“While the proliferation of the Internet of Things (IoT) is indisputable, only a few publications to date have inspired the public on issues related to energy supply, supporting the rapidly growing number of sensors and transceivers.
While the proliferation of the Internet of Things (IoT) is indisputable, only a few publications to date have inspired the public on issues related to energy supply, supporting the rapidly growing number of sensors and transceivers.
Trillion Sensor Visions: Trillion Sensor Vision Predictions
Figure 1. Forecast of the number of sensors deployed per year
(Source: Sia and Tsensor.org 2015)
Rapid IoT growth confirmed at World Material Forum 2018 in Nancy, France
And the high requirements for data storage, processing and transmission will be important issues for the sustainability of the project. Therefore, any form of energy harvesting is welcome unless absolutely mandatory.
Comprehensive potential solutions to this challenge include:
• Design ultra-low power embedded hardware platforms
• Intelligent system level power management
• Harvesting energy from the work environment to make equipment self-powered
When implementing these schemes, electronics designers must keep in mind that IoT sensors must not only measure a value (including temperature, humidity, pollution, light levels), but must also communicate that value to their system host C, which is usually wireless, to Limited power.
To achieve this, every system-level component in the design must be fully considered, including sensors, receivers, energy sources, and communication duty cycles.
This white paper will explore how ON semiconductor‘s energy-efficient solutions enable battery-free applications with state-of-the-art persistent sensor technology.
Ultra-low-power transceivers and communication protocols
The first step in the design is to select an RF transceiver that is both ultra-low power and supports wireless protocols to transmit information. An appropriate wireless protocol should support and include the following features:
• Support LAN transmission distance
(about tens of meters indoors)
• Low power consumption through architecture
(Short frames and low Tx power reduce CPU and radio power budget)
• Support for secure transmission (eg encryption)
• Provides a simple receiving mechanism (beacon)
• Provides easy-to-use hardware implementation
(e.g. direct interface between sensor and transceiver)
• High level of integration (SiP or single chip)
• Provides standardized communication protocols (IEEE or SIG type) including interoperability between sensor nodes and gateways
• Low implementation cost to support mass market availability
Fortunately, the Bluetooth Special Interest Group (SIG) and the Zigbee® Alliance now offer wireless protocols, focusing on optimizing their respective protocols for many years. We now offer Bluetooth 5 as well as Zigbee Green Power protocols optimized for short frame duration, security and transmit power.
Transfer all the information needed to join in less than 10 milliseconds (ms). The key is how to implement these protocols with energy-optimized devices and take full advantage of voltage and current resources. ON Semiconductor has designed the device based on its expertise in ultra-low power microcontrollers and hearing aid audio, with a power budget as low as 10 mW at 6 dBm. Several products are now available to solve the challenge: the NCS36510 supporting the Zigbee protocol and the RSL10 supporting Bluetooth Low Energy. The combination of protocols and smart power implementation requirements result in the following equations, shown in Figure 2.
ON semiconductor Technology: ON Semiconductor Technology
IoT Protocols ZIGBEE GP; BLE UNB SubGHz: IoT Protocol ZIGBEE GP; Bluetooth Low Energy UNB SubGHz
Harvesters: Energy Harvesters
Figure 2. “Rule of thumb” for IoT application energy consumption
Choosing an Energy Harvesting Source
The equations in Figure 2 provide us with guidelines for the energy requirements of modern low-power connectivity and communication protocols. All that remains is to choose the appropriate acquisition source and range of use. Time is another factor that must be considered. The power produced by a continuous acquisition scheme may be low, but the intent is to accumulate over time, so the gain factor is important. For example, acquiring for 1 second and transmitting for 10 ms produces a gain of 100. By comparison, acquiring for 10 seconds and transmitting for 5 ms yields a gain of 2000. Electrolytic capacitor technology fully supports energy accumulation in the range of seconds.
