Extending operational lifetime for battery-powered devices is crucial
This article is part five of our 'Match Made in Automation' Series. You can find the other installments of this series below:
- Part 1: Factory automation realizes boost from new technologies
- Part 2: How are magnetic rotary encoders used in industrial automation?
- Part 3: Explore options for choosing an optical rotary encoder for motion control and position sensing
- Part 4: AI takes on growing role in HVAC system efficiencies
Batteries are everywhere, in our phones, our watches, vehicles and any number of other devices. Many people are familiar with the nightly routine of charging a device for the next day.
Commercial and industrial applications rely on batteries. Remote sensors have become essential infrastructure in both consumer and industrial environments. However, while batteries offer flexibility and portability, they also have long-term service needs. There will be ongoing replacement costs and regular maintenance visits.
Battery power delivers flexibility and portability
Within the industrial domain, battery-powered IoT devices are simple to install. Typically equipped with wireless connectivity, they are quick to deploy since no wired connections—power or data—are necessary. Installing line power outlets and network cabling involves significant effort and labor costs. By comparison, with many battery-powered sensors featuring zero-provisioning techniques, they can take just minutes to install. However, the ongoing "truck roll" labor costs of sending an engineer to replace the batteries will significantly exceed the price of the batteries.
Battery replacement may even be impossible. An asset tracker onboard a vessel sailing from Asia to Europe could be at sea for a month or more. It may then be delayed in port during offloading before continuing its onward journey by train or truck. The tracker's usefulness is closely aligned to the battery life. Prolonging battery life has become a crucial design goal.
Uncovering the architecture of an asset tracker
Reviewing the key components of a battery-powered, wirelessly connected asset tracker is a good place to start when looking at ways to extend the tracker's in-service performance.
A microcontroller (MCU) is usually at the heart of any tracker, providing computational capabilities, interfacing with sensors, and processing and packaging measured data for onward transmission. Typically, an MCU is selected based on several technical criteria, including clock speed (an informal measure of computing capability), internal flash memory, peripheral functions (data conversion, serial bus connectivity, clocks, timers, etc.), security functions, and power consumption. More advanced MCUs are highly integrated, with functions such as a wireless transceivers and neural network accelerators. These elements may consume more energy when being used but should be easily disabled when not.
External sensors will be needed to capture and measure various parameters, which will be application dependent. For an asset tracker, this will likely include temperature, humidity, shock, and vibration forces. Geolocation data requires a global navigation satellite system (GNSS) receiver, typically provided by a module with a built-in antenna. The sensors used may or may not be designed with sleep modes, while the GNSS receiver will almost certainly have a low power setting.
The battery life of an asset tracker can be calculated based on the battery's capacity (mA hours) and the tracker's average consumption. The tracker's physical dimensions and weight will dictate the maximum battery size possible, putting a limit on the capacity.
Tips to prolong the battery life of an asset tracker
Determining the power profile of a battery-powered device requires investigation. First, like any MCU-based device, the current flow will be dynamic. Peaks will occur during periods of increased computational workload. Consumption will also go up when peripherals such as sensors, location modules and transceivers operate.
While measuring or estimating the maximum current consumption using a digital multimeter or data sheet is possible, the results are not particularly useful in calculating the potential battery life. Measuring average current flow over time provides a more accurate basis for calculating battery life. Achieving this requires a purpose-designed tool for embedded power profiling.
An example is the Nordic Semiconductor Power Profile II kit (PPK2). Used with the companion nRF Connect software tool, the PPK2 can measure and, optionally, deliver currents from sub µA up to 1 A. With the PPK2, embedded developers can visually understand the device's power consumption profile in real time, an essential step in reducing the average current and prolonging battery life.
Power profiling in action
The real-time current consumption view from the Nordic Semiconductor PPK2 power profiler app indicates the average and maximum current on a time axis. (Source: Nordic Semiconductor)
Once equipped with a real-time display of power consumption, the embedded development team can implement various techniques to lower the device's average current profile, including:
Duty cycle: The device's duty cycle significantly influences the average power consumption. How often does the MCU need to read the sensors? The use case and the parameters measured will, to some extent, determine this. For example, the shipping container asset tracker probably only requires geolocation and environmental measurements every 30 minutes or less at sea.
Try experimenting with different duty cycle settings and see the differences in average current.
Task scheduling: Can the pace of performing sensor readings and subsequent processing be spaced out to avoid significant current peaks? Wireless transceivers typically exhibit prominent energy peaks as they initiate a link to the host. Reading sensor data and immediately placing them back into sleep mode before initializing the wireless transceiver could soften power peaks and reduce the average current profile.
Some wireless MCUs allow the radio functions to be suspended independently of the MCU status. Investigating the options available from different MCUs is recommended.
Sleep modes: All MCUs have a variety of sleep modes. So do most sensor ICs and wireless modules. It is worth researching which modes are most effective in lowering the average power consumption profile. There may be a time penalty when waking from a deep sleep mode. The extra time may be required to stabilize oscillators before the firmware is ready.
Some micro-electromechanical systems (MEMS) accelerometers can enter a sleep mode as soon as measurements have been made. The results are placed into ultra-low-power I2C or SPI registers accessible to the host MCU.
MCU selection: Engineering teams should consider the trade-offs. It takes computational power to process sensor data, implement security encryption algorithms and communicate the results. An ultra-low power MCU may have a higher power profile if it needs to run harder for longer. However, over-specifying an MCU might incur consistently higher peak currents, even for shorter periods.
There are several trade-offs to be made between peak active current, deep sleep current and computational capability. EEMBC–a non-profit consortium–provides useful benchmarks for MCU performance and energy consumption. Its website can be a good starting point for making comparisons.
Energy Harvesting: Another approach to extending operational lifetime is to store harvested ambient energy in a rechargeable battery. Stored energy can supplement the primary battery. Potential energy sources include solar, vibration, thermal or a combination of several. Constraints will exist and may include the position or dimensions of a photovoltaic panel. Profiling the average energy harvested against the device's usage and the battery's state of charge characteristics requires careful analysis. In some applications, replacing the rechargeable battery with a supercapacitor is possible.
Prolonging battery life
Developers can employ several techniques to extend the battery life of an embedded design. Each use case will have slightly different energy requirements, so power profiling to understand real-world consumption characteristics is an important starting point.
Follow Avnet Silica on LinkedIn
