Timing changed on Arduino script, latest version of ESP-Home script added

changes.md

@@ -0,0 +1,46 @@
+# Changes
+
+## 2026-04-27 — Arduino firmware refactor (`Gaugecontroller/Gaugecontroller.ino`)
+
+### Non-blocking VFD multiplexer
+`vfd::refresh()` previously held each digit for 2000 µs via `delayMicroseconds`,
+which capped the effective stepper pulse rate at roughly 500 Hz regardless of
+`maxSpeed`. It now tracks `phaseStartMicros`/`phaseActive` and returns
+immediately while the digit is still being held; the main loop runs at
+microsecond cadence again and the configured `maxSpeed = 4000.0f` steps/s is
+actually achievable.
+
+### Fixed-buffer command parser (no more `String` heap churn)
+Replaced `String rxLine` with `char rxBuf[128]` and converted the entire
+command pipeline to take `const char*`:
+
+- `processLine`, `sendReply`, `vfd::parseCommand`
+- All `parse*` functions: `parseSet`, `parseSpeed`, `parseAccel`, `parseEnable`,
+  `parseZero`, `parseHome`, `parseSweep`, `parsePosQuery`, `parseCfgQuery`,
+  `parseLedQuery`, `parseLed`, `parseBlink`, `parseBreathe`, `parseDflash`,
+  `parseVfd`, `parsePing`.
+
+`parseSpeed` / `parseAccel` / `parseSweep` use `strncmp` + `atof` because the
+default AVR-libc `sscanf` doesn't support `%f`. No allocations on the command
+path; the Mega's heap no longer fragments over time.
+
+### Cached `ledNeedsSwap[TOTAL_LEDS]`
+Per-LED RGB-vs-GRB swap flag is now precomputed once in `setup()` from
+`gaugePins[].ledOrder`. `encodeForStrip` is a single array index instead of
+walking the gauge table on every LED read/write.
+
+### Cached step direction per gauge
+Added `Gauge.lastDir`. `setDir()` skips the DIR-pin `digitalWrite` when the
+direction hasn't flipped (the common case during a step run) and adds a 1 µs
+DIR-to-STEP setup delay only when it actually flips.
+
+### Cleanups
+- Removed the `absf` helper; use `fabsf` consistently.
+- Removed the `+ 0.0001f` epsilon in the trapezoidal braking-distance divisor.
+  `parseAccel` already rejects `accel <= 0`, so the divisor is always positive.
+- Fixed the `<r> <ig> <b>` typo to `<r> <g> <b>` in the protocol comment for
+  `DFLASH`.
+
+### Build verification
+`arduino-cli compile --fqbn arduino:avr:mega Gaugecontroller`:
+17758 B flash (6%), 1845 B SRAM (22%).