H2D Combo: Inconsistent first layer despite tramming + ABL (bad mesh?)

Bambu Lab H2 Series Bambu Lab H2D
1

Hi,

I’m reporting an issue with first-layer quality on my H2D Combo. I first noticed it during regular prints, then confirmed it with a full-bed first-layer test. For context, I have around 650 printing hours on the H2D, and previously logged about 1500 hours on an Ender 3 V2. On the Ender I did extensive modifications and repairs, and I regularly performed manual calibrations that consistently resulted in reliable, high-quality prints-so I’m comfortable diagnosing first-layer and bed-leveling related issues.

  • Conditions and material

- Filament: Bambu Lab ABS (most tests were done with ABS because it’s my main production material)

- Filament dried in the printer and stored in AMS

- AMS humidity kept below 5%, filament is re-dried regularly

- Bed cleaned with dish soap and additionally wiped with 99,9% IPA

  • Symptoms

- Uneven first layer across different areas of the bed

- Typical pattern: one corner shows too much material (nozzle too close to the bed), another corner shows gaps between lines (nozzle too far from the bed)

- This affects real prints because I use the whole bed area for production parts

  • Initial steps taken (no improvement)

- Checked for firmware/system updates

- Ran automatic bed leveling (also at high bed temperatures)

- Performed XYZ axis maintenance and lubricated the lead screws

- Tightened the screws at the back of the ceramic nozzle mount

- Adjusted drive belt tension

- Calibrated Flow Dynamics and Flow Rate

  • Next attempts (worse after cleaning)

- Cleaned the nozzle using the Cold Pull method

- Re-ran automatic bed leveling

After these steps, first-layer quality became worse.

  • Nozzle replacement and recalibration (slight improvement, issue remains)

- Replaced the nozzle with a new one

- Re-calibrated Flow Dynamics and Flow Rate

- Performed a full printer calibration

I saw some improvement, but the issue is still present.

  • Manual tramming + more leveling (no solution)

- Performed manual bed tramming strictly according to the Bambu Wiki instructions

- Ran automatic bed leveling again (also at high temperatures)

Unfortunately, the first-layer issue still occurs.

  • Additional observation (wiper plate)

I also noticed that during the preparation procedure the nozzle tip does not move over the center of the metal nozzle wiper plate, but instead cleans closer to the rear edge. The plate shows wear marks where the nozzle contacts it (the plate is not bent). Support confirmed that this behavior is normal.

  • Support feedback and comparison tests

Support suggested:

- the results are “not too bad,” maybe slightly under-extruded at edges/center

- testing with PLA Basic (I did: first layer was better, but small issues were still visible, especially in the rear-right corner)

- checking nozzle installation (confirmed OK with photos)

- setting first-layer flow ratio to 1.05 (did not solve the issue for me)

- possibly slowing the first layer (test: first-layer infill speed 80; results became much worse, with large areas showing gaps between lines)

I also tested ABS in a different color to rule out a specific filament batch/color — the symptoms are similar (different corners behave differently; some areas look “too close,” others “too far”).

  • My hypothesis (mesh / leveling measurement)

Based on the symptoms, this looks like an incorrect mesh or incorrect automatic bed leveling measurement:

- front-right: too much material, nozzle collecting buildup, rough texture typical of Z being too low (bed too high)

- back-left: too little material, visible gaps between lines typical of Z being too high (bed too low)

Since mechanical tramming was done correctly, it seems like the leveling measurement is being performed or interpreted incorrectly.

  • Side effect: worse real prints

While troubleshooting, I printed my standard product again (same file I always use; only filament color changes based on orders) and print quality deteriorated significantly:

- first-layer issues

- print artifacts

- poor wall edge quality

I have comparison photos: previous prints (orange) were satisfactory, current prints (blue) are clearly worse.

  • Latest actions (current status)

While waiting for a reply, I performed full maintenance and calibration again — the problem still persists.

If anyone has experienced something similar on the H2D (especially with ABS and high bed temperatures):

- what else should I check regarding a potentially incorrect mesh (sensor, procedure, temperature effects, specific diagnostic tests)?

- are there known cases where auto bed leveling produces a wrong mesh even after correct manual tramming?

Thanks in advance for any specific suggestions.

2

Nice troubleshooting write‑up!

From my experience, when printing full‑plate models it is more often than not necessary to do a proper heat soak to get a consistent first layer. I exclusively use CFX plates and I always let the bed heat soak for 15–30 minutes before the print starts to stabilize the temperature across the whole surface.

Have you ruled out uneven bed heating as a cause yet? For example, have you tried doing a 15–30 minute heat soak before you started experimenting with tramming and ABL to see if the first layer behavior changes?

1 Like
3

Thanks for the tip. I actually read about it once but I missed it. I’ll start testing and come back with the results later.

4

Heat soaking helped; there’s a significant difference. There are still some irregularities in the rear corners, but they’re small enough to be acceptable. Thank you very much for your help! I have a question: do you know if Bambu Studio allows you to set the bed preheat time before printing, or will it be necessary to add this to the initial g-code?

5

Not yet, they ignored my request from start and concentrated on those who lives in 120V grid for now.

This is how I solved it so far in both my X1 Carbon and H2D
Messages doesn’t work and there are some code before and after just to show where I put this code in my machines startup G-code.
The masked word is Swedish for end :wink:

;===== z ofst cali end =====

M400
;M73 P99

M141 S[overall_chamber_temperature]

; ===== Start av 15 minuters temperaturstabilisering =====
M117 Temperaturstabilisering i 15 minuter...  ; Visa meddelande på skärmen
M400 S900                                                       ; Vänta 15 minuter (900 sekunder)
M117 Temperaturstabilisering klar                  ; Uppdatera skärmens meddelande
; ===== ■■■■ på 15 minuters temperaturstabilisering =====

M104 S{nozzle_temperature_initial_layer[initial_no_support_extruder]} A

;===== mech mode sweep start =====

This is a 15 minutes delay, use 1800 if you need 30 minutes and you can erase the messages because they are not shown. On the display it will still display that it’s in ABL.

3 Likes