Register for updates
Register for updates
Back to top

Guide To EH & PMICS

Can energy harvesting be used to power edge IoT devices?

The Internet of Things, or IoT, a phrase first coined by British technology pioneer Kevin Ashton in 1999, is not an industry sector, or an application, or a technology. It’s a global phenomenon whereby billions of ‘things’ (as distinct from people) are linked across the internet. They communicate with each other and with applications that may be hosted locally or in datacentres thousands of miles away. And the phenomenon is ubiquitous, already touching almost every aspect of our lives.

It’s not just about fitness trackers and smart fridges, some of the first applications pounced upon by consumers and the media. IoT networks are growing in agriculture, in healthcare, in industrial facilities, in transportation, in our homes, in offices, and in our cities’ infrastructures. Almost everything seems to be getting ‘smart’.

Towards a battery-free,
self-sustaining IoT

The devices at the edge of the networks are primarily sensors and, less frequently, actuators. They’re sometimes referred to as edge devices or IoT endpoints. Many are wired to hubs or routers that in turn connect to the internet. But there are a host of applications where wired connections are either not desirable or not possible. For example, asset tracking devices would be encumbered if tethered to fixed points, and sensors on agricultural tractors can’t be hardwired.

The goal for many designers is to create self-sustaining sensors that are either battery-less or whose operating life exceeds that of their original battery, thereby reducing battery waste. For most IoT devices, the primary contributor to power consumption is not the associated microprocessor that makes it ‘smart’ but the wireless link that’s used to transfer data to and from the device or the sensor element itself.

Top 10 IoT Application areas 2020

IoT ubiquity creates a system design challenge

Consider the dilemma faced by designers of wireless chips and modules. Some IoT applications involve sending tiny amounts of data infrequently. Smart utility meters are one example. Others, like security cameras, some of which now have embedded artificial intelligence functionality so that they only respond to “interesting” events, send a lot of data frequently or continuously. The communications requirements, both in terms of data rate (bandwidth) and operating range will be significantly different between these scenarios, as will the power requirements of the edge devices.

The accessibility of internet connections will vary too. Hubs and routers may be wired to the internet or connected to local telecom exchanges via optical fibres. Short hops may be all that’s needed to connect wireless edge devices to these hubs. In other instances, a cellular radio connection may be the only viable option for connecting sensors in remote locations to server-hosted applications.

As a result of these varying requirements, several wireless protocols are commonly adopted, sometimes in combination, to achieve the best trade-off between range, bandwidth, quality of service (reliability of the link), and cost. Some protocols have been developed and accredited by standards bodies specifically to address IoT applications. This blog post by BehrTech gives an overview of six popular protocols and summarises their suitability for various applications. The WiFi and Bluetooth, including Bluetooth Low Energy (BLE), are dominant for short-range communications but Zigbee has made strong inroads, particularly in the smart home sector. Bluetooth Mesh extends the range of the underlying protocol because each device only has to be within range of another, rather than directly within range of a hub.

Power consumption

Data rate, power consumption, range and cost are key trade-offs.
Courtesy of BehrTech.

Wireless protocols

The choice of wireless protocols can be confusing but patterns of usage are emerging.
Courtesy of BehrTech.

So, here’s the ubiquity challenge for IoT wireless products. No technology is adopted unless it’s cost-effective. Becoming cost-effective means manufacturing at scale and economies of scale are harder to achieve when the applications for a given technology are fragmented. Many wireless chip and module manufacturers have, therefore, taken the approach of developing multiprotocol products based on software-defined radios, so that one product can work in many applications.

For example, Nordic Semiconductor makes switched multiprotocol chips that combine Bluetooth Mesh, Thread and Zigbee. Silicon Labs offers a particularly wide selection of multiprotocol chips and modules, and wireless modules from Murata and u-Blox offer IoT designers a choice of single and multiprotocol products for both short-range and long-range communications.

Power consumption of
wireless devices

The primary determinants of power consumption for any wireless link are its range, the bandwidth used, and the duty cycle of the signal in the application. Each application is different, but the overview table gives typical figures for the various technologies while in operating mode, as distinct from standby.


Typical power consumption of wireless links in
IoT applications vs. data rate and link range.
Courtesy of Voler Systems.

