The dancing nozzle problem

[Supramaker 0]

Today I tried to print a kind of grid, several mm thick.

While printing the bottom layers, lines are layed to fill areas that are often small and have narrow corners (between the holes).

So lines are printed one after the other, but as the nozzle approaches the narrow corners, the pattern dictates that the lines get shorter and start cornering around more often. At some point, the nozzle seems to "dance" slowly on a small area. Then I can literally see the plastic flowing in excess, a drop appears. Then the nozzle is moved away going through the huge blob, and drags rests of it around. An ugly picture. Stringing is the smallest concern.

 

 

This is due to the fact that the nozzle effectively moves slower and slower as it executes turns in short order, but appareently Cura does not take this into consideration and does not compensate the flow. The flow simulation confirms: no variation. The same applies for speed display.

 

 

This is not a pure cosmetical issue. With several such layers, the excess material accumulates. The nozzle hits the blobs violently when traveling and even knocks the piece off the bed.

Incidentally, I used 5 bottom layers and 5 top layers, exacerbating this effect (not knowing what would happen).

 

This problem can be mitigated by a very slow printing speed, so that the flow in the problematic spots is not very different from printing at regular speeds then. But I am not happy printing at 10 mm/sec, when I can do it 10x faster at least.

While experimenting, I tried coasting with high values (0.3) and I could indeed eliminate some such spots but only when they happened to be at the end of the line, otherwise they were clearly recognizable. However this is not a general solution. I also tried different line patterns, and even a flow equalization ration of up to 400% - no effect.

 

What is the proper way to tackle this problem? (Not just mitigating it)

 

PS:

Would Klipper be aware of cornering lines? (I don't use it).

[Slashee_the_Cow 272]

It would really help if you could provide the Cura project file (.3mf, in Cura go to File > Save Project). Without it, it's hard to test things and figure out what works.

 

This could be a problem with jerk (how quickly it will decelerate/accelerate at corners) being too high, making it not slow down at corners at all (and therefore not needing to change the flow) - and also some violent shaking of the printer.

 

It's also worth noting that many printers' motherboards only load a few lines of gcode at a time, so if there's a lot of tiny moves close together at a high speed, it might not be able to keep up (which would make it pause momentarily and leave a glob).

[Cuq 198]

Such pattern depends of a new parameter introduce in Cura 5.5 call small  top/bottom surface. Try to change this parameter. But on my pointvof view you have also an issue with your flow. Try to reduce to 96 - 98 % 

[Supramaker 0]

4 hours ago, Slashee_the_Cow said:

It would really help if you could provide the Cura project file (.3mf, in Cura go to File > Save Project). Without it, it's hard to test things and figure out what works.

 

You can reproduce this behaviour with ANY model actually. I have been seen that happening from the first piece I ever printed, but I did not try to troubleshoot it. It was not necessary, as usually the layers are buried later on if the effect is not so pronunciated. So I used to ignore that, things are not perfect in the 3d printing world.

 

It just happens that with this model due to its shape there are many such spots, while at the same time several bottom layers are defined, using the notoriously leaking PETG. 

 

But I will provide a project file, why not.

 

5 hours ago, Slashee_the_Cow said:

It's also worth noting that many printers' motherboards only load a few lines of gcode at a time, so if there's a lot of tiny moves close together at a high speed, it might not be able to keep up (which would make it pause momentarily and leave a glob).

 

No pause, just slower. I observed cautiously many times while printing. Do not expect a Duet 2 WiFi to have that weakness (the idea would be digressive).

[Supramaker 0]

3 hours ago, Cuq said:

Such pattern depends of a new parameter introduce in Cura 5.5 call small  top/bottom surface. Try to change this parameter. But on my pointvof view you have also an issue with your flow. Try to reduce to 96 - 98 % 

 

I also tried larger values of that parameter without success. The point is, this is not a small bottom. It is a small area in the bottom.

 

Don't be surprised to know that the first thing I did was to decrease the flow to 90%. It looks inded as a case of overextrusion. Result: light signs of underextrussion appeared, accompanied by the same blobs. So it got worse.

 

The material flowing at this spots in excess is not accounted for anywhere. It just flows "without permission". Of course a lower flow means a slightly lower pressure, and the blobs would be smaller, in theory. Practically? No.

[Cuq 198]

It's strange that the parameter I mentioned generates paths like your screenshot, should be like that :

 

There's something weirde about your model. But without sharing your project, it's hard to understand your problem.

[Slashee_the_Cow 272]

1 hour ago, Supramaker said:

You can reproduce this behaviour with ANY model actually. I have been seen that happening from the first piece I ever printed, but I did not try to troubleshoot it. It was not necessary, as usually the layers are buried later on if the effect is not so pronunciated. So I used to ignore that, things are not perfect in the 3d printing world.

Reductio ad absurdum: Not sure I'd see the same behaviour if I printed a bouncy ball out of TPU (damnit, now I really want to do that).

 

Plus the 3mf also contains all of your quality/print profile settings so it's possible to see if any of those are affecting it.

[Supramaker 0]

Alright, please find the project file attached, including the corresponding picture. In the picture, the blobs are deformed by the nozzle passing through them.

 

Be aware that in my case, 110% flow is usually the correct rate (e-steps calibration has been done, but using it as 100% would yield underextrusion in many cases).

