Microphone Design and Simulation

MEMS microphones are being rapidly adopted in consumer electronics. Smart devices now use two or more MEMS microphones to improve directional sensitivity and employ Active Noise Cancellation (ANC) for better sound quality. Microphone arrays are also being used in other consumer based products, with multidirectional functions to improve performance. To capture a significant piece of this fast growing market, leaders are adopting novel “hybrid” design methodologies that replace time-consuming build and test methodologies and significantly enhance strategies that rely solely on FEA based technologies for simulation.

mems microphones

Figure 1 MEMS microphone market forecast application, millions units. (Yole Development)

Design Challenges

Sensitivity, noise, linearity, dynamic range, reliability, and yield are all critical parameters for MEMS microphone design. Design engineers need to optimize the performance based on these criteria, while also reducing cost, form factor, and accelerating time to market.

Design strategies that rely solely on FEA based technologies and build and test are outdated, miss critical insights, and are far too slow. New methodologies that combine best-in-class behavioral and circuit level simulation with classic FEA approaches achieve far better results in a fraction of the time. Working with industry experts, Coventor has created a new hybrid design paradigm for microphones. An overview the design platform is presented in Figure 2.

mems design platform

Figure 2 Coventor’s Novel Hybrid-Simulation MEMS Microphone Design Platform

A New Hybrid-Design Methodology

Rapid Design Exploration with Advanced Behavioral Models
First, a microphone model is constructed in MEMS+ using elements from a parametric library. The 3D model is loaded into MATLAB, Simulink, or Cadence Spectre to simulate device performance vs specs: frequency response, sensitivity, maximize sensitivity while avoiding electrostatic pull-in. Note: While this example shows capacitive sensing, piezoelectric sensing can also be modeled.

mems microphone model

Figure 3. 3D Microphone Model in MEMS+

Flexible MITC finite-shell elements are employed to model the non-linear mechanics of the membrane. The elements have add-on electrodes that model the capacitance and electrostatic force acting between the membrane and fixed back-plate. The electrode model includes perforation holes and uses conformal mapping to accurately model fringing-fields in the capacitance and force computation. For clarity, perforations in the back plate are not shown.

Add Detailed Analysis with FEM/BEM Field Solvers
Next, fluid damping values are computed with Coventor’s FEA based tools – Designer and Analyzer, and these values are added to the MEMS+ model. To simulate noise and sensitivity analysis, the MEMS+ models are then run in Cadence Spectre.

Figure 4 Coventor’s FEA solver Analyzer is used to compute the damping values for the perforated membrane, which are transferred into the cavity model. The computed fluid damping coefficient is included in the model flow resistance term to enable noise analysis in Cadence Spectre.

MEMS + IC System Simulation
The MEMS designer can now also include the sensor electronics for a full MEMS + IC simulation. This same model can alternately be run using Simulink to investigate the performance of the microphone at the system level, for example (for ANC).

Figure 5 Cadence example schematic showing Microphone and output stage with Cadence Spectre noise and sensitivity response

Co-simulation of the sensor electronics and the sensing membrane can be achieved by transferring the MEMS+ model into Cadence Virtuoso. Figure 7 shows an example schematic comprising the MEMS+ model connected to a simple RC network, thus allowing the noise and sensitivity of the microphone to be simulated using a standard Cadence Spectre noise analysis. As the models are parametric the designer can easily adjust the design to explore the performance envelope. Yield analysis can be undertaken using Monte Carlo simulation or optimization algorithms employed to target a performance metric, for example minimize noise and maximize sensitivity.

Field solvers from CoventorWare Analyzer suite are also useful for investigating and verifying can also be used simplifying assumptions in MEMS+. For instance, designers can check that the perfectly clamped anchor points used in the MEMS+ simulation were indeed a reasonable simplification. This verification can be achieved by simulating and comparing the pull-in voltage in both MEMS+ and Analyzer.

As a further example, designers can investigate the effect on damping of the interaction between the membrane pressure relief (or “pierce”) holes and the perforation holes in the back plate. This is illustrated conceptually in Figure 6, which shows how the total damping changes with variation of relief hole radius.

Figure 6 Conceptual variation in damping value with increasing pierce hole radius

A Complete Platform for Microphone Design

Coventor’s products, MEMS+ and CoventorWare, provide a platform for MEMS microphone design, addressing critical design challenges that cannot be solved with point tools. These challenges range from simulating the complex multi-physics of the sensing element and associated electronics to system level modeling for ANC, package analysis and manufacturing yield enhancement.

This hybrid approach has numerous advantages. First, because the underlying physics of the MEMS microphone are included in the MEMS+ models, microphone designers no longer need to create a reduced-order or equivalent network models from FEA data and/or analytical expressions. This saves valuable time and resources. Second, because these models are inherently parametric, rapid exploration of design changes and process changes can be immediately realized.

In addition, utilizing a platform that integrates with best-in-class simulators like Cadence Spectre and/or Matlab/Simulink provides designers the best combination of accuracy and capacity. As Cadence and the Mathworks make speed and capacity improvements to their algorithms, these get automatically multiplied by the additional algorithm improvements and functionality made by Coventor.

Comments are closed.