Inclinometry was done as follows using the 'fixed' TMU inclinometer:

    Datasets  El Dirn     HST     mean leg temperatures   Humidity
                          start     start   middle   end       %
  ( 20021005  90  cw      02:15      0.0      0.3    0.7     100 )
    20021228  90  cw      09:20      0.0      0.8    2.0   100-20 var

The demand azimuths, as contained in the file azfilecw.dat were not those recorded during this run by the INCLINE task. The demand and recorded azimuths and their differences are shown here. On average they differ by 0.03 degrees. Inclinometry data taken most recently on 19 Oct show no such effect. This difference is more than can be explained by changes in the interim to any parameter (such as IA) of the TPOINT model. Besides ! - inclinometry is taken in EN(coder) (or MOUNT) coordinates, and should be independent of any pointing model. Additionally, the azimuth differences show large excursions from this average. On several occasions errors of between 0.07^o and 0.09^o occur, but most notably of all are the measures around azimuth 96 which are as much as 0.77^o in error.

New code was written to interpolate the new data onto the demanded azimuths, and in order to make comprehensible comparisons with previous data. The temperature data from the inclinometer on the TMU were good so the new data are also temperature-corrected in the usual manner, and the resulting model is compared with the previous model of Oct 05 :

Many 'spikes', some as much as 30" in amplitude, are seen in the F1 component of the model-difference plot above. The largest occurs at azimuth 95 - the same place as the 'positioning' error noted above. Many 'spikes' occur at track joints, but not all. These are *not* the same 'spikes' as have been seen in times past when the front antenna wheels have been underloaded : the distribution, amplitude and width of the spikes are quite different. The plots below give further details, with the F1 (yaw) component of the model broken down into its component parts : viz the LY and RY tilt measures:

      az=0-100       az=80-160       az=150-200       az=200-280       az=270-360

The details show that the LY and RY excursions are most often anti-correlated, with many small (2") excursions occurring every few degrees. The anti-correlated excursions are the signature of a torsional force, namely that the the telescope is accelerating as the inclinometry data are being taken; i.e. that the telescope has neither reached its demand position nor stopped, as is required by the INCLINE task. Is the antenna having a 'hard time' getting positioned near a track joint? What of those excursions not near track joints?

The data are no good for generating a new track model so I am leaving the Oct 05 track model in place. The formal rms scatters in the F2- and F3-differences are 0.4" and 0.5", respectively, and are quite typical (see the last such analysis), so I suspect that the track profile per se has not changed significantly.

The more important concern raised, though, is that antenna control is not working as expected, and that the antenna is not on-source when it ought to be. The plots above would not support a claim that this effect can explain the large pointing excursions seen recently -- one of 10" was seen last night for instance at azimuth 373^o+1^o, where these new inclinometry data look relatively 'quiet' -- but the nature of this fault may be quite variable in azimuth-space from run to run or moment to moment.

20021229
A repeat CW run was performed. The azimuth positioning errors are very similar to those of UT20021228. The errors from these two runs may more easily be compared here. The resulting model (not that we'll use it) is compared with that of 05 Oct here and shows very similar 'spike' patterns to that above.

20021230
Possible errors in the 'UKIRT' TCS installed on 18 Dec were identified by RK and modifications made. Partial inclinometry in the az range 10-123^o showed the correct acquisition of each demand azimuth and the absence of 'spikes'. [ Ignore the false filename, and note that no RLB has been done, and that the units of TY, LY and RY here are in the original mV (20mV=1").]

All appears restored to the inclinometry process.

16 pointing data this night had raw rms scatters in (daz,del) of (1.6",1.9") - a good omen on which to end the year.