[abehrman]

Description

Multi-tool printers with a purge chute or bucket, like the Snapmaker U1, can prime each tool there after a tool change instead of on a prime tower. The tool still gets a pressure-normalizing prime, but nothing has to be built on the bed. This adds that as an opt-in feature and sizes each prime by how long the tool was parked.

This replaces my earlier drafts that tried to avoid the prime tower by flushing into infill or skipping the prime for support-only filaments. Those left the nozzle unprimed, and a tool needs a prime to restore pressure after a park.

How it works

• Tool change prime G-code (printer setting toolchange_prime_gcode) runs after each real tool change when Smart tool change prime (toolchange_prime_enable, off by default) is on. It only runs when the prime tower is off and the printer isn't single-extruder multi-material.
  • Placeholders: prime_length, prime_speed (limited by the filament's max volumetric speed), prime_first_use, retract_restore_length, next_extruder, previous_extruder, new_filament_temp.
• Restores the retraction in the chute. A parked tool still holds its tool change retraction (10 mm on the U1). The block extrudes it in the chute first, on a line marked ; restore retraction and sized by retract_restore_length. The G-code writer then counts the tool as unretracted, so it doesn't unretract again on the part.
• Sized by idle time, as a step. The block is tagged, and the G-code processor sizes its prime extrusions by the tool's estimated park time:
  • a park of at least the idle time gets the maximum length: by then the nozzle has cooled on standby;
  • a shorter park gets the minimum length: the nozzle is still hot and only needs its pressure back;
  • below the skip time the prime is commented out, but the retraction is still restored.
  • The tag comment records why, for example (idle 42s, 3.10mm), (first use, …) or (after pause, …).
• Full prime after planned pauses. A tool parked across a pause or filament change (PAUSE, M600, M0, M1, M25, or the pause and color-change tags) gets the full length, because the slicer can't know how long the pause lasted.
• Time primes on the printer (toolchange_prime_on_printer, off by default). This covers pauses the slicer can't see: a pause pressed on the printer, a filament runout, a long reheat.
  • The G-code is emitted untagged, and the printer's own macro sizes the prime from real time.
  • The U1 profile calls SMART_TOOLCHANGE_PRIME, which is included in doc/printer_settings/snapmaker-u1-smart-toolchange-prime.cfg and proposed to the firmware in Add SMART_TOOLCHANGE_PRIME to prime a tool by its real park time u1-klipper#21. It times each park with print_stats.total_duration and primes fully after any pause.
• First use. The starting tool counts as primed by the start G-code. Each later tool's first use is reported as prime_first_use, and the U1 template leaves that to the firmware's SM_PRINT_PREEXTRUDE_FILAMENT.
• Snapmaker U1 profile: the template ends the same way as the firmware's SM_PRINT_PREEXTRUDE_FILAMENT. Inside SAVE_GCODE_STATE/RESTORE_GCODE_STATE it:
  1. moves to the discard position;
  2. restores the retraction and primes;
  3. retracts 0.5 mm;
  4. cuts, wipes and knocks the blob off (INNER_CUTOFF_BASE_DISCARD, INNER_ROUGHLY_CLEAN_NOZZLE_BASE_DISCARD, INNER_DISCARD_FILAMENT_BASE_DISCARD);
  5. leaves the chute Y first with MOVE_TO_XY_IDLE_POSITION_EXTRUDER.

Recommended defaults (process settings, "Tool change priming"):

Setting	Default	Why
Minimum prime length	10 mm	About 24 mm³ of 1.75 mm filament. The stock U1 prime tower purges 45 mm³ (about 18.7 mm) per tool change at 0.20 mm layers, and 2 mm was too little on a test print
Maximum prime length	20 mm	Matches the stock pre-extrusion; purges filament that sat in a cooled nozzle
Idle time for maximum prime	30 s	Time for a parked nozzle to cool below the flow temperature: measured on a U1 PETG print (220 °C, standby 70 °C), below 180 °C after 24–30 s and about 135 °C at 60 s
Skip prime below idle time	0 s	Always prime unless you raise it

Fixed since the first version

A test print on a U1 lost position on a tool change: the head hit the frame and missed the dock by several inches. The first version had two problems that likely contributed. I can't prove which one caused it.

• The new tool's 10 mm tool change retraction was unretracted on the part right after the prime, leaving a blob in the same spot on every swap.
• The head left the chute with a fast diagonal travel instead of the firmware's Y-first move.

This version fixes both, as described above. The prime length is also a step now instead of a linear ramp.

The U1 profile change is now in Snapmaker/snapmaker-profiles#1. #1128 moved the presets into the submodules/Profiles submodule, so fdm_U1.json is no longer in this repo. That PR adds toolchange_prime_gcode to the U1 profile and bumps the profile version to 02.03.04.05. Merge both, then update the submodule pin.

The version bump matters. Without it, the app keeps its cached copy of the U1 profile, so an updated build still slices with the old prime G-code. That happened on a test print: the 2 mm prime ran while the tool was still retracted 10 mm, so nothing came out in the chute, and the 10 mm unretract then left a blob on the part.

No breaking changes: everything is off by default, and nothing changes for printers without tool change prime G-code.

Screenshots/Recordings/Graphs

None. The UI change is six settings in a new "Tool change priming" group on the process "Multimaterial" page, plus the printer's tool change prime G-code field.

Tests

• New tests/fff_print/test_toolchange_prime.cpp (7 cases): sized as a step by idle time, the retraction restored in the chute with no second unretract on the part, skipped below the skip time, not emitted for single-extruder multi-material printers, off unless enabled, full length after a pause, and left to the printer when timed on the printer.
• All 7 pass, and so does the full fff_print_tests suite (58 cases), in a Release build on macOS arm64. That run was on my local branch, which has this change plus unrelated work.
• This branch's changes were cherry-picked cleanly from that local branch and are otherwise unchanged.
• SMART_TOOLCHANGE_PRIME was checked in a Jinja simulation of two prints, including a pause pressed on the printer. It hasn't been tried on a real U1 yet, and neither has this version of the slicer side.

Supersedes #1021 and #1022.

🤖 Generated with Claude Code

[abehrman]

Thanks. Addressed in b5b0dad:

1. Skip vs idle: fixed. A park at least as long as the idle time always gets the full prime, matching the macro. The tooltip says so. New test: idle 0.001 s with skip 36000 s gives the full prime on every block.
2. Estimates: they come from the first pass, so they count every prime at full length. That overstates by at most the minimum-to-maximum prime per short park, a few seconds and a fraction of a gram. I added a code comment instead of re-running the time model on the rewritten lines.
3. E mode: for absolute-E prints, an M82 is now emitted after the block, before G92 E0. New test with use_relative_e_distances = false and an M83 template.
4. Profiles: the profiles now live in the Profiles submodule (after the Snapmaker main merge), so the U1 preset change goes there, not into this PR. Until then the feature stays off by default and is inert.
5. Label: a reused tool with no recorded park time is now labelled "park time unknown".
6. Cast: now uses static_cast<int>.

All [ToolchangePrime] tests pass (9 cases).

🤖 Generated with Claude Code