New Design Flows
A new breed of intelligent sensors has emerged aimed squarely at IoT edge subsystems. In this article, Greg explores what defines a sensor as intelligent and steps through the unique design flow issues that surround these kinds of devices.
The evolution of the basic sensors to intelligent electronic sensors is creating a revolution in how we gather useful data from the world around us, analyze that data to make decisions and connect together vast intelligence systems to enable new solutions and to accomplish tasks that we have never been able to perform before. This opens up the market to new ideas for existing companies and start-ups. Suddenly, sensor-based product development is exciting again and we see design teams like yours tackling new challenges from IoT to Industrial IoT in order to deliver new solut
Basic sensor systems typically consist of a sensing device with some signal conditioning that outputs an analog signal. But the technology is available today to create intelligent sensors and they are the key for to making innovative IoT edge systems.
Most would agree that the intelligent sensor must contain these key elements (Figure 1):
• A sensing device that measures physical parameters from the real world as an analog input to the system.
• Signal conditioning circuitry that transforms the sensing device signal into a signal that the system can use. This circuitry can include tuning or amplification.
• The analog-to-digital circuitry converts the analog signal to digital so that the processor block software can interpret the signal, run routines and perform sensor calibration.
• A computational block, such as a processor or DSP, which analyzes the sensing device measurements.
• A communication block, such as a wireless transmitter, that exchanges information with the larger intelligent system.
SENSOR MARKET IS GROWING
Designers can create electronic sensors using a variety of technologies including silicon photonics, CMOS, fluidic chip and LEDs. But, MEMS sensors are the most interesting technology to explore due to their wide footprint in the intelligent sensor market. According to the Yole Développement’s “Status of the MEMS Industry 2017” report, the MEMS sensor market in 2016 exceeded $11 billion in sales (Figure 2).
In 2003, Knowles created the first MEMS microphone and due to its form-factor and resistance to heat that allows surface mounting, the product found its way into smartphones. Then in 2005, smartphones with accelerometers came into the market and accelerated the current MEMs sensor growth. In 2016, there was a big increase in the use of RF MEMS filters due to the complexities of 4G/5G communication, according to the Yole Développment report. But, the average selling price (ASP) of sensors has fallen to under $1 USD.
If we examine overall sensor sales, the consumer market—which includes smartphones, drones, smart home devices and wearables—is by far the biggest market for sensors. The automotive market comes in second with driver assistance, safety and self-driving technology that are laden with sensors. In third place, the industrial market is driving sensor purchases for the Industrial IoT.
Given that the ASP of MEMS sensors is approximately 60 US cents a unit, how will vendors make money in the IoT marketplace? If they are a market leader and they ship billions of sensors, then volume is a revenue factor. Or mergers and acquisitions can broaden the market. However, there are other approaches that vendors are taking to increase revenue. One such approach is sensor fusion.
Sensor fusion means developing a product that combines multiple sensors, a processor and intelligent software to create a high-value system that is more accurate than using the individual sensors. A good example is the InvenSense (TDK) ICM-20728, the world’s first integrated 7-axis MotionTracking device (Figure 3). This device contains a 3-axis gyroscope, 3-axis accelerometer and a pressure sensor in a single-chip platform solution with an onboard digital motion processor and firmware algorithms.
Software opens the gate to new revenue paths. For example, vendors can offer a portfolio of application-specific products at different price points where the hardware remains the same, but the intelligent sensor functionality is solely controlled by the software. Because the sensor is connected to other sensors and the Internet, the vendor can move into services. These services can include data fusion to optimize systems or to calibrate sensor systems remotely, providing data analysis or performing remote diagnostics and maintenance.
INTEGRATED DESIGN FLOW
From hobbyists to large companies, designers are taking their new IoT ideas to market by taking advantage of intelligent sensors. A new breed of designers has arrived and they are making hardware design trendy again.
This new breed of designers are reshaping design flows and they have new expectations. They typically work in small teams and require integrated design flows to quickly and easily produce a functioning device while spending as little money as possible. They require the capability to develop a proof-of-concept for system validation in order to capitalize on the opportunity of the IoT market. Design teams need to rapidly implement products using integrated design flows that allow them to quickly develop all the pieces needed for the sensor-driven IoT edge device, including: sensing elements, analog circuit interfaces, analog-to-digital logic, digital logic and RF—all at a low cost compared to traditional IC and systems design.