The power requirement is anything from 150 microwatts to 400 milliwatts. Remember, that may just be for a few seconds once every few weeks or continuously, depending on the application. Particularly at lower powers, several micro-energy harvesting technologies can be deployed to complement primary batteries so that the batteries don’t have to be changed so frequently, to charge secondary batteries, or to eliminate batteries by using capacitors or super-capacitors as energy buffers.

The main technologies that are suited to micro energy harvesting are solar, indoor light, vibration, temperature gradient, and electromagnetic or wireless. These energy sources are often transient and will not necessarily be available when the sensor or other IoT device needs to send or receive data. This creates the need for the energy buffer, a storage device, which may be either a capacitor or rechargeable (secondary) battery, and a power management integrated circuit (PMIC). The PMIC processes the energy from the harvester, manages the charge delivered to the buffer, then delivers power to the load when it’s needed. A PMIC designed for use with energy harvesters is called an EH PMIC.

Micro energy harvesting: the economic challenge

If battery usage is to be reduced or sensors are to become self-sustaining, micro-energy harvesting systems must be seen as an economical alternative to batteries. Many edge IoT devices can run for months or sometimes years on a coin cell, something like a CR2032. These cost just a few US cents each in high volumes, so the hardware costs of any micro-energy harvesting system will be higher. However, when you remove the need to change batteries, particularly where sensors are in hard-to-access locations, and you consider the environmental benefits of eliminating battery waste, the cost-benefit analysis takes on a different hue. You only have to consider the retrueal cost of, say, a 20 mile journey to change a 20-cent battery every now and again to appreciate just how much could be saved over the operating life of a self-sustaining IoT sensor.

Most environments have more than one potential source of energy available. Take an industrial machine as an example. It may well be exposed to sunlight much of the time if it’s in a plant with windows or transparent/translucent roof. It will usually vibrate to some extent, thereby creating kinetic energy, and there will be temperature differences to be exploited by attaching thermal gradient harvesters to parts of the machine. It may also be operated in an area in which there is artificial light. Driving greater adoption of energy harvesting for IoT edge devices means finding new ways to harvest energy from the optimum number of sources available within a given environment.

How much energy can be harvested from the technologies available?

The amount of energy that can be produced by a harvester will, of course, depend on its size and specification but, as a starting point, it’s helpful to consider the relative energy densities of the different kinds of harvester.

How much energy

Outdoor solar leads the energy density table by a considerable margin but in industrial settings, harvesting thermal energy and kinetic energy from vibration are the second and third most promising technologies. Radio frequency energy harvesting is at the bottom of the list, so is unlikely to be useful for any IoT applications except to power the most frugal of devices.

Solar cells with an area of 35-40 square centimetres are capable of delivering around 0.5 Watts (at <20% efficiency) and are available for less than 1 USD each in volume. Piezoelectric harvesters are typically at least an order of magnitude more expensive and produce far less energy but, in the appropriate application, may still save the considerably greater expense of callouts to change batteries.

The low efficiency of solar cells, when used indoors, has recently been addressed with new technology from Epishine. The Swedish company claims its evaluation kit, which includes an indoor-optimised photovoltaic panel with six or eight cells, a power management chip, and a supercapacitor as the energy buffer ‘can deliver sufficient output current to power most low-power radios such as BLE, Zigbee, LoRa and similar’. Lightricity is a UK company with a similar offering.

Advances in all of the technologies cited here, and others, mean that micro energy harvesting is now viable for reducing or eliminating batteries in a growing number of IoT edge applications, most of which adopt wireless technologies for data transfer from sensors.

How do I calculate the energy needed for my IoT edge sensor?

The sheer breadth of applications and the varying environmental conditions in which edge IoT sensors may be used make it difficult to be precise about how much energy will be needed to power devices.

As mentioned earlier, where a wireless communications link is involved, the radio will often be the most significant contributor to energy consumption not least because low power microprocessors and microcontrollers not only consume little energy when operating but invariably provide sophisticated sleep modes that reduce power consumption to a few microwatts when their processing power is not needed. Even at full-throttle, chip design techniques such as sub-threshold switching may keep processor power consumption down to a small percentage of that demanded by the device as a whole. Recently, the desire to carry out more processing at the IoT edge, rather than transmitting everything to a server-hosted application, drove the development of tinyML processors that can run machine learning algorithms while consuming just a few milliwatts. While the tinyML Foundation states that its technology is ‘enabling a variety of always-on use-cases and targeting battery-operated devices’, anything operating at microwatt to milliwatt power levels is potentially a good candidate for energy harvesting and there is much discussion of about this topic in the tinyML community.

