Layer shifts at same height, distance and gcode

[SergioBA 0]

Hi.

 

I've been experiencing this issue since January. I use cura since 2.x version (about 3 years), and never had a layer shift before this year.

I have 5 printers (ender3, custom prusa steel, anet a8, artillery X1) and I have checked pulleys, belts, motor vref, etc. Everything seems OK.

In about 3 years 3D printing never had a layer shift before, not even when I got my first printer, the Anet A8.

And now all of my 5 printers suffer this problem.

 

The problem has been increasing since January. Everytime I slice a STL, there's a big chance of getting layer shifts, which are not visible in the preview.

If I print that gcode again, it will happen exactly the same a hundred times. Then I re-slice it and usually the problem is solved

 

I'm a 100% sure it's a slicing problem. If the gcode is "good" it wil never fail, when the exact same model with the exact same settings failed before.

It has been tested with the hundreds and hundreds of face shields I recently printed for the Covid-19 situation.

 

This is driving me crazy, any solution known?

Thank you in advance.

[gr5 2,172]

This is caused because the printer is accelerating an axis at a higher rate than the printer can achieve.  That is the guaranteed truth.

 

HARDWARE

This happens on several of your printers but 90% of the time the printer needs oiling (too much friction) or belt tightening (pulleys slip on shaft) or raising the current to the stepper (to get more torque) or lowering the torque to the motor (drivers overheat and shut off for 1ms).  I have seen all 4 of these issues.  Many instances of all 4 on my own printers and other printers.  But for it to happen on all 4 printers seems unlikely unless none have been oiled recently or temperatures in your printing room have increased at the same time as probelsm started.

 

FIRMWARE

It should be impossible for Cura to send a movement that exceeds the printer's ability.  Marlin lets you set max jerk and max acceleration and someone at the factory who designed your printers SCREWED UP.  After assembling 100 printers they should run them at maximum speeds taking sharp corners (think printing the points on a star) and make sure the printer never skips and if it does they should lower max jerk and max acceleration.  So if the problem is not hardware then it is firmware.

 

SLICER

It's easy to blame the slicer because a different slicer won't show off the problem.  And because you can indeed fix the problem with slicer settings.  Basically there is one movement somewhere at one particular speed or speed change that is causing this.  If you insist on only playing with the slicer (because it's well, easier than fixing the damn firmware), then here are some settings you might play with:

coasting - this changes the extruder speed to zero suddenly and even though the issue is not with the extruder, it can mess up the planner.  I wouldn't be surprised if turning this on fixes the problem (with coasting the printer slows down more before the coast) or turning this off fixes the problem (firmware realizes it needs to slow down sooner for a corner).

Infill pattern (some patterns have sharper corners)

Speeds!  The higher the speed, the more likely it's going to shift.

Short line segments - not a slicer setting exactly but the code in Marlin is pretty crappy.  If you get too many small angle changes in a very short segment (say change angle by 45 degrees for 0.1mm and then again by another 45 and then by a third 45 all separated by 0.1mm) - then Marlin might think it's fine to run through the corner at full speed as each change in angle is small but in reality they add up to a big angle change in a very short (1ms) time.  Marlin is too stupid to notice this.  This is particularly a problem with prints like "yoda" where you have complicated, rough, surface with tiny triangles in the CAD model.

 

Look: the reason it always happens at the same spot is that there is some high speed sudden movement at that spot.  It might look like an ordinary corner to you but maybe there is a speed change or coasting move or some other sudden shift that is just enough to cause a layer shift (likely an infill pattern).

 

I've seen this many times - it prints almost everything fine and ALWAYS does a layer shift at the same spot.  Yes it can be fixed in the slicer but it's not the slicer's fault.  It's the crappy firmware.  Try printing slower.

 

Marlin is buggy.  I've been through the code many times. It's a hack.

[SergioBA 0]

3 hours ago, gr5 said:

This is caused because the printer is accelerating an axis at a higher rate than the printer can achieve.  That is the guaranteed truth.

 

HARDWARE

This happens on several of your printers but 90% of the time the printer needs oiling (too much friction) or belt tightening (pulleys slip on shaft) or raising the current to the stepper (to get more torque) or lowering the torque to the motor (drivers overheat and shut off for 1ms).  I have seen all 4 of these issues.  Many instances of all 4 on my own printers and other printers.  But for it to happen on all 4 printers seems unlikely unless none have been oiled recently or temperatures in your printing room have increased at the same time as probelsm started.

