Called BHI385, it also includes a 32bit processor running embedded AI software “creating a complete programmable system”, claimed the company. It “accurately captures dynamic movements such as smashes, swings, and jumps [enabling] motion analysis for high-impact events across sports, fitness and gaming”.
Features include six and 9 degree-of-freedom sensor fusion, step counting, activity recognition, and single, double and triple tap gesture detection).
“Self-learning cyclic gesture recognition allows it to detect repeated motion patterns and adapt them to the individual, not only recognising what is being done, but also how well it is being performed,” said Bosch. “This enables quality-of-movement evaluation in sports or rehabilitation training.”
All this said, neither the data sheet nor the software is yet available to asses the algorithm capabilities in detail – the device is not available until Q3.
Examples of typical consumption have been given: <500 µA at 50Hz for sensor fusion, <50µA for simpler accelerometer-based algorithms, or 8μA in stand-by.
Measurement range options (all 16bit) are ±4, ±8, ±16 and ±32g for the accelerometers, the latter limited to the 28g, and ±125, ±250, ±500, ±1,000 and ±2000°/s for the gyros.
The offset over life, after soldering, is ±25mg for the accelerometers and ±0.5°/s for the gyros.
Interfacing is a mix of SPI and I2C, with output rate options between 12.5Hz and 1.6kHz for the accelerometers and 12.5Hz to 6.4kHz for the gyros.
Operation is across 1.71 to 3.6V for the logic, 1.71 to 1.89V for the IO, and -40 to +85°C.
Packaging is 2.5 x 3 x 0.95mm 20pad LGA.
There is some more information on the Bosch BHI385 product page.
In February, ST announced a 2.5 x 3mm inertial measurement unit with six accelerometers for sports wearables.
