0 Introduction The phenomenon of material structures releasing strain energy in the form of elastic waves through dislocation, slip, microcrack formation, crack propagation, and fracture under external or internal forces is called acoustic emission. Acoustic emission is generally a weak sound wave with a frequency of about 20 kHz to 20 MHz. The acoustic emission signal originates from the defect itself. Defects of the same size and nature will exhibit different acoustic emissions due to differences in their location and stress state. Therefore, acoustic emission can be used to detect the extent of defects or monitor operating machinery, such as for bridge structural safety and partial discharge detection inside large power generation equipment. Understanding the acoustic emission signals from defects and establishing characteristic models of these signals allows for long-term monitoring of defect safety, a key advantage of acoustic emission technology over other methods. Acoustic emission sensors differ from accelerometers; they rely on the resonant deformation of the piezoelectric crystal itself to convert the magnitude of the surface vibration sound waves of the measured object under stress wave action into a corresponding voltage signal output. Because sound energy attenuates due to reflection and scattering at interfaces between different media, as well as propagation within the same medium, the actual sound signal received by the piezoelectric crystal is already very weak, and interference signals in the field may obliterate it. Therefore, this weak millivolt-level voltage signal sensed by the piezoelectric material must be filtered and amplified before transmission to reduce the impact of interference signals and improve the signal-to-noise ratio. Since piezoelectric crystals have poor load-carrying capacity, the amplification factor of the input signal must be increased while ensuring accuracy. This requires the preamplifier circuit to not only have a large input impedance, high common-mode rejection ratio, and low noise figure, but also meet the requirements for high-frequency response characteristics, possess a linear and large dynamic range, and ideally be able to automatically adjust the gain to avoid insufficient amplification for small signals and severe saturation for large signals. Acoustic emission technology is commonly used for in-service monitoring of outdoor metal pressure vessels and engineering facilities. Under different operating conditions, the intensity of elastic waves varies greatly, and the acoustic emission preamplifier must be able to automatically control the gain. Due to the difficulty in supplying AC power and its susceptibility to strong electromagnetic interference, battery power is preferable. However, existing acoustic emission preamplifiers on the market lack automatic gain adjustment and are all AC-powered, making them unsuitable for strong electromagnetic field environments. Therefore, this paper develops an acoustic emission monitoring system with digital automatic gain control. The acoustic emission preamplifier is battery-powered to avoid interference inherent in AC power supplies. This system features strong anti-interference and power management capabilities, adapts to harsh environments with high magnetic fields and high voltages, automatically samples at set times, and can operate unattended for extended periods, achieving satisfactory results in the field. 1. Structure of the Acoustic Emission Monitoring System with Digital Automatic Gain Control. The system consists of an acoustic emission sensor, a preamplifier, a photoelectric conversion circuit, optical fiber, a high-speed data acquisition card, and computer software, as shown in Figure 1. The selection of the acoustic sensor's detection frequency band is one of the key factors in improving the sensor's detection sensitivity. The sensor uses a WD wideband response differential acoustic emission sensor from Physical Acoustics, Inc. (USA), with a frequency response range of 100 kHz to 2 MHz and an operating temperature range of -65℃ to 177℃. Considering the harsh electromagnetic field environment of the metal pressure vessel, the acoustic emission sensor and preamplifier are double-shielded and placed on the pressure vessel. The high-speed data acquisition card and computer are located approximately 100 meters away from the pressure vessel. The amplified acoustic emission analog electrical signal is converted into a modulated optical signal, transmitted via optical fiber to an external photoelectric converter, and converted back into an analog electrical signal before entering the computer's sampling card as a digital signal. The automatic gain control signal from the computer is also transmitted back to the preamplifier via optical fiber to achieve automatic gain control. This greatly reduces strong electromagnetic interference and ensures the reliability of the detection signal. To save battery power, the system includes a power management function, enabling the system to automatically enter sleep mode or timed sampling mode. Field testing showed that the piezoelectric sensor outputs a signal between 2 mV and 10 mV. The optimal linear operating range of the photodiode MF359 is 5 mA to 65 mA; due to single-supply