We present three complete seasons and two half-seasons of SDSS r -band photometry of the gravitationally lensed quasar SBS 0909+532 from the U.S .
Naval Observatory , as well as two seasons each of SDSS g -band and r -band monitoring from the Liverpool Robotic Telescope .
Using Monte Carlo simulations to simultaneously measure the system ’ s time delay and model the r -band microlensing variability , we confirm and significantly refine the precision of the system ’ s time delay to \Delta t _ { AB } = 50 ^ { +2 } _ { -4 } \textrm { days } , where the stated uncertainties represent the bounds of the formal 1 \sigma confidence interval .
There may be a conflict between the time delay measurement and a lens consisting of a single galaxy .
While models based on the Hubble Space Telescope astrometry and a relatively compact stellar distribution can reproduce the observed delay , the models have somewhat less dark matter than we would typically expect .
We also carry out a joint analysis of the microlensing variability in the r - and g -bands to constrain the size of the quasar ’ s continuum source at these wavelengths , obtaining \log \ { ( r _ { s,r } / \textrm { cm } ) [ \cos { i } / 0.5 ] ^ { 1 / 2 } \ } = 15.3 \pm 0.3 and \log \ { ( r _ { s,g } / \textrm { cm } ) [ \cos { i } / 0.5 ] ^ { 1 / 2 } \ } = 14.8 \pm 0.9 , respectively .
Our current results do not formally constrain the temperature profile of the accretion disk but are consistent with the expectations of standard thin disk theory .