Low-power application design with MEMS accelerometers

Low-power applications take advantage of MEMS accelerometer sensors for increasing battery life.

Sensors are becoming less and less power hungry and embed features that help decrease overall system power consumption. For example, motion-activated wake-up allows the entire system to sleep when the user is not using the device. Nevertheless, there are also other ways to make use of a MEMS accelerometer to decrease overall power consumption.

Starting from the MEMS accelerometer sensor itself, it should be flexible in its operating modes. As depicted in Figure 1, there is a well-known trade-off between a sensor’s resolution and its output data rate on one side and the current consumption on the other side — the higher the resolution or the data rate, the higher the current consumption and vice versa. Fortunately, there are sensors on the market that are able to operate at a few microamps and consume a couple of nanoamps in power-down or standby mode.

For high-demanding applications the operating mode of the sensor can be changed on the fly, utilising higher resolution and data rates only when it is really needed. Some sensors are even able to perform this mode switching automatically. The user configures the resolution and data rate needed in the active state and defines a condition for enabling it. Then the sensor is switched to the inactive state, where it still measures data but at a very low data rate and resolution, waiting for the condition (a motion event) to switch it back to active state.

Figure 1: Sensor parameters impacting battery life.

Another good design practice is to utilise a low power supply level, because lower power supply levels mean lower current consumption. That’s why for low-power applications a 1.8 V power supply is preferred.

In some designs, power cycling of the sensor can be used. The sensor’s power supply is activated only when motion data is supposed to be measured; otherwise the sensor is powered off. This can even be achieved by supplying the sensor from a microcontroller’s pin, as shown in Figure 2. When applying this technique the power consumption budget needs to be calculated properly, because with every start of the sensor the user needs to configure it and wait until its outputs are settled to provide the correct data.

Figure 2: Sensor’s power supply control by microcontroller pin.

The majority of MEMS accelerometers are digital sensors, which means they internally convert measured analog signals to digital data. Benefits include a smaller bill of materials due to the integrated AD converter and lower susceptibility to signal distortions. Thanks to embedded interrupt generators, MEMS accelerometers can generate a trigger signal when certain user-parametrised conditions are met. This is where motion-activated wake-up comes from. The microcontroller (MCU) configures the sensor to generate a wake-up trigger and goes itself into a very low-power sleep mode. When a motion is detected, the sensor will generate an interrupt signal; the MCU then receives the signal, switches into an appropriate operating mode and finally handles the situation that has occurred.

Digital sensors can also take over tasks related to motion processing normally done in an MCU. An MCU could do the same job of course, but with much lower power efficiency — the MCU in the milliamp range and the sensor in the microamp range. Detection of movements like free fall, single- and double-tap (user actions similar to a mouse click), portrait/landscape orientation detection and others are realised by an internal logic sensor. An MCU does not need to make any computations; it just waits for an interrupt trigger and reacts to the movement only when it occurs.

Digital sensors often integrate configurable filters which are conditioning acceleration data just measured. These can be low-pass, high-pass or even anti-aliasing filters that pre-process data for the MCU and offload it even more.

A data buffer, in most cases FIFO type, embedded in a sensor will lower current consumption too, because it will allow the MCU to read data less frequently. This way the MCU will be able do other tasks, stay asleep for longer and save time needed for serial communication with the sensor.

Serial communication between the sensor and MCU contributes to overall power consumption too. For very low-power applications dealing with every microamp, serial communication could have a significant impact. Most MEMS accelerometers communicate over SPI or I²C interfaces. An SPI interface is more efficient in terms of power consumption for three reasons: first, there are no pull-ups on communication lines causing extra current consumption; second, it supports higher data rates; and third, it has less overhead in serial protocol.

Regardless of the interface, significant reduction of serial communication can be achieved if instead of polling the sensor, ie, continuously asking for the status of new data availability, the application rather utilises data-ready interrupt. Data-ready interrupt is automatically generated by a sensor when it has finished data measurement and conversion and a new set of data is ready to be read by the MCU. When this interrupt is activated, the MCU can immediately read output data from the sensor in one single-read operation.

As stated already, a lower sensor output data rate means lower current consumption. A so-called single data conversion mechanism allows the data rate of the sensor to match perfectly with application needs, as shown in Figure 3. Using this mechanism, measurements are started either by an external trigger signal routed on a sensor’s pin or by register write initiated from the MCU using serial command. Data acquired this way is then stored inside the sensor. The sensor can also initiate a data-ready interrupt signal informing the MCU that data conversion has been completed and data is now available to be read by the application. Thanks to this feature, data rates even smaller than 1 Hz or basically any other rates beside the predefined ones are achievable.

Figure 3: Single data conversion mechanism.

We have discussed the features of a MEMS accelerometer that are important for low-power applications and also ways how to utilise them in system design. STMicroelectronics’ latest ultralow-power 3-axis MEMS accelerometer, the LIS2DW12, brings flexibility in designing new applications with accelerometer sensors thanks to its low current consumption down to 1 µA, number of operating modes, wide range of output data rates, rich set of embedded digital features, high temperature stability, and enhanced features like digital filters and FIFO buffer. Advantages include motion-activated functions and user interfaces; smart power saving for handheld devices; motion detection for appliances; and impact recognition logging for wireless sensor nodes.

Image credit: ©stock.adobe.com/au/Creative-Touch