More Than One Sense
In our vehicles, on our wrists and smartphones and in the Internet-connected devices encountered in industrial and consumer settings, sensors are finding ways to work with greater energy efficiency, improve our health, and adapt to highly individual differences.
While for every “round table” type feature EECatalog publishes I find myself wishing that I could be in the same room with all of the expert panelists at once, it’s especially true this time. Although I think if you walked into that room you might be initially puzzled to hear talk of everything from hearing aids to coaching to how one person glances at a watch in a different way than another—plus meta data, motor parts, the field bus wars and more. Once you realized the topic was sensors and MEMS, however the wide-ranging nature of the conversation would make sense.
Our thanks to Dr. Wolfgang Schmitt, Marketing Strategist, Bosch Sensortec GmbH; Per Slycke, Vice President, Motion Tracking, Fairchild Semiconductor; Ian Chen, Sensor Solution Division, Freescale; Mahesh Chowdhury, Director of Strategic Platforms and IoT, STMicroelectronics; Rich Collins, Segment Marketing Manager, ARC Processors and Subsystems, Synopsys; and Ron Lowman, Strategic Marketing Manager for IoT, Synopsys.
EECatalog: How will connectivity standards affect the ability to fully capitalize on the IoT? With regard to connectivity standards, what’s your advice for getting off the dime with regard to issues where the industry has not been able to resolve differences?
Ian Chen, Freescale: One way of getting off the dime is to reject any connectivity dogma. For example, Freescale’s Intelligent Sensing Framework allows our sensors to work across the entire offering of our MCU portfolio while abstracting the method of connectivity to the embedded programmer. With this approach, a designer could make one IoT design work across various connectivity approaches. Thread, the wireless networking protocol aimed at making the connected home happen, is one such standard that Freescale supports.
Mahesh Chowdhury, STMicroelectronics: Bluetooth Smart is one of the connectivity standards that we believe to be suitable for cost-efficient low-power wireless connectivity solutions for IoT applications enabled by smart MEMS sensors, computing and connectivity. The steady increase in the adoption of Bluetooth Smart devices for a large variety of such IoT applications indicates that things are moving along smartly.
Per Slycke, Fairchild Semiconductor: Interoperability is ultimately in the interest of everyone involved. Once the market takes off this will become less of a pre-occupation and everyone will just use what the market has decided should become the de facto standard, which is not necessarily the most technologically advanced or “best” solution. I think the fast adoption of Bluetooth Smart (BLE) is a great example of this in action right now.
Ron Lowman, Synopsys: Wearables have seen the adoption of Bluetooth Smart as a de facto standard for a couple of reasons. Companies such as EM Microelectronic have benefited from that with low-power Bluetooth capabilities. Implementation cost, including die size, stack size and power budget, is significantly better in Bluetooth Smart than WiFi. And it’s pervasive, as it’s already on all our personal devices (mobile phones and tablets). Ethernet and WiFi protocols weren’t initially designed for “things,” and the protocols defining “the field bus wars,” such as Modbus, were not designed to be streamed to websites. However, myriad standards organizations are tackling this problem very proactively, and their efforts will provide an open source platform and open source abstraction layer that will enable developers and designers to focus on their key value generation. Interoperability will be a reality coming not from a single solution, but [rather] from a small array of solutions to fit different IoT segment needs.
EECatalog: How are “cognizant of energy use” practices affecting your company’s offerings? What is your advice for achieving differentiation in your customers’ eyes as well as for your customers’ customers through how sensors and related offerings handle power concerns?
Dr. Wolfgang Schmitt, Bosch Sensortec: Our sensors target for the most part mobile devices, and throughout our portfolio we have more than once been the first in the CE MEMS industry that managed to set the benchmark for lowest power consumption. Most recently we launched the industry’s 1st CE IMU (i.e., a combined accelerometer and gyroscope) to come below the critical 1mA level of current consumption, something that allows customers to experience longer battery lifetimes on their smart phones, smart watches and other mobile devices.
Our BHI160 (IMU) and BHA250 (accelerometer) target smartphones and incorporate a so-called fuser core (i.e., a dedicated processor), where the sensor data fusion software is already being processed at sensor level. The fuser core is power-optimized for this specific function. What’s more, the sensor data fusion does not need to be processed by the relatively power-hungry application processor of the smartphone.
