Optimal live-z setting not the same everywhere on smooth bed ?
 
Notifications
Clear all

Optimal live-z setting not the same everywhere on smooth bed ?  

  RSS
Husflid
(@husflid)
Active Member
Optimal live-z setting not the same everywhere on smooth bed ?

As a frequent reader of this forum I am very grateful for all the knowhow I’ve been able to pick up in the steep learning curve since I built my MK3s six months ago.  

The accumulated experience of the members of this forum is just as amazing as their willingness to patiently share it with all the noobs! Thank you to all!

 
I have made many perfect prints and the quality of the MK3s still impresses me - my previous printer was a chinese clone and the difference speaks volumes.
 
Now, I have met an issue where I hope to get some help:
 
Recently I did some maintenance on the printer, including replacing noisy bearings on the Y-axis. After this, I ran a full XYZ calibration  with no apparent problems. Noise was gone and the bed seemed more stable.The prints were also better.
 
But now, starting a print with a relatively large (100x150mm) bottom surface, I've noticed that I don’t seem to be able to find a live-z setting that works for all the bed. I can get a perfect first layer in one area, but harp strings ( nozzle too high) in another area while I get rough surfaces ( nozzle too low) in two corners.
 
I have repeated the built-in z-calibration routine two times, but see no improvement.
 
My impression was that the mesh levelling - I use the 7x7 setting - should compensate for the inevitable minor variations across the bed, but this seems not to work in this case.
 
Attached photos shows my calibration test print:
(Note: The pictures are overlapping, there are total 5 columns and 5 rows of squares)
 
1. Front left
 
2. Front right
 
3. Back right
 
4. Back left
 
 
The squares in the right hand corners are obviously squished too much. And in the center/left there seems to be an area of 3-4 squares showing harp strings, like if the bed is lower in this area.
 
There are no visible dents in the bed, so any uneveness should be very limited.
 
Shouldn’t the mesh levelling be able to compensate for this?
 
 
Anyone suggestions for what might cause this and what would be the remedy?
 
Thanks
Ole
This topic was modified 10 months ago by Husflid
Posted : 24/01/2021 7:34 am
Husflid
(@husflid)
Active Member
Topic starter answered:
RE: Optimal live-z setting not the same everywhere on smooth bed ?

Since I posted this, I’ve found other threads to the same issue -“uneven bed “ and I will be looking into some of the suggested remedies - incl perhaps the Nylock mod.

Still I am wondering why the PINDA 7x7 mesh levelling can’t compensate?

Ole

This post was modified 10 months ago by Husflid
Posted : 24/01/2021 7:40 pm
technogeeky
(@technogeeky)
Eminent Member
RE: Optimal live-z setting not the same everywhere on smooth bed ?
Posted by: @husflid
 
1. Front left
 
 
2. Front right
 
3. Back right
 
4. Back left
 
Thanks
Ole

  

I have a Prusa MINI+ and used a similar technique to troubleshoot issues with the first layer. I ended up printing a single sheet across the entire bed, 1 layer high with a filament that made issues easy to see. I found that I could increase the print speed after the nozzle went away from the corners, to 200% which saved me time.

In any case, for certain squares in particular: (front right, bottom left square) and (back right, bottom right square) - the problem you are having looks to me *very* similar to the problem I solved and through my investigation I concluded that it wasn't basic Z height but instead was Z rigidity that was the problem.

bed bouncing problem

The key realization that led me to believe it wasn't "too low" Z height is that the banding/bulging is not oriented along the direction of nozzle travel. In your case, your filament was laid down from top-left to bottom-right (or vice-versa) which should result in bulges oriented the same way. But the bulges are perpendicular to the axis of nozzle motion.

 

My hypothesis was that the entire bed was flexing or bouncing up and down. I found that the pattern of ridges was more or less repeatable across prints. I took a look at it with a microscope:

microscope image

For me on the Prusa MINI+, the fix was to more tightly screw in the mounting screw/nut on the *bottom* of the bed (the screw facing upward). I noticed that the banding seemed to occur above the mounting points., and tightened those screws first - in this case: (back left, middle left, and back right).