To calculate the likely energy requirements for an IoT device, here are two of the most useful sources of information:

  1. IEEE Sensors Journal has published a useful paper entitled ‘The Power of Models: Modeling Power Consumption for IoT Devices’. This is particularly valuable because it looks at how to model wireless sensor power consumption at a system level, rather than just considering processors or radios. It describes a methodology for simulating consumption by looking at the four key elements: the processor, the sensor, the radio, and the system management.
  2. Since 1997, the Embedded Microprocessor Benchmark Consortium (EEMBC) has benchmarked microcontroller performance and power consumption. EEMBC a not-for-profit organisation that has evolved to produce 15 benchmarks, including the IoTMark™. The IoTMark involved benchmarks for IoT edge nodes where ‘grid power is unavailable and human intervention to change batteries must be kept to a minimum’. It has benchmarks for both Bluetooth Low Energy and WiFi edge devices. EEMBC proposes a measurement framework based around the STMicroelectronics PowerShield, an extensible platform for probing the power consumption of embedded systems. The organisation has even created a bill of materials for the platform, in partnership with DigiKey. A schematic for the IO is available, and Gerber files for the printed circuit board can be downloaded. For a comparison of purely microcontroller performance, EEMBC’s CoreMark® data is useful too.

It’s also worth noting that The Power Sources Manufacturers’ Association (PSMA) has links to several useful articles and if you search ‘Energy Harvesting’ on the EDN technical media site, you’ll find more interesting material, particularly articles by Rich Quinnell.

The STMicroelectronics PowerShield platform is useful for
probing the power consumption of embedded systems

 

Design considerations for energy harvesting power management

The energy harvesting power management integrated circuit – the EH PMIC – is the glue that holds the self-sustaining sensor module together. As described earlier, the most viable energy harvesters for IoT edge applications are those that exploit sunlight, indoor light, movement, and thermal gradient.

One of the technical challenges is that each type of energy harvester has different electrical characteristics. Photovoltaic and thermoelectric harvesters produce a continuous trickle of direct current at low voltage. Thermoelectric harvesters produce a continuous trickle of bi-polar DC current at a low voltage and so are low impedance. Photovoltaic harvesters also produce a DC voltage, but with current, and so impedance, that varies with the level of light.

Most EH PMICs are designed with a single input and to connect to a single type of harvester. Most are also limited by their architecture and have fixed input voltage ranges, so can’t work with all harvester types. These two factors alone may make some self-sustaining edge IoT sensors uneconomic. Readily available energy may be wasted when the energy harvesting system is designed for a single harvester type, if interface circuits are needed the cost, size and complexity can be prohibitive, and if there is a desire to use a variety of energy sources, a dedicated EH PMIC is needed for each one, again adding cost and complexity.

This recent article on Fierce Electronics noted how much more attention is now being given to energy harvesting as a way of power IoT edge devices and it mentions the traditional analogue power management companies that offer PMICs. But, arguably, the younger, niche players in the EH PMIC sector are leading the way in developing more effective and efficient devices for these applications.

e-Peas offers both EH PMICs and complementary low-power microcontrollers. Each EH PMIC is designed to work with a specific type of harvester: photovoltaic, thermal, vibration, or radio frequency (RF). Most parts can deliver up to 550mW output, somewhat less for RF harvesting. In some instances, external resistors and rectifiers are needed to condition the input to the EH PMIC.

Nowi Energy has a small EH PMIC designed for consumer applications. It’s optimised for harvesting energy from indoor and outdoor light and can be modified for use with movement energy harvesters with the additional external circuitry between the harvester and the device’s input

Trameto’s EH OptiJoule PMIC technology is based on a different approach. Its EH PMICs have inputs that autonomously adapt to whatever harvester is connected to them – light, thermal gradient, or movement harvester – without the requirement for external interface circuits. Versions are available with up to four inputs and any combination of harvesters – of the same or different types – can be connected to each. The PMIC automatically maximizes the power delivered to the buffer in response to the energy available from each source.

 

In the same way that RF module manufacturers have often adopted a multiprotocol radio approach in the design of their devices, by employing OptiJoule EH PMICs they can use a single PMIC for multiple applications, even when the energy harvesting source has not yet been decided or where the application can benefit from adopting multiple harvested energy sources. This flexibility comes without requiring a separate PMIC for each energy source and without the complexity, cost and space penalty of external interface circuits.

Newsletter sign up