Performance of and problems with the current system

The proof of the pudding is in the pointing. Track models have been created irregularly since 1988, but only 4 times since February 1992. R.m.s. errors in azimuth and elevation are normally less than 2", and have, for some extended periods, been as low as 1.3" in azimuth, and 1.7" in elevation. Recent tracking experiments (see below) show that track errors of up to 5" may still occur as a result of poor resolution near bumps, and the overall track errors may impose more general pointing degradation. As part of the Pointing Project we aim to reduce pointing errors in each co-ordinate to 1".

However, as may be deduced from the tortuous descriptions in Sections 4 & 5, the acquisition of inclinometry data in the past has been anything but routine. We may identify many inherent problems or sources of systematic error that should be eliminated with any new system :

• Moving the (two, old) inclinometers from one location to another is time-consuming, both in the actual effort required for the transfer, and in the time needed for the meters to settle thermally to their new environment, and poses some risk to the instruments themselves.
• The mounts on the A-frames were latterly of high quality, but the mounting arrangements on the TMU were crude and cramped by the limited amount of space.
• Part of the cabling had to be installed at the start of each inclinometry run. This was messy and time-consuming. The new system will use permanent cable routes.
• The use of the RxH microprocessor was inconvenient and compromised its other role as a part of the holography system.
• The behaviour of the inclinometers with temperature is essentially unknown. It was clear from experience that their voltage output changed after a change of location, and this was attributed to the accompanying change in ambient temperature. This problem may be overcome either by having sufficient instruments that may be left permanently in one location, or by calibrating their performance against temperature and including a temperature probe alongside the inclinometers. (The new inclinometers have built-in temperature probes).
• The complicated sequence of reference measures during an inclinometry 'run' was necessitated by unpredictable sudden or gradual changes in output voltage (so-called zero-point changes or drifts). Whether or not these are attributable to inherent instabilities in the old meters is unclear - it is possible that the meters are susceptible to shocking when the antenna moves over a track joint (see section 3.3 above). In any case, it is hoped that the new meters will prove more stable over the hour or two needed to take enough data to construct a track model, and that reference measures may be reduced or even eliminated - Can data be daten on-the-fly ?
• The presence or magnitude of any hysteresis required one CW and one CCW run, but probably two of each, since a comparison of the runs in the same direction would yield the repeatability of the system against which to compare the a CW-CCW pair. Again, it is hoped that repeatability will be easier to obtain with the new system, and hysteresis more easily quantified.
• The time required to take one full-circle sweep of inclinometry data was typically 3 hours with the inclinometers in one of the two locations. Other constraints during the available daytime at JCMT, and the need for repeat measures (as above) invariably meant that two days were needed to acquire sufficient data to construct a model free from systematics. This is too slow to react to an emergency such as an earthquake.
• Tracking data taken of Uranus on 940708 showed the need for increased resolution around the bumps. Errors of 5" are seen in these locations. Increased resolution means additional measurements and a proportional increase in the time taken to acquire the data. Increasing the resolution to 0.1degree around each of the 56 bumps involved will more than double the number of data required. This would be intolerable with the curent system.
• Typical scatters in measurements made with the current system are :
  
  a) repeated measures of reference azimuths during data-taking : 0.3"
  b) large scale variations between CW runs or CCW runs : 2"
  c) large scale variations between CW & CCW runs : 3"
  
  
  A new system should reduce these (last two) errors essentially to zero, and perhaps eliminate the need for so many reference measures.