Aims. In this paper we study the impact of wavefront sensing with an ELT-scale elongated LGS using on-sky data obtained with the AO demonstrator CANARY on the William Herschel telescope (WHT) and the ESO Wendelstein LGS unit. CANARY simultaneously observed a natural guide star and a superimposed LGS launched from a telescope placed 40 m away from the WHT pupil.
Methods. Comparison of the wavefronts measured with each guide star allows the determination of an error breakdown of the elongated LGS wavefront sensing. With this error breakdown, we isolate the contribution of the LGS elongation and study its impact. We also investigate the effects of truncation or undersampling of the LGS spots.
Results. We successfully used the elongated LGS wavefront sensor (WFS) to drive the AO loop during on-sky operations, but it necessitated regular calibrations of the non-common path aberrations on the LGS WFS arm. In the off-line processing of the data collected on-sky we separate the error term encapsulating the impact of LGS elongation in a dynamic and quasi-static component. We measure errors varying from 0 nm to 160 nm rms for the dynamic error and we are able to link it to turbulence strength and spot elongation. The quasi-static errors are significant and vary between 20 nm and 200 nm rms depending on the conditions. They also increase by as much as 70 nm over the course of 10 m. We do not observe any impact when undersampling the spots with pixel scales as large as 1.95″, while the LGS spot full width half maximum varies from 1.7″ to 2.2″; however, significant errors appear when truncating the spots. These errors appear for fields of view smaller than 10.4″ to 15.6″, depending on the spots’ elongations. Translated to the ELT observing at zenith, elongations as long as 23.5″ must be accommodated, corresponding to a field of view of 16.3″ if the most elongated spots are put across the diagonal of the subaperture.