Outline of Design

Figure 1 shows schematically the proposed design of MAPPIT 2. The instrument may be divided into a number of subsystems:

 
Figure:   Principal elements of MAPPIT 2 (not to scale). The telescope beam from the AAT coudé train enters at left, where it also passes through an atmospheric dispersion corrector, Dove prism (for field rotation) and field lens (these elements are not shown). Also required in practice but not shown is an acquisition/guiding TV which views the reflection from the aperture plate at the main coudé focus, and a pupil viewing TV which views the NRM/slit mask. The diameter of the collimated beam is 25 mm. The actual number of lenslets will be about 40.

Beamsplitter

The wavefront sensor must view the same slit of the aperture as will fall on the mask to form speckles or fringes on the science detector. Thus the wavefront sensor and the mask must both be in planes conjugate to the telescope pupil (or the dominant atmospheric turbulent layer), and must be optically superimposed. This may be achieved using a conventional beamsplitter, as shown. However, for greater efficiency it would be preferable for the wavefront sensor to receive all the light except the narrow band used by the science detector arm. With a suitably re-arranged layout, this may be implemented as a later development, employing a narrow band mirror (rugate filter) or possibly using the reflection from the narrow-band filter itself.

The Wavefront Sensor

It is proposed to use a Shack-Hartmann sensor, which will consist of a 1-dimensional row of about 40 lenslets, each of order 0.5 mm in size. Each lenslet subtends a subaperture of cm) or slightly less on the primary mirror. The image spot formed by a lenslet has a displacement from its nominal position which reveals the average phase slope across the corresponding part of the aperture. From the set of phase slopes at roughly spacings, for each exposure of ms the full wavefront phase profile can be found, and hence the instantaneous point-spread function (PSF) of the telescope/atmosphere combination (eg Marais et al. 1992). Real-time estimates of the atmospheric turbulence parameters and plus residual focus errors can also be found from the sensor data.

The Shack-Hartmann sensor can use broad-band light, thus allowing the system to obtain adequate signal (rms phase errors < 0.3 rad, corresponding to 100 detected photons per lenslet spot) for objects down to about 9.0 mag. This value assumes observation of a red star, with 10 cm apertures, 10 ms exposure time, a 50% beamsplitter, and the AAO Thomson CCD as detector. Since the system is 1-dimensional, only the spot displacements along the line of the array are needed, so the illuminated area of the CCD can be binned along columns to give a 1-dimensional output array. Originally introduced for NRM observations (Buscher et al. 1990), column binning results in CCD readout times of order 10 ms, enabling the detectors to be run without a separate shutter and achieving 100% duty cycle. At the same time, the concentration of detected photons from a whole column into a single pixel reduces the effects of readout noise on the data.

For square lenslets (the preferred outline) forming a diffraction-limited spot on the CCD, the spot FWHM is , where is the wavelength and f/d is the focal ratio of the individual lenslets. In order to obtain spots large enough for Nyquist sampling by CCD pixels it is necessary to use focal ratios of 100 or more. The alternative of precisely aligning a smaller lenslet spot on the pixel boundaries of a quadrant detector is not feasible for the number of lenslets required. The displacements which must be measured are small, since a phase slope of radians across one lenslet subaperture produces only a shift of the spot equal to its diffraction FWHM. The predominant effect of the atmospheric seeing is to move the spots around, not to broaden them. However, the spots will be broadened somewhat by chromatic aberration, residual wavefront curvature over the subapertures, changes in the atmospheric phases during the sampling time, and object structure if the target object has a size comparable with the seeing disc.

For maximum speed of the CCD readout from the wavefront sensor, and to fit within available CCD sizes, the spots should be placed as close to each other as possible, while still allowing adequate spacing to cope with maximum expected phase slopes. It can be shown that a value of radians phase difference across one subaperture allows for the maximum effects of seeing, with some reserve for the spot-broadening effects mentioned above. A key parameter of any lenslet array, independent of the optical system used to feed it, is the quantity . It gives the spot separation / spot size (FWHM) , and to accommodate the maximum phase difference of radians across a lenslet, this parameter should be 4.4 (at the longest usable wavelength). This is equivalent to 8.8 pixels per lenslet on the detector. Minor telescope tracking errors may cause larger displacements of the spots, but in this case all spots move in a correlated manner, so they can be followed as a group if sufficient extra pixels are allowed at the edges of the detector.

The interferometry subsystem

The wavefront sensor input comes from a slit across the telescope pupil. It will have a maximum length of 3.3 m for observations in which it is essential to avoid the central obstruction, or 3.9 m for those in which a central gap can be accepted. The beamsplitter passes light from the same slit to either an array of holes for formation of NRM fringes on the science detector, or to a slit for use of slit speckle interferometry. An interference filter selects the required narrow wavelength band. NRM will probably be preferred for objects bright enough to give adequate signal, because the baseline redundancy of speckle interferometry reduces the relative amplitude of the higher spatial frequencies in the data. The science detector will be another CCD, which should be operated with readout synchronised to the wavefront sensor readout. Column binning will again be needed for adequate readout speed, and is allowable because the fringes or speckles have only 1-dimensional structure.

© Copyright Astronomical Society of Australia 1997