[ww8l]

Closes #50.

What

Icom ID-51 Plus (and the 50th Anniversary edition) programmed from its own microSD card, the ID-52's arrangement. The radio saves and loads one .icf (MENU › SD Card › Save / Load Setting › ALL). The export builds a new SetYYYYMMDD_NN.icf from the newest file the radio saved, so the operator's own file is never modified.

• Memories: 500, with channel lists as banks A–Z (100 each). CHIRP's id51plus map, checked against the radio's own file and Icom's CS-51PLUS for every memory. Uses the ID-5100's 49-byte record, with this radio's multiplier rule (5 kHz first) and step table.
• Settings: 241 menu settings plus the manual position, including all eight D-PRS items. Each was measured in CS-51PLUS by changing one control and diffing the save. The table and schema come from one generator. Not offered: Home CH (display-only in CS) and the clock (not in the file).
• The original ID-51 (33900001) and the Plus2 (33900003) are refused by name.

Shared changes

• radios/icom_card.rs: the .icf card-folder rules now shared by the ID-52 and ID-51. Behaviour is unchanged, but "newest file" now checks files newest-first.
• SK::EnumBits: an option list inside part of a byte, for ATT(AM).
• The ID-5100's settings helpers take per-radio ApplyRules.
• export_noting / export_settings_noting: a setting the encoder declines now reaches the write report as a note.
• RadioDriver::unstorable_frequency: channel_fit excludes a channel the radio's memory cannot hold. The ID-51 uses it for 8.33 kHz airband channels.

Verified

• On a real ID-51 Plus 50th Anniversary, every ladder rung passed:
  1. identity copy
  2. one-field rename (proves the #CD checksum)
  3. full 50-memory codeplug in banks HOME/GMRS, through the real exporter
  4. band probe (108.000 AM / 173.995 / 380.000 / 478.995 / 146.520 / 446.000), all six landed
  5. settings spot check (Backlight, Beep Level)
• Desk gates against the radio's own file (the #[ignore]d tests in icom_id51/real_file.rs): the file round-trips byte for byte, and all 500 records re-encode identically. The 87 memories re-encode from the channels they describe, and the radio's own settings re-apply with zero bytes moved.
• npm run ci is green locally (906 tests, clippy).

Not verified

• Two conservative rules apply without measurement on the ID-51: the NMEA/logger sentence limits (at most 4 / at least 1, taken from the ID-5100), and MY call signs stored as typed. A profile outside the limits leaves that group alone and the report says so.
• The app's UI for this radio (reading settings from the card in the profile editor, the Write to microSD dialog) was not clicked through.
• The ID-52's write path now runs through the shared helper; its tests pass, but no ID-52 card has been written since.

🤖 Generated with Claude Code