Reasons behind the ZEROSET and SNAFU procedures

  ZEROSET

ZEROSET is the procedure that must be run when the telescope control system has been started up, or if for some reason the encoders have become corrupted during operation. It is a technique for initialising the Ferranti encoders. These are incremental devices and must be given an absolute reference from which to work. The ZEROSET process involves sweeping the telescope across hardware markers on both axes. When these markers are detected, the encoders are zeroed by the hardware, if this mode is enabled by having typed ZEROSET. The hardware detector is positioned so that the ``zeroset position'' is at the zenith.

There are errors inherent in this process, but these can be minimised by the correct operational procedures.

1. The angle within which the telescope is ``detected'' on each axis has a finite size of a few arcsec. This is why it is possible to position the telescope and keep it stationary ``in the zeroset position'' with the green zeroset lights above the RA and Dec dials lit. However, the encoders will be initialized by the hardware if the telescope is anywhere within this region and this leads to zeroset errors comparable to the extent of the zeroset region. For this reason, the technique of setting the telescope at this position and then running ZEROSET is unsatisfactorily crude. The technique capable of giving the highest accuracy and the most repeatable results is to 1) ZEROSET at the User Interface, and then 2) move the telescope slowly through the zenith in both axes, but always going in the same direction. In this way the leading edge of the detector area will be used as the reference, and this is clearly much better determined than any random point within the region.
2. The speed with which the telescope is driven past the detectors is now not very important. Although the detection/encoder-reset is done by the hardware, and there is some small error associated with this process, tests indicate that the error is practically independent of speed, but depends on direction. Markers have been placed on the speed control potentiometers to indicate a recommended speed at which the ZEROSET triggering can easily be observed.

SNAFU

The SNAFU procedure is designed to perform a somewhat similar task to ZEROSET. It does however have various subtleties which should be appreciated if it is to be used properly.

Originally SNAFU was used to update only the encoder index errors (ID and IH) of the telescope pointing model, which are simple zeropoint offsets of the encoder readings. These are deduced by pointing the telescope at an object whose position is precisely known. More recently, it was considered desirable to introduce the possibility of updating the collimation error of the telescope as well. This error (CH term in the pointing model) is the offset between the mechanical and optical axes of the telescope and ideally should not change if the telescope configuration is stable. The collimation error is indistinguishable from the index error in Declination, being a constant angular offset. But in Right Ascension a Cos(Dec) term is introduced by the projection onto the sky. The error in RA as seen on the sky is given by:

E(RA) = Index Error Cos(Dec) + Collimation error

As the Declination increases, the effect of the index error diminishes and it is a good approximation to ascribe the pointing error solely to the collimation error. Note though, that cos 850.09, but cos 700.34. The SNAFU process can therefore be split into two separate operations.

1. The telescope is set on a star at high Declination and the correction observed is attributed to the collimation error in RA and the index error in Dec. The hour angle of the star observed is not critical (it can be +3 hrs before other effects become important) but it is essential that the Declination be greater than 85, or else the approximation that cos(Dec)0 implicit in the method is invalid.
2. An object at low declination is observed, and only the index errors in both axes are updated.

The SNAFU software therefore has two modes of operation, and depending on which mode it is in, the coefficients of the pointing model that are updated are either (ID and IH) or (ID and CH). The mode of SNAFU is set by the Task Common variable HATERM. At the moment, access to this variable is somewhat cumbersome and only possible through the use of INPSCF on the telescope control system terminal. The default (e.g. startup) value for HATERM is 2, which means it is in the mode where the index errors ID and IH are corrected. The consequence of this is that it is safe to perform a single star SNAFU at any Declination (< 45) at any time without changing the value of HATERM. In fact, with improvements in the performance of the telescope, serious problems with collimation do now rarely occur, and most of the time doing a single-star SNAFU is sufficient. We have more problems with the encoders than with collimation. However, after an instrument change, a full 2-star SNAFU should be performed. This involves changing HATERM to value 1, and then SNAFUing on a star with Dec >85. Again, the star used for this need not be on the meridian, but it is important to get high declination. The value of HATERM should be returned to 2 before any SNAFU is repeated at a lower declination. Since the high-declination SNAFU is only an approximate solution, if the corrections needed to bring the stars onto the reference point are large (>10 arcsec) when performing a 2-star SNAFU, then the whole procedure should be repeated at least once. In this case the declinations chosen for the two stars should be (HATERM=1 , Dec > 85), (HATERM=2 , 30<Dec < 40). This avoids a 180 rotation of the dome between SNAFUs. Reference Point for SNAFU  

It is equally important to be precise and consistent about what reference point is used for SNAFU. At Cassegrain, some point on the A&G box TV is used. It should NOT be an arbitrary point, since the software assumes that the SNAFU position is fixed relative to the rotator axis and uses this information to perform aperture offsets and corrections for rotator angle changes. A preference has emerged for SNAFUing on the spectrograph slit. The position along the slit must be defined, and this is the intersection of the slit and the single aperture dekker. This convention must be strictly adhered to, because the rotator axes will be determined relative to this point. It also has the inherent advantage of ensuring that the slit mirror, as opposed to the field mirror, is used to view the object. If this position is not optimum for placement of the object on the detector, then use an aperture offset to move to the required position afterwards (i.e. define an offset using SNAFU/A and thereafter use the BEAMSWITCH A button to move to that point). A change of reference point requires a 2-star SNAFU to be performed, since it implies a collimation shift.