The choice of energy harvesters for self-sustaining sensors
Editor’s note: This is the second of two articles reviewing recent advances in self-sustaining sensor node technology. The articles focus on integrating diverse renewable energy harvesting methods and exploring diverse energy harvesting methods and power management techniques. Part 1 covers design principles and Part 2 (below) covers design choices. These innovations address challenges of intermittent energy supply and ultra-low-power operation, paving the way for robust, long-lasting sensor networks in smart infrastructure, environmental monitoring and industrial applications.
The implementation of energy harvesting for low-power IoT nodes faces a few challenges.
There are few standards available for reference, so each installation relies heavily on a custom design approach. Although many energy harvesting techniques have been well understood for decades, the adoption of renewable energy for micro-powered devices is still in its infancy.
As a result, there are concerns over reliability, maintenance, and operating life. In addition, integrating energy harvesting with existing power sources may be complex and expensive. All these factors raise questions about the real return on investment that is achievable.
When the decision is reached to deploy energy harvesting, perhaps the most significant factor in the choice of technology is the environment in which the harvesters will operate.
Photovoltaic (PV) sensors need light, which may not always be present. Vibration sensors, perhaps embedded in road surfaces, may only produce energy when traffic is present. Thermoelectric sensors rely on the presence of a temperature difference, which is again likely to be far from constant. The table summarizes the most common industrial energy harvesting technologies and the typical power levels available from each.
(Source: Power Supply Manufacturers Association)
In some applications, it may be desirable to employ more than one kind of harvester to account for changing environmental conditions, for example, day and night. However, this complicates energy management. Harvester output levels can vary by orders of magnitude, as can the impedance (current and voltage) they present to the power management device or system.
Self-sustaining sensor node building blocks
Consider the building blocks that make up a system for powering a sensor node using micro-energy harvesting.
The performance of energy harvesters of all kinds is slowly improving. Recent developments include indoor photovoltaic harvesters that can produce energy from exposure to indoor light levels as low as 50 lux. Some claim that this enables tiny battery-less asset trackers to operate perpetually from this source of energy.
A research paper has proposed a way to extend the battery life of an IoT-based heart rate monitor by using a piezoelectric device to harvest kinetic energy from the heartbeat. A similar goal may be achievable with thermal energy harvesting using body heat as the energy source. The harvesters are thermoelectric generators (TEGs), and some suggest that flexible versions could be worn as body straps or even integrated into clothing.
Acoustic energy harvesting could be useful in noisy environments, such as in the presence of industrial machinery or HVAC systems. Coupled Helmholz resonators, electromagnetic, piezoelectric and triboelectric harvesting technologies can be used, and researchers have recently developed a nanomesh acoustic energy harvester (NAEH) that harvests sound energy at a peak power density of 8.2 W/m2.
As with so many emerging technologies, the challenge for designers of IoT devices is to keep up to date with the latest developments in harvesters while ensuring that these innovations are technically and economically viable in real-world applications. There is growing interest in radio frequency (RF) energy harvesting, a form of electromagnetic harvesting. The levels of harvestable energy are particularly low with this technology, but it has been demonstrated to be viable for powering extremely low-power sensors or tags over short distances, sometimes just triggering a wake-up function in a device. The RF energy could come from WiFi, cellular radio or dedicated transmitters.
How to manage harvested micro-energy
In most instances, a power management integrated circuit (PMIC) will be used in conjunction with an energy storage device, either a battery or capacitor, to manage the flow of energy between the harvester, storage device and the load.
Early PMICs were made for specific types of harvesters, but more recent devices will operate with a variety of harvesters, selectable by the IoT device’s design engineer. Some will even work with several harvesters of the same or different types connected simultaneously, dynamically optimizing their output for best performance.
The efficiency of energy-harvesting PMICs is typically in the 85% to 95% range, but it's also important to consider the quiescent current of the chip because, at the energy levels involved, this can have a significant impact. A total quiescent current of under 500 nA is desirable, but best-in-class PMICs now boast figures in the sub-100 nA range. Low quiescent current is essential in enabling cold start for a micro-energy harvesting system in the absence of an external power supply.
Maximum Power Point Tracking (MPPT) is now standard for most PMICs but may alternatively be done within the MCU if supported. One other factor to consider is how many external components may be needed so that the PMIC can work effectively with the chosen harvesters. For example, some require large external capacitors or inductors, which add cost and complexity; others may need power conditioning components between the harvesters and their inputs.
PMICs also provide essential hardware support for sensor node MCUs or SoCs to implement adaptive duty cycling, whereby a sensor node autonomously adjusts its active and sleep periods (the duty cycle) in real-time based on the availability of harvested energy and the current state of its energy storage.
A complementary technique, reinforcement learning (RL)-based energy allocation, may also be executed by the sensor node’s processor with the PMIC acting as a configurable hardware enabler. Here, machine learning optimizes various resource parameters (e.g., transmission power, duty cycle, etc.) by rewarding actions that maximize long-term energy efficiency. For example, an RL agent may learn to prioritize data transmissions during high-energy periods while minimizing interference over the sensor’s wireless link.
Batteries or capacitors for storage?
The answer depends on the application. Batteries generally offer higher energy density, meaning they can store more energy per unit volume and weight than capacitors, and this aspect of their performance has seen considerable progress recently. They store more energy per unit volume than capacitors, but they are more limited in the rate at which they can deliver that energy to the load.
Also, batteries have a limited number of charge/discharge cycles before their performance deteriorates, which is rarely an issue for capacitors. Both supercapacitors and thin-film batteries exhibit leakage currents that degrade stored energy, but PMICs are also used to mitigate this problem.
The duty cycles of IoT sensors, the availability of harvestable energy, the characteristics of the sensor, and the wireless communication protocol selected are among the factors affecting not only the energy harvesting system overall but also the choice of storage device. The decision on energy storage medium needs to consider capacity, energy density, voltage, discharge profile and temperature.
Data sheets are a starting point, but it’s vital to check the claims of battery and capacitor manufacturers by testing their products under simulated real-world conditions. The "normal conditions" from which datasheet figures are derived may not apply. As a result, it's easy to over- or under-specify energy storage components, resulting in higher costs, either for the components themselves or, in the case of batteries, because they need to be changed earlier than planned.
Summary
Self-sustaining sensor nodes are designed to operate with minimal or no battery replacement by harvesting energy from sources like solar, RF, magnetic, and piezoelectric inputs, which reduces both maintenance costs and environmental impact. Achieving this requires a holistic, energy-efficient design, where every component from the sensor and MCU or SoC to the wireless communication protocol is selected for low-power consumption, since wireless communication is often the largest energy drain by far.
MCUs and SoCs should support deep sleep modes and autonomous low-power peripherals. The choice of wireless protocol significantly affects energy use, as does the node’s duty cycle, which must be matched to the application’s data needs and energy budget.
For energy storage, batteries offer higher energy density but wear out faster, while capacitors deliver energy quickly and last longer but store less. Both need real-world testing.
In addition to their primary roles of managing the energy flow within the sensor node, PMICs provide hardware support for advanced features that allow sensor nodes to dynamically adjust their activity and optimize energy use based on real-time conditions. All these elements must work together to ensure reliable, long-lasting operation even with the intermittent and unpredictable energy sources that characterize micro-energy harvesting.
PART 1: How energy harvesting underpins the self-sustaining IoT sensor node