Give away medical masks when you place an order. learn more

Meeting Power Demand with Energy Harvesting in IoT Sensor Nodes

Expectations for billions of smart sensors pushing data to the Internet of Things (IoT) depend on successful deployment of MCU-based wireless sensor nodes, or motes, capable of operating for years without battery replacement. Energy-harvesting techniques can deliver the necessary power from ambient energy sources. Yet, the ability for engineers to achieve a "zero power" design depends on careful attention to low-power operation of MCUs and RF transceivers, both during active sensor-data collection and while sitting idle between measurements. Designers can meet these requirements using power-saving techniques and ICs from semiconductor manufacturers including Atmel, Linear Technology, Maxim Integrated, Microchip Technology, Renesas, Silicon Labs, STMicroelectronics and Texas Instruments.

Smart IoT devices will continue to emerge in applications as diverse as automotive, fitness and medical arenas. Despite their broad difference in usage, however, IoT mote designs share a similar architecture (Figure 1). Here, energy transducers and power-management circuits serve as the power supply to the load, while MCUs wirelessly transmit sensor data to other motes, data aggregators, systems, or even smartphones and tablets in the wireless sensor network.

Figure 1: While an optimized power-management subsystem (highlighted) provides maximum power from ambient energy sources, "zero-power" operation demands the lowest possible power consumption from the processor and RF transceiver used to collect and transmit sensor data (Courtesy of Linear Technology).

In this architecture, power-management designs based on specialized ICs, such as the Linear Technology LTC3588-2, Maxim Integrated MAX17710 and Texas Instruments BQ25504 form the heart of energy-harvesting power supplies. On their input side, these subsystems provide impedance matching and conditioning required to maximize output from solar cells, piezoelectrics, thermoelectric generators (TEGs), and other transducers used to convert energy from diverse ambient sources into useful power (Figure 2). Devices such as the TI BQ25504 even support dynamic load-matching through maximum-point power tracking (MPPT) to maintain maximum output from fluctuating ambient sources. In turn, these subsystems supply the load with required voltage and current levels — and even signal the processor when fading ambient sources or depleted energy stores cause supplies to fall below minimum levels.

Figure 2: Despite the trickle of energy available from some ambient sources, optimization of energy conversion and minimization of power consumption can ensure continued operation of IoT motes for years and even decades (Courtesy of Cymbet).

With the availability of specialized ICs, energy-harvesting power supply subsystems can deliver significant power even from the trickle of energy available from ambient sources. Ultimately, however, a successful MCU-based wireless sensor design relies on building the system with the lowest possible power requirements and using the most power-conserving MCUs and RF transceivers available. In practice, achieving minimum overall power consumption requires a detailed examination of the power associated with different operational states of the MCU and transceiver.

Power profile

In a typical IoT smart sensor application, a mote might sit idle for long periods of time between peak periods of activity for sensor data acquisition and wireless transmission. As a result, power consumption for the underlying wireless sensor system can exhibit periodic peak periods of power demand separated by extended periods of quiescent operation (Figure 3). In fact, peak-power demand can typically exceed instantaneous energy output from ambient sources, requiring use of energy-storage devices such as supercapacitors and thin-film batteries to meet peak-power requirements.

Figure 3: The typical power profile for a wireless-sensor node exhibits sustained periods of low-current requirements interrupted by high-current demand that can easily exceed instantaneous power harvested from ambient sources (Courtesy of Silicon Labs).

Within these peaks, wireless communications typically accounts for most power consumption due to its relatively extended time of operation (Figure 4). In many cases, the use of more basic communications protocols with minimum overhead can shorten this period of high-power consumption. At the same time, the use of more efficient RF transceivers can significantly reduce their peak current demands. For example, the Atmel AT86RF233 2.4 GHz requires only 13.8 mA Tx current at maximum transmit power and 11.8 mA Rx current while featuring deep-sleep current of only 20 nA. Although higher-frequency 2.4 GHz transmissions support higher-data bandwidth, sub GHz radios, such as the Texas Instruments CC1000 RF transceiver, feature greater range and lower-power operation. In 433 MHz transmit mode, CC1000 achieves current as low as 5.3 mA at 0.01 mW transmit power output or 26.7 mA at 10 mW.

Figure 4: Radio communications can dominate power profiles, but designers can use basic communications protocols to shorten the duration and more power-efficient RF transceivers to lower peak-power demand (Courtesy of Texas Instruments).

