Get email delivery of the Cadence blog featured here
Micro-electro-mechanical systems (MEMS) sensors will play crucial roles in many Internet of Things (IoT) devices. In a newly archived webinar, experts from ARM, Cadence, and Coventor show how to design a MEMS vibration sensor (right) using Cadence® and Coventor® tools, and then implement it in a system-in-package powered by an ARM® Cortex®-M series embedded processor.
The webinar is actually part two of a webinar series that began in July 2014. Part one presented an industrial IoT example in which a factory is moving from manual dials to wireless sensor nodes that can monitor and regulate pressure in hundreds of pipes. Part one shows how to create a single-die system on chip (SoC) with temperature and pressure sensors and an ARM Cortex-M0 processor.
Part two, described in this blog post, shows how to create and add a MEMS accelerometer to monitor vibrations in pipes. This vibration sensor works with the SoC described in part one, but is on a separate die, and is combined with the SoC into a system in package (SiP). The webinar reviews ARM, Cadence, and Coventor methodologies for designing multi-sensor solutions. Part two also considers the use of more powerful ARM Cortex-M series processors.
In brief, the webinar walks you through the processes of MEMS design and verification, MEMS plus SoC optimization and integration, and smart sensor fusion (in which data from several sensors is combined in useful ways). “All three companies bring complementary technologies for multi-sensor design, but it’s the integration of these technologies and tool chains that makes the flow ready to use, so you can get to tapeout and into the IoT market as early as possible,” said presenter Ian Dennison, solutions group director at Cadence.
Integrated Tool Flow
The new webinar uses a tool flow that is very similar to the flow used in part one, except for the addition of Coventor MEMS+, a tool for MEMS creation and analysis. Cadence tools include Virtuoso® Schematic Editor, Virtuoso Analog Design Environment (ADE), Virtuoso AMS Designer, Virtuoso Digital Implementation (VDI), Spectre APS, and SiP Layout. ARM provides Cortex-M RTL, tools for software and firmware development, and physical IP.
Chris Welham, worldwide applications engineering manager at Coventor, described the integration of MEMS+ with the Cadence Virtuoso platform (see diagram below). Users first design a MEMS device, using components such as plates and suspension beams from a library. They can then create an instance model in Virtuoso and a p-cell (parameterized cell). A schematic symbol can be brought directly into the Virtuoso Schematic Editor, and the device it represents can be simulated using Spectre APS. Users can view 2D results in Virtuoso or 3D results in MEMS+.
Coventor MEMS+ integration with Cadence Virtuoso tools
Wellham provided more details about the vibration sensor. It is built from an accelerometer and composed of a mass suspended by four suspension beams. One pair of comb capacitors senses the motion of the mass. Another pair of comb capacitors applies a feedback control force. This force is generated by a sigma-delta controller and is used to maintain the neutral position of the inertial mass.
To connect the MEMS sensor to the circuit, the user defines electrical terminals in MEMS+. This creates a pin on the symbol in Virtuoso. These pins are used to apply acceleration to the sensor. The MEMS+ model maps mechanical values, such as displacement and force, to voltage and current in Spectre.
Models are parameterized, and parameters can be changed in Virtuoso, so there’s no need to go back to the original model in order to run a “what if” experiment with different parameters. This allows an automatic optimization of the MEMS device and the analog circuit together.
Since the vibration sensor is on a separate die, Dennison reviewed the various ways that an SoC and a MEMS sensor can be placed in one package. One approach is to place the dies side by side on a silicon interposer. Another is to stack dies using bond wires or through-silicon vias (TSVs). Each approach has its own advantages, limitations and requirements.
Dennison showed how Cadence SiP tools allow the integration of MEMS sensors and SoCs. One useful tool is an SiP Wire Profile Editor. This helps users manage the many rules for die stacks and wires. Cadence also has a full suite of tools for package design. “It may be that the total pinout for MEMS and SoC will fit into a standardized package, but it is quite likely you’ll need to design your own package cavity and package substrate,” Dennison said.
Tim Menasveta, CPU product manager for Cortex-M0 and M0+ processors, talked about the need for a “sensor hub.” A factory sensor, he noted, must collect many types of data, but systems that continually sense data will drain battery life. A sensor hub processes data and makes decisions on how the processed data is communicated out of the sensors. Sensor fusion algorithms run on the sensor hub.
Menasveta showed an energy consumption graph that depicts the system energy spent on gathering sensor data. If smart sensors can decide for themselves what sensor data to transmit, and smart sensor fusion algorithms are deployed, the power savings can approach 90%.
Menasveta showed how an ARM Cortex-M7 processor could be used in a sensor hub. The M7 has a higher level of DSP and floating-point processing capability compared to other members of the Cortex-M series. Menasveta reviewed the strong points of the M0, M0+, M3, M4, and M7 processors for IoT applications.
The webinar is available online. A short registration is required.