Sensors and headers

Two subsystems on the Bitsy are not soldered down: the distance eyes and the servo arm plug into headers. That is a design decision with three reasons. Moving parts and exposed parts BREAK — a header turns a broken sensor into a swap instead of a rework. They are the expensive parts — one classroom board serves many sensors. And they are the parts students handle — a header is the mechanically honest interface for hands.

The distance eyes port is four pins for an ultrasonic ranger: VCC, TRIG, ECHO, GND. The MCU raises TRIG, the module chirps at 40 kHz, and ECHO goes high for as long as the reflection takes to return — distance is time-of-flight, read on GP2, triggered on GP3. The servo port is three pins — SIG, VCC, GND — and SIG carries PWM: a pulse every 20 ms whose width, 1 to 2 ms, commands the arm's angle. Any GPIO can source that PWM on this MCU, which is exactly why the old rule 5 clause about PWM-capable pins needed no validator.

These two ports are also the LEGACY DIALECT you met in PS-1: their pins maps name the rails literally — "VCC": "3V3", "GND": "GND" — documenting the connector's physical pin order. The newer parts wire rails silently. Since the library grew its required-signals check, BOTH dialects are held to the same standard: omit VCC from a servo header and the validator answers with "J2: header_1x3 requires pins.VCC" — the rule that was once stated-but-unenforced, now doing its job on the exact ports it was written for.

Both dialects, side by side, as two entries of one peripherals array — each valid, each complete:

[
  { "id": "eyes", "part": "header_1x4", "refdes": "J1",
    "pins": { "VCC": "3V3", "TRIG": "GP3", "ECHO": "GP2", "GND": "GND" } },

  { "id": "screen", "part": "header_i2c", "refdes": "J3",
    "pins": { "SDA": "GP4", "SCL": "GP5" } }
]