[BUG] Delayed extrusion after travel when using 0.6 mm nozzle (and 0.4 mm as well)

[rasssta]

Printer type - [MINI]
Mini

Printer firmware version - [e.g. 4.0.5, ...]
4.4.1+4194

Original or Custom firmware - [Original/Custom]
Original

Optional upgrades - [e.g. Filament Runout Sensor]
Filament runout sensor
0.6 mm nozzle
WiFi

USB drive or USB/Octoprint
Octoprint. But I have the same issue when printing from USB.

Describe the bug
I have an issue with scars and holes in perimeters, that I believe started to show when I switched to 0.6 mm nozzle. I'm not actually 100% sure it's related to the nozzle since the issue only shows in certain prints, so it's not really clear when it begun. But the problem also seem to appear with 0.6 mm nozzle for the MK*, see prusa3d/Prusa-Firmware#2760. Prusa support has been involved in that ticket and referred me to create a new one in this repo.

The issue is that extrusion is delayed after certain travel moves by the print head, which causes scars and holes in the print. I've tried to slice with PrusaSlicer, SuperSlicer and Cura and I get the same problem with all of them. I've also tested to remove the filament sensor but the problem persists. PTFE tubes have been replaced to new ones.

It seems to be mainly related to extrusions that are made after the print head is moved from the left hand side to the right hand side of the sheet.

In the print example (picture attached) I placed 4 towers, one in each corner of the sheet. The tower on the right was on the bottom right, the top tower was on the top right, the left one on the top left and the lower one on the bottom left.

But if I move the right hand towers close to the middle I still get the same issue with them, whereas if I move the left hand towers to the middle they print just fine. So it's not just about the actual placement of the object on the sheet, but rather the combination of placements and the print moves it creates.

The problem also occurs with single objects where the same type of movement is performed, in the print, from left to right on the sheet.

I have made some videos where I've filmed the print head and extrusion wheel and the extrusion wheel starts moving at the exact same frame as the print head is lowered (after z-hop), for the parts of the print where I get the holes/scars. I haven't compared it to the non-hole objects but I would assume that the extrusion should start when the print head is lowered?

How to reproduce
Swap to 0.6 mm nozzle. Put an object in each corner of the sheet which has only 1 perimeter.

Expected behavior
No holes or scars in the print.

G-code
See next post.

Crash dump file
None

Video
None. But phots will be attached.

NOTE: If you are affected by this issue, please print the test gcode that Prusa Support has uploaded further down in this issue #2997 (comment) and post your result(s) in this issue, please include filament name as well.

[rasssta]

[rasssta]

Gcodes:

Cura: OPM_rettest_2.txt

PrusaSlicer: rettest_0.6n_0.2mm_PLA_MINI_15m.txt

[rasssta]

I just changed back to a 0.4 mm nozzle and I get the same error so it's not related to the nozzle size

[rasssta]

I have created two demo/test STLs where each tower is marked with I/II/III/IIII or none (which is center tower, I will call it number 5) in the 5 tower STL.

STLs (rename to .stl):
Rettest_4tow_0.4mm.txt
Rettest_5tow_0.4mm.txt

4 tower test:
Print order: Tower 3, 4, 2, 1
Tower 4 has a lot of gaps

A lot of stringing for tower 2

All other towers are overall okay.

5 tower test:
Print order: Tower 1, 3, 5, 2, 4
A lot of stringing for tower 4

Gaps are now on tower 5 instead

[rasssta]

Reverted to FW 4.3.4+2835. Same issue.

[art-0]

Your problem is related to retraction length. Each type (PLA, PETG) of filament needs a different one. Change the retraction length e.g. to 1

also read about linear advance e.g: https://forum.prusa3d.com/forum/hardware-firmware-and-software-help/linear-advance-for-the-mini/#post-618399

[rasssta]

I'm well aware of both retraction length and linear advance. Both have been extensively tested and tuned without any difference in regards to this problem.

[rasssta]

Adding another STL with 2 towers that is faster to print:
Rettest_2tow_0.4mm.txt

[Prusa-Support]

Thanks for testing and reporting.

