Darker layers always on the same layers

[redslifer 2]

Hi there.

One of the huge problems I have and I finally wanted to share here too, since I saw much other people with the same problem using cura recently, is that there are darker layers always in the same spots of a print.

 

In this case, the benchy, the darker layers are just the layers under the first railing layer, starting from the end of the letters on the back of the benchy.

The darker layers are also a bit overextruded in all the directions and they are also very weak. The print usually breaks, or splits, on that section just when removing it from the bed.

I tried different materials, colours and settings without any difference, more or less they all become darker and split in that section.

I will post the gcode so anyone can help me determine what hapens precisely on those layers that's different from the others. Even just the speed, coasting or whatever can affect it..

Thank to anyone who'll help here.

 

3DBenchy.gcode

 

 

Edited January 23, 2019 by redslifer

[baxi 2]

I have the exact same problem. I did a lot of testing with it (Cura 4.0).

This is a layer shift of 1 wall-width size.

It shifts from the end of the benchy to the front.

Until now I was not able to find a solution for this.

In preview and reloaded gcode there is no shift.

But it is caused by the lowest layer of the top which is in the back of the benchy.

By changing the number of top layers this line moved exactly as expected.

Turning the benchy 90° also changed nothing. So it's not an axis problem.

Many people have the same problem.

Knowing what the problem is I looked to other prints I did with tops laying inside the model, and they had the same problem.

So this bug is not a benchy specific problem.

Benchies printed with other slicers also don't have this problem. It's Cura specific.

[burtoogle 513]

My guess is that you are getting underextrusion following long un-retracted travel moves. Try setting the Max Comb Distance With No Retract to 20 or similar.

[baxi 2]

TY I will try it, but I am sure that it is not underextrusion at all. The extrusion is quite good, just the layer is shifted. 

[gr5 2,172]

under and overextrusion look similar like layer shifting if it's only 20% underextruded or 20% overextruded.  Most things look pretty good if the under or over extrusion is < 20%.

 

Note that Z axis issues (Z axis doesn't move consistent distance every time) can also cause horizontal lines in a print.  But this looks different.

[baxi 2]

I changed comb, no change as expected.

I calibrated the printer in many different kinds.

I can for sure exclude any axis or extrusion problems.

The full layer shift here is proven, so is the top layer as reason.

I printed already a lot of technical parts with now "lower top layers", and they all came out amazingly perfect.

Including Screws and nuts which work perfect without any need to refinishing them.

[burtoogle 513]

Hello @baxi, please attach the gcode file that exhibits this problem.

[baxi 2]

1 minute ago, smartavionics said:

Hello @baxi, please attach the gcode file that exhibits this problem.

 

I will, but later.

Right now I try to construct some part in which this problem will be shown better.

And smaller, easier to do research with.

I will then post STL and gcode here.

[burtoogle 513]

30 minutes ago, baxi said:

 

I will, but later.

Right now I try to construct some part in which this problem will be shown better.

And smaller, easier to do research with.

I will then post STL and gcode here.

 

Thanks. Actually, the STL isn't much use on its own. Far better is to save the project and attach the .3mf file as that has both the model and the settings.

[baxi 2]

1 minute ago, smartavionics said:

 

Thanks. Actually, the STL isn't much use on its own. Far better is to save the project and attach the .3mf file as that has both the model and the settings.

Ok I will. I always store the .3mf files for all tests.

[baxi 2]

Until now I could not create a good square sample.

So I attach the benchy files. 

It's 4 files from 2 tests. Test no. 3 (very slow 1st layer to get the first layer very good).

And test no. 4 which only changed parameter is the thickness of the top changed from 10 to 3 layers.

I got the expected result, that the shif-line went up ca. 0.9mm.

 

#4_F#4_as#3_3TopLayers_3DBenchy.3mf #4_F#4_as#3_3TopLayers_3DBenchy.gcode #3_F#4_noCoasting_flow96_ILS10_3DBenchy.3mf #3_F#4_noCoasting_flow96_ILS10_3DBenchy.gcode

[burtoogle 513]

Hello @baxi, thanks for the files. I still think that you are likely to get underextrusion due to the long un-retracted travel moves so please try setting max comb distance with no retract to something like 20. Also, don't use the sharpest corner z-seam alignment as it rarely does what you want (model dependent). Instead, use the user specified location to fix the z-seam where you want it.

[burtoogle 513]

Re-reading @redslifer's words, if there is delamination then that would suggest too much cooling fan or too low temps.

[baxi 2]

I will try the above soon and give feedback.

I don't think that I have too much cooling as I still have a very small fan on it. In opposite I still have the typical marks for too small cooling at the benchy and other parts.

 

Meanwhile I made the benchy with 0.3 line width. This had a lot of positive effects on other parts of the benchy, but esp. this shift got much worse. In fact it was the only part which got worse.

Edited May 1, 2019 by baxi

[baxi 2]

So printed 2 tests.

#9 MaxComb 20 + Z Seam User Specified --> The layer still shifted, but went up to highest layer of top.

#10 MaxComb 20 + Z Seam Sharpest Corner --> Layer shift like before in 9, but print quality over all got very bad.

