Abstract: To address curve drift, poor real-time performance, and high false-alarm rates in online tool wear monitoring for unmanned CNC machining lines, this study develops a novel monitoring system based on spindle current signals. A compact 3D-packaged hardware acquisition circuit is designed to enable high-fidelity acquisition of cutting load signals. In addition, an FFT-based frequency-domain feature extraction method and a PMC-instruction-based synchronization correction strategy are introduced to achieve accurate curve alignment and robust condition monitoring. Continuous machining experiments were conducted on a Doosan PM240 CNC lathe. The results show that the proposed system effectively suppresses lateral curve drift, reduces hardware cost by approximately 60%, and lowers the false-alarm rate from 10.15% to 1.47% compared with the conventional method. The system can reliably identify abnormal conditions such as tool wear, chipping, and tool breakage. Owing to its low cost, high robustness, and easy deployment, the proposed system is well suited for engineering applications in automated machining of mechanical structural components.