In this white paper, Silicon Labs (also known as “Silicon Labs“) will help developers gradually understand the components of sensors and discuss the role of each component in development, including its impact on performance. We will also explore some specific challenges and priorities that designers should consider when starting any sensor project design. Click the Read the original button at the end of the article or copy the link to get the complete content: https://www.silabs.com/whitepapers/smart-sensor-design-and-anatomy-lesson
The current technological world has made us aware that sensors can convert data into capabilities that simplify our lives, thus forming the backbone of the Internet of Things (IoT). Modern sensors come in almost every shape and size, and you may currently have devices that integrate one or more sensors. Essentially, the job of a sensor is to acquire data such as light, temperature, or pressure, and respond appropriately using that data. For example, when the room temperature falls below a specified threshold, the thermostat turns on the heater. Sensors are not a new phenomenon; after all, thermostats have been in use for nearly 140 years. However, with the surge of connected devices, networked sensors have sparked a technological revolution, elevating their status in our lives from convenience to necessity, making our homes more comfortable, cars safer, coffee supply timelier, and businesses more efficient.
Smart sensors have permeated almost every corner of enterprises, and the most common types of IoT sensors currently under development include:
Analysis of Smart Sensors
Battery-powered sensors are often installed on printed circuit boards (PCB), which affects many components that make up the sensor, including antennas and RF functions. We won’t spend too much time discussing PCB, but it’s important not to underestimate its impact on sensor performance.
The battery is another key element in sensor design as it must provide the necessary energy while having enough capacity to keep the design running throughout the required lifespan. The battery also affects the size requirements of the device. Ideally, simply using a larger battery would meet the increasing performance demands or extended lifespan. However, considering the growing consumer demand for miniaturized products, large batteries are not practical. For example, if a door or window sensor requires a D battery, it would be limited in use due to its size; even the size of AA or AAA batteries can pose problems for consumers. One trend we see is that while devices are becoming smaller, battery life is actually increasing.
The next important component of battery-powered sensors is the wireless system on chip (SoC). It is crucial to optimize the SoC in the following aspects:
-
Receiving Sensitivity
-
Transmitting Power
-
Having an Appropriate Processor Speed
-
Having Sufficient Flash and Random Access Memory (RAM) to Run Communication Protocols
You will also need sensing elements to perform the functions originally designed for the sensor. Another important component is the software running on the sensor, which is responsible for managing all other components and driving communication inside and outside the sensor.
The final component is the casing. The shape and size of the casing are influenced by the battery size. It also affects RF performance—certain plastics, and even the dyes used when coloring these plastics, can affect the device’s RF transmission.
Comparison of Common Battery Options for Sensor Development
Table 1 compares some battery types that developers can choose from when building connected sensors. We can see that the CR123A lithium battery has a voltage of 3.0V, suitable for powering low-voltage circuits with a single cell. This battery has a relatively large capacity and a larger size. The second battery in the table is the slightly smaller CR2 lithium battery, which also has a rated voltage of 3.0V, but a slightly lower capacity and smaller size. The third type is the CR2032 button lithium battery, which is the smallest in the list and very suitable for small volume sensors.
Figure 1: Comparison of Common Battery Options in Smart Sensor Development
The last two battery types (alkaline AA and alkaline AAA) both have quite a large capacity, but are larger than the CR2032 button battery. These batteries can only produce 1.5V of voltage, meaning two batteries are needed to generate the same 3V voltage as the button lithium battery. Therefore, unless the design requires a very large capacity, choosing button batteries has the most direct advantage as it supports the smallest form factor design.
Button lithium batteries also have another characteristic to consider: the maximum pulse load or discharge that this type of battery can provide. In the above Figure 1, you can see that the CR2032 has a very low maximum pulse discharge capability, as do other button lithium batteries. Although this value indicates the maximum pulse load that button batteries can provide, their capacity will also significantly decrease accordingly.
Button lithium batteries have a higher internal resistance, see the two examples in Figure 2. On the left is a circuit powered only by the battery, and on the right is a circuit powered by both the battery and the power management IC. The peak current on the button lithium battery can cause or produce a voltage drop, leading to circuit shutdown or even reset. Additionally, drawing peak current from the button lithium battery can significantly reduce battery life. Traditionally, these drawbacks have been addressed by adding a very large capacitor. The left image shows the pulse discharge produced by a button lithium battery with a parallel large capacitor for high-power RF transmission. In this example, the pulse peak is 38.5mA. Any current exceeding 15mA will reduce the lifespan of the button battery, which is not ideal. The right image shows the same high-power RF transmission, but here, the same button lithium battery is managed by Silicon Labs EFP01 energy-saving power management IC (PMIC). Not only did the peak discharge reduce from 38.5mA to just 11.8mA, but a smaller 47µF capacitor can be used.
