Introduction

The good image quality of the NTT is in part due to the active control of the primary and the secondary mirrors. The primary (M1) mirror is supported by 75 actuators and three fixed points. The force applied to each of the 75 actuators can be adjusted and thus the shape of M1 can be modified. The secondary mirror (M2) can be moved in X,Y,Z, where the X,Y motion of M2 is used to correct for decentring coma (in fact only the one along the axis of gravity is used; the movement corresponds to a rotation around the centre of curvature of M2) and the motion in Z controls the focus. The active supports of M1 are used to compensate for various deformation effects in the telescope structure and the mirrors, and for effects due to inhomogeneities of the air temperature in the dome. Some of these effects are elastic and can be empirically calibrated for each position. Others have inelastic components and are more difficult to predict. Confusion is sometimes found about the difference between active optics and adaptive optics. Adaptive optics can correct for turbulence in the atmosphere by means of very fast corrections to the optics, whereas active optics only corrects for much slower variations. Thus, whereas adaptive optics can reach the diffraction limit of the telescope, active optics (as on the NTT) only allows the telescope to reach the ambient seeing.

There are three different procedures to set the NTT Active Optics System (AOS). The first is to use a default setting which has been calibrated for the zenith position. The default setting is obtained through an initialization procedure which is run every afternoon. The second method is to correct the default setting for gravitationally induced deformations, using predefined look-up tables. These tables include corrections for decentring coma and defocus, but not for higher order effects. The third method is to do a full wave front analysis, the so-called image analysis, and to calculate the mirror settings from this. This method can be used either in open or in closed loop.

Image analysis

The image analysis systems (there is one at each Nasmyth focus, located inside the instrument adapter/rotators) consisting of a Shack-Hartmann grid and a CCD camera to record the image. This image is displayed in the RTD (Real Time Display) assigned for the Technical CCD in use. The pupil image corresponding to a particular star (or the guide star when doing image analysis in parallel mode while observations are in progress) is transformed by the grid into a regular pattern of dots. The position of each dot has been calibrated with an internal light source passing through a pinhole generating an artificial star. The wave front distortions can be obtained from the displacement of each dot from its calibrated position. From this, a software program determines the telescope aberrations. The procedure solves for: defocus, spherical aberration, coma, astigmatism, triangular coma and quadratic astigmatism.

A low order Zernike polynomial is fitted to the map of the displacement vectors. The accuracy or validity of the solution is estimated from the rms. residual deviations with respect to this polynomial. If the rms. is poor, the corrections are normally not applied to the mirror.

The results of the image analysis are displayed in the Graphic User Interface of the Main Active Optics Panel and in the Active Optics Engineering Interface.

The optical quality of the telescope is given in the Transverse aberration field. The important parameters are:

• rms. This was defined above. It gives a good estimate of the seeing conditions (local plus atmospheric). If the rms. is large te results are largely determined by atmospheric and dome seeing and therefore it is not recommended to reset the mirrors. Under very good seeing conditions the rms. is below 0.12, while for very bad conditions the rms. can be higher than 0.20. The recommended limit is 0.16 beyond which no corrections should be applied to the optics.
• d80. This is the diameter within which 80% of the light would be concentrated without seeing effects, and is thus a direct measure of the intrinsic optical quality of the telescope. The various components that contribute (quadratically) to d80 are shown. Defocusing is not considered in the total sum. On nights of excellent seeing, the optical quality may be adjusted to d80 < 0.1arcsec, but twice this value is still acceptable.
• M2 movements. This field gives the corrections applied to the position of the M2 unit.The focus offset between the image analyser and the instruments has been accurately calibrated. Normally it is not necessary to check the focus after a correction.

After image analysis is completed, and if, the rms. is high, the correction procedure must be executed. Also note that setting the focus may shift the image by a few times 0.1 arctic: this should be avoided for e.g. narrow slit spectroscopy. So, it is recommended to execute the corrections in between exposures.

A poor rms. can be an indication that the air inside the dome is not well mixed. The standard integration time for the image analysis is 30 sec. On this time scale the wave front aberrations generated by the free atmosphere are effectively averaged out unless the wind is very low. This may not be true for wave front aberrations generated by the local air in the dome, depending on parameters, such as, temperature inhomogeneities, the wind speed and the air flow pattern in the enclosure. Active optics will only correct the low temporal frequency contents of these local air effects. Under unfavourable conditions, for example when the primary mirror is more than 1 degree warmer than the ambient air, these effects can lead to a severe degradation of the image quality.

As it was mentioned above, the image analysis system can be used in parallel mode, during the science exposures. In this mode, a dichroic is inserted in front of the guide probe which deflects most of the light of the guide star to the Shack-Hartmann grid. The corrections are calculated during the exposure, and after the exposure is finished a decision can be taken whether or not the corrections should be applied. The force setting will take approximately three minutes. This mode requires a sufficiently bright guide star which is not always available. Also, the parallel mode presently only works with zero rotator offset. In imaging mode this will in most cases be acceptable, but in spectroscopy a rotator offset may have been applied, either to align the slit with the object(s) or to orient the slit along the parallactic angle.

Practical Considerations

The AOS is initialized every afternoon by the telescope operator. A full image analysis will be done at the beginning of the night, when it has become sufficiently dark. This will generally be immediately after the taking of twilight sky flat fields and takes 15-20 minutes. This analysis is done not only to improve your images, but also to monitor the telescope and detect possible problems. The observers may decide to shift these measurements to later in the night if they conflict with urgent observations, but the telescope operators are instructed to do this test every night.

When the seeing conditions are good, it is advisable to make a new image analysis every time the telescope is moved of more than 15-20deg in altitude when close to the zenith (altitude > 60deg) or ~10deg when at lower altitudes. Note that a change in azimuth has no effect on the optics, and that these values can be relaxed when the seeing is bad. When staying for a long time on a same field, keep a careful eye on the image quality of the EMMI/SUSI frames. Images elongated by more than about 10% (to check use the MIDAS command CENTER/IQE or the Real Time Display functions) indicate residual astigmatism (in the presence of some defocus) which is generally the first indicator of imperfect settings. In any case, the judgement of the night assistants can be fully trusted: do not hesitate to ask their opinion, and follow their recommendations.

For the majority of NTT observations, whether imaging or spectroscopy, better seeing will give improved signal to noise ratio. Under good conditions, the time spent on an extra image analysis will generally be a good investment. It is recommended to use the parallel mode as much as possible, and to apply the corrections when appropriate, using the guidelines above. However, there are several conditions where little or no improvement can be expected. The first is when the seeing is poor (significantly above 1 arcsec). If the seeing is poor, the default setting is normally sufficient. Second, if wind shake of the telescope is important (especially when observing into the wind, at wind speeds of 10m/s or more). Finally, if there is little wind the solutions will not be good: this typically happens when the wind drops below 2-3 m/s. In the last case, either try doing an image analysis with the telescope pointing into the wind, or wait for the wind to pick up again. Problems may also occur when M1 is significantly warmer than the ambient temperature.

If the automatic mode is used, it should be used for every preset because the corrections are differential relative to the previous values. Thus, if the seeing conditions deteriorate strongly and image analysis becomes unnecessary, the telescope should be pointed close to the zenith before disabling the automatic correction mode in order to return to the default set-up.

There used to be a small offset between the telescope focus at the image analysis camera, and the focus at the instrument. This is not the case anymore, and a telescope focus is generally performed only just after the begining of night.