BASIC KNOWLEDGE Energy harvesting: how it works, and why power electronics decide whether it's viable

Energy harvesting can enable maintenance‑free, sustainable IoT and sensor networks by replacing batteries and their recurring service costs. This article summarizes the realistic power available from common ambient sources, describes how modern PMICs integrate boost, MPPT and nanopower regulation, and explains why cold‑start thresholds, quiescent current and buffer selection are the critical design variables; it also offers practical guidance on materials, circuit and layout choices that support battery‑free operation.

Energy harvesting powers electronics by converting ambient light, heat, vibration, or radio waves into electricity, usually at the scale of microwatts to a few milliwatts, and the parts that decide whether a deployment works are the power-management chips that condition that trickle rather than the energy source itself.

A single harvesting power-management integrated circuit, such as Texas Instruments' BQ25570, combines a boost charger, maximum-power-point tracking, and a nanopower buck converter inside a 3.5mm x 3.5mm package, and it’s built to extract microwatts to milliwatts from high-impedance sources without collapsing them. Laboratory boost converters have gone further, self-starting from 128 mV and holding operation at inputs as low as 7.5 mV

Energy harvesting sources

Four ambient sources cover almost every deployment: photovoltaic, thermal, kinetic, and radio frequency. Photovoltaics dominate because indoor and outdoor light offer the highest power density per unit area and the simplest integration, which is why trackers such as Mordor Intelligence place light harvesting ahead of the other categories.

Thermal harvesting uses the Seebeck effect across a temperature gradient, common where machinery or the human body supplies a few degrees of difference, while kinetic harvesting converts vibration through piezoelectric, electrostatic, or electromagnetic transducers, and radio-frequency harvesting rectifies ambient signals from broadcast, cellular, and Wi-Fi bands, the approach behind a battery-free rectifier that researchers at the National University of Singapore demonstrated in 2024.

However, it’s numbers that draw the line between which of these sources are viable and which aren’t. Outdoor silicon photovoltaics deliver on the order of tens of milliwatts per square centimeter in direct sun, but a small indoor cell under office lighting drops that into the low microwatts, roughly three orders of magnitude lower. Thermoelectric generators across small gradients, and vibration harvesters, typically produce microwatts to low milliwatts.

Radio-frequency harvesting sits at the bottom of the range in most settings, which is why it has stayed in the research and niche tiers rather than mainstream deployment. Those figures sit far below what a wireless sensor draws during its active moments, and that mismatch is the constraint the rest of the system is designed around.

RENEWABLE ENERGY

24/7 renewables outcompete fossil fuels on costs

Cold-starting from millivolts

A harvester that has been dark or still holds no charge, so the converter has to start itself from whatever the source produces at that instant, often a few tens of millivolts. That is below the threshold at which conventional switching regulators operate. Harvesting front ends solve this with dedicated cold-start circuits: a low-voltage oscillator or charge pump brings the chip to a point where its main boost stage can take over.

The BQ25570, for example, uses a hysteresis-based start-up threshold, beginning to charge only once its input clears a set minimum, then running down to far lower voltages once established. The Guangzhou University converter cited above reaches self-startup at 128 mV through a multi-stage circuit, then sustains output at 7.5 mV, which shows how far the startup voltage and the running voltage can diverge.

Leakage is the other half of the problem. A chip that consumes more standby current than the harvester supplies will never charge anything, so quiescent current in the nanoampere range is a hard requirement rather than a specification to tune later. The cold-start voltage ultimately becomes the figure that decides which chip a designer can use, because a part that needs 600 mV to wake is useless on an indoor cell that only ever reaches 400 mV, regardless of how efficient it is once it’s running.

Tracking a moving maximum

Every energy source has an internal impedance, and maximum power transfers only when the load matches it. For a solar cell that optimum sits near a fixed fraction of its open-circuit voltage, roughly 70% to 80% under most conditions, while a thermoelectric generator peaks closer to 50%. Maximum-power-point tracking (MPPT) holds the converter at that operating point as conditions change.

Harvesting MPPT works differently from the version in a rooftop inverter. Rather than continuously perturbing the load, the BQ25570 periodically disconnects its boost charger for 256 milliseconds to sample a fraction of the source's open-circuit voltage, refreshing that reference roughly every 16 seconds and regulating to it in between. Sampling rather than continuous tracking costs a little harvested energy during each sample window but consumes far less power overall, which is the correct trade at the microwatt scale. A host microcontroller can override the sampled reference to run a more elaborate algorithm where the energy budget allows.

ENERGY STORAGE

CATL launches world’s largest energy storage testbed

Why every harvester needs a buffer

The gap between trickle input and bursty output is the reason a storage element exists at all. A sensor node spends most of its life asleep, drawing nanoamperes, then wakes to read a sensor and transmit, a burst that can demand tens of milliamperes for a few milliseconds. No small harvester can supply that peak directly. A buffer, charged slowly and discharged quickly, bridges the gap, a power architecture CAP-XX and others describe as standard for the category. The design target is energy-neutral operation, where the average power harvested over a full cycle meets or exceeds the average the node consumes, so the buffer never runs dry across the day-night or duty-cycle swing.

The buffer can be a rechargeable cell, a conventional capacitor, or a supercapacitor, and the choice shapes the whole design. Capacitors tolerate effectively unlimited charge cycles and simple charging, requiring only current limiting and overvoltage protection rather than the constant-current, constant-voltage a lithium cell demands, and that simplicity is important when the goal is a device that outlives the battery it would otherwise replace. The same harvesting PMICs that manage the source also manage this storage element, clamping it below its safe ceiling and disconnecting the load before it drains past a usable floor.

None of these stages tolerates the losses that are acceptable in line-powered electronics. A 90% efficient converter, for example, wastes a tenth of a watt on a 1 W load without anyone noticing. The same converter wasting a tenth of its input on a 100 µW harvester can put the entire node below its operating threshold. Power electronics built for the grid or for a phone charger don’t transfer to this regime, which is why a distinct class of nanopower parts exists, and why the conversion chip rather than the solar cell determines whether an energy-harvesting product is viable.

(ID:50884805)