Solved Very stubborn print quality issue

[Vinculum]

Here's the deal. I have several very large (900 * 1200 * 1500mm) custom CoreXY machines. They have been running for 3,5 years non-stop 24/7.
One of them was taken out of service to get an upgrade and do some testing.
Just before I did that I wanted to print a file as a pre-upgrade reference. But it has some very severe lines in the print. See exhibit A:

I couldn't find the issue so i just went and did the upgrade hoping it would dissapear with it.
What I did:

Replaced:

• X & Y motors.
• Belts (now GT2 9mm original Gates)
• Pulleys
• All bearings
• New X rails and carriage
• New extruder (Sherpa with 0.6 CHT nozzle)
• New electronics: Duet 3 6HC, 3HC and Pi4 SBC
• 4 new Z-motors (geared 5:1 nema23)
• All new wiring

Lubricated:

• All rails & carriages (All are MGN15 from HepcoMotion)
• Ballscrews
• Everything that moves....

Did the same print and here is exhibit B:

I stopped the print after I saw that.
After that I did a lot of testing. Tried everything, upped the microstepping, changed currents etc. But nothing.
Since I replaced the entire XY gantry I assumed it was something in Z because of the regularity in height and because that was mostly untouched. Checked the ballscrews, checked motors and couplers, lubed everything like crazy. It runs soo smooth but with the same issue still. What you should now is that every (geared) Z motor is coupled to a ball screw and the flange is coupled to a lineair rail that locks it in a straight line. The screws are straight (and have been for the past 3,5 years) but that doesn't matter because their movement is restricted any way by the rails. Again, as it has been working al this time and still does on all the similar machines that are still running.

Printed with and without mesh bed leveling made no difference. I can't find it!
Then I tried to use a different layer height. I present exhibit C :

Same simple cube (200x200x200mm), same settings & speed, only the layer height is different. As you can see 0.15, 0.2 & 0.3mm.
The stripes change!! I might be mistaken but if it is a mechanical issue where something happens every X mm than it would not change with a different layer height, right?
The gap between the lines is about 2mm for the 0.15 layer height, 2.5mm for 0.2 and 3mm for 0.3mm layer height.

I am out of ideas and out of hope. Please help me!

[rjenkinsgb]

Possibilities I can think of are either a jamming reel feed, or dodgy hot end thermistor?

[engikeneer]

@hbm-3d how stable is your bed temperature, and are you using PID or bang bang? Given that the lines seem to be time-dependent, not z-dependent, those are the first things I would look at.

[rjenkinsgb]

I considered that, but looking at that first full torso print, the lines are the same all the way up. Bed temperature effects seem unlikely to be that even at the full height from the bed.

Possible if it's a heated chamber rather than bed alone?

[Macgyver]

if you do a print that is say 20mm square but 100mm high, and then do a similar print with the same layer height but do it 50mm square and 100mm are the layer lines the exact same on both prints? Goes to checking cooling and temp control.

My gut on it as well was Z-axis issues as you had said the problem existed before the upgrade and is the same after the upgrade. So it would have to be something that is unaffected by the upgrade. You had stated that you never really touched the Z-axis and to me, it looks like a weak coil in your steppers.

To ask an odd question is the result the same on both sides of the print, It looks like it is, but just want to confirm the squashed layer is uniform on all sides or is the layer deformed at different layers on opposite sides?

[Vinculum]

@rjenkinsgb
Well the bed is PID tuned (also tried BangBang) but seeing that it is 900x1200x10mm it doesn't fluctuate even if i wanted it to.

Did pid tuning on the hotend as well but as said, it did this also with a E3D volcano (now dragon).
There is no heated chamber and room temperature is regulated and stable.

[Vinculum]

@macgyver
I wil try that (will be monday, day off tomorrow).

My gut is also Z since it was largely unaffected. But the motors were replaced as well so a defective coil can't be it since it was with the old motors and with the new.
Also the bed hangs on 4 independent motors so 1 defective motor would tilt the bed resulting in uneven lines, not the consistent ones i have now.

[Dad003]

what is powering the Z axis ? belt or leadscrew or ballscrew ? this look like a regular pattern , you might have a leadscrew nut wore out or a bent screw . a bent screw will do that everytime it complete a turn

[rjenkinsgb]

@hbm-3d
Are the lines parallel or spiral? It's difficult to tell from the photos, but I think they are parallel?

[Vinculum]

@rjenkinsgb They are parallel.

[Vinculum]

@dad003 above I said that the ballscrews are locked in place by linear rails. Also if that were the case than it would not change with the layer height is my guess.

[rjenkinsgb]

What do the inside of the square prints look like - flat, ridged with the ridges matching outer ridges or ridges alternating in & out??

[theruttmeister]

@hbm-3d given that it seems to be an error based on layer number, my guess would be some sort of compounding error in either firmware or maybe mismatched slicer settings.
Steps/mm on Z or E causing some sort of trailing offset.

I'd start by trying to confirm that the machine really is doing what is commanded in Z and E.

It's that or you have really really bad CPE on the ballscrews.

[oliof]

I know this sounds weird but it could be drag on the filament spool. I recently had this with a cardboard spool on a printed spool holder -- just too much friction and the filament needed a good yank from the extruder to move. If you're not using a bearing based roller, it might be worth looking into one. (would also explain why this is less expressed with 0.3 layer height because more material being pushed means more continuous pull on the spool).

[Vinculum]

@rjenkinsgb As far as i can see they are matching the outside.

[Vinculum]

@oliof Good idea but the spool holder is the same as before it had issues.
Also, it is not bearing based but it is also guided all the way so there is no yank, the tension is fairly steady throughout.

[Vinculum]

Ok the plot thickens. I printed a simple square, no infill, just 3 perimeters @ 150mm/s with 0.15mm layer height.
The only difference is the size, 100x100, 200x200 & 400x400.

Here is the result (forget the writings):

I'm lost...

[oliof]

what extruder and hotend are you using? bowden or direct?

[Vinculum]

@oliof Direct. Phaetus dragon with bondtech cht nozzle and sherpa extruder with ldo motor.

Ok here's what I think now. It must be extruder related somehow. It isn't the Z-axis or it would not change depending on layer height or even object size as it does now.
It is definitly something driven or mechanical since it is way to consistent to be something like material unevenness.

The main indicator is that the pattern size increased by object size or vice versa. The smaller the object, the larger the gaps become. Bigger object, so more material extruded, is lower interval. It almost looks like it is overextruding every X meters of filament.

[oliof]

please check that the drive gears on the sherpa are still aligned to each other and inline with the filament path (I've seen some lose position elsewhere which can cause weird issues). Also make sure your hotend fan is still working correctly. Since you have identical printers next to this, maybe switching out the extruder and hotend from another one to this is easier/faster to ensure they are or are not part of the issue here.