A TESS model gives a conditional 95% upper bound of 3.1% on HD 74423's mean pulse pattern change from 2019 to 2025

We test the full complex orbit modulation of the main pulse in HD 74423. We use public TESS data and archived fits for 12 sectors from 2019 to 2025. The endpoint comparison uses 82,347 exposures. Each sector has its own pulse frequency. A model with 11 main pulse peaks gives ten complex side-peak coefficients after the central pulse amplitude and phase are removed. The orbit light signal sets the phase reference. We compare visit means with equal weight for each sector. The estimated relative root-mean-square complex pattern difference is 0.800%. A joint Gaussian confidence ellipsoid gives a conservative conditional 95% upper bound of 3.025%, or 3.1% when rounded up. The zero-change Wald statistic is 17.557 in 20 dimensions, with nominal p = 0.617. SAP flux gives an upper bound of 4.201%. A model with 15 pulse peaks gives 3.518%. Tested baseline models, orbit terms, and block lengths give no clear endpoint change. Some within-visit checks depend strongly on processing and block length. The limit applies to two visit means under the estimated residual covariance model. It does not prove continuous stability or a fixed stellar surface pattern. This extends a prior amplitude-ratio test to the full main pulse pattern, including phase. The same observing sample supplies both studies. Original FITS files, source records, fitting and analysis code, full covariance matrices, figures, numerical checks, and repeat instructions are included. Fits rebuilt from the first and last sector inputs match the archived coefficients and covariance matrices exactly.

HD 74423 · TIC 355151781 · TESS · stellar pulsation · binary stars · tidally trapped pulsations · complex amplitude · pattern stability · confidence bound · reproducible analysis