Instrumental Artifacts in Gravitational Wave Ringdown Spectroscopy: Identification, Validation, and EdGB Constraints from GWTC-3

We present an open, reproducible gravitational wave ringdown spectroscopy pipeline applied to 13binary black hole merger events from GWTC-3. Using coherent multi-detector extraction, Bayesianmodel comparison via nested sampling, and a multi-stage systematic validation protocol, weidentify a previously undocumented instrumental bias affecting ringdown analyses ofGW200311_115853. A mains harmonic at 300 Hz, measured at 9.4 times the broadband noise floorin the H1 detector, suppresses whitened strain power near the predicted quasi-normal mode (QNM)frequency of 288.4 Hz, causing the free-frequency model to migrate to 232 Hz and producing aspurious Bayes factor ln B₂₀ = +6.4. We demonstrate this mechanism through three independenttests: spectral line amplitude measurement, a 50-trial time-slide background estimation (findingthat one off-source slide exceeds the on-source value), and a signal injection calibration showingB₂₀ is insensitive to injected GR signal power across all tested SNRs. For GW190521, we find weakevidence for a longer-than-GR damping time (ln B₁₀ = +2.293, δτ/τ = +0.748 [+0.262, +0.978] at90% CI), which is inconsistent in magnitude with any known modified gravity prediction and mostlikely attributable to merger transient contamination. Mapping the GW190521 posterior ontoEinstein-dilaton-Gauss-Bonnet (EdGB) gravity yields a 90% upper limit of √αᴇᴅᴊ < 3.4×10⁷ km,formally valid but deep in the non-linear regime (ζ 18) where the perturbative framework breaks ∼down. General relativity is consistent with all tested events. We provide our pipeline as open-sourcecode and characterise the conditions under which mains harmonics produce false Bayes factorsignals in QNM spectroscopy.