Multiple H2D's With Excessive Z Plane Skew Error

Bambu Lab H2 Series Bambu Lab H2D
1

We’ve purchased two H2D’s and need a 3rd asap but are reluctant. We bought these specifically to print large sections of vacuum fixtures that are rectangular in shape with tall vertical walls.

The parts we’re printing are roughly 295mm x 275mm x 275mm. Its important for us to get fairly straight square walls but we are seeing some skew on our first test prints. We ran the V3D Calilantern Mk2 calibration print at the 200mm size and found the same skew results we were seeing in our test prints.

Here are the results:

Bambu H2D Skew
Bambu H2D Skew1902×983 144 KB

The ZY (front to back) seems far off for the height we are trying to print resulting in the top of the part being about 1mm off in the Y direction when compared to the base on printer #1. This was checked with the part base sitting flat on a precision surface using a machinist square up against the wall.

I have contacted Bambu Labs multiple times about this and their reply was:

“We sincerely apologize, but at this time there is no way to resolve this issue. The Z-axis accuracy is determined by the Z lead screws, which are fixed to the printer’s frame and cannot be adjusted.

Currently, there is also no method to improve this issue by modifying the G-code.”

I then requested to warranty the printer through the 1yr manufacturer’s warranty to take my chances on a replacement being better and they said:

“If replacing the machine would resolve the issue, we would certainly be willing to do so. However, this is not a problem that can be solved by a simple replacement.

For a large machine like this, slight Z-axis tilt due to long-distance shipping is within expected tolerances. With current manufacturing processes, we regret to say that we cannot guarantee a machine with a perfectly 100% straight Z-axis.”

This is understandable but my frustration is that there is no way to compensate for the error mechanically or in software. I don’t expect a printer shipped across the world in a box to hold perfect skew but there needs to be some means of tuning this out.

Two of the same printers seem to show the most skew error in this plane so I’m curious about other H2D’s out there.

Does anyone know if the bed can be manually adjusted to tune this out? I’m worried that if I try, either the bed mesh calibration will have too much of a delta or that the controller will overcompensate / or correct my manual tilt.

Any help will be much appreciated!

1 Like
2

Supposedly you should be able to do a belt teeth skip then redo bed mesh calibration completely. I never tried it on the H2Ds though.

image-10
image-10777×1036 59.7 KB

This should allow you to adjust each lead screw freely. But it’s still quite a challenge tbh

3

Hmm, that’s interesting. Sounds like a pain but might be worth a try.

4

Are you sure this isn’t a print bed tramming problem? The “self leveling” that the printer does is anything but “leveling”. It compensates for the Z axis error caused by a plate that’s tilted, but the bottom surface will still be skewed relative to the walls. On the build plate the walls will be vertical and at right angles to the top, but not the bottom. Put the print on a flat surface and it’s going to look like it’s tilted over…

1 Like
Precisely repeating layer drift same height h2d pc
5

Similar issue on the X2C which was extensively discussed in two or three threads and I vaguely recollect someone figured out some kind of a clever solution to workaround bambulab’s closed source and general lack of interest in the topic. Maybe worth a look.

I know it’s not a solution for your H2D, but worthy of note: you can do z-axis skew compensation in Klipper.

Full official guide: Skew correction - Klipper documentation

So… if you happen to also have a klipper printer, maybe do your critical z-axis skew prints on that.

1 Like
6

This sounds worthy as the first step. I’ll print up a dial indicator mount and check the tram after heat soaking to get an accurate reading of the bed plane position!

1 Like
7

Put a model on the build plate that’s just a big square (outline, only). Watch the video feed while the printhead is moving to print it and you should be able to see how much the build plate is moving up and down.

It’s subjective, but if the movement is really noticeable, it’s probably too far out of tram if you’re looking for dimensionally accurate parts.

The more the plate tilts for “leveling”, the more trapezoidal your walls are going to be.

I’d note my H2C could really use a tweak. I can see the plate moving up/down a couple of mm as the printhead moves from left to right. Haven’t printed anything where it’d matter, yet.

1 Like
8

I looked into this a while back with the X1C guys talking about using the M1005 gcode and actually tried it on the H2D to no avail lol. Damn closed-source!

1 Like
9

I think the indicator will give me a real good sense of how far off it is. Having two printers with the same issue ordered 6 months apart makes me think they’re all like this. I’ll definitely report my readings and correction results once I get around to it.