operation, the midpoint can be set at 35 mA. The input level of the voltage-to-current conversion circuit should be between 0.165 V and 2.145 V, with a midpoint level of 1.155 V. Therefore, the amplified sensor voltage signal should vary between ±1 V. The total maximum amplification gain is 500 times. Obviously, a single-stage amplifier cannot handle the amplification of such a high-frequency signal. Typically, the distributed capacitance of the connecting cable affects the sensitivity of the piezoelectric sensor, and the voltage amplifier measurement system is very sensitive to changes in cable length, requiring recalibration when the cable is replaced. Here, the piezoelectric sensor is located close to the preamplifier, the cable length is very short and its position remains constant, the distributed capacitance of the connecting cable is controlled, and the voltage amplifier can function as a charge amplifier. The system circuit principle is shown in Figure 2. The AD8067 forms a high-impedance differential input circuit, picking up the millivolt-level voltage signal from the piezoelectric sensor, with a signal gain AV of 10. Differential amplification reduces common-mode electrical noise introduced by the sensor and its cable; in addition to amplification, the AD8067 converts the high output impedance of the piezoelectric sensor to a low output impedance, i.e., impedance matching and conversion. The AD8369 is a digital variable gain amplifier, which can achieve a maximum signal gain of 100 while maintaining high-frequency characteristics. The high-frequency transformer T4-6T matches the output impedance and converts differential output to single-ended output. Its operating frequency range is 20 kHz to 250 1 Hz, with an insertion loss of approximately 0.3 dB. The AD8029 is a high-frequency, low-power voltage-to-current converter with a quiescent current of only 1.6 mA. It converts the output voltage signal of the AD8369 into a current signal, driving the photodiode MF359 to convert the analog current signal into a modulated optical signal. This signal is then transmitted via optical fiber to the TTI-TIA525 photoelectric converter at the computer end, where it is converted back into an analog voltage signal and sent to the sampling card. The MAX6301 and MC14013 form a power management circuit, similar to a power switch, which activates the power supply to the preamplifier under computer control. The MAX6301 has a maximum quiescent current of 4 μA, and the MC14013 has a maximum quiescent current of 60 μA, resulting in minimal battery drain. When the computer sampling detects a low piezoelectric sensor signal, the AD8369 is automatically set to a higher gain; conversely, when the piezoelectric sensor signal is high, the AD8369 is automatically set to a lower gain. Figure 3 shows the block diagram of the digital control signal fiber optic transmission circuit. 1. Key Chip Selection and Analysis The basic requirements for the first-stage amplifier are: high input impedance, high gain bandwidth, and single power supply. Most differential amplifier integrated circuit chips cannot simultaneously meet these conditions. After comparison, the AD8067 was selected as the first-stage amplifier. It is a high-speed, low-noise FET input, full-amplitude output amplifier with an input impedance of 1000 GΩ; a bandwidth of 54 MHz at a gain of 10 dB; a slew rate of 640 V/μs; a high common-mode rejection ratio (-106 dB); and a maximum DC bias voltage of no more than 1 mV after laser tuning. Its supply current is only 6.5 mA, making it very suitable for single-supply, high dynamic range, and low distortion applications. Resistors R12 and R15 and capacitors C17 and C18 form a filter to filter out unwanted frequency signals. The AD8369 is a single-supply, dB-scale linear digital VGA (Variable Gain Amplifier) ​​with a 45 dB gain adjustment range in 3 dB steps. It operates from low frequencies to 600 MHz, with a bandwidth gain ripple (flatness) of less than 0.15 dB over a ±20 MHz bandwidth within a 380 MHz range. With a 200 Ω input impedance, dynamic input impedance matching achieves a low noise figure of 2.2 dB. At VOUT = 1VP-P, fin = 70 MHz, and a 1 kΩ load, the two-tone third-order intermodulation distortion is -69 dBc. Gain control of the AD8369 is achieved through a digital interface (serial or parallel) to reduce circuit design complexity. The internal circuit structure of the AD8369, as shown in Figure 4, consists of a 7th-order R-2R trapezoidal resistor attenuation network, a fixed-gain amplifier, a 3 dB switched attenuator, a complementary current source output network, a bias circuit, a gain step control circuit, and a digital interface. When the AD8369 is operating, the digital interface (parallel or SPI serial port) receives a 4-bit binary gain control code. The high 3 bits are used to control the transconductance unit and the trapezoidal resistor attenuation network, achieving gain adjustment in 6 dB steps and a maximum of 42 dB. The lowest bit is used to control the 3 dB switching attenuator, which, together with the preceding resistor attenuation network, ultimately achieves digital gain adjustment in 3 dB steps within a 45 dB gain adjustment range. The AD8369's output circuit uses a fully differential configuration