MEMS Microphones
Introduction, Integration, and Implementation
Image Source: bestforbest/Stock.adobe.com
Current consumer demand for ever-increasing mobility and optimized form factor for electronics have led to advances across the integration of disparate circuit components and constant performance improvements of a single device that serves multiple applications. With the onset of the pandemic and the explosion of video conferencing or video/voice communication, connectivity from anywhere with multiple applications is a necessity. Microelectromechanical systems (MEMS), a class of components that saw rapid growth in the 1970s and 1980s, enable smaller, more powerful, and ever-increasing mobile devices for consumers. Integrating and implementing MEMS microphones in particular provides advanced acoustic capabilities in these devices.
MEMS Overview
What Are MEMS?
MEMS are devices that integrate mechanical elements, sensors, actuators, or switches on a common chip. Although definitions vary, MEMS devices typically have dimensions less than 1mm in length and are built on a silicon substrate due to their fabrication process. MEMS devices inherit and use a wide variety of integrated circuit (IC) semiconductor surface micromachining fabrication processes (e.g., lithography, wet and dry etching). Thus, MEMS devices leverage the maturity and infrastructure of the semiconductor industry, resulting in low-cost, high-reliability, and high-performance devices. Although the fabrication process is shared between MEMS and ICs, differences exist. MEMS fabrication processes deviate by including deeper etching, a release process to allow mechanical structures to separate from the bulk substrate, and consideration for stiction and parasitics. Overall, a different set of fabrication conditions must be considered for MEMS with a focus on mechanical properties (e.g., Young’s modulus, residual stress, and fatigue limit) is essential.
MEMS are commonly found in electronic devices that are used in everyday applications. Notable MEMS devices include accelerometers, gyroscopes, pressure sensors, and many others. MEMS play a key role in the general trend of device miniaturization and the cutting-edge performance capabilities of current devices. One such class of MEMS devices that will be explored in depth is the MEMS microphone.
MEMS Microphones
MEMS microphones are acoustic devices that are relevant to all audio applications that require small packaging dimensions, high performance, low power consumption, and reliable performance. MEMS microphones operate as transducers that convert acoustic pressure into electrical signals. MEMS, like many other device components, can be divided into different types based on signal (analog compared with digital), output (single compared with differential [analog MEMS]), and operating principal (capacitive compared with piezoelectric).
These types of MEMS microphones are listed in Table 1 for reference, with notes on their advantages.
(Source: Author)
MEMS Microphone Applications
Laptops, cell phones, video teleconferencing stations, and noise-canceling headphones are examples of devices that use at least a single MEMS microphone. In many cases, devices such as in-ear headphones, would not be where they are now without MEMS microphones. Advances in MEMS microphones provide noise cancellation, audio recording, and audio playback with minimal distortion and high-performance capabilities in an extremely small form factor.
Integrating and Implementing MEMS Microphones
Size, Weight, Power, and Cost
Capacitive MEMS microphones are characterized by six different features:
- Packaging and enclosure
- An acoustic port
- A membrane
- A back plate
- A back cavity
- A front cavity
These six characteristics affect how well the microphone performs (Figure 1). When examining size, weight, power, and cost, fundamental limits and trade-offs exist due to how these features affect the microphone. When compared with general microphones, MEMS microphones are extremely compelling due to better size, weight, cost, and power envelopes. Application use cases however, must be considered and dictate microphone selection.
Figure 1: MEMS microphone general design and structure. Mechanical design determines the performance of the MEMS microphone. Highlighted features of the MEMS microphone all play a role in how the MEMS performs. (Source: Author)
Dynamic Range
The dynamic range of a MEMS microphone is characterized by the acoustic overload point (AOP) and the noise floor of the device (Figure 2). A key factor to high-performance microphones is to ensure that the dynamic range does not limit performance of the device and covers the full range of signal levels desired. When properly designed, the dynamic range of the microphone will indicate the measurement range of the microphone from the softest audible signal to the loudest without significant distortion (or sound pressure levels [SPLs]).
Figure 2: Illustration of dynamic range. Note that typically the usable dynamic range of the microphone will be less due to application requirements that bound both at high SPL (due to distortion requirements) and low SPL (due to SNR requirements). (Source: Author)
The AOP of a MEMS microphone describes the distortion performance of the MEMS microphone at higher SPLs. The AOP provides the SPL where the total harmonic distortion (THD) exceeds 10%. A higher AOP is better and directly translates to the ability to capture louder sounds more accurately at a given distortion level. If the device under design requires high-quality audio performance at higher SPLs, a large AOP is required. In certain datasheets, the AOPpeak or the AOPRMS may be specified. In these cases, a 3dB value can be added to convert AOPRMS to AOPpeak.
To calculate the dynamic range for an analog MEMS microphone, the maximum voltage swings on the output voltage are needed. Always verify that the voltage swing matches the dynamic range.
For digital MEMS microphones, consider the analog-to-digital converter (ADC) when discussing dynamic range. The ADC needs to have enough bits to fully represent the full signal swing. To calculate the ADC level (for a pulse code modulated signal), use the following equation.
Noise and Distortion
When using a MEMS microphone, account for noise and distortion. Noise and distortion represent two different factors for design performance. Distortion in a MEMS microphone is generally given by THD as a percentage that represents the ratio of energy found in higher harmonics to the fundamental frequency or first harmonic. The equation for THD is found below.
Note that the number of higher-order harmonics (n) must be specified when describing THD and should be accounted for when examining any given microphone. THD is measured at the output of the microphone; thus, a lower THD level means that the output is a more accurate representation of the desired acoustic signal. Careful design is essential to minimize the distortion depending on the application and acoustic signal profile. Certain algorithms may be more susceptible to distortion; likewise, certain notes may sound more distorted to the human ear at the same THD level.