For clarity, it seems that the print in the first picture doesn't match the .gcode file prepared in Prusa Slicer.
The Prusa Slicer G-code was not using pre-settings. The retraction is a bit exaggerated and LA is 0.04 (I would expect it to be around 0.12 or so for PLA + 0.6 nozzle).
I don't want such little things to make the problem look even worse.

I could reproduce the problem in similar print scenarios though.

I recommend focusing on reducing the print complexity so that more users can easily reproduce the same print scenario without worries.

For example:

1. Use Prusa Slicer genuine pre-sets, stock 0.4 mm nozzle printer, Generic PLA filament, 0.2 QUALITY print profile.
2. Use rounded shapes to reduce speed and flow alteration around the corners.
3. 0% infill and 1x perimeter for better visibility of the issue
4. Painted seam line towards the center, so that the nozzle won't cross the perimeters during the travels.

test.zip
notes

• objects and perimeters printed counterclockwise, starting with the front-right corner

result

• all objects are affected
• the first object (front-right) is most affected (after travel from-left-to-right)

1st in the sequence

2nd in the sequence

3rd in the sequence

4th in the sequence

If more users can reproduce and confirm the issue, that would be appreciated.

A shaking or deformed hotend PTFE tube could be a concern! It would create more artifacts like this.
Consider cleaning the hotend and reseating the hotend PTFE tube if your results are noticeably worse with the same settings.
Before my test in the pictures I did that, just to be on the safe side. 🙂

Michele Moramarco
Prusa Research

[posen4u]

Thanks for testing and reporting.

1. Use Prusa Slicer genuine pre-sets, stock 0.4 mm nozzle printer, Generic PLA filament, 0.2 QUALITY print profile.
2. Use rounded shapes to reduce speed and flow alteration around the corners.
3. 0% infill and 1x perimeter for better visibility of the issue
4. Painted seam line towards the center, so that the nozzle won't cross the perimeters during the travels.

test.zip notes
...
If more users can reproduce and confirm the issue, that would be appreciated.

A shaking or deformed hotend PTFE tube could be a concern! It would create more artifacts like this. Consider cleaning the hotend and reseating the hotend PTFE tube if your results are noticeably worse with the same settings. Before my test in the pictures I did that, just to be on the safe side. 🙂

Michele Moramarco Prusa Research

I sliced your provided test file on my MK3s, using the Prusament PLA, 0.2 QUALITY and Prusa Slicer default 0.4 printer profile printing with Prusament PLA Blend My Silverness. Note, my printer is running in stealth mode. I tested this twice, first having the model centered on the bed and then move each tower to the corner of my larger printbed.

• Test 1, towers centered on bed with the same distances as on the mini bed.

• Test 1, All 4 towers (placed in their representative corners)

• Test 2, towers moved further out the corners for MK3 size.

• Test 2, All 4 towers (placed in their representative corners)

• Comparison bewteen test 1 and 2

Print files

• nocrossing_mk3s.zip

[rasssta]

For clarity, it seems that the print in the first picture doesn't match the .gcode file prepared in Prusa Slicer. The Prusa Slicer G-code was not using pre-settings. The retraction is a bit exaggerated and LA is 0.04 (I would expect it to be around 0.12 or so for PLA + 0.6 nozzle). I don't want such little things to make the problem look even worse.

Regarding the retraction I don't actually recall which settings I used back then as I've tested so much stuff back and forth by now.

But I was actually a bit surprised that 0.12 is the default LA setting when I started to look into LA, and did the linear tuning test print found here https://marlinfw.org/tools/lin_advance/k-factor.html and got the following result:

0.04-0.05 seems to be the best according to the test.

EDIT: My bad! I was under the impression that the printed values were multiplied x10 for some reason. Will try with 0.40 instead of 0.04.

[rasssta]

1. Use Prusa Slicer genuine pre-sets, stock `0.4 mm nozzle` printer, `Generic PLA` filament, `0.2 QUALITY` print profile.

