The wearable computing market looks poised for take-off. According to International Data Corporation (IDC), shipments of wearable computers, including smart watches (see Fig. 1), sports and activity trackers, healthcare monitors and smart glasses, will triple in volume in 2014 over 2013 to 19.2 million. Led by activity and fitness monitors, IDC predicts that this emerging market will jump to a robust 111.9 million units by 2018.
Fig, 1: Distance runners use smart watches to monitor performance.
What is remarkable is that this growth is increasing despite the fact that many wearable computing solutions on the market today rely on less-than-optimal silicon solutions. Designers of lower end wearable computing systems often rely on multiple discrete ICs to accomplish their goal. Inevitably these jerry-rigged solutions drive up cost, power, and increase device footprint.
Alternately, developers of higher-end wearable devices build their solutions around processors originally targeting smartphone or tablet applications. More often than not, these products are not optimally matched from a performance or power standpoint to relatively targeted and highly power-limited wearable applications.
The Need For Efficient Silicon
To maximize battery life, ensure end user satisfaction, and drive the wearable computing market to the next level, designers need silicon devices that offer a highly optimized tradeoff between performance and low-power usage. To address this need, a number of silicon vendors have recently begun to offer system-on-chip (SoC) or ASSP solutions optimized for the wearable market. These devices support the need for constant data collection and data analysis with dedicated, always-on sensor hubs and, in some cases, embedded coprocessors that support mathematical functions. But these highly-customized solutions are generally more expensive than off-the-shelf MCUs, a key consideration in the cost-sensitive consumer markets that wearable computing products compete.
Ideally, system designers building wearable computing and other Internet of Things (IoT) designs need devices that can deliver the robust processing capabilities required to support efficient data acquisition while consuming minimal power. To lighten compute requirements on the host processor and reduce system power requirements, these new processors must efficiently control and collect data from multiple sensors by offering multiple digital interfaces on-chip. And to keep systems affordable yet retain design flexibility, they must offer a high volume, off-the-shelf price structure, but feature enough on-chip memory to meet multiple application requirements.
Some of the first devices to target this gap in the silicon supply chain were announced earlier this year by LAPIS Semiconductor Co Ltd. To help manage multiple sensors, collect data and analyze the results, the company developed two new MCUs based on its original high-performance eight-bit RISC CPU core and a 16-bit coprocessor for arithmetic operations as shown in Fig. 2.
Fig. 2: MCU manages multiple sensors, collects data and analyzes the results.
Operating at up to 4 MHz, the MCUs add 64 KB of flash memory for program storage, 4 KB of RAM for program use and 8 KB of dual port RAM for data logging. The on-chip 16-bit coprocessor can support advanced data analysis by performing multiplication, division, and square root functions.
The ability to talk to a broad array of sensors is essential in these applications. There are two types of analog-to-digital converters (ADCs) built-in to these MCUs. The RC oscillation type ADC is designed for high accuracy temperature and humidity types of measurements while the three channel 12-bit successive approximation ADCs (SA-ADCs) are used to measure the output voltage of any analog sensor. For interfacing with digital sensors and wired/wireless communication devices, the MCUs come with SPI master, I²C master, UART, and synchronous I/O ports. They also feature general purpose I/O and PC slave interfaces for connecting to a host processor.
Finally, a low profile and compact footprint are essential in wearable computing applications. The two new MCUs come in a thin, 48-pin wafer level, chip-scale-package (WL-CSP) or 48-pin quad flat pack (QFP), respectively.
Advances In Miniaturization
Of course smaller footprints across the entire component spectrum are crucial to success in this emerging market. While wearable-computing-product users demand smaller, thinner components, the design challenges to further miniaturization in some component categories are especially imposing. Take transistor manufacturers as an example. Miniaturization is extremely difficult when manufacturers decrease package dimensions in a transistor because smaller dimensions place severe limits on element size. As transistor size shrinks, on resistance increases making it highly difficult to maintain the performance of small-signal transistors.
