The guiding system in SIS computes the instantaneous centroid of a given guide star and corrects for its random motion at frequencies up to 20 Hz.
The normal way of using active guiding with SIS is to correct the slow drifts in telescope tracking with the Telescope Control System (TCS) and then correct, to first order, the fast random motions due to atmospheric turbulence with an independant tip-tilt mirror in front of the instrument. The electronics associated with the active mirror directs the low frequency (< 0.3 Hz) error signals to the TCS for correction.
It is certainly possible to compensate both low and high frequency motions with the tip-tilt mirror alone, but only for short exposures. In the case of a noticeable drift, the mirror would progressively deviate from its mid-point position and would rapidly reach the limit of its movement (the maximum amplitude is a few arcsec).
It is also possible to have the error signal for slow corrections computed independantly with the Cassegrain bonnette guiding system (acting on another guide star). This could potentially be a source of trouble but does give an improvement in the final resolution similar to the ``normal'' guiding mode. This should, however, be considered only as a back-up mode in case of problems with the communications between SIS and the TCS. As a result, the procedures given below refer to the ``normal'' mode of observation.
Figure 3.1 summarizes the possible configurations for guiding with SIS.
The SIS guide probe that is used to find a suitable guide-star has a small field of view (3") divided into 4 quadrants by an array of 4 lenslets, each feeding an avalanche photo-diode (APD). The guiding loop attempts to equalize the signals measured by the 4 APDs, over an adjustable integration time, by activating the tip-tilt mirror.
Figure 3.2 shows the guide probe window displayed when the ``GUIDE-PROBE'' form is activated. Since the guide probe location in the focal plane versus position its position on the CCD is well known, when you enter the CCD coordinates of a star, measured on a previous exposure, into the guide probe window, the target coordinates of the probe are calculated. They are then displayed when the probe is in place, after you have selected ``Start Guiding''. For instance, entering CCD coordinates (848, 1304) will yield probe coordinates (2063, 3872) and the star should fall very close to the center of the probe. To maintain this mapping accuracy, after each mounting of a CCD the procedure CALIB should be run. This is normally done during the set-up.
Figure 3.3 shows the available domain for the guide probe motion, compared to the science field with a CCD like Loral 3 ( 15 m pixels). This form appears on the Pegasus session and is updated whenever the guide probe is moved. The accessible field is about . The scale for the probe motion is: 1 step = 7.5 m in the focal plane = 9.4 m on the CCD = 0.054" on the sky. The field of view for the probe is 3", divided in 4 quadrants. The ``garage'' position reads ; for the probe coordinates.
For precise centering of the guide star it is better to use the arrow control boxes, so the active mirror will be near its middle position when active guiding is started. Use the slide bars to select step sizes (no more than 2 or 3 units) for each axis. Allow some time for the display of the star position in the window to update or look at the oscilloscope which gives a real time display.
Figure 3.3 shows that the present guide probe is quite large. For some programs in multi-slit spectroscopy locating the guide probe in the science field may be tolerated. On the other hand, it is better to avoid any occultation for direct imaging programs (see figure 5.12). This can be achieved with the X coordinate of the guide probe close to the maximum value of 4000, i.e. virtual Y coordinate for the CCD close to 2500. We recommend that observers choose guide-stars for each of their fields in advance. Sometimes, a slight decentering and/or bonnette rotation may also be needed in order to avoid occultation.
The SIS guiding system was tested on various stars after the change of detectors (from photomultipliers to avalanche photo-diodes). In the field of H1413+117 (a typical high galactic latitude field) only 3 stars were bright enough for guiding attempts. Below are listed their R magnitudes and measured total fluxes (in counts/s above the sky level of 2800 counts/s):
We observed image improvement relative to the Cass bonnette guiding for the first 2 stars with a guiding integration time 0.1 s and still for star 40 with an integration time of 0.2 s. However, this corresponds to the practical limiting magnitude; the S/N ratio in each quadrant was only with this integration time. A practical rule for choosing a suitable guide star is to have at least 1000 counts/s above the sky background, which corresponds roughly to and .
Guiding on brighter stars is more practical and allows a range of integration frequencies up to 100 Hz. The gain in image resolution, versus Cass bonnette guiding, is of the order of 0.1"- 0.2" (typically, 0.5" - 0.6" versus 0.7") but we do not yet have a large enough body of statistics to study the dependance of image improvement on guide-star brightness. The degree of image improvement also certainly depends on the turbulence regime when you observe. We can, however, recommend the following values for the integration time of the tip-tilt mirror command:
The best images obtained to date have image qualities a little below 0.4", which is similar to HR Cam performance. Long exposures (30 min) show resolutions similar to the short (30 s) exposures, and demonstates the good reliability of the stabilizing system.
Another point, important for the multi-slit spectroscopy mode, is that the image position is very stable on the CCD. For a given field, the centroid of a star moves by less than 1 pixel (0.0866") in 2 hr. The flexures between the guide probe and the detector are thus negligible.