Fits with 20 cosine terms give 82 primary eclipse times and unstable secondary times for NN Ser TESS sector 51

We test eclipse timing for NN Ser in public TESS sector 51 light curves. The main analysis uses 20-second SAP flux and QUALITY = 0. It uses paired windows of plus or minus 800 seconds and symmetric templates with 20 cosine terms. Each event has its own intercept and slope. The model assumes uniform integration over each cadence interval. We release 82 primary eclipse times. Their median error is 15.7 seconds when the shared sector template is held fixed. A local line gives an orbital period of 11238.91067 seconds. Its block-deletion error is 0.07074 seconds under the same template assumption. The secondary fit has a local error of 5.44 seconds. Competing minima and block deletion make its time unstable. We test this failure with 2000 independent Gaussian simulations at the actual sampling and weights. The simulations contain no phase signal. Of 1886 fits with an interior maximum, 72.9 percent have local timing errors below 6 seconds. Their median error is 5.03 seconds. In 116 of 2000 simulations, the maximum fit improvement reaches the observed value. This noise model does not include correlated noise or a phase curve. A fixed trapezoid gives only a weak secondary depth measurement. These fits do not establish a planet or orbital eccentricity. The package includes the original FITS files, source hashes, archive query records, event table, code, figures, numeric checks, and reproduction instructions.

NN Ser · TIC 172608008 · TESS sector 51 · eclipse timing · secondary eclipse · Gaussian noise · parameter identification · public data · primary eclipse times