Holes with PETG-HF

[GuiltySpark343]

Hi, I am having problems with the new Bambu Labs PETG HF. I get weird holes on some places of a big thread I’m trying to print - please check the pictures. I still have some regular Bambu Labs PETG Basic left, which doesn’t cause this problem.

Does anyone have a recommendation on how to fix this for PETG HF?

Printer model used:
Bambu Labs A1. Tried a 0,4 mm stainless steel nozzle with about 200 hours in it as well as a brand new 0,4 mm hardened steel nozzle. Didn’t make any difference.

Slicer settings used:
Default Bambu Labs preset “0,20 mm Standard @BBL A1”
Filament type was set to default “Bambu PETG HF” or “Bambu PETG Basic”, depending on what was fed to the printer.

Type of filament used
Grey Bambu Labs PETG HF and white Bambu Labs PETG Basic. Both dried for 16 hours at 65 °C in a Creality filament dryer. Stored in and printed from an air tight enclosure with fresh dessicant.

PETG HF with holes on the underside of the thread (lower portion of the picture was touching the build plate while printing)

PETG Basic without holes

PETG HF with “crack”

PETG Basic without “crack” (altough not perfect)

3mf-File of the thread in PETG HF. Only difference for the PETG Basic part is that the material was set to the default “Bambu PETG Basic”.
joystick_mount_lower_adapter_v7.0_test_PSta_MBaPETGHF_N4.3mf (186,4 KB)

[DeVerm]

You need to calibrate the filament for your printer/nozzle.

[GuiltySpark343]

Thank you for your advice. I always do the recommended Bed Leveling and Flow Dynamics Calibration before each print. I also did the “Motor Noise Cancellation”, “Vibration Compensation” and “Auto Bed Leveling” that can be started via the printer’s touch screen.

But I assume you mean the “Flow Dynamics Calibration” and the “Flow Rate Calibration” that can be started via Bambu Studio. I din’t do this before, so now I did.

Flow Dynamics Calibration was pretty straight foreward:

I went with k = 0,035.

Then I did the Flow Rate Calibration. I must say that i found the result confusing:

Does this look normal? “5” had the smoothest “inner” surface, but the outline looked terrible. “-10” is the only one with a barely acceptable outline, but the “inner” surface has gaps between the lines. The corners look terrible throughout.
I went with “5” as i think this test is about the “inner” surface. This increased the flow rate from default 0,94 to 1,034. The following flow rate fine tuning didn’t change that.

The resulting thread looks better than with default flow rate. the holes above the thread are still there, but much smaller.

For comparison, here is the original with flow rate 0,94 (before calibration)

You can see the gaping holes just above the thread.

Unfortunatley, the flow rate increase comes with a serious downside when looking at top surface quality. Here is the XYZ-Cube with flow rate 1,034 (after calibration)

The surface is all rough with zits and blobs everywhere.

Here is the result with the default flow rate 0,94 (before calibration)

Much smoother surface.

I could to some more experiments and see how much i can turn the flow rate up before the XYZ-cube’s top surface starts to become ugly. But I don’t like the idea to have to decide wether to have no blobs on the surface or to have no gaping holes on the outline of my print. Is this really how it is? Feels bad.

[EnoTheThracian]

@GuiltySpark343

But I don’t like the idea to have to decide wether to have no blobs on the surface or to have no gaping holes on the outline of my print. Is this really how it is? Feels bad.

In “Quality” “Advanced” settings, you can set a dedicated “Top surface flow ratio”.

[GuiltySpark343]

Thanks for the suggestion. Unfortunately I can’t see this option in my Bambu Studio v.19.5.51.

What slicer and version is your screenshot from?

[EnoTheThracian]

The same. You probably need to enable Dev Mode:

[GM_Frank]

I think we’re experiencing the exact same issue. This is my post

Does your holes appear right after printing an overhang wall? from your model it seems so.

[GuiltySpark343]

@EnoTheThracian: Thanks for the advice. It worked, switching dev mode on made me see that option.

@GM_Frank

The pictures of the threads in my post made on 13th Sepember were shown upside-down, so whats above the thread in the pictures was below the thread during printing.
I just checked the order in which the lines are printed and it turned out that the area that had holes in the surface were printed directly after the thread. It really seems like i’ve encountered the same effect like you did.

Now I checked the printing speed. Doing the overhang, the print head slows down to 10 mm/s and then accelerates very quickly to 200 mm/s. Flow during overhang is down to 0,78 mm3/s and then increases very quickly to about 18 mm3/s (hard to read that color scale). It seems like in reality the flow can’t increase quickly enough to handle such fast acceleration. This also explains why drastic flow increase (what I did) or drastic speed decrease (what you did) help.

