Powering the Internet of Things via Energy Harvesting

The push is on to add Internet capability to everything—often called the Internet of Things (IoT)—and the challenge for design engineers is to figure out how to power each of these IoT nodes. If there is an AC power line or conventional long-life battery available, the decision is greatly simplified, of course. The circuitry in the much-heralded, intelligent, Internet-enabled thermostat from Nest, for example, has inherent access to low-power DC of the HVAC control line, as well as an internal battery that the source can keep charged.

But such power is unavailable in many cases, and energy harvesting looks like an attractive, free, no-headache alternative once it is designed in and installed. However, there are many facets and limitations with such “free” energy, beyond the initial component cost. First, before even looking at harvesting as an option, it’s critical to put together the obvious: a power budget. How much current and voltage will the IoT node need? What power is needed at each phase of operation: wake up, active and shut-down?

Equally crucial, what is the duty cycle or update rate of this IoT device? This is crucial because for most IoT nodes that go into extreme low-power sleep mode, the largest power use is in the transition from deep sleep to active mode. As a result, it is important to minimize the number of wake-up cycles, and have fewer such cycles while transmitting longer data bursts in each cycle.

It’s also important to keep the fundamental nature of harvesting in mind. The transducer is a source of energy that comes at random times, and in random, usually minuscule amounts. We collect this energy when it is available, much as we would collect rainwater in a barrel, but we spend this collected energy as power, meaning that we must be able to provide enough of the collected energy at the minimum mandated rate to operate the IoT node’s circuitry. (See Figure 1.) The ratio of energy-collection/time to power-spend time is very high—typically several orders of magnitude. We have to be conscious of this as we go through the energy-harvesting and circuit-power-requirements budgets.

Figure 1: Energy harvesting is characterized by a long period of random, erratic energy sourcing, then a brief period of power delivery to operate the load.

Once the power requirements have been established based on both the circuitry of the node and the use pattern, the detailed consideration and design of harvesting can begin. Any harvesting design really has three sub-functions in addition to the source transducer itself: the circuit that extracts the energy from the transducer, the energy-storage element (battery or supercapacitor) and the harvesting-management circuit, which controls the flow of energy into the battery and the flow of electrons as power out of the battery. (See Figure 2.)

Figure 2: A data-logging and reporting system using harvesting as its power source is composed of the power-related front end, the sensors being monitored, a low-power microcontroller, and an RF link (adapted from Texas Instruments).

Harvesting Source Options

The most common sources that can be harvested are light, vibration (including sound), heat, RF and air/water flow. While each of these is the result of a fundamental physical parameter, knowing which one to successfully access and capture is just the beginning of the design-evaluation process. The difficult question involves the actual transducer embodiment one can use, and how that transducer will be able to provide energy in some form which, in turn, can be converted to electrical energy.

For light the answer is obvious—a solar cell will convert it into electricity. For others, however, the answer is more complex. With vibration the obvious choice is a piezoelectric transducer, but often an electromagnetic coil and magnet arrangement can be used.

Regardless of the choice, keep in mind that harvesting sources generally have low energy potential and consequently low conversion efficiency. Again, the solar cell is a good example since the cell’s light-to-electricity conversion efficiency is on the order of 10% to 15%, so the 1,000 W/m2 maximum solar energy reaching the Earth’s surface translates into generating 100 W/m2 at most, and that is at optimum solar conditions. Most other harvesting sources also are in a 10% typical conversion-efficiency range.Table 1 shows some sources and key performance parameters.