MEMS design for better audio
In fact, the success of image recording on mobile platforms calls attention to a less visible but equally important element in a consumer’s multimedia experience, the audio quality. This is defined by the quality of MEMS Microphones being used.
Throughout the relatively short history of camera-equipped smart phones, audio quality has not improved at the same rate as video quality. One reason for this was the absence of consumer demand. Quite simply, when you are making a voice call, a good microphone primarily benefits the receiver of the call and not the owner of the phone. This makes it difficult to market a good microphone to the buyer of the smartphone.
Microphones in smartphones today do more than capture voice for transmission. They also work as audio sensors in a very low power mode to support voice activation and control. And they provide high quality audio when the phone is used for video recording. Video recording mode especially drives microphone requirements toward higher acoustic performance. It can be very frustrating when an ultra-HD resolution smartphone video is paired with a low-quality audio recording.
Since the smartphone owner directly hears the performance of the microphone when they play back video, microphone quality can be a differentiator. If the audio is poor it may ruin the recording, even if it is in 4K video resolution.
Acoustic overload point (AOP)
The popularity of video recording (and sharing) places a greater priority on a microphone’s acoustic overload point or AOP. Beside the well-known Signal to Noise Ratio (SNR), the AOP is the most important quality indicator for a microphone.
AOP defines the sound pressure level (SPL) at which the microphone reaches a total harmonic distortion (THD) of 10 percent. As a measurable characteristic, AOP is a good start in evaluating how well a microphone performs at high sound pressure levels, including high noise environments such as concerts and night clubs. In some usage scenarios, even wind noise can cause a microphone to reach its AOP.
In the past, phone manufacturers used 120dB SPL as a baseline AOP level for most microphones. The AOP requirement recently went up to 130 dB SPL and higher. The additional 10 dB results in superior acoustic performance for the end user even in loud concert environments. In addition, higher robustness against wind noise is achieved. Audio algorithms on smartphone level can better process the audio signals without the occurrence of artifacts.
Micro machines at work
Smartphone microphones are MEMS (Micro Electro-Mechanical Systems) devices fabricated in high volume using semiconductor production processes. The typical design combines a MEMS sensor and an ASIC – see figure 1. The sensor includes a membrane that moves as a result of acoustic pressure, creating an electrical signal that is amplified for analog microphones or processed by the ADC for digital microphones in the ASIC.
Fig. 1: The typical microphone design combines a MEMS membrane and an ASIC.
Infineon enables high SPL microphones with chipsets that it sells to microphone manufacturers. The solution consists of the MEMS microphone element and an ASIC with analog or digital output. The MEMS microphone, which converts the audio to an electrical signal, is basically a DC biased capacitor, where movement of a membrane (or diaphragm) caused by audio pressure changes the voltage over a capacitor plate or plates – see figure 2.
Fig. 2: MEMS microphone capacitive sensor principles: A) Single back-plate and B) Dual back-plate.
The main challenge of handling the pressure level of loud sounds is the large mechanical movement of the membrane which will cause distortion when the membrane is displaced to its extremes. The second challenge is to design the ASIC to handle the large signal that the MEMS element generates.
One way to make a high SPL microphone is to reduce the sensitivity of the microphone by making the membrane stiffer or by reducing the bias voltage.
This would reduce movement of the membrane and thus make the generated signal lower. However this also reduces another key parameter, the Signal-to-Noise Ratio (SNR), accordingly. A high SNR is important in many use-cases and describes the margin between the microphone’s own noise, which should be as low as possible, and its sensitivity, which should be as high as possible.
The sandwich effect
Another approach is to implement a MEMS element which places the moving membrane between two capacitor plates as shown in figure 3. This produces a differential (compared to single-ended) output, which has several advantages. A dual back-plate MEMS microphone minimizes distortion due to its symmetrical construction. The same principle is used for high end studio condenser microphones.
Fig. 3: Infineon’s dual back-plate MEMS design for high AOP microphones.
A differential element is more readily managed through the audio processing chain (pre-amp, ADC, etc.), which potentially reduces power requirements for the ASIC. It also reduces RF interference, resulting in fewer signal processing steps.
A dual back-plate device is more robust against wind noise due to the fact that higher AOP Manufacturers of single back-plate devices typically use a filter to eliminate low frequency wind noise, with subsequent impact on audio quality. This filter removes bass, which is particularly important in recording music. After all, it really is all about the bass.
A result of the above is that a dual back-plate device shows a much more linear behavior until it reaches the AOP. Compared to conventional single ended microphones the SPL where audible distortion begins (2 percent) is pushed out by approximately 10 dB. This has significant impact on the quality of the audio signal as shown on the graph of figure 4.
Fig. 4: Measured THD on microphones from a major smartphone tear-down.
Infineon-based high SPL microphones use a dual back-plate design. This achieves a very high AOP while matching or exceeding the SNR of alternatives in the market today. Audio testing has shown that with excellent audio playback and listening conditions, THD greater than 2 percent is noticeable. Thus, achieving an AOP greater than the current minimum industry level of 130 dB SPL with less than 2 percent THD is significant.
Infineon commissioned a study by an independent institute to assess microphone performance according to the latest ITU-T recommendation utilizing POLQA (Perceptual Objective Listening Quality Analysis). The results confirm the findings with the subjective listening test box and demonstrate the outstanding performance of microphones based on Infineon’s dual back-plate MEMS Technology.
There are many elements that affect the final performance of the silicon microphones in a smartphone, including the design of the audio processing chain and the overall microphone array. Multiple variables also affect the overall power consumption of the microphone array, though the reduction of signal-processing complexity made possible with a dual-plate design can be an advantage.
In the end, the best starting point for design of the microphone path is a listening test. Available test boxes pictured in figure 5 can be used to compare low and high AOP microphone elements and provide a baseline for development.
Fig. 5: Portable live demonstrator of high and low AOP microphones.
About the author:
Andreas Kopetz is in charge of MEMS Microphone Marketing at Infineon Technologies AG – www.infineon.com