Hello,
About 10 days ago we found that our H2D prints significantly oversized parts in the Z direction. A 190 mm PLA part measured 193 mm. The issue is reproducible, and caliper measurements show the bed moves ~0.15 mm too far per 10 mm.
We know that after delivery in May 2025, Z dimensions were slightly undersized (199.2 mm for a 200 mm part).
We checked the belts and downgraded the firmware. With firmware 01.01.00.00, parts return to 199.2 mm. All newer firmware versions produce oversized parts.
This suggests incorrect Z-axis resolution/steps in recent firmware. Is this a general issue with current firmware or specific to our machine?
Could other H2D owners print a tall part and share their results?
Weâve contacted Bambu, but theyâre currently on holiday.
I contacted BL on 11.02.26 and they asked for pictures, which I attached here now as well.
BL didnât reply since I provided pics on 12.02. - hopefully only because of Chinese New Year.
With the new firmware we manually moved the bed 7x 10mm and the caliper showed 71mm. The same movement is very accurate with the old firmware.
Therefore the identical test project resulted in 199.2mm vs. 203mm with old vs. new firmware.
Currently we remain on the old firmware 01.01.00.00, but obviously thatâs no long-term solution.
I attach two 3MF files of our test print if someone wants to try it out.
We usually print with 0,6mm nozzles, more infill & walls and use significantly lower print & acceleration values for strength and quality of our functional parts.
But the results are the same (+/- 0,2mm) with a 0,4mm nozzle and other print settingsâŠ
I just did a measurement with a dial gauge on my H2D my manually moving the bed up and down. and a 20mm displacement in Z measures as 19.950mm on the dial gauge.
I have not tested this on printed parts yet, but for now it seems that itâs not having the deviations that you are encountering.
Wow, waited 10 days for the most generic reply from BL customer support which is doubting in the validity of our caliper testing method and doesnât even get the core of the problem (old vs. new firmware leading to 199,2mm vs. 203mm print result with all other parameters staying the same):
Hello, there
Thank you for your feedback.
I sincerely apologize for the troubles you have encountered. After confirming with our development team, the firmware does not affect the printing accuracy of the Z-axis.
This issue may be related to filament shrinkage and the measurement method of the caliper.
Therefore, we strongly recommend that you try using a new roll of filament and print a 30mm height cube to check whether the error remains at 0.8mm.
If you encounter any further issues, please feel free to contact me here.
Best, regards.
Bambu Lab Customer Support
Pretty disappointing.
Will use the âEscalate your caseâ button in the email and hope that the next level of customer support can at least identify the problem and give me a valid argument how this problem is not connected to the firmware.
We had to finish urgent orders the last 2 weeks, therefore we didnât have the capacity to test this until now and maintained on the old firmware 01.01.00.00.
Yesterday we updated to the new 01.03.00.00 firmware and the deviation is back.
A Ă30Ă100 mm test part resulted in Ă30.03Ă101.4 mm.
A caliper test of the heatbed movement again showed 71 mm for a commanded 70 mm movement (vs 69.7mm with the old firmware).
This indicates that the deviation is still present and scales linearly (our 200 mm height results in approximately 203 mm).
A factory reset after the firmware update, as well as vision encoder calibration, did not resolve the issue.
Bambu Lab has not provided any help in the last two weeks despite âescalatingâ the request and their promise to reach outâŠ
Thanks for your testprint.
Seems like your printer is in the ânormalâ range of printing slightly too small, like ours with the old firmware.
Which is still a shame that one canât improve it by adjusting the steps of the printerâŠ
Yes it is super weird and the first time we ever encounter something like that.
Our three P1S and our P2S work fine and print slightly too small âas usualâ for Bambu Lab in our experience.
I contacted BL again and also our German reseller of the machine.
Will check for âAuto-Recovery from Step Lossâ tomorrow, we werenât aware of that option yet.
Although I doubt that it gets deactivated by a firmware update, but you never knowâŠ
@Milosh_Keat
I donât believe this is the root of our issue. We are not experiencing any nozzle scraping, and the problem appears to be a firmware-related linear deviation of printed parts or bed movement.
Today we received the following response to yet another reminder we sent to Bambu Lab (after they had previously said they would contact us again once they had investigated the issue further):
At this point, the support experience has been extremely disappointing. For five weeks now the responses have consistently missed the actual problem we are describing. We have not been this disappointed with a company in a long time.
Given the current situation, it is difficult for me to see how we could continue purchasing or recommending Bambu Lab products in the future. Which is a shame, cause we really enjoy our 5 BL printers for the last 2 years.
We have also reached out to our German reseller, who responded quickly and said they would look into the matter. However, I am not sure how much they can realistically do without support from Bambu Lab themselves.
Fun fact:
Our newly purchased Snapmaker U1 prints a 200 Ă 200 Ă 200 mm PLA calibration cross within 0.1 mm accuracy in all three dimensions. In contrast, our H2D prints the same calibration cross 0.5â1 mm undersized in all three axes when using the older firmware.
This is despite purchasing and using the âŹ100 Vision Encoder, which is advertised as providing âbelow 50 ”m motion accuracy.â
Notably, the vision encoder was the only solution Bambu Lab suggested back in May 2025, when we initially contacted support about this dimensional inaccuracy â and unfortunately, it did not improve the situation at all.
Donât have context on all of it, but I wouldnât expect the vision encoder to help your machineâs Z matter - it only claims to calibrate the xy axis movements, and thus is out of scope.
I imagine if this changed between firmwares, they altered the z motorâs steps/mm and others could replicate
Hopefully
