Energy Harvesting Starting To Gain Traction

Tens of billions of IoT devices are powered by batteries today. Depending on the compute intensity and the battery chemistry, these devices can run steadily for short periods of time, or they can run occasionally for decades. But in some cases, they also can either harvest energy themselves, or tap into externally harvested energy, allowing them to work almost indefinitely.

Energy harvesting has been on the drawing board for powering semiconductors for years, but so far it has seen only limited uptake. Solar, hydroelectric and geothermal energy are used on a large scale, while light, heat, wind, vibrations and radio waves have seen limited use in much smaller devices.

Fig. 1: Energy sources include light, electromagnetic, thermal, kinetic, and more. Source: Energy Harvesting Technologies for Achieving Self-Powered Wireless Sensor Networks in Machine Condition Monitoring: A Review

In the IoT world, energy harvesting has held the promise of reducing or eliminating the need for batteries. This is particularly attractive for devices where replacing batteries is difficult, such as livestock sensors, smart buildings, and remote monitoring, as well as for applications such as wearable electronics and for tracking moving goods. But so far, it has seen only limited use.

This is partly due to small and often unreliable input sources and energy levels. In addition, converting the energy from the environment to electricity requires designs with very high efficiency, which can make these approaches cost-prohibitive. If an operation requires a constant flow of electricity, energy storage will be another consideration.

Comparing a design for a smart IoT sensor that uses a coin cell battery to one that uses energy harvesting shows the coin cell design to be much simpler. But while a design using coin cell battery does not require the additional energy harvesting circuitry, it will require the battery to be replaced at some point.

Based on theoretical calculations, most IoT specifications predict the battery would last from a few to more than 20 years. Those estimates typically do not consider battery leakage and drainage, however. Consumer-type coin cell batteries have much lower quality and reliability compared with industrial grade IoT batteries. This explains why an industrial grade battery costs are an order of magnitude higher. It also helps explain the allure of energy harvesting, which can eliminate the need for batteries, or automatically charge the batteries for the lifetime of the devices.

“Energy harvesting technologies are evolving,” said John Stabenow, director of product engineering, Siemens Digital Industries Software. “As the demand for IoT increases, there are a great deal of incentives to provide battery-less IoT solutions. The benefits are obvious. Without the need to change thousands of batteries every few years, the savings can be substantial. IoT sensors are getting smarter, smaller, and ultra-low power. By using system design modeling and simulation, the optimal power budgeting and circuit design of a smart sensor can be developed. As a result, energy harvesting technologies can be fine-tuned to meet the specific power requirements.”

An ideal use case is a combination of efficient energy harvesting for a device with ultra-low power consumption in order to achieve sustainable operations. To put things in perspective, here are some examples of the energy available from various sources and estimated energy requirements of various devices:

Efficiency considerations
Solar power is a popular energy source, but using solar panels to convert solar energy to electrical energy is inefficient, which is why solar panels are so large. Sunlight is converted to electrical energy using photovoltaic (PV) semiconducting materials. Multiple PV cells are joined together to form a module or panel. Using a technique called maximum power point (MPPT) enables maximum efficiency when converting power from a PV module.

In a battery charging situation, the MPPT algorithm compares the PV output and the battery voltage level, then sets the optimal voltage level for charging. The MPPT algorithm performs most effectively during cold weather and/or when the battery is mostly discharged.

“One of the most renewable energy sources to harvest energy is solar,” said Kasra Khazraei, principal applications engineer at Infineon Technologies. “A solar system consists of two main components, solar panels and power electronic circuitry. The efficiency of the majority of solar panels is below 20%, with an average of around 17% to 18%. A breakthrough in the physics of solar cells is necessary to increase these numbers meaningfully. That is why a lot of efforts have focused on improving the efficiency of power electronic circuits to increase the overall energy yield of a solar system.”

Analog Devices, for example, developed a tiny package (1.63mm x 1.23mm) to harvest energy from a single solar cell. Using that approach, the company claims to have achieved an efficiency level in the 85% to 90% range, enabling almost all the energy harvested by the solar panel to be utilized.

“An efficient design of an energy harvester that targets ultra-low power applications, such as in-car detectors, livestock tracking, smart farms/cities, and wearable consumer devices imposes very low power consumption requirements – in the range of a few µW — and cold start-up voltage below 500mV when PV cells are used, and below 100mV when using a thermo-electric generator (TEG). In general, the MPPT embedded algorithm is helpful, as it maximizes the extraction of energy from both of the sources,” said Alessandro Nicosia, technical marketing group manager of STMicroelectronics. “Additionally, a robust PCB design is needed to protect the system against noise and environmental disturbances that might cause false triggering of battery over-charge and over-discharge thresholds, and unprevented internal circuitry stand-by phases and battery/load power-delivery continuity.”

