Inconsistent Z axis steps/mm

I have built a custom CoreXY printer and I have run into a weird issue. I bought T8x8 lead screws from BIQU and paired them with a 3D printed nuts for the Z axis. Previously, the accuracy was almost perfect, but now I cannot get a consistent Z height. 20 mm cube is 19,2 mm, if I change the steps/mm to compensate, then 100 mm tower is 102,5 mm tall.

I have tried flipping the lead screws upside down with no improvement. Thus, I would say the lead screws are not at fault. There is no backlash in the 3D printed nuts either.

Both lead screws are coupled to a stepper using a spring coupling and supported on both ends by a pillow bearing. The bearings have a little bit of backlash but gravity helps with that.

I have tried multiple adjustments and cannot seem to figure it out, so if you have any ideas, even crazy ones, they are all welcome. Thank you.

... adjust it for the 100mm-cube or bigger ... and then check, if the first layers are squished more than the later ... or something like "elephant foot" or such ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

There is a bit of elephant foot as well as the layers being a bit more squished. But I would say nowhere near enough to make a 0,8 mm difference.

I will do a few more test prints, though.

What are the acceleration and jerk values for the Z axis? If you're trying to make it do things too fast the motors might skip steps.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I have dual lead screw moving the bed. Each lead screw has its own motor with its own TMC2208 driver running at 400 mA. The limits are: velocity 30 mm/s, acceleration 200 mm/s2, jerk 0,3 mm/s.

I have previously tried increasing the motor current (with the intention of mitigating potential step loss) with no improvement.

I just completed two test prints at the ideal 400 steps/mm (ideal for 8 mm/rev screw, 200 steps/rev motor, 16 microstepping).

First are "stairs" where there are steps 1 mm tall until 20 mm height. 1 mm, 2 mm, 3 mm are near perfect, then it goes wrong: 3,9 mm, 4,8 mm, 5,7 mm, 6,6 mm, 7,5 mm, 8,4 mm, 9,3 mm, 10,2 mm and the rest until 20 mm stays at about XX,2 - XX,3 mm.

The 100 mm tall tower is 99,2-99,3 mm tall.

It seems that there is an additive error instead of a multiplicative one. But I have no idea where it could be coming from.

Are you using PID control on the bed temperature?

Do both screws lift the same amount or does the bed tilt when you run these tests?

Pictures of prints might be useful here.

Edited 1 time(s). Last edit at 08/29/2023 09:38AM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Yes, I do use PID on bed.

I have not noticed any sort of tilting. When printing larger/wider models, the height was equally wrong on all 4 sides.

Pictures of the stairs and the tower are attached to this message.

Are you letting the bed heat up before you start the prints? Your controller may report the target temperature is reached long before the bed actually stabilizes.

Can you post a picture of the printer?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

One picture is from the front, the second is a detail on one of the Z lead screws.

I always heat the printer up a few minutes before starting a print to let the heat spread. The results are identical to when starting a print right after another has finished without letting the printer cool down.

The printing area is 350x350x270, overall size is 610x610x605.

Seems like I have solved the issue. Previously I was using Marlin 2.0.9.1. Now I have upgraded to Marlin 2.1.2.1 and the issue is gone. Now all prints (with the exception of one) are accurate down to 0.1 mm in Z height. There probably was a bug in the previous version that got trigerred only when certain configuration was enabled, because the printer used to print accurately, but at that point it was not fully configured yet.