Freezer Sampling and Aliasing
Freezer Sampling and Aliasing
The original field note below is preserved verbatim. Its statement assigning a frequency to a three-second transient is superseded by the corrected periodic-component explanation that follows.
A 10-second freezer schedule samples at 0.1 Hz, giving a Nyquist boundary of 0.05 Hz. To illustrate aliasing, assume a repeating component with a three-second period: its frequency is about 0.333 Hz, and sampling it at 0.1 Hz can make it appear near 0.033 Hz, a false slow wobble lasting about thirty seconds. Sampling that component requires more than twice its frequency: the boundary is about 0.667 Hz, or one sample every 1.5 seconds.
A three-second door transient is not necessarily a repeating three-second signal. Its duration alone does not establish a single frequency or a sufficient sampling rate; the node must observe the event dynamics that matter. This distinction is why the sampling floor and any door-open trigger belong in the contract.
For conversion resolution, a typical 12-bit, 3.3 V analog-to-digital converter (ADC) has a 0.806 mV step and a 74.0 dB ideal quantization signal-to-noise ceiling for a full-scale sine wave. These ideal conversion figures do not compensate for a missed event.
Technical boundaries. The aliasing calculation assumes a repeating component with a three-second period; a three-second door transient does not establish a single frequency or sufficient sampling rate. The ADC ceiling is ideal and applies to a full-scale sine wave.