Feedback for the Creality Print 7.3 / K2 Pro team — reproducible issues with evidence, from a power user

I own a K2 Pro (firmware 1.1.8.3, CFS, single nozzle) and I run it hard: presets tuned per material, gcode inspected after every job, Moonraker queried directly when the app misbehaves. The hardware is genuinely good — but the software keeps costing me hours. This post is the feedback I wish the team had: every issue below is reproducible, with evidence. If even two of these get fixed, my workflow halves in friction.

1. The exhaust fan never turns off after a PETG print

  • Generic PETG factory profile ships complete_print_exhaust_fan_speed = 80. After the print ends, that fan output stays pinned at 84% (verified via Moonraker: output_pin fan1 = 0.84) while the machine sits idle and cold, until the next job or a reboot. The console shows the chamber cooling loop finishing (cur_temp = 43 → 40, end_temp!) — but the fan pin is unaffected by it.
  • Users shouldn’t have to power-cycle a printer to stop a fan. Suggestion: a bounded post-print timer (e.g. 10 min), or stop when chamber reaches ambient, and make the after print fan speed visible in the UI, not just the during one.

2. Invalid profile values are silently discarded

  • I set seam_position = "rear" in a custom process profile. The slicer accepts the file, laminates, and quietly prints with aligned — the gcode config block says ; seam_position = aligned. No warning, anywhere. (rear is a PrusaSlicer-era enum; this slicer wants back.)
  • Silent fallbacks are the worst kind of bug because the output looks plausible. Suggestion: validate embedded 3MF settings and user presets against the enum, and log/warn in the console when a value is dropped.

3. The gcode metadata can name a preset while using another preset’s values

  • A project declares filament preset X; at slicing/send time the CFS slot imposes its own profile values. Result: the gcode header names preset X while the actual values (bed temp, flow cap, fan delays) come from a different, factory profile. We proved it in one job: the unused filament slot resolved its custom preset perfectly while the used slot got factory values under the custom preset’s name.
  • Whoever debugs a failed print from the gcode alone gets lied to. Suggestion: one source of truth, and an explicit warning when a slot/profile override replaces the project’s declared preset.

4. User profile sync can delete local profiles

  • With “sync user presets” enabled, logging into the account replaced locally tuned profiles with the account’s cloud versions — silently. Tuned profiles are hours of work. Suggestion: merge with conflict prompt, never a blind replace, and keep a local backup of the previous state.

5. Cloud transfer is unreliable, and there is no LAN send in the app

  • Jobs sent through the cloud regularly fail to land on the printer; I currently slice, then upload the gcode directly through the Moonraker API (/server/files/upload) because it is the only path that works every time. The printer is on my LAN; the app refuses to use that.
  • A simple, robust LAN send option in Creality Print would remove an entire support category on its own.

What works and is worth keeping

The K2 Pro print quality itself, the CFS RFID flow, and the fact that Moonraker is exposed at all — that last one is the only reason most of the above was diagnosable. Please keep it open and documented; it turns frustrated users into useful bug reporters.

Happy to provide gcode files, Moonraker dumps, or exact repro steps for any of these if a maintainer wants them.

3 Likes