RF energy harvesting designers are coming up with ways to convert energy with high efficiency, as well. This approach is useful for applications such as battery-less consumer products and wireless charging for electronic devices. One potential RF energy breakthrough use case is the retail electronic shelf label (ESL) application offered by Powercast.

Traditionally, human workers change the price tags in a grocery or department store. But prices change frequently, especially when there is a sale, which can make this process labor-intensive. There have been many attempts to reduce labor intensity, including using wireless price tags. A drawback to the wireless price tag approach is that each tag needs to have a wireless receiver powered by a battery, and those batteries need to be replaced every few years. Powercast’s idea is to use wireless RFID price tags. Recently, a major department store chain started to deploy robots to keep track of inventories inside the stores. These robots can be programmed to scan the battery-less RFID price tags with RF signals in order to update prices. The new prices, shown on the RFID tags, will remain unchanged until the next scan. The Powercast RF-to-DC converter chips used in the RFID price tags measure 1 x 0.6 x 0.3mm, and support frequencies from 10MHz to 6GHz with 75% conversion efficiency.

Fig. 2: RF energy harvesting functions include RF to DC conversion and voltage monitoring. Source: Powercast

“The strength of the RF energy sources varies. Therefore, the chips used to convert RF energy to DC need to be highly efficient, preferably in the 70% to 80% range. Antenna design is also important to maximize energy harvesting,” commented Charles Greene, CTO of Powercast, “additionally, the distance of the device from the energy source, such as a dedicated transmitter or Wi-Fi router, is important as the harvested power is inversely proportional to the distance squared. For example, a wireless game controller is more power intensive and it needs to be kept within a foot from the power source. A keyboard is less power hungry and it can be kept within 6 feet of the source. IoT sensors will function well up to 80 feet.”

Low power considerations
While energy harvesting is important, it doesn’t reduce the value of a low-power design, particularly in edge applications. “It’s always about maximum compute power at the lowest power consumed,” said Sailesh Chittipeddi, executive vice president at Renesas Electronics America. “The notion of power efficiency and power consumption in new systems is becoming far more important to people. And that is driving a change in behavior, especially in the industrial segment.”

Energy harvesting complements that shift, and simpler solutions are usually better then complex ones. A 16-bit MCU, for example, consumes less power than a 32-bit MCU. Similarly, a 4-bit MCU consumes less power than an 8-bit MCU. Rightsizing the design means there is less energy to generate in the first place.

Recently, more chipmakers have been developing ultra-low power MCUs that operate in the nanowatts (nW) range. Some ultra-low power MCUs can operate at 1.8V and consume only 150µA per megahertz in active mode, while sleep mode consumes only 10nA. If memory content needs to be retained, the sleep mode current would increase to 50nA with 2µS wake up time. This trend is very encouraging to energy harvesting development.

“Until recently, energy harvesting system designers would often simply characterize the power and energy needs of their system and then select large enough energy harvesters and storage to deliver this reliably,” said James Myers, distinguished engineer at Arm. “This works well, but it meant these systems could often be large and expensive. Today the approach is shifting toward application size or cost constraint, which flows into power and energy budget, and the system needs to be designed for that. Luckily, we have a huge range of low-power components available to us now, and if they don’t fit then a custom SoC can be built that integrates and is in tune with your needs. Ultra-low power processors are especially useful to this domain as they allow intelligent tradeoffs with energy-intensive activities like radios, actuators, and non-volatile memory usage. They can even allow adaptation to intermittent power availability in storage-less harvesting systems.”

Emerging standards
The International Electrotechnical Commission (IEC) published a series of standards to address semiconductor devices for energy harvesting and generation in relation to vibration, thermal, and electromagnetic energy sources. The standards also address test and evaluation methods, test methods for flexible thermoelectric devices, and linear sliding mode triboelectric energy harvesting. [1]

Separately, the EnOcean Alliance, a 500 member, non-profit organization, supports the ISO/IEC 14543-3-10 (known as ASK and used in Europe) or 14543-3-11 (known as FSK and used in North America and Japan). It is an open, harmonized radio standard that describes the radio parameters (physical layer 1 in OSI). This standard is optimized for self-powered wireless devices.

The EnOcean Alliance is separate from EnOcean Inc. The seven promoters of the alliance include BSC Computer GmbH, Eltako, EnOcean GmbH, NIFCO Inc., IBM, Microsoft, and T-Systems Multimedia Solutions. The members of this global network have created an interoperable, maintenance-free standard with a certification program for applications in smart homes, smart buildings, and smart spaces.