 

It took some time as I wanted to be sure that my statements still stand and I spent hours doing experiments. No matter how you look at it: In the spots where the bottom layer pattern is "squeezed" and in the connection points of the hexagons (there is a small area there in the middle that wants to be filled by itself, slowing down the nozzle) there is excess flow.

 

But I admit: TPU balls won't suffer from this defects 🙂

 

And I still don't know how to cope with that. 

 

Perhaps I will try other slicers? Rather not, they will have their own quirks, which I will have to learn the hard way again.

 

DancingNozzleTest.3mf DancingNozzleTest.gcode

[Lommm 0]

in fact, the model contains a huge number of errors. They can be partially corrected and the slicer will accept the model. But it makes sense for you to change something in your modeling.  It shouldn't be like this.

11.stl

[Supramaker 0]

3 minutes ago, Lommm said:

n fact, the model contains a huge number of errors

 

I cannot follow, with the best of intentions. 

 

In Cura, using the Mesh Tools, I get a message "The model is watertight" when doing a check.

 

I don't know how to recognize further errors in the model, give my a hint. I would willingly learn to correct them.

That looks to me like tiny  details in the mesh, which would be ignored by Cura (in my understanding).

 

Do you see any link between those errors and the issue I am adressing?

Because no matter what model I use, the depicted behaviour is almost unavoidable: When building a bottom or top layer with a pattern of lines, very often the nozzle will move doing turns in a limited area, without having adapted the extrusion rate to this effectively slower speed, resulting in overextrusion. That happens in the inside of the print, but cannot always be ignored.

[Lommm 0]

12 minutes ago, Supramaker said:

 

excuse me, please. I accidentally answered in the wrong topic, and I don't know how to move or delete it.

[Cuq 198]

Ok so yes Small Top/Bottom On Surface generates this strange paths. And 110% for the flow it's not normal even if it's standard for you.  By trying to regulate under-extrusion, we end up generating over-extrusion elsewhere.  Personally, I think that a 100% flow calibration should be required  (you can have between 98 and 102% more means that you often have another problem ).

 

[Supramaker 0]

47 minutes ago, Cuq said:

And 110% for the flow it's not normal even if it's standard for you.

 

As I said: going down to 90% (which is huge if 110% is the correct value) does not diminish the amount of excess material, on the contrary: The blobs can be seen even more noticeable, as the surrounding is scarcer.

 

We already know: The results of a e-steps calibration are not mandatory to set flow rate. Of course, the number of steps per mm must be set correctly in the config, but there are other factors like diameter of the nozzle and/or of the filament, or whatever else. I noticed that using the 100% defined by the calibration, I was getting holes in models made of just one wall. Then I did a "real life" flow calibration, using approriate models in a set with varying flow rate. The result was 110% and that has proven to be fine until now.

 

It dawned on me that there is no fix for this issue, which happens to hit me severely with this particular piece. I can still do mitigation or tricks, like using just one top/bottom layer with gradient infill combined with coasting and ironing (for the topmost layer), but I hate when the nozzle strips the blobs on the solidified surface, causing vibrations and mechanical stress, not to speak about knocking the print off the bed.

 

Another kind of "solution" would be to define per-modell settings blocks containing the top/bottom layers only with their own parameters, like snailspace speed. Unfortunately, coasting is not available for per-model settings.

 

 The best solution would be to anticipate the flow decrease based on the number or frequency of turns ahead.

 

I am honestly quite surprised that this issue has not been addressed by Cura developers, considering that it is so basic and could potentially affect every print.

 

Seems like most Cura experts are on holidays.

 

I will take a look to Klipper asap.

[Cuq 198]

2 hours ago, Supramaker said:

Seems like most Cura experts are on holidays.

 

I will take a look to Klipper asap.

 

It's thanksgiving .. you are very unlucky !  And Klipper is certainly the best solution for you, I am honestly quite surprised that you have not already switch to this solution, considering that it is so basic and could potentially solved all your issues.

[Supramaker 0]

3 hours ago, Cuq said:

I am honestly quite surprised that you have not already switch to this solution, considering that it is so basic and could potentially solved all your issues.

Touché!

However I am quite new to 3d printing. My self-made printer is about 4 weeks old (building it took 2-3 months, mainly due to 3d printing parts needed), and is still work-in-progress. I want to gather experience without Klipper or other magic first. But yeah, Klipper is one of my many ideas in the project.

[Supramaker 0]

23 hours ago, Cuq said:

And 110% for the flow it's not normal even if it's standard for you.

 

I took your advice seriously and went through the process again.

It turned out that all my flow calibrations were made with small, filigree parts. They do need a higher flow than ascertained by the esteps measurement. But the larger the prints, the closer the flow should be to the "official" 100% rate. So I was biased. I have never printed a piece weighting several hundreds grams yet and if I did, 110% would be too much.

 

In short, you were a couple percent right, depending on the weight 🙂

 

[Cuq 198]

To be honest, I expect to be one hundred percent right, depending of your futur years of experience in 3D printing 😁

[Supramaker 0]

On 11/24/2023 at 6:13 PM, Cuq said:

To be honest, I expect to be one hundred percent right, depending of your futur years of experience in 3D printing 😁

You will have to wait asymptotically long. Or blow yourself very, very large.