An Introduction to IoT Components

Sensor nodes on the Internet of Things enable formerly unimaginable levels of remote monitoring and control. From locomotives and jet engines to baby monitors and home appliance controls, new applications are challenging the imaginations of designers — both in startup companies and within giant corporations. Which will be the next billion-dollar product line? They wonder. Remarkably, dramatic results can many times be obtained with low-cost building blocks. This article reviews some of the readily available semiconductor components, and suggests some data gathering nodes which can put them to use.

Introduction

If it does nothing else, the vast collection of sensor data gathering tools called the Internet of Things (IoT) will offer extremely granular electronic monitoring systems. Enabled by arrays of inexpensive components — intelligent sensors and data analysis tools — new smart buildings will provide not only dramatic energy savings but also improve the productivity of occupants. Self-driving cars will provide safety levels surpassing their human-piloted counterparts. And healthcare data banks will improve diagnostics and relieve infirmities.

IoT monitoring lends itself to dramatic transformations in otherwise traditional businesses. General Electric Aviation, for example, no longer sells jet engines: GE “rents” them. Each rental comes with a service contract. Using thousands of sensors to collect terabytes of engine data, and examining historical data across an entire jet fleet, GE technicians can predict wear-and-tear on engine components — often months in advance of a possible failure. Pre-emptive maintenance “bursts,” have proved less costly than ordinary maintenance “routines,” the company says

The sensor monitoring model can be harnessed to automobiles, factory automation, machine tools, individual health and fitness monitoring — as well as jet engine maintenance.

There is a Smart Office Building example: Some 40% of the electricity used by an office skyscraper is devoted to interior lighting. But the lighting controls will not simply turn the conference room lights on or off. By monitoring ambient lighting, recognizing who is in a room, and also how it is used, the lighting controls can not only save electricity, but also adjust the lighting of a room to increase the comfort and productivity of the people who use it.

While it becomes easier to imagine the energy savings the IoT enables for industrial settings, note that not every IoT monitoring station requires a billion-dollar service contract. In fact, the IoT itself has been inspired by the availability of low-cost components.

These include $2.50-processors, $1.50-sensors, and $1.00 communications ports. This has encouraged hobbyists and startups to come up with much more clever devices than Internet-enabled garage door openers. The Nest Labs, for example, sold its plans for an IoT-enabled thermostat for $3.2 billion.

So accessible is the hardware for IoT sensor nodes that at least one market research firm has speculated that more than half of the IoT revenues in 2020 will come from monitoring systems that have not been invented yet. Would you like to build one yourself?

Anatomy of a Wireless Sensor Node

We don’t need a jet engine sensor to control the lighting of a conference room in a smart building. But the topology of a smart building control node is likely to be very similar. There are three main components for a wireless sensor node:

• Sensors (often including sensor signal conditioning circuits in the same package)
• Microcontrollers (including embedded memory, and power management in the same system-in-package)
• Wireless (or Wired) Communications (the communications circuitry required for the sensor node to transmit data to a local network processor and/or the cloud computing resource).

The components in an IoT node will vary in sophistication, depending on the application. While engineers have a tendency to over-specify, they are usually reined in by cost considerations. But the basic topology of a wireless sensor node always includes these free elements (see Figure 1).

Figure 1. The Anatomy of a Wireless Sensor Node

Source: Electronic Design

Consider commercial office environments and consider what a conference room thermostat needs to do: As important as it may seem for the comfort and productivity of the conference room’s occupants, an environmental monitoring device — for temperature, humidity (air quality) and lighting — the thermostat/controller may need to take only a small number of readings over the course of an hour, and make an even smaller number of adjustments. The reason is simple: Even unregulated, room temperatures will not change very rapidly. (It will take the sensors several minutes to register a change.)

A ½-horsepower electric motor driving a conveyor belt in an automated factory, on the other hand, will need closer monitoring to produce the proper torque. The load on the machinery could prove taxing, forcing the motor to generate more torque, consuming a lot of current in the process; its coils could groan and smoke.

