Image depth cuts change the 2018–2025 ZTF period fit for J1813+4251 with a fixed primary eclipse shape

We test the eclipse clock of the known 51-minute binary ZTF J1813+4251 with public ZTF detections from 2018 to 2025. We fit an exposure-averaged primary eclipse and let the broad orbital profile change by filter and year. The full sample has 2,666 detections. A subset has 1,632 detections from images with limiting magnitude greater than 21.0. The linear periods differ by 0.281 milliseconds. Paired resampling of occupied 30-day blocks gives a conditional 95% range of 0.048 to 0.539 milliseconds for the full-minus-subset difference. The depth threshold was selected after inspection. Removing one shallow-image detection lowers the full period by 0.124 milliseconds. The deep-image linear period is 3069.64400269 seconds. Its quadratic fit does not give a secure decay measurement. Injections at the actual dates show low sensitivity to the published rate from gravitational radiation alone. These results depend on the fixed primary shape, detection sample, and noise assumptions. They do not identify the cause of the period shift. The package preserves the source response, time audit, code, fits, figures, simulations, and reproduction steps.

ZTF J1813+4251 · eclipsing binary · cataclysmic variable · eclipse timing · Zwicky Transient Facility · image depth · data quality · orbital period