Research on Embedded Multichannel Audio Conversion Module

With the rapid development and progress of information technology and the wide usage of audio signal processing in underwater acoustic signal processing, acoustic audio signal acquisition, conversion, and transmission technology always play an important role. To enhance the ability of signal acquisition and conversion with high reliability, this paper designs an embedded multi-channel audio conversion module. The module achieves multi-channel, multi-sample rates with synchronous analog-to-digital/digital-to-analog conversion (ADC/DAC) function, and additionally equips with dual network redundancy. The module uses a chip integrated with ARM and programmable logic FPGA as the main control chip. The writing of the underlying driver and application code ensures the reliable operation of the DAC chip and ADC chip and the steady of the network transmission. Through experimental verification, the audio conversion module performs well in multi-channel and multi-sample rates with synchronous ADC/DAC, and the crosstalk between channels is less than -50 dB. The dual network redundancy design and the application of the RS232 serial port ensure the high fidelity, reliability, and reprogramming of the audio conversion function, which shows the design can be used in a wide range of scenarios.


INTRODUCTION
Complex underwater environments and serious noise interference have resulted in low acquisition speed, poor audio conversion effect, and limited data transmission of the underwater audio signal processing system at present [1].In [2], a real-time audio processing module is proposed, but it does not support a multi-channel and multi-sample rates function.In [3], a multi-channel audio sample system is provided, but it does not provide multi-sample rates ability.In [4], an ADC/DAC system is proposed, but it does not provide dual network redundancy.The traditional audio signal processing system can no longer meet the development needs of underwater acoustic equipment.The networking of signal processing is an inevitable development trend.The integrity of network transmission and the fidelity and reliability of audio conversion have become important factors in evaluating the quality of audio processing systems.The embedded audio conversion module designed in this paper adopts an integrated design approach to achieve the entire module.This module integrates the main control chip, DAC chip, ADC chip, audio drive amplifier chip, LDO power supply chip, and optocoupler chip.It is characterized by high performance, high reliability, high real-time, low power consumption, and low heat production.The functions of signal acquisition, digital-to-analog or analog-to-digital conversion, and network transmission are implemented through FPGA logic control software and master control application software running on the main control chip.

Module DESIGN SCHEME
To realize multi-channel and multi-sampling rate synchronous analog-to-digital/digital-to-analog conversion, as well as transmission function, this paper designs the main control module and interface module by adopting a modular design scheme.The high-precision DAC chip and ADC chip integrated into the interface module can process multiple audio data from 12 KHz to 96 KHz or higher by differential methods, and ensure that the output signal is distortion-free.The main control chip integrated into the main control module writes the underlying driver and application code to realize the network transmission of audio data.At the same time, the dual-network redundancy design on the main control module ensures the reliability of data transmission with the outside world.The main control module is connected to the interface module through the VPX connector, the interface module receives audio data from the main control module, and the main control module controls the signal processing process of the interface module in real-time.The integration of the main control module and interface module can realize the following functions: (1) The digital signal input by the network is output through the listening channel after digital-to-analog conversion; (2) Real-time acquisition of external input analog signals is output to the network group after analog-to-digital conversion; (3) The application of RS232 serial port ensures online debugging and reprogramming in different application scenarios.

Hardware Design
The interface module integrates DAC chips, ADC chips, audio drive amplifier chips, LDO power supply chips, and optocoupler chips.It can realize digital-to-analog conversion and amplification output of multiple digital signals, analog-to-digital conversion of multiple analog signals, and acquisition of multiple discrete signals [5].
The main control module uses a domestic PSOC chip as the main control chip, integrating ARM and programmable logic FPGA.ARM is used to implement the design of application software.FPGA custom IO ports are connected to devices outside the module, enabling real-time control of audio DAC and audio ADC on the audio interface module.

Hardware design of the main control module
The control chip of the main control module uses the PSOC chip FMQL45T900 of Fudan Micro Corporation, which contains a PS part and a PL part inside.

