Area: Shared Components · TRL: 3 of 9 (proof of concept on paper) · Value-engineering target: about $150 USD · Difficulty: 3 of 5
A solar-powered, weatherproof sensor node with a common mounting, power and radio core, so any lab project that needs to measure something outdoors starts from the same tested base instead of a new enclosure and charger each time.
Exploded render · Detail render · Interactive 3D model · Concept blueprint (PDF) · General arrangement (PDF) · Calculations · Prototype build plan · Review note
Outdoor sensing projects in the lab fail for the same few reasons: water gets in, the battery runs flat in a cloudy week, the charger overheats or will not charge in the cold, or the radio link drops. FieldNode solves these once. It pairs a stock IP65 enclosure with a 6 W panel that also hoods the box, a single LiFePO4 cell, an MPPT charger with a cold-charge lockout, and an STM32WL-class LoRaWAN module, with two sealed sensor ports. Each project then designs only its sensor and firmware, and a fix found in one deployment reaches all of them.
Keeping it open and garage-buildable matters because the users who most need monitoring are the least able to buy closed commercial nodes or pay for their clouds. Every part is off the shelf or cut from aluminium bar, angle and sheet with hand tools, the design files are under CERN-OHL-S-2.0, and the node talks to any LoRaWAN server, including the lab's TwinKit gateway and The Things Network.
Monitoring is thinnest where hazards are greatest. Germany has more stations meeting the WMO Global Basic Observing Network standard than the whole of Africa (WMO). Only 108 countries, 55 % of the total, reported multi-hazard early warning systems in 2024, and fewer than half of the least developed countries did (UNDRR and WMO, 2024). The same report found that countries with limited early warning coverage have disaster mortality nearly six times higher than those with substantial coverage.
For air quality, 37 % of countries do not legally require monitoring at all (UNEP, 2021). Low-cost nodes can help close these gaps only if they survive in the field; long-running deployments report water ingress, cold-weather charging failures and radio problems as the main causes of lost data (Barrenetxea et al., SenSys 2008).
| Industry | Use |
|---|---|
| Water utilities and water resources | River, drain and well level logging (FloodGauge, WellSense, WaterWatch) |
| Disaster risk and civil protection | Slope movement, flood and heat sensing that feeds local early warning (SlopeWatch, HeatMap Node) |
| Municipal services and smart cities | Air, noise, parking and curb sensing on street poles, counts or levels only (AirStreet, NoiseMap, CurbCount, LoadZone) |
| Civil infrastructure and mining | Bridge vibration, tailings dam and pit wall tilt monitoring (BridgePulse, SlopeWatch) |
| Agriculture | Soil moisture, microclimate and grain store conditions at farms without power |
| Research and education | A documented, repeatable field platform for university and school projects |
| Country or region | Why it matters there |
|---|---|
| Sub-Saharan Africa | Weather observation is sparse: Germany alone has more GBON-standard stations than the whole continent (WMO). |
| India | Only about 12 % of 4,041 census cities and towns have air quality monitoring stations (CSE, 2023). |
| Least developed countries and small island states | Fewer than half of least developed countries report multi-hazard early warning systems; about two thirds of small island developing states do (UNDRR and WMO, 2024). |
| United States | The USGS runs about 12,165 streamgages (USGS), yet small streams, farm drains and urban culverts are mostly ungauged; low-cost nodes can fill local gaps. |
| European Union and United Kingdom | Dense official networks exist, but LoRa nodes must respect 1 % duty-cycle sub-bands in EU868 (TTN, citing ETSI EN 300 220), so an open reference that stays within the rules is useful to community sensing groups. |
The idea traces back to one of the first long outdoor wireless sensor deployments, on Great Duck Island, Maine, in 2003. Its 150 battery-powered nodes in sealed enclosures showed where field nodes fail: when the enclosures were opened, 22 % of the weather nodes had visible water droplets inside, and the multi-hop burrow nodes reached a median life of 34 days, under 45 % of their estimate (Szewczyk et al., SenSys 2004). None of those failures came from the sensors; they came from the box, the seal and the power budget. FieldNode starts from that lesson: solve the enclosure, the energy budget and the radio once, check them on paper, and let every outdoor sensing project reuse the answer.
Every outdoor sensing project rebuilds the same things: an enclosure that survives sun and rain, a small solar charger, a battery that lasts the night, and a low-power radio. Commercial nodes are closed or costly, and one-off builds fail in the field for the same few reasons.
