Robert Loewenstein University of Chicago Astronomy Dept. Yerkes Observatory Williams Bay, Wisconsin 53191, USA
The observer, whether remote or local, controls all functions including telescope pointing. Most observers use a graphical interface running on a Macintosh computer (2) . The Macintosh communicates with a site computer via the Internet or a high-speed modem. An operator is present at the site to shut down in case of bad weather, change instruments (though some instrument changing can be done remotely without operator intervention) and fix any problems that may arise. The telescope was designed to be operated remotely, and even in this early stage is most often used in this fashion.
One of the primary problems we faced in supporting remote operation was that of offsetting the telescope. Traditionally, observers have adjusted the field of view by nudging the telescope with a handpaddle control. This is acceptable when images are returned quickly enough to provide the necessary feedback, but our remote links do not have sufficient bandwidth. Furthermore, many scientific instruments, such as those using integrating CCDs, return images too slowly to make a handpaddle practical even for local observers. To solve these problems, we are implementing a control which allows the observer to precisely specify an offset in one operation, rather than the many iterations required when using a handpaddle. The user specifies an offset by directly manipulating an image of the sky, hence we call the control a graphical offsetter.
Figure 1: Offsetting a slit spectrograph using an image from the slit viewing camera.
The situation becomes more complicated if the sky rotates on the instrument. Working with a rotating field of view is unusual but sometimes it is required. Our telescope has an azimuth/altitude mount, so instruments without a rotator, such as our planned Echelle spectrograph, always have a rotating field of view. Even when using an instrument with a rotator, one will not always rotate the instrument with the sky. For example, observers sometimes orient the slit of a spectrograph perpendicular to the horizon to minimize error due to dispersion.
When using an instrument with a rotating field we must pay attention to the location of the "boresight": that point on the instrument about which the field of view rotates. Since it is the only point on the instrument's focal surface which looks at a fixed location on the sky, the object of interest must be placed on the boresight. The current position of the boresight is superimposed on the image, and the observer may move the boresight to any convenient location on the instrument (or off it�see the section on drift scanning).
By the time the user requests an offset, the image will be outdated, in that the field will have rotated about the boresight. However, the outdated image can still be used; the control system simply takes into account the rotation of the selected object since the time the image was taken, so the object is moved exactly where the observer requested. An example is shown in figure 2 .
Figure 2: Offsetting a slit spectrograph with a rotating field of view.
The graphical offsetter implements an offset by changing the object offset and boresight position but not the sky orientation. In addition, all of these parameters may be set directly by the user, and the object offset and boresight position may be set by using a handpaddle.
The data required to compute a graphical offset is as follows:
The new value for the boresight position is simply that specified by the user; no computation is required. The new value of the object offset is computed as follows:
Guiding on the observer's object eliminates the problem. However, only the Echelle spectrograph will be capable of guiding on the object; our other instruments use offset guiders, which require a separate guide star. An offset guider cannot correct the error in boresight position induced by pointing error, so the observer must correct pointing error before starting to guide.
The control system is capable of automatically correcting pointing error on request. The system searches the Basic FK5 Catalog (3) for the nearest star, slews to that star, measures and corrects pointing error by adding a constant offset to the telescope axes and returns to the object. There will be some remaining error due to centroiding the star (~0.1" for our telescope), offsetting back to the object (~0.1" for our telescope) and a negligible contribution due to error in the star position (~0.01" for the Basic FK5 Catalog). The pointing error will then gradually increase over time until guiding is turned on. Alternatively, instead of guiding, one may correct pointing error at regular intervals.
To accomplish drift scanning, it is useful for the telescope to be able to move along a great circle at a fixed rate of speed. This is implemented in the ARC control system by permitting the boresight to have a non-zero velocity with respect to the instrument and to be located far off of the instrument. The concept is illustrated in figure 3 . The position of the sky at the boresight is left fixed, but the position of the boresight on the instrument is set to move at a constant speed. As a result, the instrument moves away from the boresight at this speed, following a great circle across the sky.
Figure 3: Drift Scanning
For drift scanning to work, the velocity of the boresight must be along the columns of the CCD and must match the CCD's row shift rate. Fortunately, this is easy to set because boresight position and velocity are specified in the frame of reference of the CCD.
One subtlety of drift scanning is that the basic tracking parameters do not directly tell you where in the sky the instrument is pointing. The only sky position specified is that of the boresight, and that is fixed. Yet the observer will wish to record the changing position of the sky on the instrument. To support this, the control system can compute the sky position of any point on the instrument. The position is time-tagged, so it can be precisely associated with the data, despite delays in computation and communication.
When working with a rotating field of view, the observer must be careful to place the object of interest on the boresight (the center of rotation of the field of view). The boresight is displayed on the image, and both the field of view and the boresight may be moved by the observer. The observer must be careful to minimize pointing error when using a rotating field of view, as pointing error produces error in the position of the boresight. Offset guiding does not correct the problem.
Moving the boresight at constant speed along the columns of an imaging CCD camera permits drift scanning.