Figure 1. Hardware principle block diagram of the main control module
The PS part is a 4-core ARM CortexA7 with a main frequency of up to 800 MHz, external 1 GB DDR3 (32 bit), 512 Mb SPI FLASH, and 32 GB MMC.One RS232 interface can be expanded and led out from the front panel for burning, debugging, and printing of application software.The PS part is connected to two PHY chips through the RGMII interface to expand 2-way Gigabit Ethernet.In this design, the 2-way Gigabit Ethernet of the PS part is used.The PL is an FPGA logic processing part that can be expanded from the PL part to 2 GB DDR3 (64 bit), 1-way RS232 interface, and 1-way RS422 interface.And we connect 2 PHY chips through the SGMII interface to expand 2-way Gigabit Ethernet.The PL part can be connected to a voltage and current monitoring sampling chip to monitor the operating voltage, current, and power consumption of the module.
The main control module is implemented by using an integrated design idea.The main chips integrated into the board include PSOC, DDR3 SDRAM, eMMC, Flash, PHY, and so on.The required power supplies include 1.0 V, 1.2 V, 1.5 V, 1.8 V, 2.5 V, 3.3 V, etc.It is necessary to generate various power supplies required for each chip through a 12 V DC input power supply.The power-up sequence of the entire module is controlled by a CPLD chip, with a total power consumption of no more than 12 W.The main control module is connected through the VPX series connector of AVIC Optoelectronics.The hardware principle block diagram is shown in Figure 1.

Hardware design of the interface module
The interface module mainly implements the digital-to-analog or analog-to-digital conversion of audio signals and the acquisition of multi-channel discrete signals.The module integrates six domestic highprecision dual-channel differential output DAC chips with the model ES8156, two domestic highprecision audio ADC chips with the model ES7210, and eight independent single-channel optocoupler chips.A domestic LDO chip converts a 12 V DC power supply into a 3.3 V power supply, with a maximum current of up to 3 A, fully meeting the power supply requirements of onboard DAC chips, ADC chips, optocoupler chips, and other circuits.The hardware principle block diagram is shown in Figure 2.
In this design, four DAC chips are used as single-channel outputs, and two DAC chips are used as dual-channel differential outputs on the module.Therefore, the module can simultaneously output up to 8 analog signals.When receiving and processing data, 24-bit digital signals can be converted into differential analog output signals with a maximum sampling rate of 96 kHz.
The 8-channel independent single-channel optocoupler chip on the module can isolate and drive the input 8-channel discrete signal and output it to the main control board for collection, which can be used to determine whether there are analog signals to be collected.
The two ADC chips used in this article can simultaneously collect up to eight analog signals, with a maximum sampling rate of 96 kHz.

Software Design
The audio conversion module software runs on the PSOC chip FMQL45T900 of the main control module and is divided into PS part software and PL part software.The software of the PS part is the main control application software of the audio conversion module based on the domestic Reworks operating system.The Reworks operating system implements the dual network redundancy function of the PS part's 2-way Gigabit Ethernet network.PL part of the software is the audio conversion module FPGA logic control software.The main control application software and FPGA logic control software interact with each other through the AXI bus, and the two software negotiate and define a data address allocation protocol.After the main control application software writes the data to the corresponding address specified in the address allocation protocol, the data will be transmitted to the FPGA through the AXI bus, and the FPGA logic control software reads and further processes the data.After the FPGA logic control software writes the data to the address specified in the address allocation protocol, the data is transmitted to the main control application software through the AXI bus, which reads and further processes the data.The main control application software interacts with the FPGA logic control software through the AXI bus, mainly realizing the functions of digital-to-analog conversion and analog-to-digital conversion.

Design of digital-to-analog conversion software
The main control application software receives digital-to-analog conversion commands through an Ethernet interface.We obtain the data source information, sampling rate (not greater than 96 kHz), and audio signal output channel number according to the command received through protocol unpacking, and then write the sampling rate to the sampling rate setting address specified in the address allocation protocol.At the same time, the main control application software receives the digital signal sent by the audio signal source through the Ethernet interface and obtains the data source information through unpacking according to the protocol.If the data source information is consistent with the data source information of the digital-to-analog conversion command, matches the FIFO write address in the address allocation protocol based on the audio signal output channel number in the digital-to-analog conversion command, and sequentially writes the digital signals in the data packet that require digitalto-analog conversion to the FIFO address.The sampling rate and digital signals written by the main control application software will be transmitted to the FPGA through the AXI bus.
The FPGA logic control software reads the sampling rate through the AXI bus and transmits the sampling rate to the clock control module.After receiving instructions, the clock module samples the digital signals in the FIFO according to the sampling rate.The FPGA logic control software writes the sampled digital signal to the IIS controller for parallel to serial conversion.The serial data is transmitted to the DAC chip of the audio interface module through the IIS bus and converted into an analog signal, which is then transmitted to the corresponding output channel through the VPX backplane.The output channel number is determined by the FIFO address where the digital signal is stored.The digital signals written into the FIFO address 1, 2, 3, and 4 by the main control application software are written into one DAC chip and then transferred to the right channel of the listening channel 1, 2, 3, and 4 after DA conversion [6].The digital signal written by the main control application software to the FIFO write address 5 needs to be simultaneously written to two DAC chips and then transmitted to the left channel of listening channels 1, 2, 3, and 4 after differential output.The flow of digital-to-analog conversion software is shown in Figure 3.

