MEMS Microphones: A Bright Spot Among Commoditized Consumer Sensors

MEMS microphones have emerged as a bright spot among consumer sensors, which in general are going through a rapid commoditization and profit-squeezing trend. To understand what’s driving the MEMS microphone market, consider that the Apple iPhone 7 and 7S each have 4 MEMS microphones. As reported by System Plus Consulting, the latest iPhones have “a front-facing top microphone, presumably for FaceTime and speakerphone capabilities, two front-facing bottom microphones located at the bottom-front of the device, used for voice commands and voice calls, and a rear-facing top microphone for video recording and noise cancellation” – and all these different use cases have different requirements. It is not surprising therefore that Apple has 3 MEMS microphone suppliers: Knowles, STMicroelectronics, and Goertek. This bright spot has gained industry-wide attention, as evidenced by a report in EE Times that MEMS microphones are one of the next platforms that TSMC will offer.

Source: InfineonTechnologies, AG, “The Infineon Silicon MEMS Microphone”, DOI:10.5162/sensor2013/A4.3

But let’s say you understand the microphone market well and have a target set of product specifications, which typically include frequency range, sensitivity, Signal-to-Noise Ratio (SNR), distortion, power consumption, package size, etc. How do you design the device multiphysics – complex interactions between the mechanical, electrical, and fluidic domains – to realize these specifications? Second, let’s say you have a good MEMS structure that transduces audio input to a variable capacitance, how do you collaborate with readout ASIC designers to ensure the device output – analog or voltage – meets the product specification? How do you capture the design in GDSII and run DRC, and what are process variation impacts? What are the impacts from packaging? How do you test? In other words, will your product be successful at a system integration level?

What’s needed is a full-finite element analysis tool (such as CoventorWare, which has been used by our MEMS customers for nearly two decades and has particular value for MEMS capacitive sensors including MEMS capacitive microphones). A sensing mode analysis for a MEMS microphone is shown in Figure 1.

Figure 1: (L) Full-FEA model of a microphone; because of symmetry, only a quarter is modeled for computational efficiency. (R) Simulated sensing mode, diaphragm deformation; this modal result determines the microphone’s frequency range and sensitivity.  For illustration purposes, the z-axis is scaled by 20 times.

MEMS+ is a compact-FEA tool that not only solves multiphysics systems at speeds roughly 100X faster than conventional FEA, but also produces models that can simulate in Cadence tools for ASIC design, and MathWorks tools for system-level design. Figures 2a and 2b show a MEMS+ microphone model embedded in a Cadence schematic, which is used to simulate the microphone’s sensitivity and noise density, and therefore SNR.

Figure 2a: MEMS+ microphone model in a Cadence schematic

Figure 2b:  Simulated noise density of the microphone using MEMS+ microphone model (analysis in Cadence Virtuoso)

These are introductory examples of how to apply microphone device multiphysics and to add microphone SNR predictions.