10

I’d like to think they use a fixture when putting everything together that ensures the build platform is square to the printhead gantry. But evidence suggests that for whatever reason there’s a lot of loose tolerances in that assembly step. Otherwise, since the Z screws should never change phase with respect to each other between leaving the factory and arriving at the user’s location, every printer should arrive “balls on”. But if you can see the build plate moving, it’s not even close.

1 Like
11

Are you seeing the YZ skew on the top surface, bottom surface, or both? If you see it only on the bottom with a square top, then yes do bed tramming, if you see it on the top or top and bottom, then the issue is related to the alignment of the printhead y axis and lead screws. This will be nearly impossible to fix.

13

We bought these machines at work, brand new H2Ds, in this case they all arrived on a pallet from a truck, so it shouldn’t have been dropped in shipping.
I cant upload the image, but the best so far has been -0.19 in ZX and +0.16 in ZY on the same machine.
Not sure what we are going to do at this point, this machine was purchased for its large build volume, and the assemblies planned to print with this machine wont fit together with this much skew.

From what I can tell, the H2D is not a suitable product for anyone trying to print larger functional parts at this time until BambuLab implements some type of solution to correct the issue in software or at its source.

14

Welcome to the forum.

There are limitations to FDM 3d printing due to filament shrinkage, filament calibration and the technology itself. Most limitations can be accommodated with tolerance built into the cad model. In my opinion .2mm on a 300+mm bed is acceptable. This may not work for your use case. Unfortunately current 3D printing technology will never be as accurate as a CNC mill part.

15

How did you measure the skew?

With bed tramming you can reduce the skew. But as @JonRaymond already mentioned, those deviations are very acceptable for a 3D printer.

16

my parts are rectangular with straight vertical walls which can be checked by placing the bottom of the part on a known flat surface, then butting up a machinist square against the wall and checking for taper at the top or bottom using a feeler gauges. Unfortunately the to surface of the part is an organic shape as its a section to a mold so it cant be flipped and rested on.

When checked with the machinist square, you can see the parallelogram effect from skew because one side will touch the square up top first and the opposing side will touch the bottom first. The skew result I mentioned earlier of 0.21° resulted in about a 1mm air gap on the square for how tall of a part this is.

IMG_1038
IMG_10381920×1440 182 KB

I just printed the indicator mount and should finally be able to check the bed within the next week on both printers. I’ll let you guys know the results

1 Like
17

I’m pretty sure the skew values he’s referring to are in degrees not mm like the results I posted. So the part will increasingly be dimensionally inaccurate the taller the print is. Such as what i’ve seen in my example mentioned with a ZY skew of 0.21° resulting in a 1mm shift at the top of a 275mm tall print.

1mm if 5x the 0.2mm you deemed acceptable, which is way to much for my liking. Being provided the necessary software tools to try and tune this out is necessary if they can’t guarantee better physical accuracy upon arrival. This should be available to everyone in the slicer using custom gcode commands. Obviously mechanical means first is the the right way to to do it but there should be software compensation for fine tuning.

This isn’t anything new as its been done this way with other firmwares for a long time.

2 Likes
18

No, this isn’t a limitation of the technology itself. We have multiple printers here, including older open-source builds, that consistently achieve <0.03 degrees skew. Even other retail machines like the Core One L (0.08 degrees) and the X1 Carbon (0.03 degrees) perform significantly better.

Skew isn’t a general tolerance issue, it’s a misalignment across axes. I’m not expecting +/-0.1 mm accuracy across the entire build volume but the level of skew we’re seeing on the H2D is closer to +/-1 mm over Z height which is considerably more than other machines on the market.

At that point, the larger build volume becomes effectively unusable unless there’s a way to compensate for that skew. That’s all anyone is asking for, a simple way to compensate for skew the same way every other machine on the market does (and the way Bambu already does with XY skew compensation).

1 Like
19

I used the Vector3d Calilantern tool. Its very useful and gives you a clear and repeatable way to calibrate your printer. I encourage you to do the same to see what results you get.

20

Which printer do you calibrate with the Calilantern then? The H2 series doesn’t have any calibration parameters to set. And definitely nothing for the Z-axis.

21

my h2s has zy and zx skew in the .1 degree range but i am going to tram the bed to see if that is it. i bet they do not tram them at all for the printer once the bed is in place. this is something they should have for a setup routine when unboxing a printer. they can not fix it in software because the bed is driven by one motor not 3 or 4. but it has been shown help fix it if you tram the bed i think the x1c has knobs under the bed you can tweak, the h2 series is a bit diffrent but can be trammed.

1 Like