 

After tightening those screws, the problem disappeared and now I only have banding which is *along* the direction of the print head - which could be due to Z height or uneven filament (very likely). You can see the resulting print and how the problem is totally gone; the only changes are I switched to 200% print speed once I get away from the corner and switch back to 100% once I get back to the corner.

fixed, 100% then 200% then 100%

 

I hope this helps.

 

 

Posted : 25/01/2021 2:13 pm
technogeeky
(@technogeeky)
Eminent Member
RE: Optimal live-z setting not the same everywhere on smooth bed ?
Posted by: @husflid

Since I posted this, I’ve found other threads to the same issue -“uneven bed “ and I will be looking into some of the suggested remedies - incl perhaps the Nylock mod.

Still I am wondering why the PINDA 7x7 mesh levelling can’t compensate?

Ole

Re: the bouncing issue as I described and the PINDA levelling - the leveling can't detect it because the shape of the bed is moving while the print is progressing. The PINDA probing is done with absolutely zero contact between the nozzle and the bed, so this issue isn't visible until printing starts.

It does look like you have further leveling issues too, but it's worth a shot to attack the larger problem first.

Posted : 25/01/2021 2:19 pm
Husflid
(@husflid)
Active Member
Topic starter answered:
RE: Optimal live-z setting not the same everywhere on smooth bed (Solved, so far)

@technogeeky

Thank you for your advice and the detailed argumentation. The waves are most likely created though a combination of too low nozzle  and flexing of the bed. In my case however the heat bed was firmly fixed, but through the pressure of the nozzle I believe the whole bed assembly was flexing. The waves were more visible in the right hand corners, where the carriage is only supported by one bearing. When you press down on the corners it is easy to see that there is some flexibility. When the nozzle was adjusted higher the waves disappeared.

 

However, I now believe that my problems originated from the disassembly and reassembly of the bed, that I made to replace the bearings, in a weird combination with a poorly designed model and what looks like a bad quality filament.

What I did and learned:

1. The heat bed was deformed through my lack of care when reinstalling it on the carriage. Using a straightedge it was obvious that the heat bed had a bath tub shape, where the center was appr 0,1 mm lower than the edges. Releasing all screws and taking care in tightening them again in the proper sequence (user manual!) i managed to get close to perfect flat.

2. I then updated firmware and Prusaslicer to latest version and ran the Bed Level Correction test https://help.prusa3d.com/en/article/bed-level-correction_2267   a number of times. This showed that there was a slight tilt still in my bed, but making a Bed Level Correction of +10 on the right and -10 on the left I came to a near to perfect result.

3. I then tried to print my model again but it failed in the first layer: Good adhesion in one side, failure in the other side (and my bed was perfectly clean)!  I got the model from thingiverse, and the slicer did tell me there were some issues, but I corrected them with the NetFab option.

4. I then imported the .stl file to Sketchup to take a closer look, and indeed it was faulty: A lot of internal surfaces, gaps in the outer skin and a lot of crossing lines with no purpose. Most important it was not perfectly aligned with the axes, so I suspected that the flat bottom did not align fully with bed. This was hardly visible but could be something like 0.1 degree off. I couldnt see anything in the slicer indicating a misalignment however. After spending quite some time to clean up the model, I printed again, but I still I had first layer adhesion issues

5. I now realized that my near to perfect test print was made with a grey colour (Prima Easyprint PLA), and I tried to print the model in a white filament of same brand. Doing the 3x3 test print with the white filament failed again, with poor adhesion in several places. I then switched back to grey, without changing any setting and without cleaning the bed in between, and had a perfect print. My conclusion so far is the the white filament has some quality problems. The white and the grey were purchased at the same time , and both spools have been stored in my low humidity box (< 20 % humidity) for a month or more.

6. Printning the model in grey colour was a lot better, although not perfect yet

Summary so far: Check the flatness of your bed, don't assume models from the web are in good condition and be aware that filament quality may differ even if same brand and spec.

 

I will run more investigations to better understand the matter.

The journey to perfection is endless, but the process of getting better understanding is very rewarding

Sorry for the long post.

 

Ole

 

 

Posted : 28/01/2021 10:07 am
Share: