The sales volume of microphones exceeds 2 billion each year, which has attracted the attention of the microphone market in view of such a huge sales volume. About half of the microphone market is very inexpensive low-end microphones, targeting the toy market and other applications where size and performance parameters are less stringent. The other half is the portable, high-end application market, such as mobile phones, cell phones, digital cameras, notebook computers, etc. The giants in this market are mobile phone manufacturers, who use 900 million microphones each year. With a forecast 10% annual growth rate, mobile phones are seen as the fastest growing segment of the microphone market. As mobile phones are getting smaller and more functional, the demands on the performance of next-generation microphones are constantly increasing.
For many years, the microphones used in communications applications have been electret condenser microphones (ECMs). This microphone consists of a diaphragm, a backplate and an electret layer. The movable diaphragm and the fixed back plate constitute the two pole plates of the variable capacitor. The electret layer stores a fixed charge equivalent to a capacitor Voltage of approximately 100 V. The sound pressure causes the diaphragm to vibrate, thereby changing the capacitance of the microphone. Since the number of charges distributed across the capacitor is constant, the voltage across the capacitor varies as the capacitance changes, according to the capacitor charge formula below:
in,Qis the charge in Coulombs;Cis the capacitance in Farads;Vis the voltage in volts. As the sound pressure changes, the capacitance slightly increases or decreases, denoted as ΔCwhich causes the voltage to decrease or increase proportionally, denoted as ΔV.
Microphones in mobile applications are very small, typically 3 mm to 4 mm in diameter and 1 mm to 1.5 mm in thickness. Therefore their capacitance is also quite small, typically 3 pF to 5 pF, and in some cases as small as 1 pF.
If the signal produced by the condenser microphone is not capable enough to drive it, a buffer or amplifier is required before the signal is further processed. Traditionally, a simple junction field effect transistor (JFET) input amplifier has been used to implement preamplification of such a microphone. Figure 1 shows a cross-sectional view of a packaged JFET-based amplifier ECM.
Figure 1. Cross-sectional view of a microphone based on a JFET amplifier
With the improvement of ECM micromechanical technology, the volume of the microphone is getting smaller and smaller, and the capacitance is also decreasing. Since standard JFET amplifiers have considerable input capacitance, causing significant attenuation of the signal from the microphone unit, JFET amplifiers are no longer suitable for microphone requirements.
Fortunately, improvements in CMOS manufacturing processes have driven improvements in amplifier circuits. There are many benefits to replacing JFET amplifiers with CMOS analog and digital circuits. Preamplifiers implemented in modern sub-micron CMOS processes have several advantages over traditional JFET amplifiers:
• Reduced harmonic distortion
• Easier gain setting
• Multi-function modes, including low-power sleep mode
• The analog-to-digital conversion function enables the microphone to output digital signals directly
• Greatly improved sound quality
• Improved anti-jamming capability
Digital output microphone preamplifier
While simple JFET-based amplifiers have low power dissipation, they suffer from poor linearity and low accuracy. Therefore, the main goal of improving microphone design is to combine preamplifier and digital technology to increase dynamic range by improving linearity and reducing noise while maintaining extremely low power consumption.
Mobile phones are in an inherently noisy environment. The problem with the traditional JFET amplifier (and any purely analog) approach is that the output signal of an analog microphone is susceptible to interference from noise signals lurking between the amplifier and the analog-to-digital converter (ADC). Therefore, the ADC is integrated into the microphone, so that the microphone itself can provide a digital output to reduce noise interference.
A block diagram of the integrated digital output preamplifier and its interface is shown in Figure 2. The signal of the sound transmission unit is first amplified by the amplifier, and then converted into a digital signal by the ADC. An internal regulated power supply supplies power to the amplifier and ADC, which not only ensures good supply voltage rejection, but also provides an independent power supply for the analog part.
