A few annotation notes: G380 S2, M960 S1/S2, M622 J2 pattern

[scoofz]

Hi, great work on the annotated gcode — this is by far the most thorough public documentation of the P2S startup sequence. A few observations that might be worth reviewing:

---

1. G380 S2 — S2 parameter is documented

You mark S2 as unknown in both G380 calls, but this is actually covered in community research: S2 selects load cell-based collision detection mode. Unlike a standard G1 move, G380 uses the toolhead's load cell to abort the move if unexpected resistance is detected. That's specifically why it's used here instead of a plain G1 Z — it's a safety move with obstacle detection, not just a positioning move.

---

2. M960 S1 P0 / M960 S2 P0 — correction, your annotation is likely correct

I initially suggested these might map to LED channels on the P2S rather than lasers. On reflection this was wrong — the P2S only has a logo lamp (controlled via M960 S5) and chamber lighting managed separately. S1 and S2 don't correspond to any visible light source on the P2S. Your "likely a no-op from shared gcode template" annotation is correct — these are X1C laser commands that do nothing on the P2S.

---

3. M622 J2 — worth noting the pattern

You flag this as "possibly a third branch" in two places (flow calibration and bed leveling). Looking at the pattern across the full sequence, J0 = flag false, J1 = flag true, and J2 seems to consistently appear as a second "true" branch that runs similar code. In the bed leveling block notably, J1 uses G29 A1 and J2 uses G29 A2 — so A1 vs A2 might indicate first print vs subsequent print within the same session, or single vs multi-material mode. Worth investigating.

---

Great repo overall — already referenced it from scoofz/P2S-start-gcode and the OrcaSlicer PR #13079 that applies the two bug fixes you documented.

[shrx]

Awesome, thanks a lot for your input! Will include the additional context in the documentation.

[scoofz]

With pleasure :)

[shrx]

I went through your comments.

Re: 1. You said: "You mark S2 as unknown in both G380 calls, but this is actually covered in community research: S2 selects load cell-based collision detection mode." Can you link to a discussion that confirms this? I couldn't find anything, no source explicitly documents what S2 means. OpenBambuAPI says the S parameter is "custom to Bambu and unknown."

Re: 3. I could not find any supporting evidence for either "First print vs subsequent" or "single vs multi-material" conditional based on the J2 parameter. So again, I'm inclined to leave this as is until more concrete documentation becomes available. The only apparent correlation appears to be that J1 selects A1 and J2 selects the A2 parameter in bed leveling.
Note that on the H2D/H2C (which have dual extruders), the J2 block has a TPU conditional where it falls back to G29 A1 when TPU is in the first layer. This hints A1/A2 might relate to leveling thoroughness, but it's still speculation.

[scoofz]

Fair corrections on both points — I was overreaching.

On G380 S2: I don't have a source that explicitly documents S2 as load cell detection mode, and you're right that the P2S toolhead doesn't have a load cell in the traditional sense (Z homing is done via nozzle contact through a strain gauge in the bed, not the toolhead). I conflated S2/S3 differences seen across models with an assumption about the mechanism. Without a concrete reference, "unknown" is the correct annotation.

On M622 J2: Your H2D/H2C observation is much more solid than my speculation. The TPU fallback from G29 A2 to G29 A1 in the J2 block is real evidence pointing toward A1/A2 as leveling thoroughness modes rather than print session state. My "first print vs subsequent" hypothesis had no supporting evidence — should have flagged it as speculation rather than presenting it as a reasonable interpretation.

Sorry for the noise — your approach of only documenting confirmed information is the right one.

[bfarnam]

There is a newer version of the machine start gcode for the P2S dated 2026/05/26. This version is substantially the same with one very big change to a variable name:
{filament_type[initial_no_support_extruder]} has been replaced by {filament_type[initial_no_support_filament_id]}