North Pole Height Sensor — Design Summary (v2)

ToF lidar, 3.6° FOV — narrow enough to avoid seeing the pole shaft, unlike wide-beam ultrasonic options (SRF04, JSN-SR04T) considered earlier.

IP65 sealed enclosure — no DIY weatherproof box needed.

Operating range -20°C to 60°C — covers winter use.

Ambient light immunity 70klux — not a concern; the piece runs at night, and the active IR emitter works best with no competing ambient light anyway.

±1cm-class accuracy at relevant ranges — comfortably inside the “nearest few cm” target.

Aesthetics rule out any external bracket or arm — nothing can protrude from the pole's silhouette, including underneath the sign board.

Sensor housed inside the finial ball at the very top, aimed down through a small discreet lens port. No visible hardware; wiring runs down through the shaft interior.

The beam is tilted a few degrees off vertical (not straight down) so it clears both the shaft and the sign board's width on its way past — since the ball sits above the sign, the beam must clear both obstructions stacked on the way down.

At the shallow angles involved, the slant-distance-to-height correction is negligible (cos of a few degrees ≈ 1).

Open item: exact tilt angle needs two measurements — how far the sign board's face projects forward of the shaft, and the finial ball's diameter (to confirm the sensor + wiring bend radius physically fits inside it). Fallback if it doesn't fit: embed the sensor in the sign board itself instead, near its bottom edge.

Originally planned a sealed pressure/switch mat in the ground at the standing spot. Dropped after team pushback — the pole has no awning, so full weather exposure (snow burial, ice on the switch mechanism, de-icing salt corrosion) made a horizontal ground mat too risky.

Replacement: a single small diffuse/background-suppression proximity sensor (or a second TFmini Plus) mounted low on the pole — roughly shin/knee height — facing outward, tuned to a short fixed trigger range (~10–20cm).

Legs/feet stay consistently close to the pole even when someone's upper body sways, making this a reliable “are they actually at the pole” signal.

Mounted vertically on the pole face rather than flat on the ground — precipitation sheds instead of accumulating, avoiding the mat's core weather risk.

Combine with a stability filter on the top sensor's readings (buffer ~1–2 sec, require low variance) before updating the display — the side sensor gates position, the stability filter gates timing.

SRF04 / wide-beam ultrasonic — beam cone too wide, sees the pole/mount hardware.

JSN-SR04T (waterproof ultrasonic) — waterproof, but even wider beam than SRF04.

Sharp GP2Y0A710K0F (long-range IR) — 100cm minimum range likely too long for this geometry, poor resolution at distance, triangulation sensitive to any window angle.

Kinect — discontinued hardware, needs a full PC + drivers, not weatherproof, doesn't solve headwear.

Two-post beam-break array — rejected, didn't want a second post.

Capacitive sensing along the pole — rejected: rain/snow/frost cause false triggers, thick winter coats reduce coupling reliability.

Ground pressure mat — rejected after team pushback: no awning means full weather exposure, risk of weather damage/failure.

Camera / computer vision — considered for pole occlusion + positioning, but set aside: privacy/consent concerns for a public installation involving kids, plus needs calibrated geometry for real height accuracy and its own IR illumination for night use. Revisit only if there's a specific reason to want imaging (e.g. displaying/saving a photo as part of the piece).

Idea: reuse the pole's planned LED strip as the sensing mechanism. Sequentially light small segments up the pole and read a co-located photodiode/phototransistor at each segment for reflected light — a person's body reflects strongly only at the height currently lit.

Sequential (not simultaneous) activation largely avoids crosstalk between segments without needing physical baffles.

Cost is much lower than commercial background-suppression sensors — bare photodiodes/phototransistors run cents–low dollars each vs. $30–80+ per commercial BGS unit — even after adding amplification/multiplexing circuitry, likely still an order of magnitude cheaper for the ~50–100 units needed at 2–3cm resolution.

The scan itself can double as show content — “lights climb to reveal your height” fits the piece thematically, so the sensing motion and the visual effect can be the same thing.

Real remaining work: transimpedance amp or phototransistor front-end per channel, analog mux or many ADC channels to read them, a sealed window per sensor point (more penetrations than a single top sensor), and baseline/threshold calibration to handle ambient light drift.

Genuinely competitive with the single top-sensor approach now that the cost assumption was corrected — worth pursuing if there's appetite for the extra electronics/firmware build; the tilted TFmini Plus + side proximity sensor remains the simpler build if not.

Headwear (hats/hoods) will read as part of a person's height with any top-down or topmost-surface detection method — accepted, since it's within the “few cm” tolerance already targeted. This applies equally to the LED-strip array (reflects off whatever's closest, hat included).