Figure 2. Block Diagram of a Digital Microphone System Using the ADAU1301 Microphone Preamplifier
T The preamplifier is fabricated using a CMOS process using two operational transconductance amplifiers (OTA) in accordance with the method of using matched capacitors to set the gain in the instrumentation amplifier structure. This structure with MOS input transistors has a very ideal characteristic of close to zero input admittance for capacitive signal sources. Due to the use of capacitors for gain setting, high gain accuracy (limited only by the lithography process) and high linearity inherent in multi-layer-multi-layer (poly-poly) capacitors is ensured. The gain of this amplifier is easily set by metal mask programming, and its gain can reach 20 dB.
The ADC is a fourth-order, single-loop, single-bit sigma-delta modulator whose digital output is a single-bit oversampled signal. Using a sigma-delta modulator to implement analog-to-digital conversion has several advantages:
• Noise shaping moves the spectrum of quantization noise into the high frequency band, far outside the useful frequency band. Therefore, the circuit system can achieve high accuracy without strict matching requirements.
• The ADC uses a single-bit sigma-delta modulator, making it inherently highly linear.
• In a single-bit, single-loop modulator, only one integrator has stringent design constraints. The outputs of the inner loop integrators are all noise shaped, thus relaxing their design requirements. This reduces power consumption.
A potential problem with higher-order sigma-delta modulators is that they are prone to instability when the input exceeds the maximum stable amplitude (MSA). When a high-order sigma-delta modulator (>2) becomes unstable due to overload, it cannot return to stable operation even if the input falls below the maximum stable amplitude. To address this potential instability, a digitally controlled feedback system modifies the sigma-delta noise transfer function, forcing the modulator back to steady state operation.
The system enters low power consumption when the system input clock drops below 1 kHzsleep mode. At this time, the current consumed by the system is reduced from 400 µA to about 50 µA, allowing the user to save power when the microphone is not needed. Only 10 ms start-up time is required to return to normal operation from low-power sleep mode.
Because the system has a failure analysis function, a specialtest modeAllows engineers to access different nodes inside the circuit. By switching these internal nodes to the DATA pin during startup, this allows the engineer to access the special preamble applied to the DATA pin for failure analysis.
The three main noise sources of CMOS preamplifiers for condenser microphones are flicker noise (1/fnoise), the broadband white noise of the input transistor, and the input bias resistor R (which sets the DC operating point of the amplifier)BIASThe resulting low-pass filtered white noise. Considering the insensitivity of the human ear to low frequency components, A-weighting is used.
The power spectral density of 1/f noise is inversely proportional to the die area of the transistor.Referred to the input, the magnitude of the 1/f noise is given by
In the formula,Kfis a process-dependent constant, fis the frequency,WandLare the width and length of the MOS chip, respectively,Coxis the gate capacitance per unit area. Therefore, by increasing the area of the input transistor, it can be reduced by 1/fThe magnitude of the noise.Input-referred white noise and metal-oxide-semiconductor transistor (MOST) transconductance valuesgminversely proportional
In the formula,kis the Boltzmann constant,Tis the absolute temperature. When the MOST enters the strong inversion mode, gm≈ 2Id/Veff ,inIdis the leakage current, the effective voltageVeff = Vgs – Vththat is, the gate-source voltage minus the MOST threshold voltageVth. By designing the input pair tube to be very wide, when it enters the weak inversion mode of operation, the MOST is forced to operate in a bipolar-like mode. At this time,gm = Id/(nVT),innis the slope factor (the aspect ratio of the MOST tube, the typical value is 1.5),VTis the thermal voltage. Therefore, the white noise performance can be optimized by maximizing the aspect ratio of the MOST transistor.
Connect the input bias resistor to a capacitive source, so its noise is low-pass filtered.Assuming that the noise is white noise after low-pass filtering, the cut-off frequency of the low-pass filter is much smaller than the frequency of the audio segment, and the total noise power can be obtained askT/C,inCis the capacitance value of this node.