In response some manufacturers have attempted to solve this problem by experimenting with new approaches to high precision packages and transistor processes along with optimization of the transistor's internal structure as displayed in Fig. 3.
Fig. 3: New approaches to high precision packages and transistor processes allow miniaturization without increasing on resistance.
Engineers at ROHM, for one example, have found a way to reduce package size while lowering on-resistance. As a result the company's 0.6 mm x 0.4 mm VML0604 transistor package trims board space by 50% over its predecessor of less than two years ago. Simultaneously, the new device offers an 82% reduction in on resistance from 1.4Ω to 0.25Ω (VGS = 4.5 V) in the 30V class.
These transistors are part ROHM Advanced Smart Micro Device (RASMID) family of ultra-compact components that includes transistors, diodes, capacitors, and LEDs. Leveraging many of the same advances in process technology and miniaturization, the company introduced the 03015-size chip resistors in 2011 that measured 0.3 mm x 0.15 mm and in 2012 added 0402-size Zener diodes. Earlier this year it added 0402-size Schottky barrier diodes.
Lowering power consumption to extend battery life is naturally a high priority across the growing wearable computing market. One way designers are making gains in this area is by migrating to Bluetooth low energy (BLE) technology. This new variation of the popular Bluetooth communications standard supports the same communications range of the original standard while consuming significantly less power. BLE is also increasingly attractive because it is natively supported by most smartphones. Given these advantages, analysts expect BLE will be widely adopted by portable device developers in the health and fitness, security, and consumer markets.
The benefits of BLE can be seen in some of the early devices coming to market. LAPIS Semiconductor has developed a BLE IC that consumes 9.8 mA during transmission, 8.9 mA during reception, and an average of 8 µA during 2s, intermittent operation as described in Fig. 4.
Fig. 4: New Bluetooth Low Energy technology (BLE) uses less current.
That translates into an approximately 50% reduction in current when transmitting or receiving packets. Other leading semiconductor developers, i.e., ARM, are also using BLE technology in their development platforms to simplify the creation of low-power wearable Bluetooth designs.
Many of these advances in miniaturization and low power operation promise to pay similar dividends in the development of a wide variety of sensors. For example, new photoelectric pulse wave sensors (Fig. 5) for the medical and health care industry will use miniature LEDs and photo detectors to measure changes in the volume of blood in blood vessels.
Fig. 5: Wearable pulse wave sensors measure the changes in the volume of blood in blood vessels.
These wearable sensors will be used to measure pulse rate to give a more accurate measurement of caloric consumption. They will also be used to provide valuable feedback on heart rate to maximize exercise efficiency.
Similar technologies are allowing developers to build ultra-thin accelerometers measuring 0.7-mm tall and consuming as little as 2 µA of current. At the same time developers are bringing to market optical sensors that track UV exposure, pulse rate, and blood oximetry in a 2-mm² package that combines UV-sensitive photodiodes with a MCU and ADC. Look for similar advances in the development of proximity and ambient light sensors, magnetometers and other devices.
While wearable computing sits at the trendy edge of the electronics industry today, expect dramatic changes in the years to come. Market analysts say devices once considered eccentric, like smart glasses, will move to the enterprise and play a key role in boosting the efficiency of technicians, engineers, and health care and maintenance personnel. Crucial to that evolution will be the development of a new breed of ultra-compact semiconductor devices that bridge the current gap between high-performance and low-power operation and transform wearable computing into the high volume electronics market of the future.
Wearable Heart Monitor System Teams Up With Adidas' miCoach App
About the Author
Satya Dixit is the Senior Director, Systems and Applications Engineering at ROHM Semiconductor. His team defines and develops new products, customer solutions, and system architectures. They also perform IC validation and help introduce new products. Satya addresses the needs of multiple markets, including mobile, consumer, PC, automotive, and industrial segments.