Near Field Communications devices (NFC) can be used as sensors on an IoT node. In addition to processing transactions at point-of-sale (PoS) terminal, NFC devices can be used as the front ends of IoT data-gathering nodes. Simply swiping an NFC card in the proximity of a card reader will release the turnstile at bus or train terminal, unlock a hotel room door, or start the lights in your kitchen when you return home from work. From the point of view of the consumer, swiping low-energy card is simple and easy. But, used a sensor front end to an IoT data gathering operation, NFC devices can track hotel occupancy rates, the use of public transportation, and residential energy consumption,

But in each example, the task of the sensors, microcontrollers and communications circuitry is very much the same: The sensor will take a reading on the temperature produced by the HVAC system (heating, ventilation, air conditioning), or the conveyor belt speed produced by the motor drivers attached to (say) an industrial programmable logic controller (PLC). The signal conditioning circuitry will amplify (and, often, linearize) the output of the sensors so that it can be read by the microcontroller. (The sensor signal conditioning circuiting may include an analog-to-digital converter, though increasingly it is on-chip with the microcontroller.)

The room environmental monitor or motor control may adjust in response. The communication portion of the IoT sensor node will likely leave a record of the transaction and/or use its connection to the cloud computing resource to further analyze the data generated. The microcontroller may confirm that the value is correct, or suggest a corrective.

Microcontrollers architected around ARM processor cores, or the still-lively 8051, are among the most popular. Software development kits and firmware samples for a wide range of these controllers are readily available. As a general rule, however, processor performance is dictated by clock speeds, bit-width and local memory — and higher performance demands higher power consumption.

Regardless of what type of sensors are employed (ambient light, environmental temperature, motor speed, etc.), regardless of the processor bit-width of the microcontroller (8-, 16- or 32-bit), regardless of the communications loop (WiFi, ZigBee, or Blue Tooth), the sequence of collecting and evaluating data from a remote sensor loop remains mostly the same for anything you put in service to the IoT.

A word about timing, or the sequence of events: No matter what the sensor node is used for, it will likely spend much of its time (as much as 95% of it, in fact) asleep. A change in the sensor’s environment (movement, temperature, pressure, etc.) will bring the sensor to life. It, in turn, will wake the microcontroller, which asks, searching its memory for something familiar, “What have we got here?”

Most of the time the answer will be something to the effect of “nothing to worry about.” The processor will inform whatever is attached to its communications port… and go back to sleep. And that chain reaction will be much the same, regardless of whether you’re monitoring an industrial machine tool, slamming on your car’s ABS brakes, or checking on your home’s security with a smart phone.

To be sure, the proliferation of sensor nodes will be dependent on a wide array of components and functions supporting the entire chain, from sensor to data stream. These additional components (Analog and Connectivity) include:

• Sensor Signal Conditioning
• Power Management
• Near Field Communications (NFC)

Sensor Signal Conditioning

In a large majority of IoT applications, there may be no separate signal conditioning semiconductors. The signal conditioning circuits are either integrated with the sensor module package, or with the microcontrollers. Whether visible or not as a separate component, the signal conditioning is necessary to ensure the sensor can be understood by the node’s microcontroller.

In motion detectors, one example, a MEMS device will require at least two levels of signal conditioning: one to bias the micro-electromechanical system, and amplify its output; another to simplify its output (or make the sensor output easier for the microcontroller easier to interpret).

MEMS suppliers frequently will combine a signal conditioning ASIC — a “Sensor Fusion” DSP — in the same semiconductor package with the sensor element. The sensor fusion element streamlines the output of the MEMS sensor so that the waveforms generated by the sensor are easier to interpret.

In modern smart phones, the in-package sensor fusion element will supplement other motion detectors, to serve as footstep counters and/or provide other location-based services (like site-based ad servers) based on the on-going knowledge of where the user seems to be going and where he has been. Passing a Starbucks? Your cell phone may invite you in for a coffee. The on-going technology challenge for sensor makers is to reduce the size, cost and power consumption of the sensor package (see Figure 2).

Figure 2. System-in-Package technology still dominates among MEMS Sensors. Sensor fusion makes motion sensor data easier to interpret.

