3 Description of the observations

Observations reported here took place in october 1997 at the 1.52 m telescope of the Observatoire de Haute Provence (OHP). In this section we describe the instrumental configuration, the conditions in which this configuration had to work, the conditions driving the selection of the observed sources, and the observation procedure.

   
3.1 Instrumental configuration

The coudé focus of the 1.52 m telescope feeds an AO bench called BOA. BOA is an 88 actuator system with a Shack-Hartmann wavefront sensor. Since BOA is not designed for infrared, its transmission in this spectral domain is not optimal (estimated to roughly 5%).

Figure 2: Optical configuration for the observation run. Dotted lines stand for laser light beam used in OPD servo-loop while full lines stand for IR beam coming from telescope trough the AO system

The working set-up actually used for observations is schematically described in Fig. 2. It differs from the one in Fig. 1 by adding a servo-loop to maintain the OPD at zero. On the path of the none coronagraphic output, a laser beam is launched slightly off-axis and gives air-wedge fringes on a 2-element detector. The difference between the two outputs provides an error signal directly proportional to the OPD shift. A piezotranslator that moves a flat mirror placed in one arm of the AIC closes the OPD servo-loop. Optimisation of the close-loop parameters has been carried on in a quiet environment prior to the observations and led to a cut-off frequency of 2 Hz which has been judged sufficient for working in a quiet environment like the one expectable in a coudé room.

The camera, brought by the STScI group, is a Nicmos 3 array of $256\times 256$ pixels equipped with a K band filter ( $\lambda=2.2\;\mu{\rm m}$, $\Delta\lambda=0.4\;\mu{\rm m}$). The size of the pixel was 0.11 arcsec/pixel and the field of view was limited to 11 arcsec. The acquisition mode of the camera enables us to acquire 7 frames per second saved in cubes of 100 or 200 images. As a conservative value the global transmission of the focal assembly and of the telescope has been estimated to about 2%.

   
3.2 Specific conditions during observations

The conditions of turbulence during the observations were rather challenging. The Fried parameter r₀, estimated from the open-loop Shack-Hartmann data, varied from 1 to 10 cm. For the best 20% of the 10 nights of observation we derived a mean value for r₀ around 5 cm. Estimated outer scale L₀ varied from 2 m to 15 m with a mean value of 3 m. Since the value of the outer scale measured during the run is comparable to the telescope diameter the mean square residual of the phase for an estimated r₀ is indeed lower than predicted by the Noll theory (Noll 1976).

The coudé environment was not as quiet as expected. Vibrations and acoustic effects increased significantly the frequency and the amplitude of the OPD variations.  The error signal from the OPD control-loop showed frequencies as high as 50 Hz and amplitudes as large as $\frac{\lambda_{\mathrm{visible}}}{2}$. For the whole run, we have estimated a mean amplitude variation for the error signal of $\frac{\lambda_{\mathrm{visible}}}{4}$ rms at the laser wavelength which gives in infrared (K band) an amplitude variation of $\frac{\lambda_{\mathrm{IR}}}{15}$ rms.

In spite of many attempts we could not get rid of an important noise level in recorded images. The recorded images were also blurred with an additional "wave'' signal. This high noise and the poor overall throughout of the focal assembly decreased the K magnitude limit for a single snapshot (exposure of 0.14 s) to K=4for a $3\sigma$ detection.

3.3 Selection of observed sources

Since this run was primarily intended to test AIC operating "on the sky'' in ground based situation, known single stars and binary stars had to be considered. Single stars are useful to show the extinction capabilities of AIC, though they might have faint neighbouring features to extract. Binaries are used to evaluate capabilities regarding detection of companions and close-sensing around the central star.

Because of the large number of actuators in the AO, the visible magnitude needed to close the loop of the AO was limited to V=7. The K magnitude has been limited to K=2in order to reach a convenient Signal to Noise Ratio in images needed to reliably evaluate departure from uncomplete nulling.

In the case of known binaries, two restrictions apply: the angular separation and the magnitude difference between the two components. The goal being to test the close-sensing capability down to a fraction of Airy radius separations (roughly 0.4 arcsec here), we limited our samples to binaries with separation below 2 arcsec. It turned out that only 3 binaries met the criteria: 5 Lac, HD 211073 and 72 Peg. The other observed stars are either single stars or stars suspected to be complex by Hipparcos.

   
3.4 Observation procedure

As in classical imaging techniques we need images of stars (target star and comparison star), skys, flat-fields and darks. We also need, as in typical AO observations, recording data for a comparison star under the same conditions than the target star. Such data allow to increase the detection capability by substracting the comparison pattern from the target pattern. In doing so, the shape of the halo is removed, not the noise, but nevertheless there is a gain since residual features with spatial frequencies comparable to those of the companion are then eliminated (residual effects of aberrations, averaged speckle noise).

A typical observation for a target star begins with calculation of the turbulence parameters with BOA in open loop. Then, BOA closes the loop and we put the star on-axis with the pointing mirror. The control of image position is not allowed by the acquisition mode of the camera. Therefore in order to ensure that no drift from differential atmospheric dispersion is happening (AO correction in visible with observation in IR), the observer has to periodically switch from blind acquisition mode to view mode, where pointing correction is performed if necessary. After images with star on-axis (nulled), images with star off-axis are recorded (twin images, no nulling). Then calibration of the extinction is achievable and photometry of a companion is available. Moving the telescope of half a degree, enables us to record sky emission close to the star. This procedure has to be repeated for comparison stars. Flat field is determined by recording images of a wide and uniform warm source illuminating the field of view of the camera (see Sect. 4.1).