^2

[KevinOConnor]

On Mon, Jul 02, 2018 at 06:29:53PM +0000, chrzr wrote: Hello! I've been using Klipper since the early 2018 and I'm very happy with it. However, in the last two months I had a few problems during printing. For example today, I wanted to print a part, which seemed fine in the Simplify3D preview. After around 30%, I went to the printer and discovered that the nozzle was scraping on the buildplate and that the part had been ripped off. I stopped the print and the z-axis still wanted to move against the plate instead of moving to z = 10mm. I've never encountered such a behavior with my previous printers running Marlin. I have already looked at the klippy.log, but I could not find anything suspicious.

I don't see anything unusual in the log. This doesn't seem like a software issue. It could be hardware. Things you could check - if the Z actually changed direction, then it may be a bad z driver (or it is overheating), or some debris on the board could be shorting some pins on the driver or mcu. You might want to carefully inspect the printer board.

…

-Kevin

[chrzr]

Sorry for not repsonding earlier. Had some busy days this week.

I tried a few things in the meantime:

• Adjusting the acceleration, speed and jerk settings -> This did not help
• Checking the Smelzi board and drivers for debris -> All ok
• Checking the driver voltages. Z driver voltage was a bit low (0.65V), so I adjusted it do match the voltage of others (0.95V) -> This also did not help
• Using a different slicer again (Slic3r Prusa Edition this time with the same settings/profile as before) -> Printed without problems. I have no idea what made the difference ¯_(ツ)_/ ¯

As I had already ripped the control box apart, I went ahead and installed the MKS Gen 1.4 with TMC2130s. This did cost me some hours, as I had to configure the pins for the SPI connection. Afterwards I used the momentum and installed a E3D V6 hotend which also waited to be installed.
So, currently I'm running Marlin again and everything is fine, but I plan on returning to Klipper as soon as possible! :-)

Thanks for all the responses and tips!