Solar based energy harvesting
Taking the RSL10 Bluetooth 5 radio or the NCS36510 Zigbee system-on-chip (SoC) as an example, we can calculate that during the protocol transfer (lasting up to 10 ms) we will need around 10 mA. For transfers per second, we can increase the gain by a factor of 100. If the transmission happens every 10 seconds, the gain will be 1000x. This means we can set a current source of 10 mA / 100 = 100 _A or 10 mA / 1000 = 10 _A for the solar collector.
Interestingly, solar cells like Ribes Tech’s FlexRB-25-3030 provide 16 _A at 200 lux (lux) or 80 _A at 1000 lux. This is exactly what is needed.
Working voltage: working voltage
Working current: working current
Maximum voltage: maximum voltage
Maximum current: maximum current
min: minimum value
typ: typical value
max: maximum value
The JV curve shown in the figure was measured at 6500 K with a 1000 lux light source fluorescent tube
Figure 3. Electrical Specifications of Ribes Tech’s FlexRB-25-7030, FlexRB-20-6030
Using a solar cell like Ribes Tech’s FlexRB-25-3030 will allow us to provide an autonomous sensor that sends Bluetooth Low Energy or Zigbee frames with a duty cycle of 1 to 10 seconds.
Common lighting conditions
Most solar cells are characterized by two sets of lighting conditions: 200 lux and 1000 lux.These conditions cover a wide range of everyday lighting as shown in the table below
Table 1. Common Lighting Operating Conditions
*Measured using Velux’s Luxmeter app running on an iPhone® 6.
**This is not recommended as a working condition for an actual configuration.
Continuous measurement field
Based on the information mentioned earlier in this document, we can begin to study the behavior of the entire acquisition system. There must be an energy preload phase so that the device harvests energy from the solar cell before triggering the first communication. The next section (Technical Challenges) will detail various tips and guidelines for a successful implementation. Once the device has captured and stored enough energy, the microcontroller (MCU) must set the communication parameters, Tx power, channel selection and temperature measurement.
This activity will be obtained when the MCU is mostly active, so the buffered energy must be high enough to transmit as many beacon frames as possible.
Figure 4. Conceptual energy consumption view under given lighting conditions
With a 2-second duty cycle, we can implement a battery-free sensor node capable of measuring slowly varying parameters (e.g., humidity, temperature, atmospheric pressure, room parameters, light intensity).
Technical challenges and implementation
The first step is to select a communication and data processing IC that can support the desired communication protocol and has an available energy budget dictated by the harvesting device. In most cases, the selected device is required to support power-efficient standby and deep-sleep modes to save power when no operation is required. To simplify power delivery, devices with a minimum input voltage or wide input voltage range are preferred. This way, a simple buck or linear regulator can be used to regulate or limit the system voltage.
Similar requirements apply to the sensors used in the system. If sleep mode cannot be provided, power gating can be implemented to disable sensor power when sensing is not required.
The next section to consider is energy storage and energy management for powering sensors and microcontrollers. In order to store the harvested energy, various methods are possible. Which method is most suitable depends on the requirements of the target application. Typically, capacitor or battery based schemes can be used.
Capacitor-based solutions generally have lower overall capacity than battery solutions at the same volume because of their lower energy density. This makes batteries more suitable for sensors that need to stay on for extended periods of time without a light source.
Ragone Plot of Electrochemical Devices: Ragone Plot of Electrochemical Devices
Fuel Cells: fuel cells
Lead-Acid Battery: lead-acid battery
NICd Battery: NICd battery
Lithium Battery: Lithium battery
Double-Layer Capacitors: Double Layer Capacitors
Aluminum-Electrolytic Capacitors: Aluminum Electrolytic Capacitors
Power Density: Power Density
Source US Defence Logistics Agency: Source US Defense Logistics Agency
Figure 5. The Ragone diagram helps us choose the appropriate energy storage technology
The challenge with battery-based systems is that they often require more complex energy management systems. This includes charge and discharge control and battery protection against overcharge and overdischarge. This increases the system complexity as well as the bill of materials (BOM) (cost), as such energy management systems typically involve switching regulators (additional passives) and result in a more complex IC due to the required functionality. The complexity of the chip and the requirement for high energy efficiency and low quiescent current often result in rather expensive IC solutions.
