EYE ON NPI - ST LSM6DSV32X 6-Axis IMU with 32 g Accelerometer

This week's EYE ON NPI gyres and gimbels through the wave. It's the latest LSM6DS 6-axis accelerometer + gyroscope IMU from STMicroelectronics. (https://www.digikey.com/en/product-highlight/s/stmicroelectronics/lsm6dsv32x-6-axis-imu). This latest update to the LSM6DS line is packed with features such as I3C, step counter, Qvar electrostatic sensing, finite state machine, AI 'learning core' and embedded sensor fusion so not only are you getting an IMU but a full user interface and sensor management hub. We've been stocking and using the LSM6DS series for many years: starting with the LSM6DS33 (https://www.adafruit.com/product/4480) a great low cost 6-DoF to the LSM6DS3C-TR replacement (https://www.adafruit.com/product/4503) to the fancy-pants LSM6DSOX (https://www.adafruit.com/product/4438). These are pin-compatible, and nearly-firmware-compatible sensors that are compact, sensitive and precise. In particular, what we look for in a good IMU chip is the gyro offset and drift. When making an inertial measurement unit (IMU) especially one used for orientation calculations, the quality of the gyroscope has a big impact on the jitter/speed/drift. (https://learn.adafruit.com/adafruit-sensorlab-gyroscope-calibration/comparing-gyroscopes) Accelerometers are pretty good quality these days, we've figured out how to make them decades ago. And magnetometers are not as important - they don't update often and are used for orientation correction. Gyro's have the most error possibilities. In general, the more you pay, and the newer the sensor, the better performance you'll get in the form of low offset and low drift. The LSM6DSV32X has similar gyro/accel characteristics from the LSM6DSOX and ISM330DHCX - although we'll note the min/max is not documented, just the typical values. And there's some ISM330 parameters such as Angular random walk and Bias Instability, that are not specified. That said, as long as you're good on calibration, and the specs are within your limits, the embedded functions are where the LSM6DSV32X really shines. Like many ST accelerometers, there's a built in pedometer (https://github.com/adafruit/Adafruit_LSM6DS/blob/master/examples/adafruit_lsm6ds_pedometer/adafruit_lsm6ds_pedometer.ino) as well as a tilt / motion sensor. The first is good for step counting, the second lets the sensor go into a low power state and set an interrupt pin when the device is moved or tilted to wake up the rest of the UI. The more advanced functions include a 'machine learning core' (https://www.st.com/resource/en/application_note/an5804-lsm6dsv16x-machine-learning-core-stmicroelectronics.pdf). We use that term lightly because it's more like a 'decision tree' where you can set up test conditionals post-normalization/filtering to classify simple motion states such as "running", "sitting" "lying down". These won't be as flexible as a true machine-learning algorithm that uses sample data to create a model, but you also don't have to spend any power or computation. Another advanced feature ST has been adding to many of their chips is Qvar electric charge variation detection sensing. This ultra-sensitive capacitance-like sensing can be used for human presence and motion detection, as well as touch/proximity detection. The eval board comes with a few Qvar plates for trying it as a slide or touch sensor as well as a 'radar' like proximity detector. (https://www.digikey.com/en/products/detail/stmicroelectronics/STEVAL-MKI240KA/22522578) This could be nice to use if you've got a microcontroller that does not have touch sensing peripherals, since now you don't need a separate chip. The new feature we're most interested in is the "Sensor fusion low power" block - which can output a game rotation vector in quaternions, that gives the chip attitude (https://en.wikipedia.org/wiki/Orientation_(geometry)) direct from the sensor, without needing to continuously read/buffer/analyze the 6-DoF data. We'll note that there's no magnetometer, so it won't have 'absolute' orientation against the earth, but for many use cases a rough orientation - especially one that is self-driving and low power - will suffice. We'll note that for getting started, the LSM6DSV16X library on GitHub has examples for this and the other embedded features. (https://github.com/stm32duino/LSM6DSV16X) If you want to try out the new ST LSM6DSV32X 6-Axis IMU with 32 g Accelerometer (https://www.digikey.com/short/bhq3bwhf) or it's smaller sibling, the LSM6DSV16X (https://www.digikey.com/short/524744wt) both are available and in stock from DigiKey right now. Pick the 16g version if you're doing every-day motion sensing and 32g for sports / high speed activities. The eval board is also available and great if you want to start fast and also check out the new Qvar sensor feature (https://www.digikey.com/en/products/detail/stmicroelectronics/STEVAL-MKI240KA/22522578). Order today and you will be ready to integrate the latest IMU into your design by tomorrow afternoon.