2. Use rounded shapes to reduce speed and flow alteration around the corners.

3. `0% infill` and `1x perimeter` for better visibility of the issue

4. Painted `seam line` towards the center, so that the nozzle won't cross the perimeters during the travels.

test.zip notes

I printed the test1nocrossing.gcode found in the zip and got the following result with Prusa Mini, 0.4mm nozzle, FW 4.3.4+2835, AddNorth PLA Economy White:

1st:

2nd:

3rd:

4th:

[rasssta]

Another test print, this time with my normal settings (still K0.04 for this one). 99% sure of the print order:

Prusa Mini, 0.4mm nozzle, FW 4.3.4+2835, AddNorth PLA Economy White.

1st (bottom left):

2nd (top left):

3rd (top right):

4th: (bottom right):

[rasssta]

Another test, this time with my usual settings (as in the previous post) but this time with K-value of 0.42 instead of 0.04.

Prusa Mini, 0.4mm nozzle, FW 4.3.4+2835, AddNorth PLA Economy White, M900 K0.42.

1st:

2nd:

3rd:

4th:

[rasssta]

I have confirmed that the same problem exist in FW 4.2.0+2067 and 4.3.0+2756.

[posen4u]

... MK3s tests...

Adding that I am using firmware 3.12.1-5686 on my MK3s. I did another test using a reverse bowden setup as I had it available, and that made no real difference in print quality. However I swapped the filament out for "Clas Ohlson PLA" and ran the same gcode, no issues. So for the MK3 maybe this is only a case of fine tuning the default Prusament PLA profile.

Reverse bowden setup

Reverse bowden result

Clas Ohlson PLA result (not reverse bowden)

[elHatchling]

I Tried the test.zip on a Prusa Mini+, FW 4.3.4. Bought it March 2021.
The only mod is that I am using the "BondTech Heatbreak Prusa Mini".

Prusaslicer 2.5.0, Used 0.20 QUALITY, Prusament PLA profile, 0% infill. More or less everything fromm the .3mf file in the zip.

All cylinders printed fine, with no gaps, but with some stringing.

[Prusa-Support]

Thanks for your feedback.

@posen4u It seems safe to rule out the MK3-series from this issue.

@elHatchling Isn't it that BondTech Heatbreak requires a few firmware parameters and print settings tweakings?

Retraction and Linea Advance setting can make a difference but apparently that wouldn't be all.

In two cases, it seems that the object/elements printed after a "large" travel from-left-to-right is subject to noticiable gaps while the nozzle resumes extruding.
It may be useful to collect more examples of prints made with other Prusa MINI printers in order to confirm the issue and assess its diffusion.

New reports of the test made with other MINI printers, are welcome.
Please include pictures and relevant information about hardware, firmware, print settings and filament.

Michele Moramarco
Prusa Research

[elHatchling]

Regarding the Bondtech Heatbreak. Only thing I did was to use a shorter PTFE tube, according to the instructions from Bondtech. Nothing else have changed. No firmware parameters or print settings tweaked.
Using stock Mini Firmwares when updating and using stock Prusaslicer settings for printing.

[krcm0209]

New reports of the test made with other MINI printers, are welcome.
Please include pictures and relevant information about hardware, firmware, print settings and filament.

I printed the test1nocrossing.gcode supplied by @Prusa-Support on my Mini.

Modifications from a stock machine

1. Swapped to SuperPINDA
2. Swapped to Revo Micro hotend

Versions

Firmware: 4.4.1+4194
Bootloader: 2.0.2
Buddy Board 1.0.0 (1120X017XC0434)

Stats

• X axis: 5.4 km
• Y axis: 4.9 km
• Z axis: 37.1 m
• Filament: 271.6 m
• Print time: 3d 5h

Filament

Prusament PLA Blend Lime Green

Nozzle

0.4mm

Photos

[krcm0209]

I can't really back this up with science or evidence (partly because I don't have the time right now to deep dive on this subject), but I have a feeling this behavior is in large part due to:

1. the internal tolerance of the stock bowden tube on the Mini
2. the transition from a very gentle bend radius at the left-most position to a relatively extreme bend radius at the right-most position causing a pulling effect on the filament due to a changing path length

Potentially relevant or enlightening article for those unawares: https://e3d-online.com/blogs/news/battling-bowden-tube-physics

[ryanklafuric]

Following up on this thread via a reddit post in r/prusa3d.

I printed the test gcode provided and had some fairly gnarly results. As expected, results were worst in the front right. The best result was the front left. I also suspect that this issue is related to the bowden tube, like @krcm0209 mentioned.