Capacitor-based solutions may be a more cost-effective solution in applications that do not require longer operating times without exposure to light. The storage capacitor temporarily accumulates energy from the solar harvesting device until enough energy is available to perform measurements and transmit the results. When using capacitors with adequate voltage ratings, no charging circuit is required. The open circuit voltage of the solar collector used when exposed to the expected peak brightness determines the maximum input voltage. If the capacitor’s voltage rating exceeds the open circuit voltage, no charging circuit or protection is required.
For both battery and capacitor-based solutions, the output voltage needs to be regulated to provide the appropriate voltage for the connected circuits (sensors, microcontrollers, etc.). Systems using lithium-based storage options achieve voltages above 4 V, which are often outside the input voltage range of sensors and microcontrollers. To match the supply voltage, which is typically 1.8 to 3.3 V, a step-down conversion is required. In capacitor-based systems, the voltage is linearly related to the amount of charge stored. This can cause large voltage variations throughout the discharge cycle, which not all sensors or microcontrollers can accept, so some kind of voltage regulator is needed to stabilize the power supply.
Figure 6. RSL10 Solar Cell Multi-Sensor Board
The RSL10 Solar Cell Multi-Sensor Panel (RSL10-SOLARSENS-GEVK) is a comprehensive development platform for battery-free IoT applications including smart buildings, smart homes and Industry 4.0. The board is based on the industry’s lowest power Bluetooth Low Energy radio (RSL10) with multiple sensors for temperature and humidity sensing (BMA400 – Smart 3-Axis Accelerometer, BME280 – Smart Environment Sensor and NCT203 Wide Range Digital Temperature Sensor ).
The board also features a 47__F ultra-low cost, low weight and thin storage capacitor, a programming and debugging interface, and an attached solar cell.
Since the device harvests energy from a low current source, it is important that the leakage current of the entire system is small when operating and harvesting energy. Several smart devices were selected for this purpose, including an onboard ultra-low quiescent current LDO (NCP170).
PV CELL: Photovoltaic cell
10Vmax Capacitor: 10 V maximum capacitor
RSL10 beacon with 10s + 2s timer: RSL10 beacon with 10s + 2s timer
Figure 7. Conceptual diagram of the multi-sensor board
BLE5.0 Advertising: Bluetooth Low Energy 5.0 Advertising
Figure 8. Complete system overview including sensors, gateways, and cloud services
With this list of assets, the range of potential applications is vast, let’s take a quick look at them:
Climate Control (Environment)
Window breakage detection (3-axis accelerometer)
Building automation (with both environmental and 3-axis accelerometers)
Door breakage detection (3-axis accelerometer)
Door switch status report (3-axis accelerometer)
Meeting room occupancy monitoring (with both environmental and 3-axis accelerometers)
Environmental Control (Environment)
Roof and window controls (both ambient and 3-axis accelerometers)
Window breakage (intrusion) detection (3-axis accelerometer)
Industry 4.0 / Smart Cities
Air pollution detection (environment)
Worker safety (both environmental and 3-axis accelerometers)
Security monitoring (3-axis accelerometer)
Integrated/portable sensor (3-axis accelerometer)
Bicycle/motorcycle active helmet (3-axis accelerometer)
Figure 9. Battery-free window sensor demo at Embedded World 2019
Hardware Setup and Optimization
The RSL10 Solar Cell Multi-Sensor Board uses ON Semiconductor’s RSL10 to process measurement data and transmit the results in Bluetooth Low Energy advertising packets. Data packets can be received for visualization using a smartphone or any other Bluetooth Low Energy enabled device.
Aluminum electrolytic capacitors will be used as the main energy storage. Solar harvesters have an open circuit voltage range of 3 to 6 V, so about 10 V rated capacitors can be used in the circuit without any input clamping or protection. In the circuit, the capacitor is charged directly through the solar harvesting element, requiring only a Schottky diode in series. Placing this diode prevents the harvester from discharging the capacitor. We will discuss capacitor capacity later, as it depends on several other aspects discussed in later sections.
