Typical approaches to monitoring in the machining process involve frequent readings from sensors such as vibration, acoustic emission, and cutting force sensors. In many cases, though, the sampling frequency of controller/plant historian-based spindle load signals is much lower than what is used for high frequency sensors. The focus of this paper is thus on machining-process monitoring using slow sampling spindle-load signal collected by controllers and plant historians, based on a titanium machining process in aerospace industry. For this case study, there are a total of 18 synchronized spindle loads of Program 1543, Tool 6040, and Machine T5-1, in which Run 237 and Run 241 are flagged as abnormal with onset at approximately 189 s and 303 s. These 18 spindle-load trajectories are characterized via functional principal components analysis, and are evaluated using a combination of outlier detection in the score space along with reconstruction residuals. The proposed framework allows one to retain the simplicity of using a low-rate controller for monitoring while switching from point-wise density check to trajectory-wise modeling of the machining process.