I'm going to try changing filaments, and using a different spool holder to see if that has any positive impact.

Photos of all four:

Front right (worst):

Rear right:

Rear left:

Front left (best):

Asides from this test, I've observed this gap affecting prints like this Gridfinity cube:

[rasssta]

I can't really back this up with science or evidence (partly because I don't have the time right now to deep dive on this subject), but I have a feeling this behavior is in large part due to:

1. the internal tolerance of the stock bowden tube on the Mini

2. the transition from a very gentle bend radius at the left-most position to a relatively extreme bend radius at the right-most position causing a pulling effect on the filament due to a changing path length

I'm in the same position, can't back it up with science but I have been thinking about just the same two points. I don't know if there actually is a proper solution to this issue (but only workarounds) as it seems to be related to hardware and concept (bowden).

However, I can see 3 possible directions (feel free to comment!):

1. A function in the slicer which compensates for the movement of a bowden printer (can be for Mini or Universal). Let's say you specify where the bowden tube is located on the printer (right, top, top right, etc.) and then the slicer will build some form of compensation mesh for how much of extra restart length or deretraction to use after each travel depending of position A and position B for the travel (as the problem only occurs after left -> right movements).
2. Pretty much the same as 1, but implementing this in the Prusa Mini firmware instead of the slicer. Will probably have the same end result, but only solve this issue with the Mini and not for other printers using PrusaSlicer.
3. Redesigned bowden concept for the Mini. Instead of having the tube exit the extruder upwards it might be an idea to move the extruder and redirect the exit direction 0-45 degrees downwards, hence being able to use a shorter tube and more direct angle for the tube towards the printer head. Note that this is just my average Joe brainstorming so take it with a grain of salt :-)

Also, thanks to the "newcomers" in the thread for your contribution with test prints! The Reddit thread highlights that this is an actual problem for people so its nice to have more people giving input here in the issue.

[rancor1223]

New reports of the test made with other MINI printers, are welcome. Please include pictures and relevant information about hardware, firmware, print settings and filament.

Michele Moramarco Prusa Research

Here is my result with basically stock Mini+, only with a 0.6mm nozzle. I will repeat this test with 0.4mm in next few days.

Printer type - [MINI]
Mini+ (purchased in 2021/10; 29 print days logged; the PTFE tube was never replaced)

Printer firmware version - [e.g. 4.0.5, ...]
4.4.1+4194

Original or Custom firmware - [Original/Custom]
Original

Optional upgrades - [e.g. Filament Runout Sensor]
Filament runout sensor
0.6 mm nozzle
levendig silicone sock on the heatblock

To generate the Gcode I used stock settings (for 0.6mm nozzle) with the few changes from the provided 3mf file in PrusaSlicer 2.5 for "Generic PLA" (I used eSUN PLA+).

(prints are ordered clock-wise starting with bottom right corner of the bed)
First two on the left might appear connected, but there is a gap, just filled in with stringing.

---

I'm glad this is finally being looked into, I've struggled with it for a good while (although I don't remember this being a problem back when my printer was new).

[xarvh]

Coming from Reddit, I seem to be having similar issues.

Original Mini+, replaced heatblock and using a 0.6mm tungsten carbide nozzle, everything else is original.

• 0.20mm Quality @Mini (modified), the modifications are from the .3mf
• Generic PLA
• Original Prusa Mini & Mini+, but changed Nozzle Size to 0.6mm

[WeaselWasher]

Yeah. Still got issues.

…

On Wed, Aug 21, 2024, 6:55 PM Caysen Mayo ***@***.***> wrote: bump — Reply to this email directly, view it on GitHub <#2997 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFPJPILY5PSFY7DLEHMRRTTZSVAHTAVCNFSM6AAAAAAVAHSBEWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGMBTGUZDKNRWGU> . You are receiving this because you commented.Message ID: ***@***.***>

[BrainWaveDave]

Any updates? I replaced the majority of the components in the original hot end and extruder with no improvements. I then upgraded to a Bondtech extruder, Revo hot end, Capricorn PTFE tubing (tried different lengths), Bondtech PTFE Fittings (3 tubes into 1) and had very similar results, same with centre extruder mod. I have basically ruled out most of the hardware possibilities. The only components that have not been replaced are the hot end fan, extruder motor, or the Buddy control board. In my case, I’m 99% sure it’s software-related. Although Prusa seems to think its hardware related.