The RSL10 SoC contains an integrated DC-DC step-down regulator, allowing the chip to operate over a wide range of input voltages (1.1 to 3.3 V) without the need for additional regulators. Since in very bright conditions the solar collector used may exceed the maximum voltage rating of the RSL10, a linear regulator is used as a voltage limiter. If the input voltage exceeds 3.3 V, the regulator produces a constant supply voltage. If the voltage is below 3.3 V, the regulator will pass the capacitor voltage through without regulation. Since the regulator is only used when there is “too much” available energy, there is no problem with the excess energy being converted into heat. If the demand for power increases, the capacitor voltage will drop, the regulator will no longer work and “waste” energy. However, the simplicity of the regulator results in lower quiescent current. This is essential as it will help keep available energy in low light situations. Figure 10 shows a typical operating scenario, limiting voltage at lower power consumption. When the capacitor voltage drops below 3.3 V (due to higher power dissipation), the LDO regulator stops working and lets the voltage pass directly.
Capacitor Voltage: capacitor voltage
Limited MCU Supply Voltage: Limited MCU supply voltage
Figure 10. Voltage Limiting Using an LDO Regulator
When the system is fully discharged and then exposed to light, the capacitor’s voltage will slowly rise as the charge builds up. By default, RSL10 will (try to) start up once it reaches its lower threshold voltage (~1 V). It only works if the solar collector is continuously supplying the power required for startup to maintain the capacitor voltage at 1 V. If the solar collector provides less power than required, the capacitor voltage will drop. When the voltage falls below the threshold of ~1 V, startup fails because the RLS10 will shut down. This sequence is repeated when the energy output of the solar collector is lower than the RSL10 consumption during startup.
Since harvesters typically cannot produce as much energy in all (light) conditions, a start-up circuit is required to ensure reliable start-up.
The circuit used in this demonstrator ensures that the storage capacitor is sufficiently precharged before powering the RSL10 and other devices. To ensure a successful start-up, the storage capacitor needs to maintain the energy required to fully start the system. In our use case, startup is the sequence from initial power-up to the point in time when the system can enter its deep sleep mode.
The energy required to perform this sequence can be measured. Determine the minimum size of capacitor required based on the energy required and the typical input voltage range of the microcontroller. For the proposed RSL10-based application, ~120 _J is required to start. Combined with the required voltage range of ~1.5 to 3 V, this results in a theoretical minimum capacitance of 35.6 _F. In practice, larger capacitors should be used to compensate for changes in capacitance due to factors such as manufacturing tolerances, different operating temperatures, or component aging.
Start and hold circuit
To enable and disable the power supply to RSL10, use the enable signal of the clamped LDO regulator. The enabled input is generated by two sources. The first source is generated by a voltage supervisor IC (MAX809 from ON Semiconductor) powered by the capacitor input voltage, which enables the LDO regulator once the capacitor voltage exceeds 2.63 V. The second input is used to ensure that the enable pin keeps the output voltage high enough. Depending on the desired threshold, the LDO’s default turn-off threshold can be used (>1.2 V for NCP170; measured at ~1.5 V). In this case, the output voltage of the LDO is fed back to the enable pin. If a higher turn-off voltage is required, an additional voltage supervisor with a threshold > 1.5 V can be added, which will enable the pull-down once the LDO output falls below the threshold defined by the voltage supervisor. Figure 11 shows the schematic of the start-up circuit used. U3 is the secondary voltage monitor, which can be selected based on the threshold voltage required for shutdown.