Minimizing idle power

In cases with slow-changing data, where idle time overwhelmingly dominates peak-period duration, the system's quiescent power consumption can easily represent the largest source of power loss over time. Since few ambient sources can provide the kind of sustained energy output suggested in Figure 3, power conservation remains a critical requirement to ensure sufficient power is available for any operating state.

To ensure minimal power consumption, conventional power-conservation techniques, including decreased operating voltage and lower clock frequency, are fundamental in designing these systems. In fact, MCUs designed for ultra-low-power applications feature the ability to operate at lower clock rates down to 1 MHz and supply voltages down to 1.8 V and lower. Along with lower-power requirements associated with lower frequency and clock rates, these MCUs feature reduced current requirements in low-power idle modes.

Designers can find a wide range of ultra-low-power MCU families ranging from 8-bit to 32-bit devices. For example, operating at 1 MHz and 1.8 V, the Atmel AVR ATtiny 8-bit MCU family features an active-mode current of 200 µA and idle-mode current of 25 µA and consumes < 0.1 µA in power-down mode. The Renesas 16-bit RL78/G13 1.6 V MCU family requires only 66 μA/MHz in run mode and 0.23 μA in stop mode, which includes RAM retention. Based on the low-power ARM Cortex-M0+ 32-bit core, the STMicroelectronics STM32 L0 family operates down to 1.65 V and consumes only 87 μA/MHz in run mode and 250 nA in ultra-low-power mode.

Multiple power modes

At the low supply levels available in most energy-harvesting applications, designers can implement more sophisticated strategies for limiting power consumption when they are able to control power consumption at a finer level of granularity beyond simple run and idle modes. In fact, ultra-low-power MCUs typically support several power-saving features, including multiple-power modes that allow designers to place the MCU in a low-power state while retaining some level of functionality.

For example, designers can enable different subsets of on-chip blocks using Microchip's 16-bit PIC XLP 24F MCU family multiple power modes, including:
  • Run mode (CPU, Flash, SRAM and Peripherals On)
  • Doze mode (CPU Clock Runs Slower than Peripherals)
  • Idle mode (CPU Off, Flash, SRAM and Peripherals On)
  • Sleep mode (CPU, Flash and Peripherals Off and SRAM On)
  • Deep sleep mode (CPU, Flash, SRAM and Most Peripherals Off)

As a result, power draw for these 1.8 V MCUs can range from 150 µA/MHz in run mode to 80 µA/MHz in idle mode with deep-sleep currents down to 20 nA typical.

Ultra-low-power MCUs such as Silicon Labs Zero Gecko 32-bit ARM Cortex-M0+-based MCUs similarly feature a number of modes that minimize power consumption while retaining functionality. For example, while their run mode requires only 114 µA/MHz at 3 V, Zero Gecko MCUs feature low-power sleep mode (46 µA/MHz), deep-sleep mode with RTC (0.9 µA), stop mode (0.5 µA) and shutoff mode (20 nA).

Although multiple low-power modes provide reduced-current consumption during quiescent periods, designers need to balance the power savings from sleep mode with the power lost in transitioning from sleep to active mode. For devices such as the Texas Instruments MSP430L092 16-bit MCU, however, this time is minimal. At 1.3 V, the TI MCU features active-mode current of 45 µA/MHz, standby current of 6 µA and off-mode current of 3 µA. At the same time, the MCU requires less than 5 µs to transition from off mode to active mode, and can even operate with supplies down to 0.9 V.

Ultra-low-power MCUs typically support another important power-conservation strategy by allowing designers to selectively disable clocks to unused modules and peripherals or even power them down. Atmel's SleepWalking technology in its AVR series adds intelligence to AVR on-chip peripherals that allows the peripheral itself to determine if the CPU is required. For example, an AVR MCU's on-chip ADC can wake the CPU only if a sensor reading has exceeded a specific threshold.

In summary

Efficient energy-harvesting ICs optimize energy conversion from ambient sources, but the use of ultra-low-power RF transceivers and MCUs is essential for reducing power consumption in IoT wireless-sensor designs. Available ultra-low-power MCUs provide multiple low-power modes and peripheral power management features that allow engineers to exercise fine-grained control over power consumption — enabling near "zero-power" operation in IoT designs powered by ambient energy.