After these tests, its still a mystery what's making these printers different from the working ones. Let me know if there is anything else that can be tried.

[rancor1223]

Interestingly, despite having this issue myself (posted my own example prints here as well), I haven't encountered it in ages. To begin with, it seemed to be situational, but I have literally just printed entire set of parts for Voron printer (>1kg of plastic) and haven't encountered single such issue.

Also using Revo hotend and Capricorn PTFE, but that hasn't changed from my test. If I remember later, I will try to reprint the test print, if I can still replicate the issue.

[lumen-novum]

@rancor1223 Are you printing the parts one at a time? How many perimeters are you doing? Have you changed your retraction settings?

[rancor1223]

@lumen-novum No change to retraction, I only print with the default IS profiles (0.4 and 0.6mm nozzles). The Voron parts were printed with Voron recommended settings (4 perimeters), using Polymaker ASA, printed using Prusament ASA profile. For my usual printing I would use 2-3 perimeters, but I also haven't been seeing the issue there.

[posen4u]

In my case, I’m 99% sure it’s software-related. Although Prusa seems to think its hardware related.

After these tests, its still a mystery what's making these printers different from the working ones. Let me know if there is anything else that can be tried.

I believe the cause was determined or suggested to be a (hardware) design issue, potentially fixable by software.

[lumen-novum]

@rancor1223 I think the seam issue has been improved in the newer PrusaSlicer versions, but I've noticed that if you are printing thinner objects, objects that have a long travel distance to the seam, or multiple objects at a time, the seam issue is there.

Therefore, YMMV.

[SimonTratter123]

Yes, I can relate this @lumen-novum It depends on the object an if the start after a travel is a outer perimeter or inner perimeter. When it is a outer, it looks horrible

[jfmmm]

In my case, I’m 99% sure it’s software-related. Although Prusa seems to think its hardware related.

I don't think Prusa ever come out with any suggestion at what the issue could be. If you read the whole discussion here, it's the people that came to that conclusion and from my experiment it was the right conclusion. Prusa participation in this tread was minimal and meaningless.

Prusa would probably prefer if it was software as that would be fixable and probably would already have been fixed. But as it stands now they would have to admit to the design flaw and offer an upgrade kit or something, which clearly they are not in any hurry to do.

[SimonTratter123]

@Prusa-Support please help, because printer is not usable for quality parts (Bambu gets my printjobs currently, which is not what I wanted 😥)

[WeaselWasher]

Yeah. At one point I had thought I had fixed it - but I think it was situational. Because it's back. I had replaced my hot end and ptfe tubing. Went over my extruder and routined all the screws and gears. I don't ever remember this before input shaping. I'm thinking about rolling it back if I can - I would rather have a quality output over speed and the low quality seams I'm getting now. I use it only for prints I don't care about quality for.

…

On Thu, Sep 12, 2024, 7:58 AM Simon Tratter ***@***.***> wrote: @Prusa-Support <https://github.com/Prusa-Support> please help, because printer is not usable for quality parts (Bambu gets my printjobs currently, which is not what I wanted 😥) — Reply to this email directly, view it on GitHub <#2997 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFPJPILNU2I4YWOHLRYU773ZWGTX5AVCNFSM6AAAAAAVAHSBEWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNBWGUZTQMJYGQ> . You are receiving this because you commented.Message ID: ***@***.***>

[GithubUser99999999]

Does anyone know if the MK4S is also affected by this? I don't want to buy a bambulab printer.

[theboyrossy]

I've pretty much given up on the two Mini's I have, they are now sitting gathering dust. I'm now doing all my printing on the MK3 and MK4. I really should get rid of them, but I keep hoping a solution can be found.

[SimonTratter123]

@GithubUser99999999 the Mk4 I sold in January had the seam issue. Not as bad as the mini, but still there. But it had lots of VFA too. Those two problems forced me to buy Bambu instead

[WeaselWasher]

My XL doesn't have this same issue unless you run it stupid fast and cold. Same extruder as the Mk4.