Startup-Circutry and Voltage Regulation: Startup circuit and voltage regulation
LDO as a active clamping circuit and power switch: LDO as an active clamping circuit and power switch
Storage Capacitor: Storage Capacitor
Turn-on Threshold: Turn-on threshold
(optional) Turn-off Threshold: (optional) turn-off threshold
Figure 11. Startup Circuit Schematic
In Figure 12, the behavior of the startup circuit can be observed. Before point A, the capacitor voltage rises slowly until it reaches 2.63V at point A. The voltage supervisor has an internal delay that delays the actual on-time by tD, which is between 140-460 ms for the device used. After a delay, the MCU power will be activated. When the MCU supply voltage is higher than ~1.5 V, the system can operate normally. Once the voltage drops below 1.5 V (point B), the MCU supply voltage is disabled because the enable pin threshold of the NCP170 used is 1.5 V. After this, the capacitor voltage needs to rise above 2.63 V again to re-enable the MCU power.
Storage Capacitor voltage: Storage capacitor voltage
MCU Supply Voltage: MCU supply voltage
Figure 12. Startup Circuit Behavior
The board contains temperature, air quality and acceleration sensors. All sensors support sleep mode to reduce power consumption when not needed. To avoid the changing capacitor voltage negatively affecting data acquisition, a stable sensor supply voltage of 1.8 V was used. This sensor supply voltage can be disabled to further reduce current consumption. The sensor interfaces with the RSL10 SoC via the I2C bus.
In operating modes supported by temperature sensors and accelerometers, the sensors can monitor their respective physical conditions without RSL10 interaction. In this mode, the sensor will wake up RSL10 using a dedicated interrupt line if the monitored value is out of a pre-set range or if other programmed conditions occur.
Figure 13. RSL10 Solar Cell Multi-Sensor Board
The final PCB size using the 2-layer design is 24 x 51 mm, with all components on the top side to be able to connect to the solar harvesting device on the back.
Solar collectors can be connected in the following ways:
• Use 100 mil pitch connectors on the left side of the board
• 4-pin 1mm pitch ZIF connector on the right side of the board
• Pads on both sides of the board allow direct connection to panels or other connectors
Figure 14. Battery-free multi-sensor node demonstrated at EWC 2019
Firmware Setup and Optimization
The intended behavior of the RSL10 solar panel is to measure environmental parameters and transmit them in Bluetooth low energy advertising packets. The time interval between measurements and transmissions depends on the available energy. Firmware needs to monitor available energy and adjust system power state to optimize system performance.
At system startup, the RSL10 initializes all required peripherals and clocks resources and the Bluetooth Low Energy baseband. These steps are critical to make all power states of the RSL10 available. To save power, all unused peripherals remain disabled. In addition, the power to the sensor is turned off before the actual measurement is required.
After the RSL10 is initialized, the system needs to determine if it has enough energy remaining to perform certain measurements, or if it is necessary to enter an ultra-low power deep-sleep mode to charge the storage capacitor to a higher voltage level. To determine the currently available energy, the RSL10 can measure the supply voltage. A voltage of 3.3 V indicates that the capacitor is full and the LDO is already limiting the output voltage. For supply voltages below 3.3 V, the RSL10 directly measures the capacitor voltage and can determine the energy content.
If the energy is insufficient to perform the required measurements, the RSL 10 enters its deep sleep mode. In this mode, the power consumed by the RSL10 is in the range of 62.5 nW, and the storage capacitor can be charged even in low light conditions. In deep sleep mode, the peripherals of RSL10 are disabled. A portion of RAM is reserved in order to maintain some variable system state during deep sleep mode. Waking up from deep sleep is much faster than a full boot and requires far less energy.
After a fixed period of time in deep sleep mode, the RSL10 wakes up to check that the storage capacitor has accumulated enough energy to perform measurements and transmit data. The energy threshold that determines whether a measurement is feasible is determined experimentally. If the energy level is still insufficient, the RSL10 goes into deep sleep mode again.
If the available energy is sufficient to make a measurement, the power to the sensor is enabled and the I2C interface is initialized. Via I2C, the sensor is configured to perform its measurements. Once the measurement is complete, the results are read back and copied into the advertising packet used to transmit the measurement data.
Advertisement packets containing the measurements are then sent. After transmission, the RSL10 enters deep sleep mode for the minimum required transmission interval. The sequence is then repeated, starting with determining the available energy after wake-up.
