2 Experimental setup

Following the theoretical work which established the feasibility and imaging properties of hyper-telescopes, verifications were first made through computer simulations, followed by a laboratory system and a miniature hyper-telescope tested on the sky.

Figure 3: Laboratory images of multiple artificial stars obtained with the hyper-telescope. Several pin-holes, located very close to a quartz-iodine light bulb simulated a triple (left) and a sextuple (right) star. Both snapshot images were directly recorded with a CCD camera without de-convolution or image enhancement

The latter instrument consists of a 100 mm afocal refracting telescope the aperture of which is masked with a square array of $8\times 8$holes of 0.8 mm, regularly spaced by $s=8~{\rm mm}$. An eyepiece produces an exit pupil image 8 times smaller thus containing 0.1 mm sub-pupils spaced 1 mm apart. The densification is achieved with an array of micro-lenses of similar pitch located 100 mm downstream where the beams from each sub-pupil are spread out by diffraction in such a way that their central lobe fills the facing micro-lens in the array. These lenses having 100 mm focal length provide parallel and nearly adjacent collimated beams expanded from 0.1 mm to 1 mm, achieving a densification factor of about $10 \times$. At the focus of a lens located immediately downstream, the central interference peak obtained is intensified with respect to the equivalent but non densified Fizeau array. This micro-lens array, obtained commercially, was modified by immersing its active face in silicon elastomer, a medium having a slightly lower refractive index than the lenses' material, in order to lower the optical path difference (from now on OPD) caused by unequal thickness of the micro-lenses. With its aperture diameter of $D=56~{\rm mm}$ this miniature array has $\lambda /D=2.6^{\prime \prime}$ FWHM angular resolution ( $1.8^{\prime \prime }$in the diagonal direction) and a usable interferometric field of view of $\lambda /s=18^{\prime \prime }$ for a wavelength $\lambda =700~{\rm nm}$(the centre band for the detector used). Larger arrays would require adaptive optics unless used in speckle interferometry mode. We recorded the images in the focal plane of the interferometer array using a commercial Peltier cooled CCD, camera with $9~\mu{\rm m}$ pixel size.