Further developments

 The following improvements of the instrument are foreseen:

• Improved adative stabilisation of the occulter centering.
  
• Lyot stop: as mentioned in section 2.3, the use of a pinhole-shaped diaphragm introduces residual optical defects, such as static bright speckles at the crossing of a spider arm and the second Airy ring. For an optimization of the apodizing device, we will use in the future a diaphragm matching the telescope pupil shape.
  
• The Risley prisms adjustement needs to be automated.
  
• Optimized detectors: regarding the photon-counting detector, AsGa photocathodes now available extend the photon-counting range to 900nm. They will be used on future photon-counting cameras. With their clean detection of photon-events, electron bombarded CCDs appear of interest, in spite of the rather slow readout currently available. Non intensified photon-counting CCDs are also announced.
  
• Speckle noise: In the cleaned image generated by the dark speckle algorithm, and also in a conventional long-exposure a fixed pattern of speckles tends to remain visible in the smoothed halo.
  Long integrations, whether in the dark-speckle imaging mode or with conventional long exposures, tend to smooth the atmosphere- induced speckles to the point where fixed speckles become apparent. These originate in the fixed bumpiness of the main mirror and the smaller optics, including the adaptive mirror. The fixed amplitude pattern in the pupil, caused by secondary obscuration, spiders, mirror dirt, etc... also contributes, but to a lesser degree since the Lyot stop is configured to remove much of the corresponding straylight. Attaining a more accurate measurement of static residual bumpiness in the Shack-Hartmann wave sensor should improve very much the situation. The accuracy of such measurements being rather immune to the photon noise affecting fast wave corrugations, there is hope of significant improvements. An easier but less efficient method, which can be used simultaneously, involves adding bias signals, randomly variable in space and time, to the actuators. The issue will be analyzed in more detail in a forthcoming article.