[FredM67]

Why

The universal 3phaseDiverter board (rev. 6.0 and later) is designed for the internal 1.1 V reference of the ATmega328P. An LMV321A buffers it from AREF, and the doc says the firmware must switch it on. The firmware always selected AVCC, the reference of the old 3-phase board, so a new board measured wrongly.

The change existed on the unmerged new-pcb branch (cfe0682a, 2025-12-30). It was written for the old switch ISR, which #127 later replaced with the circular channel list, so it could not be merged as is.

What

• config.h: a setting a user can understand:

  // Motherboard: OLD for the former 3-phase board, NEW for the universal 3phaseDiverter board
  // (rev. 6.0 and later). A wrong choice gives wrong measurements: recalibrate after a change.
  inline constexpr PcbVersion PCB_VERSION{ PcbVersion::OLD };

  OLD is the default, so existing installations behave as before.
• processing.cpp: _ADMUX selects AVCC (OLD) or the internal 1.1 V (NEW). initializeProcessing() now writes the full ADMUX (reference, ADLAR, V1) for the first conversion too. Before, that conversion ran with REFS = 00 (external AREF) until the ISR took over.
• The startup summary prints Motherboard: old (AVCC reference) or new (internal 1.1 V reference).
• READMEs (EN/FR): a "Motherboard: old or new" section, with the reminder to recalibrate. CHANGELOG: an Added entry and an "Upgrading" note.

Checked

• OLD: the ADMUX values in the channel list are unchanged (0x60 | channel). The grid simulator passes (clouds 11/11, no_surplus 7/7, surplus_step 12/12).
• NEW: the channel list holds 0xE0 | channel and the setup writes 0xE0.
• The simulator fails with NEW, as expected: simavr scales conversions against the 1.1 V reference while the scenarios feed signals sized for 5 V. That shows the reference really switches. Simulating the new board would need scenarios with its signal levels.
• Native tests 172/172. Embedded tests 106/108; the 2 errors are empty local test folders, unrelated.
• Still to do: a hardware test on a new board (VREF ≈ 1.1 V on AREF, then calibration).

Follow-ups

• The configurator (feat: configurator web page for the router, remote units and mk2Wifi #183) gets an "old / new motherboard" choice once this is merged.
• The new-pcb branch can be deleted after the merge.
• The single-phase firmware may need the same on the universal board. It has its own OLD_PCB, which only switches the analog pins.

🤖 Generated with Claude Code