The following pic shows the flow rate. The perspective is from below, looking through the build plate.

It seems like I have an awful lot of tweaking to do. 1.) increase flow rate to make holes after overhangs go away. 2.) decrease top layer flow rate to keep a nice surface. 3.) adjust x-y-hole and x-y-contour compensation to keep dimensional accuracy. This will still leave me with lots of issues because of the flow rate being to high in general.
This is really bad. It shouldn’t be that way.

I don’t have those problems with PETG Basic and I find it terrible that Bambu Labs got rid of PETG Basic only to replace it with holey-mc-hole-hole PETG HF. Time to source my PETG from a different brand i guess.

[GM_Frank]

Try this, change the smooth coefficient to lowest setting (1). This seems to print better for me, as it smooths out the speed change. And it does not slow down the print time very much.

[GuiltySpark343]

@GM_Frank This is so awesome! Thank you for your advice. This fixed the problem for me.

Left side: Smooth coefficient 80
Right side: Smooth coefficient 0,1

Even at smooth coefficient 1 I had some small holes remaining. I had to go down to 0,1 to make them all disappear.

A lot of layers now have this flow distribution (speed looks the same)

The accelerations and decelerations make sense to me.

However, there are also a lot of layers that look like this:

Here we have sudden decelerations for some reason. Flow goes down from about 7 to 0,7 in an instant. This will show on the print, as seen later. The slicer misses the point where it should start decelerating but it keeps on accelerating (or keeps the speed constant, hard to tell from the color plot). Then, when it comes to the overhang, it does an abrupt deceleration.

This also shows in the print.

The defect pattern on the phyiscal part matches exactly the abrupt deceleration pattern that can be seen in the slicer:

The smoother the color changes from green to purple, the smoother the deceleration is. some layers have a really long purple streak (smooth deceleration), while some layers stay grean until they suddenly change to purple).This sudden change in flow can’t be done in the real world, so I get a blob on my printed part.

Is this a bug in Bambu Lab? I can think of no reason for doing sudden decelerations every few layers while I specifically asked for smooth ones.

[GM_Frank]

Great findings! I already have a support ticket open and would like to see how they say about this. The rapid acceleration after setting smooth coefficient to 0.1 seem to be a bug

I also submitted a GitHub issue.

@GuiltySpark343 are you able to share a simplified model that can show the bug of rapid acceleration?

[GuiltySpark343]

I simplified the model somewhat by stripping away unneccessary material. Total print time is now under 15 minutes and it uses less than 2 g of filament. I also created two pictures of it, clearly showing the issue and its consequences.

3mf file:
joystick_mount_lower_adapter_v7.0_test2.3mf (43,8 KB)

Does this help on your quest getting this bug fixed?
Tell me if you need anything else.

If you feel not taken seriously enough by the devs, let me know. I could also file a bug report.

[GM_Frank]

I submitted the issue and it was assigned. But I don’t see people working on this. Maybe you can reply to the issue or submit a new one to make them take it seriously : )

[Lenyo]

Wow, great stuff guys. I’ve just gone through the same with an organic type part with many overhang walls in PETG-HF. Looking at it in ‘speed’ mode evidenced that the jockeying between overhang levels was changing the speed abruptly and leaving a very poor surface. Since the overhangs spanned the gamut from 10% to 100%+ overhang wall the only solution I found was to set them all to the lowest value, increased print time by about 10%.

The part (manifold):

Default speeds:

With speeds fixed to 15mm/s

I did play with the smooth coefficient a bit but could not get smooth transitions to show in speed mode, I only went as low as 10 though. It seems this brand new feature may only apply to certain line type transitions.

Too lazy to take before and after of the actual part but just imagine a wholly torn surface on the inside of the manifold versus a nearly perfectly smooth surface.

[GM_Frank]

try 0.1 smooth coefficient. It should give you a visible smooth speed transition.

And if still not, then it’s related to the bug, feel free to bump the thread to raise attention for developers

[Lenyo]

It does look a good bit smoother, coloring makes it hard to tell for sure, I may have to print it to see the result.

[GuiltySpark343]

@GM_Frank
How can i “reply to this issue”? Where did you submit this issue? Could you provide a link?

[GM_Frank]

This is the GitHub link I posted previously:

Smooth coefficient not working all the time. Abrupt speed changes still exist

[GuiltySpark343]

Sorry, my bad. I commented on your github issue asking for a estimate on when we can expext a fix.

I saw that someone was assigned to this issue and he agreed that this is something that needs improving. I take it as a good sign.