A brief history of the transit tracking problem
   -----------------------------------------------

In the period 1997 - 1999 much tracking was done but with the 
aim of characterizing the impact upon _azimuth_ pointing due
to the central bearing 'glitch'. Because of the specific azimuthal
locations of the 'glitch' and because the amplitude of these 
pointing excursions varied as cosine(elevation), data was taken at
selected azimuths (typically 125, 140, 225, 240) far away from transit.

Following the central bearing replacement on 08 May 1999,
the first data taken coincidentally across transit were those of
Uranus (Decl. = -16.5, elevation at transit (el*) = 53.5) on 
990618 which showed a 3" step in elevation pointing at transit. 
It was said at that time "It would be of great value to know if 
(this and other systematic) variations . . repeat". However, subsequent 
tracking tests, like those of 23 Sep 1999, were chosen without 
particular concern that there were problems specifically at transit. 

The next data to show a transit effect were the 3c273 (el*=72, 
step ~3") data of 991228, but they were taken to check the track model,
and the 'small' systematic in elevation at transit was lost in the
perceived overall improvement in _azimuth_ tracking accuracy.

On 02 April 2000, while obtaining deep imaging of a bright source
(3c279) for beam characterization, tracking data through transit
again revealed a 4" step through transit. Repeat data on 04 April 2000
confirmed this, and the effect was at last recognized in the older data. 
The effect was also seen (weakly) in the north.

Analysis of transit pointing data by Remo Tilanus
showed that the effect may have been present and deteriorating 
since 1996 - possibly coincident with the cone-bar tightening.
The amplitude of the effect seemed to increase rapidly through
March 2001 [image 5] and April 2001, and an additional
sharp feature on 19 April 2001 signalled the need for 
urgent repair. Increasing friction in the elevation encoder (or its
mount) had became the most likely suspect and the encoder was replaced
on 24 April 2001.

Subsequent intensive characterization of the new encoder included
transit tracking at several el*evations, enough to reveal a relationship
between the amplitude of the elevation pointing step at transit and
cosine(el*). The new step was usually quite sharp, as seen in these plots,
but one tracking event the same night also showed azimuthal (frictional)
smearing prevalent in the old data.

The cause of the tracking effect is some combination of hysteresis
and friction in the elevation encoder mount together with an imbalance
in the antenna loading. The impact upon _pointing_ per se is poorly
characterised : the behaviour of the antenna during transit is possibly
quite different from that when slewing from one source to another, whether
across the meridian or not, and we have started collecting data that
should define the relationship between the two effects.

While this characterization is being sought, the previous empirical
pointing correction (lookup table) has been disabled, and 
observers are obliged to point locally whenever the meridian is
crossed, and to avoid tracking through transit. With the caveat that 
a little extra time is being used each shift for additional pointing, 
this has not noticeably impacted observing efficiency or data quality.
 
imc/20010514