During a PEM injection measurement, some line noises (also existing without injection) were moving depending on time and causing the mis-projection accidentally.
I checked the silent run data to investigate the behaviors of such lines.

I made a reference ASD percentile (median, 10%, and 90%) from the first 5 minutes (black) and added many 10-second ASDs (FFT length=2s, overlap=50%, Welch's method) for every 1 minute.
Such line noises are around 50, 83, 248, 271, 300, 330, 393, 486, 540, 600, 660, 780, and 900 Hz.