A fixed column and frame model has errors up to 0.0173 seconds in 32 TESS sector 14 and 15 file pairs

We test a fixed numerical model of QLP v01 minus TESS-SPOC v1 times in TESS sectors 14 and 15. The model combines a published CCD baseline with a frame rotation and a change in position columns. Its third apparent position entry is the numerical value of a light travel time in days. This is a numerical failure model, not a physically valid position. We form the rule from earlier public files and save the formula, detector and orbit gate, sample rule, and test before acquisition of new times. The new set has 32 target and sector pairs, 28 distinct stars, and 39,008 finite exposure matches. Each sector covers all 16 camera and CCD combinations. No star offset is fitted. The RMS errors are 0.005508 and 0.003455 seconds. The largest absolute errors are 0.017286 and 0.010669 seconds. Both sectors fail the saved requirement that every error be below 0.010 seconds. There are 2,152 and 129 failed exposure pairs in three target and sector pairs. Both zero-quality subsets also fail. The new data confirm an earlier gate exception on camera 3, CCD 1, orbit 37 in sector 15. A QLP-only transfer gives the same residuals within 0.520 microseconds across the complete sample. An earlier sector 16 control has errors below 0.004525 seconds. The result supplies a compact relation and a tested transfer for the stated relative file times. It does not establish an absolute clock error, a production software cause, or a new planet. The package preserves 160 complete source FITS files, the public ephemeris, source hashes, selection and protocol records, code, all results, figures, and offline checks.

TESS · QLP · TESS-SPOC · barycentric time · archive calibration · coordinate frame · ephemeris · reproducible research