Rich Collins, Synopsys: [Ways we drive innovation with] energy efficiency range from logic libraries that enable operation at voltages at 60% of Vddnom, and Always-On functionality to reduce leakage currents during sleep states, to processing cores and subsystems that can reduce the number of cycles to complete functions by more than 25 times and power reduction of over 10x for applications such as sensor fusion.
Solutions that have enabled hearing aids with efficient energy use can now be leveraged to extend battery life for applications requiring complex processing capabilities, such as voice recognition and vision systems. In automotive, one trend is to move to more advanced vector oriented motor control to increase efficiency and also reduce the size of the motor components. High dynamic loads can be controlled at minimum power usage, leveraging the closely coupled architecture of Synopsys’ DesignWare Sensor & Control Subsystem as the peripherals such as PWM control and ADC inputs do not have delays of communicating over the standard buses found in traditional designs.
Chowdhury, STMicroelectronics: Features such as “Always-On” context awareness can be enabled when the MEMS sensor produces consistent measurements with low noise and high stability while operating at very low current consumption. Our LSM6DS3, a 6-axis inertial (gyroscope + accelerometer) sensor, consumes less than 500 microamps of current. This sensor has many embedded functions such as step counting, tilt detection and significant motion detection, which offloads this functionality from the Application Processor (AP) for power efficiency.
Slycke, Fairchild Semiconductor: While being smart about how to use energy is often a necessity, when you talk about sensors it is seldom the ultimate goal. With Fairchild’s first 6-axis MEMS IMU, the FIS1100, our goal is to enable the lowest system level energy consumption, while not making any compromises in the 3D motion tracking accuracy or functionality enabled by the sensors. It is much like cars; the best electric vehicle is not only “green,” it also is fun to drive!
Chen, Freescale: We have always been cognizant of energy use for all our sensor designs, working with customers to help optimize system power consumption. By selecting the right sensor, significant power savings can be achieved. For example, both a magnetometer and a gyro could be used to monitor slow rotations, but the power consumption of a gyro could be 10 times higher than a magnetometer. Additionally, the overall system power consumption needs to be optimized in addition to the sensor power. This may involve adjusting sampling rate to optimize power cycling of the MCU, putting the sensor into lower power standby states, or even power gating the sensor.
EECatalog: What challenges do you believe are not getting their fair share of attention?
Slycke, Fairchild Semiconductor: Today there are too few “must-have” consumer applications enabled by sensors. If you look at the app ecosystem on the major platforms, most revolve around communication and, to some degree, location. There is a lot of innovation around applications using sensors, but none are must-have applications. This lack of killer applications has left the majority of the industry in MEMS motion sensors focusing on ever smaller, lower power and lower cost solutions since it is not clear which specs would enable the “killer app.” Having to deal with varying degrees of 3D motion tracking behavior and accuracy on a variety of platforms is a huge challenge for the app developer community. In fact, it is holding back innovation. I believe that, ultimately, to enable must-have applications for MEMS motion sensors, accuracy will be key. Also, it may very well be the case that the “must-have” application will not come in the shape of a smartphone.
Chen, Freescale: Challenges around smartphone applications have captured the most attention because of their disproportionately large unit volume. As embedded IoT applications are increasingly driving sensor applications, designers are now paying more attention to performance parameters, such as temperature stability, bandwidth and linearity. Further attention is needed to understand the application of system data analytics to enable new applications and the longer-term sensor operating reliability for industrial installations.
Lowman, Synopsys: Differentiation in the space is being overlooked. With all these standards organizations there is a clear path to interoperability, but many of the solutions to support this interoperability are copycats for fear of being left behind. The solutions that will drive innovation and solve problems beyond interoperability and connectivity will require leadership. Leveraging IoT specific IP from Synopsys enables our customers and customers’ customers to spend more time and money on developing their differentiation while having the confidence that they have a partner that is leading the way.
Chowdhury, STMicroelectronics: Sensors manufacturers are now adding computational capability to sensors to generate meta-data—for example, accelerometers that produce step count as opposed to raw acceleration data. Meta data is computed on the sensor and made available through the HW abstraction and application framework. The availability of this meta-data in consistent fashion to application developers is a major challenge, which sensor manufacturers, reference platform makers, and the OS makers providing the application framework, will need to address.
EECatalog: What does the industry know now that it didn’t earlier about what it takes for a wearable to be successful… or what should it know?
