Transit photometry run on Sombra: five hours of continuous 30-second exposures either side of a predicted ingress, defocused deliberately so the star spreads across enough pixels to average out the flat-field error.
Raw light curve looked like noise. It always does. After differential correction against four comparison stars in the field the dip is unambiguous — about three millimagnitudes, ninety minutes wide, exactly where the ephemeris said.
The lesson I try to get across on residencies is that this is what detection looks like: not a picture, a plot, and one you have to work to trust.
Logged by Wren Okonkwo, visiting fellow.