…

On Thu, Sep 12, 2024, 1:29 PM Simon Tratter ***@***.***> wrote: @GithubUser99999999 <https://github.com/GithubUser99999999> the Mk4 I sold in January had the seam issue. Not as bad as the mini, but still there. But it had lots of VFA too. Those two problems forced me to buy Bambu instead — Reply to this email directly, view it on GitHub <#2997 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BFPJPII4LI2ZYUUWLWXW2LTZWH2SXAVCNFSM6AAAAAAVAHSBEWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNBXGE4DCMJZGE> . You are receiving this because you commented.Message ID: ***@***.***>

[jfmmm]

the Mk4 I sold in January had the seam issue. Not as bad as the mini, but still there. But it had lots of VFA too. Those two problems forced me to buy Bambu instead

The MK4 has a direct drive so VERY unlikely to be this specific issue that we found is caused by the kink in the Bowden tube when the printer move from left to right.

There was another issue that was slicer-related that first caused me to find this post because I had it on my XL too. They resolved that months ago. This was probably your issue as I believe it affect all their printer.

Sadly I then found that my mini has this way more annoying issue too 😕

My XL doesn't have this same issue unless you run it stupid fast and cold.
Same extruder as the Mk4.

Exactly, pretty sure you don't have it because they fixed that one pretty quickly.

The reason this one is getting ignored is because it's not fixable in software.

Someone in this thread came up with a script that compensates for the issue, but you would have to calibrate every filament you have and then keep recalibrating as the Bowden tube wears down. IMO it's not something that really fix the issue in a way that could be integrated in the firmware, sadly.

[jfmmm]

Github does an awful job of showing that this is a very long tread that been going on for a while.

The 162 hidden items Load more… button in the middle is pretty easy to miss.

Been following this tread for a while and every few months there's new people that show up with a different theory that were already discredited and no idea of what was found before.

Here the script -> #2997 (comment)

Edit: This seem like a bit of an out of context comment because I was adding on a message from someone else that is now deleted. He was mentionning this script I just linked that solve this issue for some.

[lumen-novum]

Did anyone else try the extruder relocation mod? I tried it and it didn't really make that much of a difference but I wonder if it's because I did something wrong.

[BrainWaveDave]

Prusa made another response on a similar github post, here is the comment. prusa3d/PrusaSlicer#11914 (comment)

@GithubUser99999999 Those two problems forced me to buy Bambu instead

I will probably purchase a Bambu instead, especially if the other printers have similar issues (but less severe). I’m still surprised Prusa is selling the Mini despite being aware of these issues.

[kozross]

Prusa made another response on a similar github post, here is the comment. prusa3d/PrusaSlicer#11914 (comment)

@GithubUser99999999 Those two problems forced me to buy Bambu instead

I will probably purchase a Bambu instead, especially if the other printers have similar issues (but less severe). I’m still surprised Prusa is selling the Mini despite being aware of these issues.

There was a span of several months at the start of this year where a firmware update technically put the Mini out of spec as stated on its product page. Multiple issues with the Mini were only addressed because the MK4 was affected by them too due to sharing a similar board and firmware. Heck, the Mini still has a hardware issue it has had since launch! The fact is, it has always been an under-maintained product full of problems: this is just one more symptom of the same.

[Prusa-Support]

I don't think Prusa ever come out with any suggestion at what the issue could be. If you read the whole discussion here, it's the people that came to that conclusion and from my experiment it was the right conclusion. Prusa participation in this tread was minimal and meaningless.

I don't expect you all to go through 200+ comments but, to be fair, please try that before making such bold statements. 🙏
You may display all hidden comments, a bit for a time, and ctrl + F the keyword Prusa Research to basically track my signature (and comments) and with that, the progress of the issue itself. It may not be much but I hope you may see a hint of commitment there.

As already explained and confirmed by some users, the problem is not visible in everyday printing as the specific requirement is printing distant objects/features, especially if made of one perimeter only.
Unless you have a specific need to print distant objects/features made of one perimeter only, the problem should be barely noticeable, if visible at all - yet there may be ways to mitigate the issue (check previous comments) or users can effortlessly avoid such a very specific print scenario in most cases.

