Shenzhen hicsemi Electronics Co., Ltd.

introduction

The unprecedented growth in the number of mobile phone users and the complexity of usage scenarios have made environmental noise suppression technology an urgent demand for mobile communication. Environmental noise usually includes two parts, point noise and diffuse noise, or a neutralization of both. The point noise is closer to the user, and its amplitude and frequency change quickly; Dispersive noise is farther away from the user, and its amplitude and frequency change slowly, such as background noise. Diffuse noise can be suppressed with the technology of one microphone, while point noise needs to be quickly captured and suppressed due to its fast change. Usually, two Microphone array need to be used for rapid positioning to form a spatial filter for rapid convergence and noise suppression (see Figure 1). This article will take the MTK platform as an example to introduce the design points of FM2018-380 in anti noise mobile phones.

Figure 1 Schematic diagram of the polarity of the anti noise spatial filter in handheld mode

Key points of SAM technology design

Unlike handheld applications with a microphone, array microphones need to consider the sensitivity difference and placement position between the two microphones (see Figure 2). The distance between two microphones is greater than 60mm. For near-field signals, the near-field voice signal picked up by the main microphone (close to the user's mouth) is more than 6dB larger than the reference microphone (close to the user's ear). For far-field signals (noise exceeding 0.5m), the signals picked up by the main microphone and the reference microphone are basically the same. Based on the differences between array microphones, chip processing is used to differentiate between near-field and far-field signals, preserve useful near-field signals, and suppress noise in all directions of the far-field

Figure 2 Positioning of the microphone on a straight phone

The structural design ensures that the opening sizes of the two microphones are consistent, ensuring the airtightness of the microphone cavity. If there is a hands-free call function, more consideration needs to be given to issues such as speaker and microphone vibration reduction.

Figure 3 Typical application schematic diagram of FM2018-380 on mobile phones

Design of noise resistant mobile phones

Taking MTK's mobile platforms (MT6225, MT6318, MT6319) as an example (see Figure 3), the main microphone (MIC0) and reference microphone (MIC1) each require separate bias circuits that use differential signal input to the chip. The microphone bias circuit needs to add small capacitors to filter out RF interference. After being processed by FM2018-380, the array microphone signal is output through a line (LINE_OUT) and sent to the microphone input terminal (MICP0, MICN0) of MT6225 using a false differential circuit. The false differential circuit can reduce radio frequency interference and local noise. The output differential signal of the mobile phone hands-free speaker is sent to the reference signal input terminal (LINE_IN_P, LINE_IN-N) of the chip for echo cancellation, used for echo cancellation in hands-free mode.

The commonly used control signals include reset control (RESET), power saving control (PWD_LED), and serial host interface (SHI). All digital pins of FM2018-380 can withstand a voltage of 3.3V and can be directly connected to the universal input output port (GPIO) of MT6225 without voltage conversion. The universal I/O port is used to simulate the SHI interface and control the FM2018-380. The SHI interface supports a clock signal of 400kHz at most. Select an appropriate Pull-up resistor according to the clock speed. When the clock speed is 100kHz, the resistors R5 and R6 use 10K Ω. When the clock speed is 400kHz, the resistors R5 and R6 use 2.2K Ω. The resistors need to be pulled up to the 2.8V power supply of the baseband chip.

Figure 4 FM2018-380 Control State Transition Diagram

After the phone is turned on, the reset control (RESET) remains low. After 1.8V stabilizes, the power saving control (PWD_LED) is high, and the MT6225 clock is input to the chip. After the power saving control remains for 5ms, the reset control (RESET) becomes high. After waiting for 10ms to stabilize the PLL of FM2018-380, the baseband chip sends parameters to initialize FM2018-380 through the SHI port. After initialization is completed for 108ms, the power saving control (PWD_LED) is low, Put the chip into a power-saving state (see Figure 4). When a call is made or a call is made, the baseband chip wakes up FM2018-380 by setting it high through the power saving control (PWD_LED). Based on the value of the 12th bit of parameter 0x1E51, it sets whether to download the parameters again after wake-up. When the value is 0, save the parameters and automatically restore to the working mode before power saving; When the value is 1, the software resets and the parameters are downloaded again. To switch between handheld and hands-free modes, it is necessary to reset the chip and switch the parameters of the hands-free or handheld mode.

Measure the difference between the main microphone (MIC0) and the reference microphone (MIC1) of the near field voice signal in the handheld mode, and the difference between the main microphone (MIC0) and the reference microphone (MIC1) of the far field noise signal. Calculate the Acoustic model parameters of the spatial filter according to the difference between the two differences, and adjust the threshold value of active voice detection (VAD). When speaking in the near field, VAD1 is triggered (to determine whether there is speech inside the near field pickup), and when there is noise in the far field, VAD2 is triggered (to determine whether there is external far field noise outside the pickup). When both near field speech and far field sound are present, the noise is the dominant signal, and VAD2 is triggered. However, when there is near field speech, VAD2 is not triggered. Adjust the noise spectrum in the frequency domain in a simulated noisy environment, so that the reference channel noise spectrum of the main channel is parallel, and adjust the frequency domain noise suppression parameters. Adjust the Automatic Volume Control (AVC) function. Through the SHI interface, read the reference speaker volume provided by FM2018-380 based on the noise situation, change the gain of the downlink, and if the speaker volume has reached its maximum value, read the tone parameter to change the tone of the high-frequency band of the downlink channel, improving the receiver's intelligibility.

Figure 5FM2018-380 Actual Test Effect of Point Noise Suppression

performance testing

In MTK's engineering mode audio enhancement project, turn off baseband noise suppression and echo cancellation to avoid affecting sound quality and noise echo suppression effect. Evaluate the speech communication system containing noise suppression algorithms using the ITU-T Recommendation P.835 subjective testing method, and test the speech quality processed by FM2018-380 according to the standard. The noise level is equal to the signal-to-noise ratio at the standard microphone between 0dB and 12dB, and the voice quality under noise is scored by the voice quality assessment system (VJS). Figure 5 shows the actual test ability to suppress single point noise. The test distance is from 0.3m to 1.5m, the interval is 0.15m, and the test angle is 0 °, 90 °, 180 °, 270 °. The noise suppression values of each angle and distance can be obtained.

epilogue

The FM2018-380 is based on directional distance array microphone (SAM) pickup technology and is specifically designed for mobile communication handheld devices that combat noise. It is very suitable for noise resistant designs and applications such as GSM, CDMA, 3G, etc. This article is based on this chip and takes the MTK mobile phone platform as an example to introduce the design points of anti noise mobile phones, which has strong practical value.

Design of anti noise handheld communication product based on FM2018-380 chip