1 Introduction In addition to industrial cleaning, high-power ultrasonic devices have broad application prospects in medical, military, oil transducer technology, marine exploration and development, noise reduction and vibration reduction systems, intelligent robots, and wave-driven oil production [1]. Ultrasonic devices consist of ultrasonic inverter power supplies and transducers. In recent years, the development and research of new rare earth functional materials have made it possible to manufacture high-power ultrasonic transducers, but high-frequency sinusoidal inverter power supplies that match them are still rare. At present, the high-power sinusoidal inverter power supplies on the market are all low-frequency products made of IGBTs [2], while most high-frequency inverter power supplies are square wave power supplies or pulse inverter power supplies with adjustable duty cycles. Therefore, high-frequency high-power sinusoidal inverter power supplies have become a bottleneck for ultrasonic applications, making the development of such power supplies an urgent problem to be solved. Here, the development of high-performance, high-power sinusoidal ultrasonic inverter power supplies is studied by applying hybrid pulse width modulation (HPWM) control technology, using MOSFET parallel operation mode, and using a single-chip microcomputer to form an intelligent control system. 2 System Composition The inverter power supply used for high-performance, high-power sinusoidal ultrasonic waves has a frequency of 25kHz and a power of 4.5kW. The voltage requirement is adjustable between 0 and 200V, and the frequency requirement is adjustable between 10 and 25kHz. 2.1 Design of the Scheme Figure 1 shows the system hardware composition block diagram of the inverter power supply [3]. It consists of two main parts: AC/DC and DC/AC. It includes the main circuit of AC-DC-AC, drive circuit, single-chip microcomputer control system, low-pass filter, display and protection, etc. The main circuit is directly powered by 220V mains power. The single-phase AC voltage is converted into DC by the thyristor constant current and constant voltage control module to provide a constant DC voltage for the inverter. In order to make the inverter obtain a sinusoidal output with better performance and waveform, a large carrier ratio is required. Since its carrier signal will reach 400-600kHz, only MOSFET can be selected as the switching device. However, the output power of MOSFET is small. In order to increase the output power, the MOSFET can be connected in parallel to solve the contradiction between high frequency and high power. The inverter section uses a constant-frequency triangular carrier signal to asynchronously modulate the input sine wave. The control method employs HPWM technology, converting the DC voltage into a series of equal-amplitude pulse signals. The amplitude and pulse width of these pulse signals are always proportional to the modulated sine wave. After these pulse signals pass through a low-pass filter to remove the high-frequency carrier signal, a sine wave output with the same frequency as the modulated wave can be obtained. Therefore, by simply changing the input modulated wave, an amplitude-adjustable frequency-converting sine wave output can be easily achieved. 2.2 Microcontroller Control System The power supply uses an 80C196MC microcontroller specifically designed for controlling inverters as the control core of the inverter [4, 5]. The waveform generator WFG inside the 80C196MC microcontroller occupies very little CPU time and can directly output 4 PWM signals from the P6 port for driving the inverter. The minimum microcomputer system, composed of the 80C196MC and EPROM2764, will complete the setting of the ultrasonic frequency and voltage. The system sets the carrier frequency and simulates the output of a unipolar sinusoidal constant amplitude pulse width modulation (HPWM) signal. It can display voltage amplitude and frequency, as well as provide power supply protection control. 2.3 Inverter Main Circuit and HPWM Control Method At high frequencies, the switching losses of the power transistors are extremely high, making the devices prone to damage and limiting power output. The key technical challenge of this power supply is how to obtain a high-power frequency-converting sinusoidal output under high-frequency conditions. That is, the difficulty of the inverter lies in how to reduce the switching losses of the transistors, significantly decreasing the du/dt and di/dt stresses to achieve high-frequency inversion. To achieve these goals, the inverter main circuit adopts a single-phase full-bridge topology that easily implements soft-switching technology. HPWM control is used in the control method. Figure 2 shows the inverter's main circuit topology. Figure 3 shows the drive signals of the four transistors and the inverter's output signal. The essence of the HPWM control method is still a unipolar SPWM control method. The inverter bridge output receives a three-state output voltage waveform. During the positive half-cycle of the output voltage, the pulse signal generated by the intersection of the sinusoidal modulated wave and the triangular carrier wave controls the high-frequency complementary switching of the VS1 and VS3 bridge arms; and controls the low-frequency complementary switching of the VS2 and VS4 bridge arms, i.e., VS2 is off and VS4 is on. During the negative half-cycle of the output voltage, the operating states of the two bridge arms are interchanged. VS1 is always off, VS3 is always on, and VS2 and VS4 operate in high-frequency modulation mode. In HPWM control, two power transistors always operate at low frequencies, which reduces switching losses overall, which is extremely beneficial for increasing power at high frequencies. Compared with the general SPWM control method, the two bridge arms alternately operate in low-frequency and high-frequency states in HPWM, making the operation of the two bridge arms symmetrical and the operating state of the power transistors balanced. This will extend the life of the power transistors, increase the reliability of the entire circuit, and has the advantages of high voltage utilization, low harmonic content, and low switching losses. Since each switching transistor is connected in parallel with a capacitor, with appropriate selection of filter inductor parameters, the circuit can easily achieve zero-voltage switching (ZVS) of the switching transistors, greatly reducing the stress on du/dt and di/dt, and fully realizing high-frequency high-power inversion. 