Design and Development of a Low-Cost sEMG Acquisition System with Adaptive DSP Pipeline

Abstract

Surface electromyography (sEMG) is a widely used non-invasive technique for capturing muscle electrical activity, with applications in prosthetics, rehabilitation, and gesture-based human-computer interaction. Progress in low-cost sEMG research is hindered by the high cost of laboratory-grade acquisition hardware. This paper presents the design and development of a low-cost sEMG acquisition system validated on hand grip recordings from six healthy adult subjects (S1–S6). The system integrates a custom Analog Front-End (AFE) using an AD620 instrumentation amplifier and LM324 operational amplifiers with an ESP32 microcontroller for real-time digitization and wireless transmission. A six-stage digital signal processing (DSP) pipeline — comprising DC removal, adaptive notch filtering, fourth-order Butterworth bandpass filtering (20–450 Hz), Daubechies-4 wavelet denoising (VisuShrink, Level 5), full-wave rectification, and linear envelope extraction — is applied offline. Across all six subjects, M1 SNR ranges from 50.2 to 60.6 dB (mean ± SD: 56.1 ± 4.3 dB), M3 SNR ranges from 49.1 to 58.6 dB (mean ± SD: 54.8 ± 4.1 dB), and noise floor RMS remains below 0.215 µV in every session — well below the 5 µV hardware specification benchmark. The total system cost is below NPR 15,000, demonstrating that research-grade sEMG quality is achievable with off-the-shelf components across a diverse set of subjects.