Energy harvesting innovations and outlook
Energy harvesting development is gaining momentum. More silicon products are available from companies including Analog Devices, Atmosic, EnOcean, Metis Microsystems, ONiO, Powercast, Renesas, STMicrosystems, and Texas Instruments. Products will be smaller, lighter, smarter with AI, and even lower power. Many innovations are coming out and future possibilities are only limited by one’s imagination.

“In recent years we have seen investment in wide-bandgap switch development to improve the efficiency of the power electronic system,” said Infineon’s Khazraei. “As we will see in the next 5 to 10 years, adoption of highly advanced, wide-bandgap switches made of gallium nitride and silicon carbide will revolutionize renewable energy systems. These switches enable very high-frequency power density and efficient design. A significant reduction in the size of the circuits will lower the cost to install and maintain solar systems.”

Recently, the University of Washington in Seattle demonstrated in a video a lightweight, low-power, dandelion-like sensor floating through the air, sampling temperature and humidity. According to Vikram Lyer, an assistant professor at the school, this tiny device consumed from a few to more than 10 microwatts, depending on the sampling rate. It harvested energy from the sun. With velocity of 0.87 meters per second, the 30mg device — a dandelion weighs 1mg — could travel between 50 to 100 meters in a moderate breeze. The probability of landing safely upright was about 95%. The device could potentially be used to monitor forest fires in dry weather regions. Additional research is being done to expand control and applications of the sensor.

The European Research Council (ERC) has provided a research grant of 1.5 million euros to the Chemnitz University of Technology in Germany to develop the world’s smallest battery, the smart dust battery. Based on past battery research, the team has set a goal to develop a battery capable of delivering100 microwatts of energy per square centimeter for super small computer and electronic applications. When this becomes a reality, it can be embedded in future IoT devices that rely on energy harvesting for recharging.

“Energy harvesting technologies will continue to evolve,” said Oliver Sczesny, president and co-founder of EnOcean Inc. “Legacy companies, as well as startups, will bring new ideas and innovations. For example, solar and thermal energy harvesting have a lot of research ongoing for thin, flexible, and often printable energy harvester foils. Prototypes are available and, especially for solar, mass production has already started. Those types of harvesters offer a wide band of possibilities, from powering small sensors to energy harvesting on a larger scale.”

Other concepts under development include wireless battery-free body sensor networks using near-field clothing to monitor the wearer’s physiological condition, and wireless soft sensors to measure patients’ vulnerabilities to injuries. New approaches, such as harvesting from the circuitry’s transient energy, are being tried, as well It is expected that research organizations, innovative high-tech firms, and startups will continue to come up with new ideas for energy harvesting.

“Computing systems represent information with ones and zeroes, where the information token of binary data is commonly present in CMOS chips as electric charge,” said Azeez Bhavnagarwala, founder and CEO of Metis Microsystems. “A ‘1’ is represented at a circuit node by moving the charge from the power grid of the chip to the node, raising its electric potential to the supply voltage of the chip. A ‘0’ is represented at a circuit node by draining away the charge held at it, lowering its electric potential to the reference ground potential of the chip. In both cases, these circuit nodes holding data serve as a source or a sink of electrostatic energy – an equivalent of a ‘silicon battery’ across circuit nodes. This silicon battery could become available as an energy resource within the chip, providing some of the power that memory and arithmetic components require.”

This is an important shift. “Circuit IP for CMO-based static memories – 6T SRAM, 8T register file, CAM and digital CIM arrays – have been developed to harvest transient on-chip data to improve the energy-delay of CMOS components by as much as an order of magnitude,” said Bhavnagarwala. “This improvement can take place without changing operating voltages or the CMOS process. Harvesting transient on-chip data also favorably impacts other design metrics such as the uncertainty of signal development in the presence of significant MOS device variations. Unlike conventional technologies that harvest energy from ambient sources, data harvesting methods and circuits need not be limited to only low power density applications such as trackers or sensor networks. They can power processors across a broad spectrum – from energy-starved devices at the edge to accelerators and networking hardware in the data center.”

Expect to see more of this technology in the future.

References
[1] Energy harvesting standards:

• IEC 62830-1:2017
• IEC 62830-2:2017
• IEC 62830-3:2017
• IEC 62830-4:2019
• IEC 62830-5:2021
• IEC 62830-6:2019
• IEC 62830-7:2021

Resources
https://www.enocean-alliance.org/wp-content/uploads/2020/04/SmartAcknowledge_Specification_v1.7.pdf
https://www.enocean-alliance.org/specifications/
https://www.enocean-alliance.org/wp-content/uploads/2020/04/factsheet_Certification_EN_Jul122018.pdf