The LSM6DSV320X is the first mainstream inertial sensor to house a gyroscope alongside two accelerometers, one capable of sensing up to ±16 g and one sensing up to a staggering ±320 g.
The new device can thus measure extremely high impacts without necessitating an external sensor. Additionally, as a member of our smart IMU family, it also includes a machine learning core, a finite state machine, an adaptive self-configuration, and a Sensor Fusion Low Power (SFLP) IP, enabling on-device computations. The LSM6DSV320X thus opens the way for applications like automatic emergency call, thanks to car crash detection on mobile or wearable applications, or to track any highly intensive dynamic motion on consumer devices to sense what’s happening to users and all around them
Table of Contents
Sensing acceleration
What are accelerometers?
In the most simplistic terms, an accelerometer measures proper acceleration, meaning an acceleration relative to the Earth’s gravity. That’s why manufacturers specify the range of acceleration measured using G-force or “g”. 1 g is about 9.81 m/s2, which is the Earth’s gravity. Hence, a still smartphone on a flat surface will exhibit about 1 g on one of the three axes because the surface pushes against the Earth’s gravity. Comparatively, a tablet in free fall will output 0 g since no force is applied except the Earth’s pull. And while acceleration is most often measured along three axes (x, y, and z), the scale uses ± to represent a positive or negative direction along these axes.
Why do we have low-g and high-g accelerometers?
Accelerometers measure a range between a minimum and maximum proper acceleration before they saturate and must, therefore, balance range and precision. An accelerometer measuring tilt or vibrations when taking a smartphone picture must be attuned to fine movements. The sensing element, hence, uses a suspension supple enough to detect minute changes in direction. The problem is that a significant acceleration will saturate this fine sensing element. Conversely, a high-g accelerometer features a sensing element with a stiffer suspension. Hence, while it cannot measure movements as fine as a low-g model, it can detect high impacts without saturating. That’s why engineers trying to detect both types of accelerations need dedicated solutions.
Sensing acceleration in consumer products
What are the challenges in small devices?
Today, almost all personal electronics products and IoT nodes have accelerometers with a scale going up to ±16 g, which is enough to determine the screens’ position, monitor vibration, and detect usual human activity on such a product. Using multiple accelerometers means a larger PCB to accommodate the additional components, which is impossible on small consumer products. However, there’s a demand for high-g accelerometers because they enable new applications. Concussion monitoring requires a device capable of sensing 120 g. Similarly, heavy equipment monitoring, asset tracking with damage sensing, or crash detection requires over 200 g. The problem is that detecting such high acceleration demands an ad hoc device or a certain creativity.
How has the industry dealt with this problem so far?
Typical accelerometers on consumer products never go as high as ±320 g. Manufacturers wanting to track a crash or other intense movements must, therefore, find alternatives. Some have used a combination of sensors, such as a gyroscope, a pressure sensor, a GPS, and microphones to detect signs of distress, like glass breaking. But these solutions are rare in consumer devices. They are expensive to build, demand significant expertise, massive investments in testing and development, have prohibitively expensive BOMs, and complex power management schemes. This is why the feedback ST got when we showcased the LSM6DSV320X at CES, MWC, and Embedded World has been phenomenal. It makes tracking intense impacts so much more accessible without any accuracy or power consumption compromises.
LSM6DSV320X: Sensing what happens to the users and all around them
How would the LSM6DSV320X work in everyday situations?
The LSM6DSV320X is unique because it can serve multiple functions simultaneously. If we take the example of a car or asset tracker, the low-g channel helps navigation systems detect stops or turns, while the high-g channel records drops and massive impacts. The device logs both data sets. The low-g channel provides high resolution and low noise for fine measurements, whereas the high-g channel offers an extended range for abrupt shocks. And because they are independent, saturating the high-g sensor doesn’t affect the low-g sensor’s reading and vice versa. This duality also means developers can optimize an application for each sensitivity by applying separate filtering and processing, depending on the sensor, for greater accuracy.
How did the LSM6DSV320X become possible?
We can put a low-g and a high-g accelerometer side by side because we’ve been working on miniaturizing sensors for more than 20 years. It started when ST shrank sensing elements as we improved our lithographic processes. Then, in 2015, we launched the LSM6DS3, which made waves by being our first IMU to house a gyroscope and accelerometer under one roof. 10 years later, the LSM6DSV320X features a gyroscope and two accelerometers on the same die. We also included a machine learning core, finite state machines, a Sensor Fusion Low Power (SFLP) IP, and an adaptive self-configuration IC on the same die, thus leveraging the work we started with the LSM6DS0X, the first MEMS with ML capabilities.
The presence of all this intelligence means that the LSM6DSV320X can automatically switch from outputting the results of the low-g accelerometer to the high-g alternative and vice-versa. Both constantly send information, and developers never have to worry about missing data. The new device simply reconfigures itself depending on the motion it senses, and its machine learning core can even offer contextual awareness. In practical terms, the device can do on-board processing of the data collected without waking the host MCU. Hence, thanks to the machine learning core, finite state machine, and adaptive self-configuration, the LSM6DSV320X is capable of more than just the sum of its parts.
Next steps
The best way to start experimenting with the new sensor is to grab the STEVAL-MKI109D evaluation board and the STEVAL-MKI251A adapter board, or a SensorTile.box PRO, to start running example code and demo applications. This software demonstrates that the LSM6DSV320X can work in systems that don’t necessarily demand high-g detection. The fact that it is pin-to-pin compatible with the previous LSM6DSV IMUs and still uses a compact 2.5 mm x 3 mm package means that engineers can upgrade their system and offer new features while reusing their PCB layout and initialization code. We also provide drivers and application examples on GitHub to facilitate the development process.
One example is ST’s Motion XLF library. Included in the X-CUBE-MEMS1 software package, the library provides an automatic way for developers to activate the high-g accelerometer to conserve power. We implemented mechanisms to ensure that applications would benefit from both sensors while still ensuring the lowest power consumption possible. Consequently, programmers don’t have to implement their algorithm and can focus instead on the features that will help their application stand out. The library also handles time-varying offset evolutions, discontinuity, and noise profile inconsistencies between the low-g and the high-g accelerometers, ensuring a smooth data transition between the two.