Source: NXP

The power consumption of the sensor signal conditioning package is critical in remote IoT applications, where it is difficult for technicians to string wires or change batteries. In these applications, energy harvesting devices are valuable additions to the sensor nodes in their ability to capture signals as low as 40 nA (40 billionths of an amp).

Power Management Devices

These devices tend to be divided by function. One popular group includes custom-crafted voltage controllers (PMICs), which supply a feature-rich mobile device with a variety od voltage rails. These devices have traditionally supported mobile handsets. The other kind of power management device is a much simpler voltage regulator, which supplies a small number power rails (one or two) for each of the circuits in the remote sensor nodes.

What the power management devices have in common is their need to supply regulated (or tightly fixed) voltage rails to all elements of the IoT sensor node. To conserve battery life, the voltage regulators must consume little-to-no-power of their own.

This leaves the designer a choice between linear and switch mode regulators. The linear regulators are very inexpensive and easy to implement. The switch-mode regulators are more expensive, sometimes difficult to use in a circuit, but offer the highest energy transfer efficiency. A wide variety of battery-backed power rails in remote sensor nodes could be implemented with switch-mode regulators with 50- or 100mA regulators.

The Impact for Near Field Communications (NFC)

Though they will generate and process reams of analyzable data, NFC devices are not usually considered IoT devices. (Neither are mobile handsets.) They are not generally intended as data gathering nodes. At best, NFC devices may be considered special-purpose IoT processing nodes.

Intended as credit card replacements, the NFC devices embedded in cell phones are increasingly available as dedicated transportation cards, NFC devices are intended to quickly process financial transactions. You swipe the NFC device at a point-of-sale terminal, and the appropriate fee is automatically subtracted from your bank account or added to your monthly cell phone bill.

It may be worth understanding how NFC transactions actually work — perhaps purchasing a development kit — to explore their utility for IoT data gathering. The NFC transceiver operates as a short-distance RFID tag, enabling users to perform contactless data exchanges, to access digital content, and connect electronic devices simply by bringing RFID transmitters and receivers in close proximity (e.g., 4 or 5 cm) to each other. The connection is an inductive coupling (magnetic coupling between two flat coils: one in the transmitter, another in the receiver).

The NFC reader will automatically detect the presence of a transmitter operating at 13.56 MHz. Electrically stimulated by the NFC reader, the card will respond with a series of RF data bursts: One to identify the card and its accounts; additional bursts to complete the transaction, and return to a passive (unreadable) sleep state. Current NFC standards determine not only the contactless operating requirements, but also the data formats and transfer rates.

While currently used for financial transactions, including ATMs, turnstiles, vending machines and point-of-sale terminals, they can also support raw data exchanges between mobile phones, PDAs, and personal computers. Household applications for NFC can include home door locks, lighting controls and thermostats. Swipe your card on the door post of your dark house, and it could turn on your lights.

Recently introduced versions of the NFC transceiver modules offer more transmit power than previous versions for increased data integrity and user’s ease of use. An ARM Cortex-M0 enables a fast wake up from sleep. Contactless NFC chips include coil interfaces and I2C metal connections — and sell for as little as $0.30 per chip.

Component Choices by Application (Key Specifications)

IoT components will inevitably bridge the gap from smart cities to smart home monitoring and control. Invariably there are key specifications affecting your choice of components for IoT sensor nodes. In addition to the three main building blocks — microcontrollers, sensors and communications ports — you’ll need to specify signal-conditioning components, and power management devices. A summary of relevant devices and key specifications are listed below.

• How much intelligence in the sensor node? How much intelligence in the cloud? The answer will dictate whether you use an 8-, 16- or 32-bit processor, how much memory is attached, and what clock rate you’ll use to run your IoT node.
   
• Where is the data concentrated and how is it communicated from the remote sensor node? (ZigBee, BTLE, Ethernet)
   
• How much resolution and what kind of sampling rate is required for sensor signal conditioning? Data Converters for jet engines or truck weight stations may require 16- or 18-bit resolution; machine tools could make good use of 12 bits.

Are the power sources batteries, offline power supplies, or energy harvesters? A key concern is energy transfer efficiency; 95% or more. You don’t want your voltage regulator tapping your supply voltage.