Text—JOHN KOON
Source— Semiconductor Engineering
Compile—Editorial Office
At this stage, tens of billions of IoT devices are still battery-powered. Depending on the computational strength and battery chemistry, these devices are only able to operate stably for a short period of time, with individual devices occasionally able to operate for decades. But in some cases, these devices can also collect energy on their own, or take advantage of externally collected energy so that they can work almost indefinitely.
For many years, the energy harvesting technology that powers semiconductors has been in the design stage, but today, the technology can only achieve limited energy absorption and harvesting. Solar, hydroelectric and geothermal energy are used on a large scale, while light, heat, wind, vibration and radio waves are used in smaller devices to a limited extent.
Figure 1: Energy sources include light, electromagnetic, thermal energy, kinetic energy, etc. (Source: Energy harvesting technologies for self-powered wireless sensor networks in machine condition monitoring: overview)
In the IoT space, energy harvesting is expected to reduce or eliminate the need for batteries. This is particularly attractive for devices that are difficult to replace batteries, such as livestock sensors, smart buildings, and remote monitoring, as well as applications such as wearable electronics and logistics tracking. But so far, the technology has not yet become widespread.
In part, this is due to the small and often unreliable input sources and energy levels. In addition, design methods that convert energy from the environment to high efficiency in terms of power demand can make these methods prohibitively costly. If the operation requires a steady stream of electricity, then energy storage will be another option worth considering.
Comparing the smart IoT sensor design using coin cell batteries to the sensor design using energy harvesting shows that the design of coin cell batteries is much simpler. However, while a design that uses a coin cell battery does not require additional energy harvesting circuitry, it requires the battery to be replaced at some point.
Based on theoretical calculations, most IoT specifications predict that battery life will last from several years to more than 20 years. However, these estimates usually do not take into account battery leaks and drainage. Compared to industrial-grade IoT batteries, consumer coin cell batteries are much lower in quality and reliability. This explains why industrial-grade batteries cost an order of magnitude higher, while also showing the appeal of energy harvesting, which can reduce the need for batteries or automatically charge them over the life of the device.
"Energy harvesting technology is constantly evolving," said John Stabenow, director of product engineering at Siemens Digital Industrial Software. "With the increasing demand for IoT, there is a strong driver for the use of battery-free IoT solutions. The advantages are obvious, there is no need to replace thousands of batteries every few years, and the cost savings are considerable. Intelligence, miniaturization, and ultra-low power consumption have gradually become the mainstream direction of the development of IoT sensors. By using system design modeling and simulation, optimal power budgets and circuit designs can be provided for the development of smart sensors. As a result, energy harvesting techniques can be fine-tuned to meet specific power requirements. ”
The ideal use case is to provide a combination of efficient energy harvesting for devices with ultra-low power consumption for sustainable operation. To put things in perspective, here are some examples of available energy from various sources and estimated energy requirements for various devices:
Improve energy conversion efficiency
Solar energy is the mainstream energy source, but the efficiency of converting solar energy into electricity using solar panels is very low, which is why solar panels are so large. Sunlight uses photovoltaic (PV) semiconductor materials to convert electrical energy. Multiple photovoltaic cells are connected together to form a module or panel. By using Maximum Power Point (MPPT) technology, maximum efficiency can be achieved when converting the power of photovoltaic modules.
In the case of battery charging, the MPPT algorithm compares the PV output to the battery voltage and then sets the optimal charging voltage. The MPPT algorithm works best in cold weather or when the battery is nearing discharge.
"Solar energy is one of the renewable energy sources for energy harvesting," said Kasra Khazraei, chief application engineer at Infineon Technologies. "A solar system consists of two main components, solar panels and power electronic circuits. Most solar panels are less than 20% efficient, with an average of around 17% to 18%. In view of the problem of effectively improving efficiency, a breakthrough in solar cell physics is necessary. That's why a lot of efforts are focused on improving the efficiency of power electronic circuits to improve the overall energy output of solar systems. ”
"For ultra-low-power applications such as on-board detectors, livestock tracking, smart farms, smart cities, and wearable consumer devices, the efficient design of their energy harvesters places extremely high demands on low power consumption (in the μW range) and cold-start voltages below 500mV when using PV batteries and less than 100mV when using thermoelectric generators (TEG). Overall, the MPPT embedded algorithm is helpful because it maximizes the extraction of energy from both sources," said Alessandro Nicosia, Technical Marketing Group Manager at STMicroelectronics. "In addition, a robust PCB design is required to protect the system from noise and environmental interference that can lead to incorrect triggering of battery overcharge and overdischarge thresholds, as well as unprotected internal circuit standby phase and battery/load supply continuity."
