references · updated 2026-07-19
Small-scale passive-thermal experiment monitoring
confidence: medium volatility: warm verified: 2026-07-19fresh
A reusable Raspberry Pi Zero sensor/logging methodology for validating small passive-thermal builds in the field, in the spirit of Stickney and Baer's 2002 heated-vs-unheated-plates experiment: interior temp/humidity, water temp, and a solar-irradiance proxy, logged unattended on solar/battery power.
The archive already has one precedent for exactly this kind of validation work: Stickney and Baer’s 2002 matched-plate experiment, where a heated and an unheated 12”×12” aluminum plate were instrumented side by side to measure night-sky radiative cooling directly, rather than trusting theory alone. This page documents a modern, low-cost equivalent — a Raspberry Pi Zero sensor rig — for anyone validating a small passive-thermal build (such as the miniature Double Play cat-shelter experiment) instead of relying on published data from a different climate.
Sensor stack
| Measurement | Sensor | Interface | Notes |
|---|---|---|---|
| Interior air temp + humidity | BME280 | I2C | ~$5, also reads barometric pressure; wiring: VIN→3.3V, GND→GND, SCL→GPIO3, SDA→GPIO2 |
| Water temperature | DS18B20 (waterproof-housing variant) | 1-Wire | Bare TO-92 packages are not rated for submersion — the sealed probe variant is required |
| Solar irradiance (cheap/easy) | BH1750 lux sensor + linear or ML correction | I2C | ~$1-2. Raw lux-to-W/m² conversion is rough (~54% error via simple linear regression); a random-forest correction achieved R²=0.9922 in peer-reviewed testing |
| Solar irradiance (more accurate) | Open-hardware photodiode sensor (OSRAM SFH203 + LM75 temp compensation) | Serial | More build effort, no ML correction needed; saturates above 2,000 W/m² with its PTFE diffuser |
Logging
Run Raspberry Pi OS Lite (headless, no desktop) and log via a cron-scheduled Python script appending CSV rows to the SD card — not a naive while True: sleep() loop, which prevents CPU sleep states and needs manual restart after any crash or power blip. Disable HDMI output, Bluetooth, and unused services to cut idle power draw.
Power (off-grid solar)
Worked sizing example for a Pi Zero W + sensor load (~160-190 mA average): an 8-10W solar panel with a 6,600-10,000 mAh Li-ion/LiFePO4 battery comfortably covers 24-hour unattended operation, including the full nighttime draw. Charge through a TP4056 module (simplest) or a dedicated MPPT/PWM controller (more efficient), with a boost converter to a stable 5V rail. Budget ~15% losses in the charge/boost chain. Monitor actual battery voltage (not the regulated 5V rail) for low-battery shutdown, with ~3.5V/cell as a safe cutoff — an INA3221 is a reasonable voltage/current monitor for this.
Enclosure
Humidity and heat are the top two risks to the electronics themselves, not just the experiment. Use an IP65/67-rated ABS junction box with PG7 cable glands, drip loops on all cable entries, and a Gore-Tex vent plug to equalize pressure without letting liquid water in — a fully sealed box will accumulate condensation on its own.