Q&A - webinar: "An intelligent sensor for sustainable always-aware applications"

[ILARIASOFIA.GRASSI]

This thread has the questions and answers from the An intelligent sensor for sustainable always-aware applications webinar.

Can accelerometer data collected by other tools (3rd party data loggers) be used as inputs to Unico-GUI to perform the MLC/FSM analysis?

Data collected with other tools can be used in Unico-GUI if the format is compatible, in some cases you may need to do some conversion to adapt it. For the MLC tool in the Unico-GUI, you can refer to the information contained in the MLC application note (e.g. document AN5901 for LIS2DUXS12).

Do you have any sample data-set for generic movement like Steps, Wrist movement, etc?

In our MLC repository on GitHub you can find different application examples including some sample data-set of the detected classes.

What is the difference between Unicleo-GUI and Unico-GUI? Is the configuration generated specific to a certain hardware or can it be used in other platforms?

Unicleo-GUI is the graphical user interface for STM32 Nucleo boards and X-CUBE-MEMS expansion, it allows to acquire sensor data and display outputs.

Unico-GUI instead is a sotware package for the evaluation of all MEMS sensors, it allows to be generate sensor configurations even in offline mode (without any board connected). The configurations generated through Unico-GUI (e.g. .ucf and .h files) can be used also in other software tools and platforms.

Can Unico-GUI collect raw data from sensors and write them to log file in continuous mode for long time?

Yes, Unico-GUI can interact with the STEVAL-MKI109V3 board (Professional MEMS tool) to get sensor data and log to files. This motherboard is compatible with all ST MEMS adapter boards.

Can we use the features FSM MLC and ASC with data coming from an external sensor?

No, the LIS2DU family currently does not allow using data from external sensor.

However, in LIS2DUX12 we already have the possibility to use internal temperature data. While in LIS2DUXS12 we have the possibility to use temperature and also Qvar sensing channel (for electrostatic charge variation) both in FSM and in MLC.

For the possibility to use other sensor data you can check other devices available on our website, for instance the LSM6DSV family embeds a sensor hub which allows connecting an external input through the I2C master interface.

What is the difference between LIS2DUXS12 and LIS2DH12?

LIS2DH12 and LIS2DUXS12 are two accelerometers with similar basic features (e.g. low power consumption, I2C/SPI interfaces, embedded features like free fall and wake up, FIFO, etc…), both of them in a small package (2x2 mm).

However, LIS2DUXS12 is part of a new generation of accelerometers including more advanced features including: pedometer, anti-aliasing filter, finite state machine (FSM), machine learning core (MLC), adaptive self configuration (ASC), analog front-end / Qvar sensing channel.

What is the power consumption of MLC and FSM features, compared to the sensor power consumption?

The additional current consumption required to run MLC and FSM features, is just few uA or few tens of uA, depending on the MLC or FSM configuration.

However, you should consider the advantage you will get in terms of power consumption at system level, since these features allow to set the MCU in sleep mode most of the time, reducing the overall system power consumption through the edge processing. We can say it is possible to save from 10 to 1000 times the amount of power.

Is there a guide available that explains how to set up the FSM using Unico?

A dedicated application note for FSM describe how to configure the FSM feature using Unico-GUI. For LIS2DUXS12 the document name is AN5902.

Can I use other machine learning models like neural networks instead of the decision tree in LIS2DUXS12?

No, this is not possible with this family of products, which works with decision tree model only.

However, we have other devices embedding the ISPU feature (intelligent sensor processing unit) where you can bring your custom algorithms directly in the core of the MEMS Sensor. You can check for ISPU on our website and find two products, the ISM330IS and the LSM6DSO16IS for this purpose.

Is this new accelerometer family pin-to-pin compatible with older accelerometers or older families?

Yes, all our accelerometers in 2x2 LGA package are designed to be pin-to-pin compatible to facilitate the redesign activities of those customers that want to migrate to higher performance solutions.