 

FIRMWARE

It should be impossible for Cura to send a movement that exceeds the printer's ability.  Marlin lets you set max jerk and max acceleration and someone at the factory who designed your printers SCREWED UP.  After assembling 100 printers they should run them at maximum speeds taking sharp corners (think printing the points on a star) and make sure the printer never skips and if it does they should lower max jerk and max acceleration.  So if the problem is not hardware then it is firmware.

 

SLICER

It's easy to blame the slicer because a different slicer won't show off the problem.  And because you can indeed fix the problem with slicer settings.  Basically there is one movement somewhere at one particular speed or speed change that is causing this.  If you insist on only playing with the slicer (because it's well, easier than fixing the damn firmware), then here are some settings you might play with:

coasting - this changes the extruder speed to zero suddenly and even though the issue is not with the extruder, it can mess up the planner.  I wouldn't be surprised if turning this on fixes the problem (with coasting the printer slows down more before the coast) or turning this off fixes the problem (firmware realizes it needs to slow down sooner for a corner).

Infill pattern (some patterns have sharper corners)

Speeds!  The higher the speed, the more likely it's going to shift.

Short line segments - not a slicer setting exactly but the code in Marlin is pretty crappy.  If you get too many small angle changes in a very short segment (say change angle by 45 degrees for 0.1mm and then again by another 45 and then by a third 45 all separated by 0.1mm) - then Marlin might think it's fine to run through the corner at full speed as each change in angle is small but in reality they add up to a big angle change in a very short (1ms) time.  Marlin is too stupid to notice this.  This is particularly a problem with prints like "yoda" where you have complicated, rough, surface with tiny triangles in the CAD model.

 

Look: the reason it always happens at the same spot is that there is some high speed sudden movement at that spot.  It might look like an ordinary corner to you but maybe there is a speed change or coasting move or some other sudden shift that is just enough to cause a layer shift (likely an infill pattern).

 

I've seen this many times - it prints almost everything fine and ALWAYS does a layer shift at the same spot.  Yes it can be fixed in the slicer but it's not the slicer's fault.  It's the crappy firmware.  Try printing slower.

 

Marlin is buggy.  I've been through the code many times. It's a hack.

Everything you say is reasonable, but my jerks and speeds are really conservative.

Both of my 2 ender 3 run DUET 2 motherboards, so I wouldn't blame the firmware or electronics. The other ones run custom marlin.

 

Mi newest printer is the Artillery X1 and it happens almost since de first day.

My question is why it started to happen this year, in all printers, when I had been printing for 3 years without changing the firmware or slicing settings and never had a problem.

 

I even use jerk/acceleration control with really low values.

 

As I said before, slicing again the same file exactly the same way fixes the problem.

As an example, I printed the same gcode 4 times, getting the exact same layer shift the 4 times. Then re-sliced it and printed 100 face shields without a single problem, non stop.

[DarRAven 4]

deleted

Edited May 9, 2020 by DarRAven
deleted

[SergioBA 0]

7 hours ago, DarRAven said:

You can tighten the belts

 

if it doesnt work try adjusting the currents of the motors (you should make an advanced search it wont burn all the system but if you adjust it false it will shorten the life of drivers and motors)

 

if it wont work too try to put a fan top of your mother board heating at the motor driver circuit can cause layer shifts too

Please, read my first post.

Motors are cool, drivers has fan, pulleys and belts are OK.

If I try to force Y and X axis by hand when printing I have to apply a huge amount of force for it to skip steps.

[gr5 2,172]

15 hours ago, SergioBA said:

My question is why it started to happen this year, in all printers, when I had been printing for 3 years without changing the firmware or slicing settings and never had a problem.

It could be temperature or lack of oiling but most likely some new feature in cura is running up against all of your printers.  There are a tens of thousands of non-ultimaker printers using the latest Cura (for example Creality printers) without this issue.

 

But there are 400 parameters.  If you go through them all I think you can pick out the 20 or 30 ones that could cause your issue.  You probably know this but did you know there are about 7 different printing speeds in cura?  You have to be aware of all of them.  I listed a few candidate settings like infill pattern.  There's also one - something about resolution - where Cura throws away some line segments if they are too small to matter.

 

Is it possible that you are printing the same model on all your printers due to covid19 assistance?  Like I said earlier, some models have too many sudden angle changes in very short distances.