 

So I will take out MaxComb again and try so overextrusion.

[baxi 2]

The overextrusuion test had a positive effect only one one small surface at the front of the benchy. Everything else got worse. There is no effect on the layer shift, still there.

 

The next tests will be with over infill. Maybe this changes something.

[baxi 2]

Changing the infill pattern and the top/bottom pattern did not have any impact on the shift.

[baxi 2]

Tried with 200° instead of 210°. Zero effect on the shift, made the over all print quality worse.

(210° seems the best on my temp tower for this filament..)

[baxi 2]

So the good news is, that I was finally able to remove the layer shift in the benchy with test print #15.

The bad news is, that it was sliced with Slic3r...

But this finally proves, that this is a Cura problem, and hardware problems can be excluded from the list. And as I tested so many different parameter settings, I would finally say the chance that this is a bug is very high.

I also noticed, that all other but 2 minor line problems (including 2 other shift locations) are gone in the #15 print. As it was my first print with Slic3r at all, and I did only transfer the Cura profile to Slic3r, the overall print quality is ok but worse than the benchies made with the calibrated Cura.

 

So is there a way to do a bug report about this?

 

[burtoogle 513]

27 minutes ago, baxi said:

So is there a way to do a bug report about this?

 

https://github.com/Ultimaker/Cura/issues

 

Please make sure you fill out the issue template and provide sample .3mf and gcode files, etc.

[neotko 1,417]

This an issue of any fdm. At 0.15-0.2 layers the time at that point shortens and it’s a small loop followed by different size infill. It’s easy to understand if you check the gcode preview what happens there. Since there’s no slicer out there that can compensate time va heat va flow, the temperature for that 2 layers changes making the print show a visual defect. 

 

Its the same effect that happens when a small island is detached from the model, as soon time and head stays more time on an area the difference from flow/temperature makes a visual defect. The solution is to print everything slower/cooler, that’s why this doesn’t show as much on 0.06-0.1

 

Also some materials show more this than others just because at different mm3 the filament stays ‘hot’ more time, shifting the color. So as soon the filament prints with delay (at that height suddenly it has 4 jerk slowdowns (max speed to change direction) the filament stays more time on the hotend, so is actually hotter than the filament of the lower layers that does suffer less jerks because the nominal extrusion speed is easier to achieve).

 

Chechk your gcode with a good analyzer like https://www.gcodeanalyser.com/ and you will see a big slowdown on the affected area

[baxi 2]

@neotko Nice tool thx.

But I see nothing in there.

On 5/4/2019 at 3:56 PM, neotko said:

shifting the color

It is not shifting the color, it is shifting the wall 1 line width to the front.

 

On 5/4/2019 at 3:56 PM, neotko said:

This an issue of any fdm. At 0.15-0.2 layers

I print 0.12

 

On 5/4/2019 at 3:56 PM, neotko said:

Since there’s no slicer out there that can compensate time va heat va flow

As S3D and Slic3r don't have this line shifting problem, we can exclude this topic as reason.

[neotko 1,417]

27 minutes ago, baxi said:

@neotko Nice tool thx.

But I see nothing in there.

It is not shifting the color, it is shifting the wall 1 line width to the front.

 

I print 0.12

 

As S3D and Slic3r don't have this line shifting problem, we can exclude this topic as reason.

 

You are forgetting an important factor when pointing out the slicer. Cura has control over acceleration and jerk, check a print done with one slicer and other doesn’t prove a thing. Speed vs temperature is the reason of that visual error. I was giving examples of layer heights as a guide, not a recipe. Cura is more focus-based on click-print recipes, but every object has his needs. Just print 3 benchys at the same time and see how at least 1 is perfect, and possibly 1 is meh and 1 is bad

[baxi 2]

I printed now 18 with Cura and all have the same issue.

On 4/30/2019 at 6:34 AM, baxi said:

But it is caused by the lowest layer of the top which is in the back of the benchy.

By changing the number of top layers this line moved exactly as expected.

 

I can very exactly predict where this shift will be.

And I changed almost all parameters in many ways.

 

Acceleration and jerk have been off in first 15 prints, so firmware

values were used. Slic3r used the same and made no shift.

 

Right now a 2nd test with more lower speed, acceleration and jerk values is printing.

 

If this has no effect, I will do some test with 1 layer top and one with zero.

Maybe this will deliver some usable info...

 

And after I know where and how to look I found these layer shift problem at lower top surfaces at almost all of my prints. Beside this problem, the prints are surprisingly perfect considering that my part cooling is still a crap. I print mostly tech parts in which you do not really see this problem. It is the special shape of the benchy hull which makes this problem very clear visible and measurable.

Edited May 5, 2019 by baxi

[UbuntuBirdy 56]

On 5/1/2019 at 9:12 AM, smartavionics said:

Re-reading @redslifer's words, if there is delamination then that would suggest too much cooling fan or too low temps.

Could it be the height where the heat from the bed no longer reaches the layers but the fans are at 100% (perhaps the fans reach at exact this layer 100% and the temp at the nozzle drops down until the heat element can correct the temp) and their air is reflected from the bed so the layer is cooled twice?