Many design teams employ the integrated IC design and verification solution from Tanner to create intelligent sensor-based IoT systems, including Knowles. Why? Creating a sensor-based IoT edge device (Figure 4) is challenging, due to the multiple design domains involved. But, creating an edge device that combines the electronics using the traditional CMOS IC flow and MEMS sensors on the same silicon die can seem impossible. In fact, many IoT edge devices combine multiple dies in a single package, separating electronics from the MEMS design.
The design flow (Figure 5) accommodates single or multiple die techniques for IoT edge device design and verification of intelligent sensor-based design.
IoT edge design requires that analog, digital, RF and MEMS design domains are designed and work together, especially if they are going on the same die. Even if the components are targeting separate dies that will be bonded together, they still need to work together during the layout and verification process. The design team needs to capture a mixed analog and digital, RF and MEMS design, layout the chip and perform both component and top-level simulation.
Designing the electronics and MEMS on a single die include these interesting points (Figure 5):
• Schematics can contain IC and MEMS devices. IC devices are modeled using SPICE models and MEMS devices employ behavioral models that are directly modeled in the physical domains such as mechanical, electrostatic, fluidic and magnetic. MEMS capture is supported by the MEMS Symbol Library within the schematic capture domain.
• In order to support the initial MEMS/IC simulation, System Model Builder create a MEMS model using analytical equations in SPICE or Verilog-A. Combined with the MEMS Simulation Library, this allows designers to verify that the complete design initially works as expected.
• Using the MEMS PCell (parameterized cell) library, designers can lay out the design in layout environment. The library provides basic layout generators for many MEMS devices that can be used as a starting point. The library contains active, passive, test, thermal, optical, fluidic and resonator elements.
• Designers can then generate a 3D geometrical model for viewing, virtual prototyping and to export to finite element analysis (FEA) tools.
• Using the Compact Model Builder, which employs reduced-order modelling techniques, designers can create behavioral models from the FEA results for use in final system-level simulation.
Traditionally, the MEMS portion of the design starts by creating a 3D model of a MEMS device and then analyzing the physical characteristics in a third-party finite element analysis (FEA) tool until satisfied with the results. But, the team needs a 2D mask in order to fabricate the MEMS device. How do they derive the 2D mask from the 3D model? They follow the mask-forward flow that Figure 6 shows, that results in a successfully-fabricated MEMS device.
STEP BY STEP
Start with 2D mask layout to create the device. The 3D Solid Modeler then takes the layout and a set of 3D fabrication process steps to automatically generate a 3D solid model of the device. Export that 3D model and perform 3D analysis using a favorite finite element tool and then iterate if the team finds any issues. Make the appropriate changes to the 2D mask layout and then repeat the flow. Using this mask-forward design flow, teams can converge on a working fabricated MEMS device because they are directly creating masks that will eventually be used for fabrication, rather than trying to work backwards from the 3D model.
The IC team has to be proficient in digital, analog, RF and MEMS design. Within the IC team, these domains are usually independently led by designers and consist of separate teams with domain expertise. For example, MEMS design is typically a specialty field that requires unique design skills that differ from digital design. A separate software team develops the code that runs on the processor of the IC and the PCB team creates the unique board for the IC and associated discrete elements. All of these teams exchange required information and data and the whole system needs to come together and be verified. Mentor offers a unique solution to this problem (Figure 7).
Intelligent sensors at the IoT edge provide almost unlimited prospects for companies to bring new solutions to market, with the help of an integrated design and verification tool flow.
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • NOVEMBER 2018 #340 – Get a PDF of the issueSponsor this Article
Greg is the General Manager of the ICDS division, and joined Mentor, A Siemens Business, in March 2015 as part of the Tanner acquisition. Prior to this, he was the President of Tanner EDA. Greg brings 25 years of executive and technical management experience, along with a proven track record of building strong teams and delivering predictable results. Prior to Tanner EDA, he held management and technical positions in a number of different industries and companies including Sprint, General Electric and McKinsey & Co. Greg holds a Bachelor’s Degree in Business Administration from Northern Arizona University.