Design of analog-to-digital conversion software
When the microphone connected to the audio interface module is pressed, the onboard 8-channel independent single-channel optocoupler chip will isolate and drive the input 8-channel discrete signal.
The FPGA logic control software on the audio main control module collects discrete signals in realtime to confirm that the microphone has been pressed.Then the clock module starts to control the ADC chip on the audio interface module to collect the analog signal passed through the microphone at the sampling rate specified in the protocol and to perform the analog-to-digital conversion.The FPGA logic control software writes the signal that completes the analog-to-digital conversion into the FIFO.At the same time, the main control application software regularly queries whether the FPGA has completed the analog signal acquisition, analog-to-digital conversion, and AXI bus transmission.If the query is successful, the main control application software will read the digital signal after FPGA analog-to-digital conversion from the FIFO read address specified in the address allocation protocol, and then package the digital signal according to the protocol and send it to the multicast address specified in the protocol.The analog-to-digital conversion software flow is shown in Figure 4. Verification of digital-to-analog conversion function: We simulate and output sinusoidal digital signals with frequencies of 500 Hz and 1, 000 Hz in the laboratory, and transmit the two types of audio signals to the audio conversion module through the network.After setting the sampling frequency to 12 kHz, we perform the digital-to-analog conversion.We connect an oscilloscope to two listening channels and capture the sinusoidal signal displayed in the analog signal format in real-time, as shown in Figure 5.When we verify the digital-to-analog function, an experiment of simultaneous sampling and digitalto-analog conversion of four channels was conducted.When we verify the analog-to-digital conversion function, experiments were also conducted on the simultaneous acquisition and analog-to-digital conversion of four analog signals.Both experiments have verified that the crosstalk between channels can be less than -50 dB when the module performs multi-channel audio conversion.In [7], the crosstalk is -40 dB, which indicates the system in [7] has more interference from each channel than ours.

Conclusion
This paper implements the design of an embedded audio conversion module, which integrates multiple DAC chips and ADC chips to achieve multi-channel and multi-sampling rate synchronous DAC or ADC functions.The implementation of dual network redundancy increases the reliability of data communication, and RS232 enables online software debugging and reprogramming.Experimental testing shows that the signal converted by this module has no distortion, and the crosstalk between multiple channels is less than -50 dB.The integrity of the digital/analog signal is guaranteed, and the high-fidelity conversion of the digital/analog signal is realized.This module can meet the requirements of multi-channel synchronous high-fidelity digital-or-analog signal conversion in different fields and has referential significance for the application of audio processing technology.

Figure 2 .
Figure 2. Hardware principle block diagram of the interface module

Figure 3 .
Figure 3. Digital-to-analog conversion software flow

Figure 4 .
Figure 4. Analog-to-digital conversion software flow5.Test verificationAfter setting up the software and hardware debugging environment, we conduct a comprehensive functional test on the audio conversion module.Verification of digital-to-analog conversion function: We simulate and output sinusoidal digital signals with frequencies of 500 Hz and 1, 000 Hz in the laboratory, and transmit the two types of audio signals to the audio conversion module through the network.After setting the sampling frequency to 12 kHz, we perform the digital-to-analog conversion.We connect an oscilloscope to two listening channels and capture the sinusoidal signal displayed in the analog signal format in real-time, as shown in Figure5.

Figure 5 .
Figure 5. Digital-to-analog converted sinusoidal signal Analog-to-digital conversion function verification: We connect the microphone to the audio conversion module.We press the microphone and collect sound signals at a sampling rate of 24 kHz.After the audio signal has undergone analog-to-digital conversion and network transmission, the audio signal captured on the network is shown in Figure6in the audio processing software.

Figure 6 .
Figure 6.Signal waveform diagram of analog-to-digital conversion