RF energy harvesting designers are also looking for ways to efficiently convert energy. This method is widely used in applications such as battery-free consumer products and wireless charging of electronic devices. One potential RF energy breakthrough use case is the Retail Electronic Shelf Label (ESL) application offered by Powercast.
Traditionally, workers changed price tags at grocery stores or department stores. However, prices change frequently, especially when goods are promoted, which can cause price tags to change frequently. There have been many attempts to reduce labor intensity, including the use of wireless price tags. One disadvantage of the wireless price tag method is that each tag needs to have a wireless receiver powered by batteries, and these batteries need to be replaced every few years. The idea for Powercast is to use wireless RFID price tags. Recently, a large department store chain began deploying bots to track inventory within the store. These robots can be programmed to scan battery-free RFID price tags using RF signals in order to update prices. The new price shown on the RFID tag will remain the same until the next scan. The Powercast RF-to-DC converter chip used in the RFID price tag measures 1 x 0.6 x 0.3mm, supports frequencies from 10MHz to 6GHz, and has a conversion efficiency of 75%.
Figure 2: RF energy harvesting functions including RF-to-DC conversion and voltage monitoring (Source: Powercast)
"The intensity of RF energy sources varies. Therefore, the chip used to convert RF energy into DC requires high efficiency, preferably in the range of 70% to 80%. Antenna design is also important for maximizing energy harvesting," commented Charles Greene, Powercast's CTO, "In addition, depending on the application, the distance of the device from the energy source (such as a Wi-Fi router) is also important because the energy power is inversely proportional to its distance." For example, wireless game controllers are more powerful and need to stay within a foot of the power supply. The keyboard consumes less power and can be kept within six feet of its source. IoT sensors will operate within 10 feet. ”
Reduce system power consumption
While energy harvesting is important, it does not detract from the value of low-power designs, especially in edge applications. "It's always about maximizing computing power at the lowest power consumption," said Saleresh Chittipeddi, executive vice president of Renesas Electronics USA. "The concepts of power efficiency and power consumption in new systems are becoming increasingly important to people. This has driven a shift in strategy, especially in the industrial sector. ”
Energy harvesting complements this shift, with simple solutions often having advantages over complex ones. For example, a 16-bit MCU consumes less power than a 32-bit MCU. Similarly, a 4-bit MCU consumes less power than an 8-bit MCU. Properly resizing the design means less energy is produced in the first place.
Recently, more and more chipmakers are developing ultra-low-power MCUs that operate in the nW range. Some ultra-low-power MCUs can operate at 1.8V and consume only 150μA per megahertz in run mode, compared to 10nA in sleep mode. If memory contents need to be preserved, the sleep mode current will increase to 50nA within a 2μS wake-up time. This trend is very encouraging for the development of energy harvesting.
"Until recently, energy harvesting system designers typically simply characterized the power and energy requirements of their systems and then chose an energy harvester and memory large enough to reliably provide this functionality," Arm engineer James Myers noted, "which works well, but it means that these systems can often be large and expensive." Today's solutions are increasingly shifting the focus of the problem to application scale or cost constraints, which are flowing into power and energy budgets, for which systems need to be designed. Fortunately, we now have a large number of low-power components available, and if none of them fit, we can also build a custom SoC that can be integrated as needed. Ultra-low-power processors are particularly useful in this space, as they allow for intelligent trade-offs with energy-intensive activities such as radio, actuators, and non-volatile memory usage. They can even accommodate intermittent power availability in storage-free collection systems. ”
Open radio standards
The International Electrotechnical Commission (IEC) has published a series of standards to address semiconductor devices used for energy harvesting and generation in relation to vibration, heat and electromagnetic energy. The standard also covers test and evaluation methods, test methods for flexible thermoelectric devices, and triboelectric energy harvesting in linear sliding mode.
In addition, the EnOcean Consortium is a 500-member non-profit organization that supports ISO/IEC 14543-3-10 (known as ASK and used in Europe) or 14543-3-11 (known as FSK, used in North America and Japan). It is an open and coordinated radio standard for describing radio parameters (physical layer 1 in OSI). This standard is optimized for self-powered wireless devices.
