Floating solar tracking platform

the system at a glance

The problem

Ponds, park water and other small bodies of water are hard to monitor continuously: there’s rarely power nearby, and battery-powered sensors need constant servicing. Our brief was to design and build a floating platform that powers itself from the sun, points its panel at the light, and carries environmental sensors that report to the cloud.

What it had to do

We wrote the requirements as numbers so we could test against them later:

Keep the panel within 3–4° of the light source.
Stay stable, tilting less than 10° when pushed or hit by waves.
Keep every electrical part dry.
Run on its own with minimal power draw, and report readings to an IoT dashboard.

Concepts we considered

We built a morphological chart for each function and scored the combinations in a weighted decision matrix:

Function	Options we weighed
Sensing	Camera, moisture sensor, light sensor, thermal sensor
Motion	Servo, stepper, DC motor
Frame	PLA, wood, PVC pipe
Flotation	Pool noodle, paddle board, insulation panel
Energy storage	Lead-acid, lithium-ion

The winning combination was a PVC frame on pool-noodle floats, light-dependent resistors for sensing, a lithium-ion battery, a stepper for azimuth and a servo for tilt. It wasn’t the most sophisticated option on any one line, but it was cheap, easy to build and reliable, and it scored best overall.

The design

Azimuth: a NEMA 17 stepper turns the whole mast. A stepper holds its position without drawing current, which matters on a solar budget.
Tilt: an MG90S servo on a 3D-printed pivot sets the panel angle for the season.
Sensing: light-dependent resistors compare brightness from different directions, and the Arduino steers toward the brighter side. A pH probe pokes through a hole near the centre of gravity, and an ESP32 camera module watches the water.
Power: the panel charges a 12 V lithium-ion pack; a buck converter steps it down to 5 V for the Arduino and sensors.
Layout: the electronics box sits dead centre so the centre of gravity stays low and level, with the mast rising from the middle of it.

The electronics live in a polypropylene food container. It isn’t glamorous, but it’s waterproof, cost $2.50, and got us to testing quickly.

The numbers behind it

I worked through the design calculations with the team, using a safety factor of 1.2 on everything to cover wind gusts, extra mass from wiring and fasteners (we allowed 12%), and printing variation.

Check	Required	What we chose
Tilt torque (panel weight on its lever arm)	0.096 N·m	MG90S servo, comfortably above
Azimuth torque (10 m/s wind on the panel + joint friction)	0.057 N·m	NEMA 17 stepper, far above
Buoyancy	1.56 L displaced	Four pool noodles
Stability	Righting > upsetting	Margin of 2.99
Power draw	—	About 14.3 W, 1.19 A from the 12 V pack

Before printing anything, the tilt mount went through SolidWorks Simulation: peak stress was around 0.17 MPa against roughly 50 MPa yield for PLA, with very little deflection. A thermal study of the enclosure confirmed the sun-facing lid gets hottest while the walls shed heat by convection.

Risk assessment

The FMEA flagged four main ways it could fail: the platform tipping, water reaching the electronics, an actuator stalling, and not enough power coming in. Each one shaped a design choice — symmetrical floats and a central box for stability, a sealed enclosure, oversized motors, and a stepper that doesn’t burn power holding position.

Testing

Most of the flotation testing happened in a bathtub: a basic float-and-payload test, stability, push-and-release, slosh and waves, and a water-ingress check. The platform stayed up under loads well beyond our components’ weight and settled back after every disturbance.

Requirement	Result
Tracking within 3–4°	Panel turned toward the brightest light, confirmed on the dashboard readings
Under 10° tilt when disturbed	Floated loaded and recovered from pushes and waves
Tell water conditions apart	pH sensor clearly separated tap water from orange juice

What I’d change

The biggest lesson was how much early CAD work paid off: moving parts around in the model caught interference between the tracker and the enclosure before anything was printed. Next time I’d design a purpose-built sealed housing with a rubber gasket and proper cable routing instead of adapting an off-the-shelf box.