Surface Electromyography (sEMG) is a biological potential signal to measure and record the electrical activity of muscles near the surface of the skin. By analyzing sEMG signals, it is possible to identify the contraction and relaxation patterns of the muscles, and thus discriminate the gestures that generate these patterns. However, obtaining sEMG signals is by no means an easy task, due to their weak and high-impedance source signals, especially when using dry electrodes. It is critical to use high-quality sEMG signals to recognize the patterns. In this paper, we present a compact sEMG acquisition method. Most of the existing surface signal acquisition is designed with integrated operational amplifiers or instrument amplifiers, requiring additional hardware filtering circuits, making optimizing the volume and further reducing the power consumption challenging. At the same time, the separation of components will use more intermediate wires, which will cause more bloated volume and more interference. In addressing these challenges, a sEMG acquisition system based on KS1801 and STM32F103 is designed in this paper. Specifically, a sEMG acquisition sensor circuit based on Bioelectric signal acquisition analog front-end chip KS1801 is designed better to amplify the high-resistance weak signal from the dry electrode, reduce the circuit’s complexity, and reduce the system’s size. Furthermore, digital high-pass, digital low-pass, and 50 Hz digital notch filters are applied to eliminate signal noise effectively. A dual-buffer storage structure was implemented to temporarily store filtered data, providing ample time for subsequent complex recognition algorithms to process the collected data. A Bluetooth module was integrated to facilitate the transmission of cached data to a PC. Experimental results demonstrate the effectiveness of the sEMG acquisition system in reliably capturing signals from dry electrodes. It exhibits notable advantages, including robust anti-interference capabilities, compact size, stable performance, and precise signal differentiation.