A Guide to Architecting Batteryless Devices Part 1: Energy Harvesting

Removing the battery from an IoT sensor reduces maintenance, enables deployment in hard-to-reach locations, and makes it practical to monitor more assets. But a batteryless sensor still needs a reliable source of energy and designing that source requires more than simply selecting an energy harvesting technology. 

In this three-part series, we’ll explore three interconnected elements of architecting batteryless devices: energy harvesting, energy storage, and low-power wireless communication. In Part 1, we’ll focus on the first challenge: harvesting enough energy to keep the device operating reliably, even when conditions are less than ideal. 

The goal is not merely to generate power, but to generate enough power under worst-case conditions. Designs that overlook edge cases can fail in the field when new installation types are introduced or the harvested energy source behaves differently, ultimately frustrating customers.

Best practices: 

• Design the system as a whole. The harvesting aspect of the product cannot be designed in isolation from the rest of the system. How the harvested energy is stored and used is key to how the harvester is designed. For example, consider what essential functionality is required all the time, and what functions could be as energy supply allows. And what behavior changes could be made when poor harvest conditions occur to support energy savings when needed. Depending on how much energy storage is available, short conditions without harvested energy might be less of a concern, but initial start-up use cases might drive different requirements.

• Size the harvesting element for worst-case conditions. Whether harvesting light, heat, RF, or motion, there will be common conditions where the ability to harvest energy is minimal. These are the conditions that are known to occur when the product still needs to maintain its essential functionality. Defining these worst-case conditions in detail is worth the extra effort. Details around what occurs (or doesn’t occur) at these times are critical to determine the level of margin needed in a harvesting element. For example, if a worst-case condition only occurs when a factory is shut down for the night, the requirements for measurement sample rate and connectivity may be much less. 

• Consider a dedicated harvesting IC with MPPT. Maximum power point tracking (MPPT) dynamically finds the operating point that extracts the most power from the source device at any given input amount. Energy harvesting elements have a steep drop-off in voltage as higher current is pulled. This means that pulling the most current without browning out the charge circuitry will not provide maximum power. It can seem unintuitive, but pulling the current back a little allows significantly higher power to be produced. This is the benefit of MPPT. The tradeoffs to model are quiescent current and conversion losses. Also important is to choose a solution that supports the ability to charge a completely empty storage element from the lowest possible input voltage. If your sensor ever fully depletes, it needs a path to recover. 

• Firmware should play a role in the harvest. Modern microcontrollers support very low current sleep modes, as well as variable active modes. Utilize these modes to optimize the energy needed for sampling, processing, and communicating. Also important is how the firmware supports initial power up, commissioning, and recovery events. When firmware is not energy-aware, it puts additional burden on the ability to utilize harvested energy during these singular events and prevents operation when less energy is available. This is a common issue with firmware attempting to do too much too fast on boot requiring more power than the harvester can provide. Furthermore, firmware should adapt to its available harvested power, saving power-intensive tasks for when energy is more available. 

Common tradeoffs: 

• Size vs. form factor. A larger harvester means more power, but there’s a real constraint on what you can fit into the desired form factor for a product. The good news is that a modest increase in harvester size can produce a meaningful increase in harvested energy. This is an easier path to margin than optimizing everything else in the design. 

• Efficiency vs. cost. For IoT devices, the form factor is small, which requires the harvest element to be small as well. However, not all harvest elements perform the same. High-efficiency elements deliver significantly more power but come with a higher price tag. When form factor is tightly constrained, the additional harvested energy justifies the premium. 

• Placement for energy vs. placement for sensing. The ideal spot to take the measurement is not ideal for energy. For example, a ceiling-mounted occupancy sensor is in the right place to detect people, but light levels are much lower than if it could be placed on a desk surface. That tradeoff needs to be resolved in the energy model early in the design phase, not discovered after hardware is built. 

A successful energy harvesting design isn’t built around average conditions. It accounts for the moments when available energy is at its lowest and considers how harvesting, storage, firmware, sensing, and communication work together as a complete system. 

Getting that balance right can open the door to IoT sensors that require far less ongoing maintenance and can be deployed across more assets than traditional battery-powered solutions. In Part 2, we’ll look at the next piece of that system: how to store harvested energy so the device can continue operating when energy isn’t readily available. 

Learn more about how batteryless sensing can help you monitor more assets with less maintenance.