Today’s trend is that microphones are getting smaller and smaller, resulting in smaller capacitances, and noise increases as the capacitance of the microphones decreases. However, the power of the audio band noise caused by the bias resistor also depends on the cutoff frequency of the low-pass filter. The lower the cutoff frequency, the less total noise power is retained in the audio band. To keep the noise down, the value of the bias resistor must be increased by four times the value of each half of the acoustic capacitor. Therefore, for a microphone capacitance of 3 to 5 pF, the minimum resistance is about 10 GΩ.
A good solution to achieve such a large resistance on-chip is to use a pair of diodes in antiparallel with extremely high resistance (typically 1 TΩ to 10 TΩ) near the equilibrium point. For large signals, the resistance is reduced, assuming a fast settling time after overload. Figure 3 shows that the in-band noise is the input bias resistor RBIASThe function.
Figure 3. Bias Resistor Noise
The input transistor area of the preamplifier must be optimized relative to the microphone capacitance. As mentioned before, although if the input transistors are made very large, then 1/fThe noise will decrease, but the capacitive loading of the signal source will also increase, which attenuates the signal amplitude and reduces the signal-to-noise ratio (SNR) within the useful bandwidth. There is a tradeoff here: if the input transistors are made small, the capacitive loading of the signal source becomes small, but 1/fThe noise is significantly enhanced, thereby reducing the signal-to-noise ratio at low frequencies. for 1/fNoise, the best way to achieve the best signal-to-noise ratio is to have the gate-to-source capacitance of the input transistor equal to the microphone capacitance plus the parasitic capacitance. For white noise, the best way to optimize the signal-to-noise ratio is to have the gate-to-source capacitance of the input transistor equal to one-third of the microphone capacitance plus the parasitic capacitance. In practice, the best compromise is for the gate capacitance to fall between these two values.
The bootstrap circuit can minimize the impact of the input pins on the input capacitance of the entire chip.Since the white noise referred to the output andgmProportional, all current source MOSTs are biased in the strong inversion region, thus minimizing the effect of noise.
Table 1 lists the key parameters and performance of the ADAU1301 microphone preamplifier.
Table 1. Typical Parameters and Performance of ADAU1301 (unless otherwise noted)
|voltage||1.64V ~ 3.65V||Operates over the full supply voltage range, but meets specified performance specifications at 1.8 V|
|supply current||400µA||@VDD = 1.8V|
|Maximum gain deviation||X ± 0.4 dBFS/VPeak||Xis the specific gain|
|Signal Bandwidth Lower Limit||25Hz||
|Signal Bandwidth Cap||20kHz||
|Input-referred noise rms||5µVrms||A-weighted|
|signal to noise ratio||60.6dBFS||Calculated at –27 dBFS/Pa microphone sensitivity|
|Dynamic Range||>86 dB||@ THD = 10%, depends on gain|
|input capacitance||0.1 pF||
|Minimum input resistance||15 GΩ||
|Start Time||500 ms||Time from VDD reaching 1.8 V to ASIC gain reaching 1 dB final settling value|
|Maximum wake-up time||10 ms||
|Clock frequency||1MHz ~ 4MHz||Nominal value Fclk = 2.4MHz|
|clock duty cycle fDC||40% ~ 60%||
Fully integrated digital microphone
Although this digital output amplifier fully meets the requirements of ECM, it is not fully suitable for the emerging Micro-Electro-Mechanical Systems (MEMS) microphone market, which requires a higher level of integration. There is no equivalent of an electret layer in solid-state MEMS elements, so capacitive elements require an integrated high-voltage bias source. Because the microphone unit constitutes a purely capacitive load, with no current flowing from the bias reference source, this extended version of the amplifier system may include a low-power built-in charge pump, eliminating the problem of storing the charge source.
Microphone preamplifiers designed for the mobile phone market naturally contributed to the rise of digital output microphones. Noise analysis yields a low-noise instrumentation amplifier with the required dynamic range. Low-power sigma-delta ADCs can achieve high resolution without strict design constraints.Low power consumptionsleep modeenter when the microphone is not neededpower saving mode, can prolong the battery life.specialtest modeto provide manufacturers with easy access to other internal nodes for testing, and to provide a convenient condition for testing the analog output of the preamplifier.