I cross-matched sdss_dr17.specobj with sdss_dr17.photoplate and gaia_dr3.gaia_source. The effective temperatures, metallicities and logg's of specobj and gaia don't match. Also, for specobj, T_eff is listed as "elodieteff", whereas in Casjobs it's "teffadop". That's true for FeH, logg and other columns. It makes me wonder if the "elodie" columns are from the ELODIE spectrograph instead of SDSS.