2.4 Drive Circuit The drive circuit of the switching transistors can use the latest LM5111 driver. It adopts an SOIC-8-pin package and provides independent ground and reference voltage pins for the input and output stages to support gate drive configurations with separate power supplies. The peak output current of the LM5111 chip is up to 5A. The two 5A current drive channels of the LM5111 can be independent or connected in parallel to increase the peak output drive current to 10A, so as to drive very high-power MOSFETs with extremely high efficiency. The LM5111 operates at a frequency of up to 1MHz, and its turn-on and turn-off delays are small, at 12ns and 14ns respectively. It can fully meet the requirements of this power supply. 3 Software Implementation 3.1 Main Program The drive circuit of the switching transistors can use the latest LM5111 driver. It uses an SOIC-8 pin package and provides input and output. Figure 4 shows the main program flowchart. It includes an initialization subroutine, an HPWM signal generation subroutine, a keyboard scanning subroutine, and a display subroutine. In the initialization subroutine, the 80C196MC initializes parameters such as the stack address and carrier frequency, and sets the operating modes of the microcontroller's various I/O ports, interrupts, and waveform generator. The desired output sine wave frequency can be given via the keyboard and displayed by the display program. The display subroutine can perform timed sampling of the voltage signal, and after A/D conversion, dynamically and time-divisionally display the frequency and amplitude values ​​of the sine wave. 3.2 HPWM Signal Generation Subroutine HPWM is generated by comparing a sine modulated wave with a triangular carrier wave of equal amplitude. In center-aligned mode, the counting process of WG-COUNTER in the waveform generator forms a virtual triangular carrier wave. The sine modulated wave can be implemented using a lookup table method. Since the output HPWM wave is symmetrical, only a sine function table from 0° to 180° needs to be established. To achieve sufficient resolution, a sampling point is arranged every 0.15° in the sine function table. Each data point has a 15-bit binary value, occupying 2 bytes. A total of 1200 data points are taken for half a cycle of the output sine wave, stored in the memory area starting at SIGN. Assuming the carrier frequency is fc and the output frequency is fo, N = fc/fo intersection values ​​are needed in each half-cycle of the output sine wave. The sine modulation amplitude corresponding to the i-th intersection point can be obtained by looking up the table, with its address being SIGN + 1200i/N. The triangular carrier wave is compared with the sine modulation amplitude at the intersection point to obtain the HPWM switching mode of the inverter. Whenever the triangular carrier wave reaches its peak (WG - COUNTER = WG - RELOAD) or trough, an interrupt request is sent to the microcontroller to load data. This process is repeated, swapping two phases every half cycle to obtain a hybrid unipolar modulation HPWM wave. Figure 5 shows the flowchart of the HPWM signal generation subroutine. 4. Experimental Results Using the above main circuit structure and control method, a prototype with an output frequency of 25kHz, a carrier frequency of 600kHz, and an output power of 4.5kW was developed. Figures 6a and 6b show the experimental waveforms of the filter inductor current iL and the output voltage uo under inductive half-load and inductive full-load conditions. As can be seen from the figures, the change in uo is small under half-load and full-load conditions, indicating a good load regulation rate. Figures 6c and 6d show the driving voltages ugsVS2 and ugsVS4 of the two power transistors VS2 and VS4 in the same bridge arm of the inverter bridge, and their amplified experimental waveforms. It can be seen that, considering the dead time, the two power transistors in the same bridge arm are complementary in conduction. The rise and fall delay times of the driving voltage waveforms of the power transistors are very short, meeting the requirements. 5. Conclusion The high-power ultrasonic power supply using a single-chip microcomputer intelligent control system allows for manual setting of the power supply frequency and output voltage. The HPWM signal output by the microcontroller simplifies the hardware circuit and significantly improves the system's power factor and efficiency. Simultaneously, the use of HPWM control and ZVS resonant soft-switching technology reduces switching transistor losses, suppresses high-order harmonics, and minimizes transducer losses. Experiments show that the proposed power supply exhibits excellent performance, convenient adjustment, and high reliability. It can provide a high-performance ultrasonic power supply for high-power ultrasonic transducers in various fields.