Noise in a MEMS microphone can occur from many different sources including self-noise. Like all sensors, MEMS microphones suffer from noise due to electronic components (Johnson noise), quantization noise (ADC), and others. Noise sources also come from mechanical elements (i.e., how the microphone membrane is designed, debris ingress into the device, etc.). Finally, noise can come from environmental factors; wind is one of the most obvious sources of noise. To reduce noise from environmental factors, the placement of the sensor as well as the design of the enclosure of the MEMS sensor during integration are vital. Techniques such as noise cancellation are also viable and can help improve overall device performance.
Noise in a MEMS microphone is measured as equivalent input noise (EIN), expressed as dBSPL at the output of the sensor.
Sensitivity
Sensitivity is a key factor in high-performance MEMS microphones. Ensuring that the microphone is sensitive enough to efficiently convert acoustic pressure to an electrical signal with enough SNR is an important design consideration (Figure 3). In general, sensitivity is measured from a reference acoustic pressure of 94dBSPL (1Pa); however, sensitivity is measured differently between analog and digital microphones. For analog microphones, because we are measuring voltage, sensitivity is expressed in dBV/Pa or mVRMS/Pa. For digital microphones, sensitivity is measured as dBFS (decibels per full scale).
Sensitivity must be taken into account with AOP, dynamic range, and EIN because sensitivity alone does not specify enough to judge performance.
Figure 3: A high-performance MEMS microphone should be sensitive enough to convert acoustic pressure to an electrical signal efficiently. (Source: Author)
Acoustic and Electrical Specifications
Electrical/acoustic specifications for MEMS microphones center on performance, reliability, and consistency metrics. When considering applications, reliability and consistency must sometimes take precedence over performance. For example, in applications for which multiple microphones are needed (e.g., stereo), consistency of frequency response/sensitivity may be more important than overall AOP.
Some important top-level metrics for electrical/acoustic specifications include frequency response, directivity, and power supply rejection (PSR)/power supply rejection ratio (PSRR). Frequency response refers to the sensitivity of the microphone over different audio frequencies. When under consideration, this metric should be as flat as possible or as consistent as possible. A Helmholtz resonance will appear at higher frequencies and limit the performance range of the microphone (Figure 4).
Figure 4: Notional illustration of the frequency response from a MEMS microphone. The Helmholtz resonator results in resonance or increased sensitivity at the end of the illustration. (Source: Author)
Directivity or directionality indicates the sensitivity of the microphone over the angular extent of sound arrival to the sensor. A very directional microphone will only allow sound in a small region to be registered. Using multiple microphones as an array can improve directionality through receive beamforming.
PSRR/PSR indicates the ability of the microphone to reject noise added by the power supply. PSRR/PSR is measured as the residual noise of the microphone output with a spurious input signal at the supply voltage. A higher PSRR/PSR is always desired. The equations for PSRR and PSR are as follows.
Mounting
The placement of the MEMS microphone is important when designing for overall device capability and form factor. On a component level, the placement of one or many MEMS microphones will play a direct role in how the microphone will ultimately function beyond the values in the datasheet. Depending on design decisions in placement and mounting, microphone performance characteristics may be severely altered/degraded/or perform as expected. If, for example, the microphone is placed near actuating and acoustically active components, expect poor performance including distortion and increased noise irrespective of MEMS microphone specifications. Design considerations are paramount to the MEMS device selection itself.
When determining the placement of the MEMS microphone, consider the following:
The design of an acoustic sound channel with proper acoustic sealing (channel length should be as short as possible and match acoustic port dimensions as closely as possible to avoid generating another Helmholtz resonator)
- Mounting rigidity (mounting and housing/enclosure of the MEMS microphone can affect the frequency response by propagating vibrations to the component. Minimize transfer of acoustic noise to the microphone)
- Placement distance relative to other circuit components (maintain spatial isolation of the device)
- Device packaging and distance/shape of the device surface, corners, and edges (closer to the edge and surface is generally better)
- Isolation from acoustic and RF noise sources
- Isolation from thermal gradients
- Adjacency to the desired acoustic source
For specific applications such as stereo, special considerations are needed for microphone placements. For example, in stereo applications, microphones should be placed as far apart laterally as possible.
Interface
The interface for MEMS microphones can be subdivided by whether the microphone has an analog or a digital output. As discussed earlier, the interface for analog MEMS microphones is divided into single and differential outputs. Account for the output impedance in analog MEMS microphones, which can run into hundreds of ohms. Care must be taken not to attenuate the signal by mismatching impedances following the microphone. Another consideration is the inclusion of a capacitor of at least 1µF for direct current filtering.
For digital MEMS microphones, things are a bit easier; however, good interface practice is key. Output interfaces that must be considered for digital microphones include pulse density modulation and inter-IC sound. Some good practices include source-terminating resistors.
For both analog and digital MEMS microphones, ensuring that impedances are matched and parasitics do not exist on trace lines is important for signal integrity. Rejection of possible coupling of lines to noise sources must be avoided (microstrip design of trace lines can be useful). In these cases, digital interfaces are superior (rejection of noise/interference) to analog interfaces.
Conclusion
MEMS enable smaller, more powerful, and ever-increasing mobile devices for consumers. MEMS microphones, which provide advanced acoustic capabilities in these devices, were introduced at a high level, with tips for integration and key parameters to watch—application use cases should always define what type of MEMS microphone is best. Several topics that were not discussed or deserve more attention (e.g., phase distortion) remain; further information can be found in application notes with more in-depth analysis.