Lowman, Synopsys: Our mobile phone is a very personal device. Wearables will prove to be an even more personal device. It’ll take personal experiences that are diverse to see mass adoption of wearable technology, but its potential is incredible. This personal touch requires that these wearables do not just send us info and report, but offer to enhance our lives through improved health, improved social experiences, and improved financial interactions—in ways we haven’t even thought of.
Schmitt, Bosch Sensortec GmbH: As compared to smartphones, wearables have a different business model, and that’s had a previously underestimated impact on growth. Also underestimated: the requirements and the level of sophistication that customers expect from a wearable device. End customers expect a certain level of precision, added value and good design at the same time.
Chen, Freescale: As the number of wearable devices increases, the wearable market is also segmenting itself. Wearables are no longer mere consumer activity trackers but are becoming industrial tools to help prevent repetitive motion injuries, expert coaching tools for semi-professional athletes to enhance their performance, and outpatient monitoring devices to improve patient care. The specialization of wearable devices to serve these segmented markets with different mission profiles and price-performance expectations will make the wearable market more robust.
Chowdhury, STMicroelectronics: Personalized algorithms are necessary for a good user experience for wearable devices. An example that illustrates the need for personalized algorithms is the “screen wake-up upon glance” function in smart watches. The wrist motion generated when glancing at the face of a wristwatch varies from person to person. Hence, personalized algorithms that detect this variability, and adjust the screen wake-up function accordingly, would result in enhanced user experience of the smart watch.
Slycke, Fairchild Semiconductor: That it takes more compelling functionality than step and calorie counting for a wearable to be “sticky.”
EECatalog: What principles and practices does your company follow so that its offerings are more than just the sum of their parts?
Lowman, Synopsys: With respect to the IoT we evaluated our current portfolio and made investments to optimize each of the key components required within the space. We know that leveraging existing solutions and simply repositioning them into a new market wouldn’t offer what customers require, so we developed new products to enable our customers to be leaders in the space. We ensured that all the technology is mutually complementary, for example, aligning the process technologies, number of metal layers required, minimum voltage levels and more.
Chowdhury, STMicroelectronics: First, we manufacture sensors with low sensor noise, high stability, and using very low current consumption. The second part of the equation is that we develop and deploy sensor fusion algorithms that intelligently combine data from different sensors to enable key applications such as context awareness.
Slycke, Fairchild Semiconductor: We always take the system perspective. If you understand the application and the system, the optimal components look different. We invest a lot in embedded software development because it is a huge part of the system in most applications. That leads me to ask something I wish we had been asked, which is, as 6-axis and 9-axis IMUs become increasingly prevalent in consumer applications, how do embedded designers balance the need to maximize sensor-tracking performance without placing a heavy computational burden on the host processor?
One answer is that one of the more difficult compromises embedded designers face is the challenge of delivering highly accurate motion sensing and tracking without burying the host processor in computational requirements. We believe the answer lies in an innovative approach to architectural partitioning and emerging silicon/software solutions that leverage the latest advances in sensor fusion algorithm development and unique processing partitioning to minimize the burden on the host processor. Extensive experience in motion tracking applications will likely play a key role in the rapid development of these solutions.
Chen, Freescale: Freescale focuses on providing secure embedded solutions for the Internet of Tomorrow, meaning we pay attention to the overall system as the data moves from the sensors to the cloud and offer solutions that can address the end-to-end challenges. As the secure IoT continues to grow, sensor data analytics will be a key component in driving new applications and helping companies monetize those applications by providing new benefits to consumers and companies.
Because a significant portion of tomorrow’s IoT applications originates in small or new ventures today, we are also focused on making our customers more productive through software enablement and ecosystem partners.
Schmitt, Bosch Sensortec: We are following a holistic approach. That means that we don’t understand ourselves as a pure sensor hardware supplier only. We make sure that our customers will get a solution that fits best into their product, considering the whole system, including hardware and software. We also develop sensor data fusion software, for example 9-axis sensor data fusion software. This is important because we as a sensor HW manufacturer do have the best knowledge about our sensors’ hardware and how to deal with it.
Anne Fisher is managing editor of EECatalog.com. Her experience has included opportunities to cover a wide range of embedded solutions in the PICMG ecosystem as well as other technologies. Anne enjoys bringing embedded designers and developers solutions to technology challenges as described by their peers as well as insight and analysis from industry leaders. She can be reached at email@example.com