A solar-powered, weatherproof sensor node with a common mounting, power and radio core, so any lab project that needs to measure something outdoors starts from the same tested base instead of a new enclosure and charger each time.
Full design precis: docs/02-concept.md
- IP65 polycarbonate enclosure, 150 x 90 x 200 mm, with membrane vent and two cable glands
- 6 W, 9 V class solar panel on a tilt bracket, doubling as a rain hood
- Ventilated white sun shield, fitted at hot-climate sites only
- LiFePO4 cell, 3.2 V 6 Ah, fused, with a 0 to 45 °C charge lockout
- MPPT charge and power board with switched 3.3, 5 and 12 V sensor rails and a serial programming header, reached with the lid open
- STM32WL-class microcontroller with LoRaWAN radio (US915 first variant, 915 MHz whip) and SPI flash for store and forward
- Two sealed M12 sensor ports with a proposed standard pinout (pin 1 switched rail, 2 data A, 3 ground, 4 data B, 5 analog), awaiting sign-off by the adopting projects
- Pole and wall mounting kit for 40 to 60 mm poles
TRL 3 calculations (FND-CAL-001): in the worst month (2 peak sun hours) the cell stores 7.75 Wh a day against 2.67 Wh drawn at the published 100 mW sensor allowance, and a full cell lasts 5.75 days without sun. The base node costs $139.50 in parts and weighs 2.45 kg. At sites whose design maximum exceeds 30 °C a ventilated white sun shield is fitted ($148.50 and 2.61 kg in all), which keeps the inside at or below about 52 °C in 45 °C sun so the cell can still charge. The panel is 9 V class so the charger keeps its input voltage when hot, and firmware lengthens the reporting interval at slow spreading factors to stay within The Things Network's fair use. Two requirements remain at risk: charging on very cold days and autonomy with a cold or aged cell. See the requirements.
The working bill of materials is in bom/bom.csv.
The prototype build plan (FND-BLD-001) shows, in pictures, how to make each of the fifteen components and put them together in twelve steps; nothing has been built yet. The made parts are an aluminium back plate and V-blocks, a bracket of angle clips and flat bars, a printed internal plate and a folded white sun shield; the bought box is drilled, and the electronics are bought modules wired at block level. Writing the plan made the design buildable: the V-blocks, the holes in the bottom of the box, the bracket joints and the shield fixing were redesigned, and fixings, a connector strip and sensor-supply fuses were added (FND-DDR-003, open for Amish's review). Every picture is drawn from the model, and the model checks that each part touches what it should and clears what it should not.
Lithium cells can overheat, vent and burn. Use protected cells or LiFePO4, fuse every pack, charge only within the cell maker's limits (for LiFePO4, typically 0 to 45 °C) and never leave a first build charging unattended.
Mounting is work at height: use a stable ladder with a second person present, keep clear of overhead power lines, and check the pole or wall can carry the wind load on the panel.
| Folder | Contents |
|---|---|
docs/ |
Problem, concept, requirements, calculations and design decisions |
cad/src/ |
build123d Python source, the source of truth for all geometry |
cad/step/, cad/stl/ |
Exported models for FreeCAD, other CAD tools and printing |
cad/drawings/ |
2D sketches and dimensioned drawings |
bom/ |
Bill of materials |
electronics/ |
KiCad schematics and PCB layouts |
firmware/ |
Microcontroller code |
media/ |
Renders, perspectives and photos |
build-log/ |
Dated prototyping notes |
Controlled documents follow the portfolio documentation standard. Each carries a document ID (FND-PRC-001 for the precis), a version and a revision history. Branded PDFs are built with python .kit/render.py and attached to GitHub Releases when a document is tagged, for example FND-PRC-001/v1.0.
Designed by Amish Chadha. See CONTRIBUTORS.md for roles. To cite this design, use CITATION.cff (GitHub shows it as "Cite this repository").
AI assistance (Claude) was used to accelerate concept renders, prototype documentation and first-pass sizing calculations. Design direction and all decisions are Amish Chadha's, recorded in this repository's decision records (docs/decisions/).
- Hardware (CAD, drawings, BOM, electronics): CERN-OHL-S v2
- Software (firmware, scripts, notebooks): MIT
A project of the Design Molecule lab.