The EnOcean Alliance exists independently of EnOcean Corporation. The alliance's seven sponsors include BSC Computer GmbH, Eltako, EnOcean GmbH, NIFCO Inc., IBM, Microsoft and T-Systems Multimedia Solutions. Members of this global network have created an interoperable, maintenance-free standard that includes certification programs for smart home, smart buildings, and smart space applications.
Energy harvesting technology continues to innovate
Energy harvesting is gaining momentum. Companies including ADI, Atmosic, EnOcean, Metis Microsystems, ONiO, Powercast, Renesas Electronics, STMicroelectronics, and Texas Instruments are offering a growing number of silicon products. Through AI technology, products will be smaller, lighter, smarter, and even consume less power. Energy harvesting technology is maturing in continuous innovation, and future possibilities can only be limited by imagination.
"In recent years, we have seen a lot of investment in the development of wide-bandgap switches to improve the efficiency of power electronic systems," said Khazraei of Infineon. "As we will see in the next 5 to 10 years, the 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 circuit size will reduce the cost of installing and maintaining a solar system. ”
Recently, the University of Washington showed a lightweight, low-power, dandelion-like sensor floating in the air, sampling temperature and humidity in a video. According to Vikram Lyer, an assistant professor at the university, the energy consumption of such a micro-device ranges from a few microwatts to more than 10 microwatts, depending on the sampling rate. It collects energy from the sun at a speed of 0.87 meters per second, and for a device of 30 mg (the sensor weighs 1 mg), it can travel 50 to 100 meters in the breeze, and the probability of a safe upright landing is about 95%. The device may be used to monitor forest fires in dry weather areas. The university is advancing further research to expand the control and application of sensors.
The European Research Council (ERC) has provided a €1.5 million research grant to the Chemnitz University of Technology in Germany to develop the world's smallest battery: the Smart Dust Battery. Building on past battery research, the team set a goal of developing a battery capable of delivering 100 microwatts of energy per square centimeter for ultra-small computer and electronics applications. When this becomes a reality, it can be embedded in future IoT devices that rely on energy harvesting for charging.
"Energy harvesting technology will continue to evolve," said Oliver Sczesny, president and co-founder of EnOcean, "and traditional companies as well as startups will bring new ideas and innovations." For example, solar and thermal energy harvesting has done extensive research on thin, flexible and printable energy harvester foils. Specifically for solar energy collection, a prototype of the device has been launched and mass production has begun. These types of collectors offer a wider range of possibilities, from powering small sensors to larger-scale energy harvesting. ”
Other technical concepts being developed, such as wireless battery-free body sensor networks, the use of near-field clothing to monitor the physiological condition of the wearer, and wireless soft sensors for measuring patients' vulnerable points of injury. The industry is gradually beginning to experiment with new methods, such as harvesting from transient energy in circuits, and research institutes, innovative high-tech companies, and startups are expected to continue to come up with new ideas for energy harvesting.
"Computing systems represent information in 1s and 0s, where information from binary data is usually present in cmOS chips in the form of charges," said Azeez Bhavnagarwala, founder and CEO of Metis Microsystems. "The '1' representation method on the circuit node is to move the charge from the chip's grid to the node, raising its potential to the chip's supply voltage. The '0' on the circuit node is indicated by discharging the charge held on it, reducing its potential to the chip's reference ground potential. In both cases, these circuit nodes that hold the data can act as sources or receivers of electrostatic energy, equivalent to 'silicon cells' on the circuit nodes. This silicon cell can serve as an energy resource within the chip, providing some of the power required for memory and arithmetic components. ”
This is an important shift. "Circuit IP for CMO-based static memory (6T SRAM, 8T register file, CAM, and digital CIM arrays) has been developed to collect transient on-chip data, thereby increasing the energy latency of CMOS components by an order of magnitude." Bhavnagarwala thinks. "This improvement can be made without changing the operating voltage or CMOS process. Collecting on-chip transient data can also have a beneficial effect on other design metrics, such as uncertainty in signal development in the presence of significant MOS device variations. Unlike traditional techniques for harvesting energy from environmental sources, data collection methods and circuitry do not have to be limited to low power density applications such as trackers or sensor networks. From energy-poor devices at the edge to accelerators and networking hardware in data centers, these devices can power a wide range of processors.