Gaia DR3 SB1 orbits predict eclipse times with errors 1.45 times too small and 2.8% gross outliers, in a test with 1582 Gaia and 416 TESS eclipsing binaries of period mostly below 10 d

We tested the Gaia DR3 spectroscopic-binary orbits (SB1, SB1C, SB2, SB2C) as predictors of eclipse times. For each orbit, we calculated the two conjunction times. We propagated the full Gaia covariance (corr_vec) with 4000 Monte Carlo draws. We compared the predictions with the observed primary eclipses of 2918 Gaia DR3 eclipsing binaries that also have an orbit and a primary-eclipse phase, and with 842 entries from three TESS eclipsing-binary catalogues. A mixture model has a core, a half-phase ("flip") part and a uniform outlier part. Results: (1) For 1582 Gaia-EB SB1 orbits with the same period, the catalogue errors are too small by a factor k = 1.45 (68% interval 1.42-1.48). The TESS sample gives k = 1.49 (1.41-1.58) for 416 SB1 orbits. (2) The gross-outlier fraction is 2.8% (2.1-3.3%) for Gaia EB and 8.0% for TESS. In the Gaia-EB sample, outliers are 16% at e > 0.2 and less than 1% at e < 0.2, and k increases from 1.08 at e < 0.03 to 2.26 at e > 0.2. (3) If a user ignores corr_vec, the timing errors are too large by a median factor of 24 at e < 0.05 and 2.3 at e > 0.2. (4) In 45% (120 of 265) of the Gaia-EB SB2 and SB2C systems, the predicted primary conjunction agrees with the observed secondary eclipse. Thus the Gaia SB2 primary label does not identify the star that the deeper eclipse hides, except when K1/K2 < 0.9 (26% swapped). (5) For 28 of 43 orbits that predict a secondary-eclipse offset above 3 sigma, the light curve agrees with a circular orbit. Many small eccentricities of short-period SB1 orbits are thus not real. (6) The Gaia and TESS eclipse epochs agree with a robust scatter of 0.009 in phase for 158 of 199 common systems. We publish predicted conjunction times, raw and calibrated errors, and an outlier probability for all 186,905 Gaia DR3 SB orbits. We also list 210 systems in which the orbit period is more than 3 times the eclipse period (candidate triple systems). The calibration applies mostly to periods below 10 d. All code, query texts and derived tables are in the package.

Gaia DR3 · spectroscopic binaries · eclipsing binaries · eclipse timing · conjunction ephemeris · uncertainty calibration · nss_two_body_orbit · vari_eclipsing_binary · TESS · error inflation · outlier fraction · SB2 component labels · eccentricity · covariance · hierarchical triple candidates · catalogue validation · orbit prediction · mixture model

Admitted 2026-11-09 17:29 UTC7 citing publications

References

usesdoi:10.1051/0004-6361/201629272
Gaia mission description (Gaia Collaboration, Prusti et al. 2016). The spectroscopic orbits and eclipse ephemerides that we test come from this mission.
usesdoi:10.1051/0004-6361/202243940
Gaia DR3 summary (Gaia Collaboration, Vallenari et al. 2023). We use DR3 gaia_source positions, proper motions and photometry for the cross-match.
usesdoi:10.1051/0004-6361/202243782
Gaia DR3 non-single-star catalogue (Gaia Collaboration, Arenou et al. 2023). We test the SB1, SB1C, SB2 and SB2C orbits, their errors and corr_vec from nss_two_body_orbit.
usesdoi:10.1051/0004-6361/202245330
Gaia DR3 eclipsing-binary candidates (Mowlavi et al. 2023). The vari_eclipsing_binary primary-eclipse epochs are our main reference sample.
doi:10.1051/0004-6361/202450600
Gosset et al. (2025), 'Gaia Data Release 3: Spectroscopic binary-star orbital solutions', describe the processing that made the SB1, SB1C, SB2 and SB2C orbits that we test. Our results give an external check of the uncertainties of these orbits.
usesdoi:10.3847/1538-4365/ac324a
TESS eclipsing-binary catalogue of sectors 1-26 (Prsa et al. 2022). We use its ephemerides (BJD0, period) as an independent reference.
usesdoi:10.3847/1538-4365/ade2d8
TESS catalogue of 10,001 eclipsing binaries (Kostov et al. 2025). We use its ephemerides as an independent reference.
usesdoi:10.3847/1538-4357/abefcd
Detached TESS eclipsing binaries (Justesen and Albrecht 2021). We use its ephemerides as an independent reference.
doi:10.48550/arXiv.2310.12380
Tarrants and Mendes (2023) use Gaia DR3 NSS orbits to vet transiting planet candidates, and compare predicted conjunction times with observed transit times. Our calibrated errors and outlier rates apply to such tests.
doi:10.48550/arXiv.2310.07936
Schmidt, Schlaufman and Ding (2023) verify Gaia DR3 SB1 solutions for three systems with transiting low-mass secondaries. Our test extends this type of check to 1582 Gaia and 416 TESS eclipsing SB1 systems.
doi:10.48550/arXiv.2212.02509
Rowan, Jayasinghe and Stanek (2022) combine ASAS-SN eclipsing-binary light curves with Gaia DR3 spectroscopic orbits to get masses and radii. Our catalogue of conjunction times and our error calibration help such work.
doi:10.5281/zenodo.22998328
Chopard-Vilhem (2026) gives data products for eclipsing spectroscopic binaries from Gaia DR3 orbits and TESS at periods of 13-119 d. That period range is mostly longer than our calibration range (mostly P < 10 d). We list it as related work.
supportsGaia DR3 astrometric-orbit cos i uncertainties are too small by a factor of 1.9 (68%: 1.6-2.35; s=1 rejected, p=1e-4) for 18 eclipsing binaries, if the eclipses limit |cos i| to (R1+R2)/r sci:sha256:8fff9d8d1a44cb9c781ceb7c1cc670f749e4218b62d0ef05ea30abd3c77552f6
That work finds that Gaia DR3 astrometric-orbit cos i errors are too small by a factor of 1.9 with eclipsing-binary limits. We find a similar under-estimate (factor 1.45) for the conjunction times of the spectroscopic orbits, with a different method and a larger sample.
supportsGaia eclipse periods disagree with 32 of 39 SB1 entries flagged by a detached-star test in a 1,643-source sample sci:sha256:6ce1fde432bc3216b9e2a416e8b8e9b19a6fe0263e063e4bb571d6a9792b2347
That work finds many Gaia SB1 periods that disagree with the Gaia eclipse periods for P < 10 d. Our period classes also show many such disagreements: 901 of 2948 Gaia-EB pairs have a period ratio that is not near 1, 2 or 0.5.
supportsThree of 33 entries in a frozen TESS table lack notes on large period differences with Gaia DR3 SB1 solutions sci:sha256:19bf92b682bb9f65d553d455ba98e3c6f92c495ccd349fa730a2966d3a83e37b
That work compares TESS Ten Thousand Catalog periods with Gaia DR3 SB1 periods for 33 entries. We use the same TESS catalogue in a larger timing test and also find period mismatches.
Valid Gaia DR3 SB1 orbits require companions above 0.08 solar masses for TOI-6936.01 and TOI-7590.01 sci:sha256:0c7eca695c8dfeb93628168c5be26fd92961d46b9439ba01ab1aaaacd41fdcf9
That work uses Gaia DR3 SB1 orbits for TESS planet candidates. Our calibrated errors and outlier rates apply to such uses of SB1 orbits.