If you are experiencing noticeable print quality issues, that is probably a different problem requiring troubleshooting...

... but please check the PTFE tubes and don't hesitate to contact our Support for direct help.

As a recap, please notice this problem affects Prusa MINI ^only, due to hardware limitations, and manifests in a rather specific print scenario.
For the records, this has little to do with prusa3d/PrusaSlicer#11914 (comment) ❌ but prusa3d/PrusaSlicer#9908 ✔️ as the developers are making considerations about implementing the suggested workaround via slicing post-processing script (unofficial information) -> #2997 (comment).

Michele Moramarco
Prusa Research

[murk-sy]

I would like to bump this issue somewhat.

1. When printing a somewhat full sheet of 3 sets of parts for a specific product in PETG, I can immediately see which one was printed after the travel - only a single part has holes in the seam.

2. Another example is when I'm printing a completely full sheet of dowels in PA. By now I know which exact part to check for fit because there's always only one that's problematic. That part always has bumps on the seam.

3. Another way this issue (I believe at least) is visible is when printing more perimeters. One of the parts in PETG that I mentioned above has a clearly visible difference when printing in X+ (less gaps between lines) than when printing in X- (more gaps between lines).

I've also got an XL and I've essentially stopped printing customer parts on the Minis, only ones for internal use and own products. Obviously there's some general quality difference, but seam inconsistency is something that can be very easily seen, only to be reduced if you "cheat" and arrange the parts in a single line.

To be clear, all the printers are in a humidity controlled room and I've recently replaced the hotend PTFE tubes. That hasn't affected the issue.

The proof of concept post processing script #2997 (comment) uses a linear compensation, which shouldn't be too difficult to implement.
It would require a little bit of code so the compensation isn't active during cold extrusion or when extruder is disengaged.
As far as I'm concerned, if it would be an alpha, it could just have cold extrusion prevention disabled to make things easier to test.

@CZDanol I don't really like pinging people into an issue, but could I ask you to give this a look?

[SimonTratter123]

It looks like I solved the issue. When installing a new Hotend PTFE I really pre tensioned it with a lot of pressure. Then it works. If applying only normal pressure on step 11 I have the delayed extrusion problem: https://help.prusa3d.com/de/guide/auswechseln-des-ptfe-hotend-schlauchs-mini-mini_119449 More manual pressure on Step 11 or open the top Screw more than 360 degrees to get more pressure on the PTFE tube. Please try this, it would be interesting if that really solves the issue :)

[BrainWaveDave]

I tried using Bondtech PTFE tube fittings, but it didn't seem to work, even with an E3D Revo (it has a different heatbreak design). Since there's no specific one for the hotend, Id assume it would rule it out completely. However, feel free to let me know if I am wrong.

[SimonTratter123]

Well, that is interesting. My Mini was collecting dust the whole year, because doing so many tests gave me no better results (even changing PTFE). Now changing like the 5th time by really squezzing the tube in, it works. I will never touch the Hotend again now :)

[murk-sy]

Some more feedback with photos of a part sitting on a floodlight.

I've printed a flat plate with 35 perimeters. It's a bit of an extreme example, but I often print 5 perimeters and the bizarre strength/inconsistent delamination (especially when cleaning with a heat gun) issues I've been having all this time now make perfect sense.

For context, this is the (default) print direction:

Sample: PETG brown, 3 layers, first layer having 1 perimeter, rest having 35 perimeters. The circle on the part indicates bed origin. Infill standard 45°.

Sample printed at 5-10° (issue less obvious):

Sample printed at 0° (bottom and top are parallel to X axis)

I think this should be the clearest indication yet that it's a wack consequence of the bowden shape and not anything else. Note that stopping on the X axis and printing on just Y also causes some issues. The workaround script should then improve the results but not fix the underlying issue, which would probably need some simulations to properly solve.

It should also serve as an easy way to tune fixes by just checking line consistency, and once it looks good, underextrude the topmost layer and see if the gaps consistent all the way around. Likely checkable on a collimated light inspection machine which I'm sure Prusa has lying around in the HQ.

