Check: Sampling evidence and irreversibility

Knowledge check · self-test
Check: Sampling evidence and irreversibility
Predict the Spectrum Before Capture. Label each sampling spectrum with its rate and filter setting. Without those two values, a 12 Hz peak cannot show whether the capture protected its frequency identity. Clock stability belongs to the sampling result as well. If sample spacing wanders, the frequency axis also becomes uncertain. A timestamped capture can reveal dropped or late samples that a plain array of ADC values hides. Predict: The sample rate changes from 50 to 60 samples/s. Where does the 38 Hz ripple appear? Check: It folds to |(38-60)| $=22\ \text$; a real 12 Hz shaft motion stays at 12 Hz. Predict: A digital moving average makes the false 12 Hz peak smaller. Does that prove the peak was harmless? Check: No. The ADC had already aliased 38 Hz, so attenuation does not restore the lost frequency identity. Feed in a known tone below and above the planned limit. Change the sample rate and filter, then retain the source frequency, settings, samples, spectrum, error, and decision.
Predict the Spectrum Before Capture. Label each sampling spectrum with its rate and filter setting. Without those two values, a 12 Hz peak cannot show whether the capture protected its frequency identity. Clock stability belongs to the sampling result as well. If sample spacing wanders, the frequency axis also becomes uncertain. A timestamped capture can reveal dropped or late samples that a plain array of ADC values hides. Predict: The sample rate changes from 50 to 60 samples/s. Where does the 38 Hz ripple appear? Check: It folds to |(38-60)| $=22\ \text$; a real 12 Hz shaft motion stays at 12 Hz. Predict: A digital moving average makes the false 12 Hz peak smaller. Does that prove the peak was harmless? Check: No. The ADC had already aliased 38 Hz, so attenuation does not restore the lost frequency identity. Feed in a known tone below and above the planned limit. Change the sample rate and filter, then retain the source frequency, settings, samples, spectrum, error, and decision.
Back to Sampling and Anti-Alias Limits: Aliasing and Oversampling Limits