Bluetooth Low Energy Considerations:
Selecting to transmit measured sensor data to other devices via Bluetooth Low Energy using advertising packets is the most energy efficient way. This allows the collector of the RSL10 solar panel to target all nearby Bluetooth low energy devices it scans without having to establish and maintain a connection. Additionally, the solar collector sends data in broadcaster mode, stating that it does not enable the receiver after each advertising packet sent. This saves additional power consumption at the cost of not connecting and not being able to send scan response packets, increasing the maximum limit of advertising data from 31 bytes to 62 bytes. Depending on application requirements, advertising in connectable mode may be required to allow certain devices to configure sensor node parameters such as preferred advertising interval and preferred measurement interval.
To overcome the limitation of small advertisement packets that only allow 31 bytes of data, it is possible to switch between different advertisement payloads per advertisement interval. This can be used to send custom sensor data frames in one advertisement packet, followed by Eddystone Beacon URL frames in the next advertisement packet. Eddystone URL packets can be used to link to web pages with additional information and provide downloadable applications for displaying sensor data.
Unlike connected devices for context-aware services defined by the Bluetooth SIG, there is no standardized format for transmitting various sensor data using only advertising packets.
In this way, the custom ad data frame is used to transmit sensor data to the scanning device. These devices require specialized software or applications capable of parsing and processing the content of such ad packets. In industrial use cases where the entire infrastructure is managed by a single entity, this might not pose a problem, but if applied to a market that might require collaboration of equipment from multiple vendors, it could lead to interoperability issues.
Based on the above behavior, ON Semiconductor developed the RSL10 solar cell multi-sensor board firmware.
ON Semiconductor provides an Eclipse-based environment, the RSL10 Software Development Kit (SDK), for software development based on the RSL10 platform. The RSL10 SDK contains a fully integrated development environment with a powerful editor, toolchain, documentation, various sample codes and CMSIS-Pack based software packages.
Figure 15. RSL10 Software Development Kit (SDK)
The firmware can be configured using the CMSIS Configuration Wizard editor included with the RSL10 SDK, as shown in Figure 16. By using the graphical interface to provide a detailed description of each parameter and to check the correct range of input values, the required parameters can be changed to quickly evaluate different software configurations. . For cases where more complex changes are required for evaluation, source code and example projects are provided in the CMSIS package.
Figure 16. Configurable parameters displayed in the CMSIS Configuration Wizard
Figure 17 shows the current draw of the board during a sensor measurement event, followed by the announcement of the measurement data. During this event, a total of 60 _J of energy was used to measure sensor data and publish the results. If no sensor measurements are scheduled and the board only announces, the energy consumption is reduced to 20 _J.
Figure 17. Typical Operating Cycle for Sensor Measurement and Announcement
(3 V supply, announcement interval set to 1 s, sensor measurements during each announcement interval)
Receive beacon data
The RSL10 Solar Cell Multi-Sensor Board advertises sensor data as part of the manufacturer’s advertising package specific data along with the Bluetooth Low Energy logo and the full local name of the board. This allows access to sensor data from all devices that expose manufacturer-specific data to applications, including Android and IOS devices.
In this case, connect the RSL10 USB dongle (RSL10-USB-001-GEVK) to the host PC to Display the captured sensor data. Use the RSL10 USB dongle with Python bindings and the included software Bluetooth Low Energy Explorer to create a simple script that scans for nearby Bluetooth Low Energy devices and displays sensor data when they have matching advertising data .
Figure 18. Sensor data captured by the RSL10 USB dongle displayed
With this reference platform, ON Semiconductor has demonstrated that it is entirely possible to fabricate a low-cost, small form factor sensor node, powered entirely by solar energy, with capabilities including continuous sensor monitoring and data transfer to a cloud gateway. Multiple use cases will greatly benefit from the new technologies and capabilities of the RSL10 solar cell multi-sensor platform, including smart buildings, city management and mobile health. Developers use the platform to create new innovative sensor designs that can help revolutionize the Internet of Things by filling the energy demand gap created by implementing billions of smart sensors.