Blog: energy harvesting in industrial applications

Blog: energy harvesting in industrial applications

Feature articles |
By Nick Flaherty

In this blog series so far, we have considered the energy consumption of motors measured in hundreds of kW and Power-over-Ethernet (PoE) for sensors and actuators at a more modest 100W. Energy harvesting is at the other end of the spectrum where power is in the µW range.

Where is energy harvesting useful?

The Industrial Internet of Things (IIoT) puts intelligence in sensors and actuators to better control and monitor manufacturing processes – efficiency and productivity gains result. At a minimum, this intelligence needs some data processing and storage.

A typical temperature monitoring sensor; will have a microcontroller, memory, and an interface for communications. Routing Ethernet cables or other wired communication systems, such as ones based on RS432 serial connections, may be inconvenient to the IIoT node, so wireless communication is often employed.

The IIoT node needs some power to operate, which poses additional challenges– batteries run down and take time and labour to replace. At the same time, DC from a mains adapter or PoE will only work if there is a convenient mains source.

Harvesting local energy of some type to provide a power source is the right solution – if the energy demands are low. A prime example would be ultra-low-power microcontrollers and WLAN modules that only take power in infrequent bursts, as data is accumulated and transmitted in packets at set intervals.

What type of energy can we harvest?

There are four primary sources of harvested energy: 

  • Photovoltaic
  • Thermal
  • RF
  • Mechanical vibration

The voltage and current characteristics are different in each case. All require a level of power conversion for the end-load supply rail. The energy sources are also generally of an intermittent nature. Hence, a rechargeable battery or supercapacitor is often employed to ensure supply and deal with the peak power demands.


A photovoltaic panel is perhaps the most familiar 'free' energy source, but for IIoT, illumination inside buildings is typically low – producing around 10µW/cm 2 over the panel area. The panel technology should also ideally be tuned to the spectral response of the type of lighting available, with the highest efficiency materials such as gallium arsenide (GaAs) not necessarily being the cheapest or the best-performing under low-light conditions.

A better choice for dimmed light is Copper indium gallium selenide (CIGS). It is a low-cost device, but its conversion efficiency is lower and it requires rare earth materials. The choice is not straightforward, and photovoltaic sources also need an intelligent MPPT controller to get the best performance – such as the SPV1050 available from STMicroelectronics.


Thermal energy harvesters use the Seebeck effect; A thermal gradient across materials (typically heavily doped semiconductors) produces a voltage. Power density for a thermoelectric device in an industrial environment with steep thermal gradients can reach 100mW/cm 3 . The terminal voltage is in the region of 1mV for each degree gradient, making up-conversion difficult. In light industry scenarios, where load power is low, and efficiencies high, large thermal gradients may be challenging to find. In heavy industry, however, they are much more commonplace. Making thermal energy harvesting a more suitable solution.


RF can be a source of harvested energy, but power levels are intermittent, low, and poorly defined,  perhaps 0.1µW/cm 2 for GSM and 0.01µW/cm 2 for Wi-Fi. Conversion efficiency can be relatively high, though, up to around 70 percent, with useful voltages obtained from simple RF transformer techniques. The high-power RF source proximity affects the power harvested, but is still only tens or hundreds of µW at best.

Integrated RF harvesting modules, such as the P2110B from Powercast, connect simply to an external antenna and provide regulated 2V to 5.5VDC output with a 50mA peak current capability from a storage capacitor.

Mechanical vibration

Energy can be harvested from mechanical vibration using the piezoelectric effect, with an energy density of a few hundred µW/cm 3 (as seen in Figure 1). Output voltages can be high, though, and may need clamping to avoid stress on downstream components. Piezo-transducers and harvester modules are available from companies such as Mide, with output power up to 59mW at 34V for the company’s S452-J1FR-1808XB. Figure 1 shows an IIoT node that uses mechanical vibration as a means of harvesting energy.


Figure 1: A typical IIoT node with energy harvesting functionality and a low-power wireless connection

In this example, power is derived from the piezo element, rectified, stabilised, and used to energise the DC-DC converter. The output of which wakes the microcontroller, reads the sensor data and transmits this via the low energy Bluetooth transmitter.

Storing the energy

The energy harvesting module is only part of the story when powering IIoT nodes. The DC-DC converter powering the electronics must have the lowest possible quiescent power consumption (IQ), maximising the energy accumulated on the storage element, battery, or supercapacitor. This storage element should have the lowest possible leakage. The DC-DC must also have high efficiency to make the most of the energy available, such as parts in the R1800K series from Ricoh, which feature 90% efficiency and a typical IQ of 144nA.


Energy harvesting is used widely to power the plethora of sensors that acquire and transmit data in IIoT implementations. Proximity to conventional DC power is not always possible; however, light, heat, or vibration can provide enough power on an intermittent basis, to provide the data acquisition and transmission necessary for the sensor’s task. This technique has advantages over batteries, for instance, as it reduces periodic maintenance schedules, therefore reducing labour costs.

In the next blog in the series, we will look at how to keep efficiency high at the last stage of power conversion with point-of-load (PoL) converters.

Other articles in the series

Related articles


Linked Articles