Reverse Engineering the Ecowitt WS90

Before designing our own ultrasonic anemometer, it helps to see how commercial products solve the same problems. Today I tore down an Ecowitt WS90 to understand their approach.

The WS90

The WS90 is a consumer-grade ultrasonic weather station that sells for around $150 -> $200. It measures wind speed/direction, temperature, humidity, UV, and light intensity. For the price point, it's interesting to see what engineering tradeoffs they made.

Key Findings

Mesh Reflector Instead of Solid Plate

The biggest surprise: the ultrasonic transducers reflect off a metal mesh rather than a solid plate. Most DIY designs use solid reflectors.

The mesh likely serves multiple purposes:

• Drainage: Water doesn't pool on the reflector surface
• Debris clearing: Dust, pollen, and small insects fall through
• Reduced acoustic resonance: A solid plate can ring, creating spurious echoes

The tradeoff is presumably some signal loss, but with properly tuned transducers and gain, this seems acceptable.

Vibration Isolation on Transducers

Each ultrasonic transducer is mounted with rubber isolation pieces. This decouples the transducers from the housing mechanically.

Why this matters:

• Reduces crosstalk: Vibrations from one transducer firing won't mechanically couple to adjacent transducers
• Wind noise rejection: Housing vibrations from wind don't inject noise into the acoustic path
• Temperature stability: Rubber absorbs thermal expansion differences between the transducer and plastic housing

STM32 Main Processor

The brains of the unit is an STM32 microcontroller. This is a solid choice—STM32 parts have good timer peripherals for precise ultrasonic timing measurements, low power modes for battery operation, and are well-documented.

Mysterious Plastic Extrusions

The plate holding the ultrasonic transducers has numerous plastic extrusion features molded into the surface. Their purpose isn't immediately obvious.

My best guess: multipath reflection mitigation. In an ultrasonic anemometer, you want the sound to take a direct path from transducer to reflector and back. Any secondary reflections off the housing walls create ghost signals that corrupt the timing measurement.

These extrusions might:

• Break up coherent reflections from the housing walls
• Scatter sound energy that would otherwise create standing waves
• Provide acoustic baffling between transducer pairs

This is worth investigating further. If multipath is a significant issue, similar features might be needed in our open source design.

Takeaways for LOAF

Several design decisions from the WS90 are worth considering:

1. Mesh reflectors could solve the water/debris problem elegantly
2. Rubber isolation for transducers seems essential for reliable operation
3. STM32 is a reasonable processor choice, aligns with QingStation's approach
4. Multipath mitigation in the housing design needs more research

Next step: bench testing to understand how much these features matter for accuracy.