Figure 2: Maximum Pulse Load or Discharge under Different Battery Types
WirelessSoC and Why Silicon Labs Is Leading the Way
The next important consideration for building battery-powered sensors is the wireless SoC, which is the core of the sensor. Figure 3 compares Silicon Labs’ first generation (Series 1) and second generation (Series 2) wireless development platforms: both the first and second generation platforms offer a variety of low-power, multi-band, multi-protocol product options for a wide range of applications. The second generation platform has undergone significant updates to the core components, integrating a dedicated encryption security core.
Figure 3: Wireless Gecko has larger memory and features including wireless software updates to support enhanced application and ever-evolving protocol demands
Starting from the left side of Figure 3, EFR32xG1 and EFR32xG14 are very suitable for single-protocol battery-powered devices operating in 2.4 GHz or sub-GHz. The second product in the figure is the EFR32xG13 product, which is very suitable for single-protocol and dynamic multi-protocol battery-powered devices also operating in 2.4 GHz or sub-GHz. Next, EFR32xG12 is very suitable for dynamic multi-protocol battery-powered devices operating in 2.4 GHz. Finally, the EFR32xG22 product is very suitable for single-protocol and dynamic multi-protocol battery-powered devices operating in 2.4 GHz. All of Silicon Labs’ SoC combine energy-efficient microcontrollers with highly integrated radio receivers.
The operating mode of the SoC affects battery capacity and ultimately impacts battery life. For typical wireless contact sensors, battery life primarily depends on sleep current.
Figure 4: Sleep has the Greatest Impact Compared to Other Events
You can see how significant the impact of sleep is in Figure 4. During the battery’s lifecycle, application events, data polling, and even the self-discharge of the battery or wireless upgrades cannot compare to the energy consumption during sleep. In Silicon Labs’ products, energy modes are divided into EM0 (Active), EM2 (Sleep with RAM Retention), EM3 (Stop), and EM4 (Hibernate). EM4 provides the lowest sleep current, but waking up from EM4 takes a longer time. This can be challenging to meet the certification requirements for protocol standards like Zigbee, Thread, and others.
EM2 achieves a reasonable compromise between sleep current and wake-up time. Taking EFR32xG22 as an example, comparing EM2 with EM4 mode, in some cases, EM4 consumes 130 nanoamps of current but takes 8.8 milliseconds to wake up; EM2 consumes 1.9 microamps of current but can achieve a rapid wake-up time of 13.2 microseconds.
In wireless systems, the communication range is determined by the receiver’s sensitivity and output power of the transceiver. From the perspective of transmitting from the transmitter to the receiver, this is generally referred to as link budget. The communication data rate also affects sensitivity. In Figure 5, you can see that as the data rate decreases, the receiving bandwidth narrows, resulting in increased radio sensitivity. A common technique is to adjust the transmission power in the system to match the desired optimal range without consuming more energy.
Figure 5: Adjusting the Output Power of Each Node Can Ensure Sufficient Link Budget to Provide the Desired Communication Range
The Impact of Spectrum on IoT Sensor Operation
Widely used wireless devices primarily operate in the Industrial, Scientific, and Medical (ISM) frequency bands. ISM spectrum can be divided into two bands: sub-GHz and 2.4 GHz. Compared to 2.4 GHz, sub-GHz has many advantages, including path loss. Path loss refers to the reduction of power when a signal is transmitted over a certain distance. For example, when a 2.4 GHz signal is transmitted in the air for 10 meters, the path loss is 60dB. In contrast, a 900 MHz signal transmitted for the same 10 meters has a path loss of 51.5dB. For the 900 MHz signal, the loss is reduced by 8.5dB.
Figure 6: The Industrial, Scientific, and Medical (ISM) Spectrum Can Be Divided into Two Bands: sub-GHz and 2.4 GHz
The 2.4 GHz signal has a high data rate that can easily exceed 1 MB/s. The 2.4 GHz band can also use small antennas, which are less than one-third the size of 900 MHz antennas; however, its communication range is limited. The 2.4 GHz spectrum is also very crowded and susceptible to significant interference from devices like Wi-Fi and Bluetooth. In contrast, the coverage of sub-GHz radios is much higher than that of 2.4 GHz radios—the coverage of sub-GHz radios can reach several kilometers and can operate for years on a single battery with low power consumption.