[murk-sy]

Did some measurements with a dial indicator since I couldn't think of an easy way to simulate a bent bowden tube and then measure the max and minimum length in different positions.

If you'd like to test this yourself, here's the quick and dirty model: 2024-12-07 Prusa Mini Bowden Runout jig.zip

I did the measurements in the following steps:

1. Pull away the jig and straighten out the bowden tube
2. Pull on the dial indicator and place the jig into position
3. Slowly release the dial inticator and let it settle with minimum pressure. This is the Initial measurement.
4. Push on the filament (about 0.30 mm) and let it settle back. This is the Push measurement.

Notes:

• On the far left the filament may be slightly bent and can deflect the tip of the dial indicator when pushing.
• Only 1 series of measurements was done, from 180 to 0, and then 180 was checked again to make sure the indicator didn't move.
• Measured on a Mini that is mostly used for PA, so bowden wear should be minimal.

Here's the data:

Position	Initial	Push
0	0,81	0,60
20	0,71	0,51
40	0,57	0,32
60	0,55	0,28
80	0,58	0,20
100	0,30	0,10
120	0,25	0,04
140	0,23	-0,01
160	0,20	0,01
180	0,20	0,00

The numbers seem fairly normal at first glance until you make a graph:

Edit: added inverted graph for easier presentation.

If you start at 180 and move to 0, you will overextrude a total of 0.60 to 0.80mm of filament (walls printed towards the back of the printer)
If you start at 0 and move to 180, you will underextrude a total of 0.60 to 0.80mm of filament (walls printed towards the front of the printer)

Obviously this is far from calibration data, but it confirms my hypothesis: it is wack.

To be clear: Neither measurement directly measures what the compensation should be, because it depends on extrusion pressure. The compensation that would likely work best is probably a completely different line between these two and would likely also be different depending on direction/speed.

[jareware]

I believe I might be experiencing this same issue. The simplest consistent repro I've managed is this object:

There's exactly 1 perimeter. The print head builds each layer counter-clockwise:

As the head travels across the gap between layers, it continues to extrude slightly too long (leaving strings) and waits slightly too long to start extruding again (material missing from the first few millimeters):

This isn't some pathological case that only presents when given a particularly challenging model: material missing from seams is present in all prints, just depending on seam placement it might be less visible.

With slicer settings, I've tried:

• Disabling retractions
• Increasing retraction distance significantly
• Disabling lift during travel
• Lowering speed significantly
• Lowering temperature

With the physical printer:

• Replaced nozzle
• Flashed latest firmware (also tried downgrading temporarily to last December's version)
• Tried multiple filaments
• Disassembled & cleaned extruder gear
• Replaced hotend PTFE tube
• Replaced long PTFE tube
• Did multiple cold pulls
• Tightened screws on hotend
• Checked idler gear tension
• Tried disabling input shaper
• Replaced print fan

Many of the above have been suggestions from Prusa support. Nothing seems to help.

[murk-sy]

Above post inspired me to measure printed wall thickness, which was in the past often incorrectly used, but is actually perfect to investigate this.

A quite simple model with 0,5mm walls and a gap to force seam without painting:

2024-12-09 Prusa Mini Bowden Runout test v2.zip

Measured with digital calipers with consistently the same depth. Still a single group of data, but still as expected.
Printed on a different printer than the original measurements (Mini 2)

X position	Front	Back
10	0,45	0,54
20	0,44	0,54
30	0,41	0,54
40	0,4	0,54
50	0,39	0,53
60	0,38	0,53
70	0,38	0,53
80	0,38	0,52
90	0,4	0,52
100	0,38	0,52
110	0,38	0,51
120	0,37	0,5
130	0,37	0,49
140	0,38	0,48
150	0,38	0,47
160	0,39	0,46
170	0,4	0,45

Important: Model is 275mm and the corners were not measured. All measurements are based on the absolute (printer) position.

Graph (with helpful print direction markers)

Note that back to front and front to back do not have consistent print widths either

I believe this is due to filament settling into a more optimal shape as it prints in a single X position. If that were not the case, X10 and X170 for front and back would be much closer. For Y compensation a large zigzag would likely need to be printed and measured, but I don't even really know how that could be implemented.