We Built a Wind Tunnel
That Fits in Your Jersey Pocket.
Traditional aero testing means booking a wind tunnel at $500/hour, travelling to one of a dozen facilities worldwide, and hoping the controlled airflow matches what you actually ride through. We thought there had to be a better way.
How It Actually Works
The sensor sits behind your saddle rail — 14 grams, barely noticeable. Inside, four core subsystems run simultaneously at 100Hz, cross-referencing each other to build a complete picture of your interaction with the air, the road, and your own body.
6-Axis IMU
Three-axis accelerometer plus three-axis gyroscope, sampling at 100Hz. We use a Bosch BMI270 — the same chip you'll find in surgical robotics. It tracks your body's orientation in space with ±0.1° precision, which means we can detect the difference between you riding on the hoods versus the drops within a single pedal stroke.
Barometric Altitude
A Bosch BMP390 pressure sensor measures altitude changes down to ±0.1m. This isn't just for climbing stats — combined with GPS speed data, we calculate air density in real time. Air density swings 5–8% between sea level and a mountain pass, and your CdA calculation is only as good as your air density estimate.
Strain Gauges
Four bonded foil strain gauges in a full Wheatstone bridge configuration. We went through 14 prototypes to get the bridge balance right — too sensitive and road vibration drowns the signal, too stiff and you lose the low-frequency pedaling forces. The final design resolves forces down to 0.5N across a 0–2000N range.
Sensor Fusion Engine
Raw data from all four subsystems feeds into an extended Kalman filter running on an nRF52840 SoC. The filter fuses IMU orientation, barometric pressure, strain force vectors, and GPS velocity into a single coherent model. Latency from physical event to computed CdA: under 10 milliseconds.
CdA: The Number That Matters
Your drag coefficient times frontal area — CdA — determines roughly 80% of the resistance you fight above 25 km/h. Reduce your CdA by 5% and you're looking at a 30-second gain over a 40km time trial at the same power output. We measure it continuously, outdoors, while you ride.
How We Compare
| Method | Cost | Location | Real-time | Accuracy |
|---|---|---|---|---|
| Wind Tunnel | $500–1,500/session | Lab only | No | ±1% CdA |
| Velodrome Testing | $200–500/session | Track only | No | ±3% CdA |
| Chung Method (GPS) | Free (DIY) | Outdoors | No (post-ride) | ±5–8% CdA |
| DIDI.BIKE Sensor | $299 one-time | Anywhere | Yes (<10ms) | ±2% CdA |
Our Approach to Accuracy
Validated Against Wind Tunnels
We ran 200+ hours of comparative testing at the RMIT Industrial Wind Tunnel in Melbourne and the San Diego Low Speed Wind Tunnel. Our sensor readings correlate at r² = 0.94 with tunnel CdA values across rider positions from full aero tuck to upright climbing stance.
Field-Tested on Real Roads
Lab correlation is necessary but not sufficient. We tested with 12 semi-pro riders across three months of riding in Calpe (Spain), Girona, and the Adelaide Hills. Crosswinds, traffic, temperature swings from 4°C mornings to 32°C afternoons — the sensor handled all of it within our published accuracy spec.
Open Data Protocol
Every data point is exported as a standard .FIT file. No proprietary lock-in. Pull your rides into TrainingPeaks, Golden Cheetah, intervals.icu, or your own Python scripts. We publish our data schema openly because we believe reproducibility matters more than walled gardens.
Connectivity & Integration
The sensor broadcasts on both ANT+ and Bluetooth LE 5.0 simultaneously. Pair it with whatever head unit you already own — no ecosystem lock-in.