Signed traces survive.
The phase gate fails.

Retain complex phase before squaring.

Each channel’s fixed carrier comes from its previously explored record zero. Records 1–99 are evaluated without carrier retuning, using a ±5 kHz positive-frequency band. The records vary in drive-frequency metadata; they are not identical repeated trials, and the extension is not blinded or independently preregistered.

The signed finite-DFT analytic representation round-trips all evaluated records with maximum absolute error 1.5544×10⁻¹⁵ V. Demodulation retains the complex envelope. Taking its magnitude or square discards sign and phase.

The failed threshold stays frozen.

The gate requires signed RMS residual ≤5% and phase-floor coverage ≥95%. Every record fails the composite test; all interior phase fractions are zero. Median RMS residuals range from 4.26% to 9.86% by channel.

The phase floor uses the maximum complex residual over the entire post-trigger window. Its conservatism can retain endpoint effects despite an interior guard. This run does not identify which contribution dominates, and the gate is not loosened after observing failure.

A numerical residual has a defined reference.

The report gives explicit amplitude, phase and squared-envelope truncation inequalities relative to the full finite-DFT analytic signal. These bounds do not include instrument gain, timing skew, nonperiodic boundary error or a complete physical energy map.

Five synthetic controls pass. Amplitude and squared-envelope inequalities are checked on all 891 records; empirical accepted-point phase checks have no eligible points. Independent review is pending. Development — no registry entry.

Next: freeze a boundary-treatment comparison against synthetic transients and nonperiodic carriers, then distinguish endpoint effects from omitted bands. Absolute native support and energy calibration remain open.