with two pairs of complementary current sources, resulting in a differential output impedance of 200 Ω. The voltage gain of the AD8369 can be calculated using the following formula: Where: RL is the external load resistance in Ω; n is the gain control code, with a minimum of 0 and a maximum of 15. Clearly, the parallel value of RL and the AD8369's output resistance jointly determine the AD8369's maximum gain and output signal amplitude, but the magnitude of RL does not affect the chip's gain adjustment step size. Table 1 shows the AD8369's pins and their functions; Table 2 gives the relationship between gain and gain code under different loads. The gain adjustment step size, linear dynamic range, frequency response characteristics, gain adjustment response speed, and noise figure of the AD8369 are crucial parameters for system design. When the gain control code remains constant, increasing the load RL increases the chip's gain. With a constant load resistance, the AD8369 maintains a good linear relationship with the gain control code in 3-dB increments throughout the entire 45 dB gain adjustment range. The AD8369 exhibits a relatively flat amplitude-frequency response from low frequency to 400 MHz. When the load resistance changes from 1 kΩ to 200 Ω, the noise figure decreases by approximately 1.5 dB under the same supply voltage and operating frequency conditions. When the supply voltage increases from 3 V to 5 V, the noise figure increases slightly under the same load and operating frequency conditions. When both the supply voltage and load remain constant, the noise figure increases with increasing operating frequency. However, when the AD8369's operating frequency is below 300 MHz, its noise figure remains below 7 dB. The MF359 is an end-emitting light-emitting diode (LED) specifically designed for 62.5μm/125μm fiber optic transmission. It features a power wavelength of 780 nm, an electrical signal bandwidth of 55 MHz, a digital aperture of 0.275, and a dual-lens structure for optimal coupling of optical power into the fiber. The TTI-TIA525 is a convenient high-speed fiber optic converter with a 125 MHz bandwidth, less than 3 pA/Hz 1/2 RMS noise level, a transconductance range of 14 kDa, a detection wavelength range of 400 nm to 1000 nm, a digital aperture of 0.29, a linear output power greater than 12 mW, a peak-to-peak output voltage of 2 V with a 50 Ω output load, and an output bias voltage of ±1 V. Its output voltage is directly connected to the NIDAQ6155 sampling card. The NIDAQ6155 PCI bus multifunction sampling card features four 12-bit resolution analog input channels, each with a sampling rate of 10 × 10⁶ samples/s and an analog voltage input range of ±200 mV to ±42 V, with each channel equipped with a spoofing filter; two 12-bit resolution analog output channels, each with an output rate of 4 × 10⁶ samples/s; eight digital output interfaces, 5 V TTL/CMOS level; two 24-bit counters; one digital I/O for power-on control; and four digital output interfaces for controlling the gain of the AD8369. 2. Experiments and Conclusions To verify the frequency response characteristics of the developed high-anti-interference digital automatic control gain acoustic emission monitoring system, a standard lead-core breakage signal acquisition and calibration test was conducted using the method recommended by the American Society for Testing and Materials (ASTM). Specifically, a WD acoustic emission sensor was attached to a steel plate made of the same material as the pressure vessel using a coupling agent. A 2H pencil lead with a diameter of 0.5 mm and a length of 3 mm was broken approximately 30 mm away from the acoustic emission sensor. The lead breakage signal was picked up by the acoustic emission sensor and converted into an electrical signal, which was then output to the high-anti-interference digital automatic control gain acoustic emission monitoring system. Finally, the signal was transmitted via optical fiber to a 4-channel PCI-DAQ6155 data acquisition card for signal data acquisition. The acquired data was further analyzed and displayed by computer software. Figure 5 shows the waveforms of the input signal of the preamplifier and the output signal of the TTI-TIA525. Curve ① is the output signal of the TTI-TIA525 when the AD8369 gain code is 1000; curve ② is the input signal of the AD8067. The waveforms show that this preamplifier can amplify the acoustic emission signal with minimal distortion. Field experiments show that the system features high signal output voltage amplitude, wide frequency response range, high sensitivity, strong anti-interference capabilities, and ease of use, making it suitable for condition monitoring of industrial production equipment. The main drawback is the high cost of the six optical fibers. The next step will be to utilize a low-cost, low-power microcomputer, the MSP430F, to control the digital gain adjustable amplifier, the AD8369. By integrating the MSP430F into the preamplifier, the number of optical fibers will be reduced from six to two: one to transmit sensor signals to the computer, and the other to transmit computer control signals to the MSP430F. This will allow the MSP430F to configure the AD8369's gain code and power management, significantly reducing system costs.