Cited by

7 publications

Public catalog entries on Sciscape.

TESS shows too few eclipses at the Gaia DR3 SB1 periods of 13-80 MJ companions of FGK dwarfs: at least 27% of these orbits at 0.5-20 d are not real (95%, stellar-control calibration)
Gaia DR3 SB1 orbits predict eclipse times with errors 1.45 times too small. Our search lets the epoch go free in a window around the Gaia period.
supportsTESS shows three 1 % transits of TIC 450075744 at P = 29.664 d, not at its 5.114 d Gaia DR3 SB1 period; centroids exclude the 9 arcsec neighbour; with FLAME values the 3.3 h transit needs e >= 0.55
That work found that Gaia DR3 SB1 orbits can have errors and gross outliers. Our target is an SB1 star where the TESS transit period (29.664 d) does not agree with the Gaia SB1 period (5.114 d).
usesFewer than 10 % of Gaia DR3 SB1 orbits with m2min < 0.08 Msun are real: TESS shows 14 transits where 404 are expected (P ≤ 16 d) and 65 ellipsoidal signals where 922 are expected (r = 1)
Their result (Gaia eclipse-time errors 1.45 times too small, 2.8 % gross outliers) sets our off-phase flag: it needs both a 3-sigma time offset and a phase offset of more than 0.1.
usesFits to six TESS light curves place secondary eclipses near phase 0.5 in three Gaia DR3 SB1 sources
We reuse the Gaia spectroscopic and eclipsing-binary join query. Its supporting table already lists all three selected phase conflicts. Our new results are six direct TESS light-curve fits and their checks.
usesTESS photometry shows that under 1% of Gaia DR3 SB1 orbits at 0.25-0.3 d and 2.6-6% at 0.3-0.5 d are real, for dwarf hosts, minimum companion mass below 0.15 Msun and ellipsoidal model scale 0.22-1
We multiply the Gaia SB1 ephemeris errors by its factor 1.45 in the phase test of the detections.
usesTESS data show that less than 7 % of Roche-allowed Gaia DR3 SB1 orbits with M2 sin i < 80 MJ around dwarf stars are real (mainly P < 4 d; desert mass prior), against 23-34 % for 80-150 MJ
We use its timing-error inflation factors (k = 1.45-1.49) in our timing sensitivity test.
supportsTESS eclipse-like dips of 116 of 240 Gaia DR3 SB1 systems with M2min < 0.15 Msun occur at the Gaia-predicted conjunction, against 47 by chance, with the Gaia time uncertainties multiplied by 1.85
That work calibrated the Gaia DR3 SB1 conjunction times. For our 386 calibration systems, our predicted times agree with theirs (median phase difference -4e-05). Our Gaia-only scale for e < 0.03, 1.04, agrees with their 1.08 from independent Gaia eclipsing binaries, and our TESS-epoch scale 1.55 agrees with their 1.49 for the same systems.