We know that communication range depends on transmitting power, receiver sensitivity, and data rate. However, the range can also be influenced by antenna selection, so it is essential to understand the characteristics and trade-offs of the specific design to choose the right antenna. In battery-powered sensor applications, size, radiation pattern, ease of design, manufacturability, and cost should all be considered.
In Figure 7, you can see that the left side shows a dipole antenna. This is a differential structure, typically with a length from one end to the other that is half a wavelength. This type of antenna should be kept away from the ground plane and any metal or conductive objects. Dipole antennas can easily match to a 50 ohm impedance, but at 900 MHz, its length will exceed 6 inches, making it difficult to use in small battery-powered sensors.
Figure 7: Comparison of Dipole, Monopole, and Loop Antennas
Another type of antenna (as shown in the middle of Figure 7) is a 1/4 wavelength monopole antenna. This type of antenna is also easily matched to a 50 ohm resistor. The design of a monopole antenna is very simple, and its resonant frequency can be adjusted just by changing the length of the antenna. When the physical size is acceptable, this type of antenna is a great solution. For example, at a 900 MHz frequency, if there is a ground plane, the length of a 1/4 wavelength antenna is about 3 inches. The antenna on the right side of Figure 7 is a loop antenna, which can have two sizes in terms of power: small loop antennas and large loop antennas. For devices like battery-powered sensors, only small loop antennas should be considered because large loop antennas have a circumference close to a wavelength, which is about 12 inches for devices operating in the 900 MHz band. Small loop antennas have a very narrow bandwidth, which is favorable for product selectivity, but also makes tuning critical. However, once the tuning is set, it does not easily detune due to factors such as handheld interference or nearby objects, making it very suitable for handheld devices.
Spiral antennas can be made from any conductive material. Small spiral antennas operate perpendicular to the spiral axis, and this must be considered when integrating a small spiral antenna into a device, as the antenna will protrude from the circuit board. This type of antenna can also be tricky to use because its impedance depends on many parameters, including the diameter of the coil, the pitch of the loops, the tightness of the winding, the length of the coil, and its operating frequency.
Any changes in these parameters, even nearby objects including people, can cause the spiral antenna to detune. However, from a size perspective, spiral antennas can be very small. In fact, if wound tightly enough, it can be much shorter than a monopole antenna of the same frequency. The antenna shown on the right side of Figure 8 is a chip antenna, which is the smallest antenna designed for frequencies ranging from 300 MHz to 2.5 GHz. Chip antennas have a very narrow bandwidth and must be manufactured to precise frequencies. This may be the most expensive antenna solution and is typically used for surface-mounted devices.
Figure 8: The Last Two Antennas to Consider Are the Spiral Antenna and the Chip Antenna
Sensor Software and the Brain of Smart Sensors
The software running on the sensor is crucial for creating reliable, robust, easy-to-develop, and secure devices. Silicon Labs offers a variety of SoC options for battery-powered devices, all sharing a common development environment called Simplicity Studio. As shown in Figure 9, the platform components include a multitasking operating system (OS), Vault Security, and the Radio Abstraction Interface Layer (RAIL). RAIL provides an interface layer for the underlying hardware, allowing you to simplify and shorten the development process by abstracting all the complexities of registers and the details required to set up the underlying radio hardware.
Figure 9: The Software Running on Wireless SoC Is Modularly Designed, with Higher-Level Functions Layered or Built on Lower Levels
Above this platform is the protocol stack, responsible for implementing all the complex functions of various protocols. For example, the Zigbee protocol stack, Z-Wave protocol stack, Thread protocol stack, and Bluetooth protocol stack. Above the protocol stack is the application layer, responsible for providing various API or programming interfaces to the applications above so that they can connect to the protocol stack below. This way, the upper application layer can utilize the functions provided by the protocol stack. Finally, the operating system in the program module interfaces with all these upper layers because it is responsible for timers, inter-task communication, synchronization, scheduling, interrupts, exceptions, and task dispatching.
Figure 10: Silicon Labs Wireless SoC Supports the Most Popular Protocols Used by All Major Ecosystems. Whether it’s Zigbee, Z-wave, Bluetooth, or Thread, you can develop devices for the major ecosystems
Silicon Labs EFR32 wireless Gecko products integrate power management, security, and multi-protocol support to help developers meet the requirements of IoT sensor design. For more information, please visit the link: https://www.silabs.com/wireless/technology
This article is also published in EDN China magazine
Scan the QR code below to follow Silicon Labs’ social media platforms
