projects:lawrencelightsheight [wiki]

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== North Pole Height Sensor — Design Summary (v2) ======

===== Sensor: Benewake TFmini Plus =====

  * 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.

===== Mounting: inside the finial ball, tilted =====

  * 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.

===== Position confirmation: side-facing proximity sensor (not a mat) =====

  * 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.

===== Rejected / set aside approaches =====

  * **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).

===== Still on the table: LED-strip scanning array =====

  * 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.

===== Known accepted limitation =====

  * 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).