How to design a sound level meter based on MEMS?
MEMS are microelectromechanical systems that are fabricated using the same materials (usually silicon) and etching techniques used to make microelectronic circuits. These techniques can build microscale and nanoscale structures with high precision and reproducibility. MEMS microphones are very small but highly sensitive (noise floor typically better than 30dBA). Many MEMS microphones integrate amplification and digital sampling chips at the device level (even chip level), thereby directly providing digital signals and reducing the cost of other parts of the system or instrument. In addition, direct integration of analog-to-digital circuitry at the device level eliminates electromagnetic noise coupled to analog input lines in conventional designs.
MEMS microphones are manufactured using a tightly controlled microetching process, so the individual characteristics of each MEMS microphone are extremely consistent. They are very linear (0.1% total harmonic distortion (HD) or better at 1kHz/94dB SPL) and have a wide dynamic range (typically better than 30dBA to 120dBA). In addition, MEMS microphones have little sensitivity to temperature changes, and likewise, their microphone diaphragms are so small and thin that they are more than 10 times less sensitive to vibration than electrostatic microphones. In addition, MEMS microphones are widely available in the consumer electronics market, so they are also very cheap. Their sensitivity remains very stable over time and usually does not require recalibration to stay within Type I specifications.
These advantages make MEMS microphones ideal for designs. Of course, MEMS microphones have some shortcomings to make up for if they want to design an efficient sound level meter.
Because MEMS microphones provide digital signals at the device level, it is not possible to remove the pressure sensitive cavity from the circuit and test the analog link in isolation. All relevant standards for sound level meters were written in the 1970s and assumed that the sound level meter design consisted of a single microphone cavity driving an analog processing chain or an analog-to-digital converter (ADC) followed by a digital processing chain. This requires the use of electrical signals instead of microphones to test sound level meters. MEMS microphones, on the other hand, complete the analog-to-digital conversion at the device level, which means that even though a sound level meter may have the performance required to comply with a standard, it cannot be tested using the methods specified in that standard.
Due to the very small size of the silicon structures of MEMS microphones, even tiny dust particles can easily enter the microphone cavity and damage them. Extremely high static and dynamic stresses (typically above 160 dB-SPL) can also cause damage to these small silicon structures.
MEMS microphones typically have sharp resonances in the 10kHz to 20kHz range. Correction for this resonance is required so that the frequency response of the sound level meter falls within the limits of the appropriate standard.
