SUMMARY
POINTING on 3c273 and 3c279 through transit shows general similarity to the earlier map16 tracking data.

Tracking through transit seems afflicted by a change in elevation pointing. Data using the map16 method taken on 19991228 , 20000402 and 20000404 are collected below. The plots show the variation of elevation pointing error with azimuth (left) and elevation (center). The plot on the right magnifies the central plot in the vicinity of transit.

The test reported here used the standard POINTING command to obtain the pointing data and followed 3c273 and then 3c279 through transit. The resulting pointing residuals are shown below plotted against azimuth (left) and against elevation (right). Click on the images for better viewing :

The azimuth pointing residuals are well behaved, but the elevation pointing is disturbed. Plotted against azimuth, the data for 3c273 start in the lower left of the frame, and show a change in del at transit. When 3c273 reaches azimuth 186 a slew is made to 3c279 at azimuth 170. The pointing seems to need no immediate adjustment, but subsequent behaviour shows a decrease and then an increase at transit again. The plot of elevation pointing residual vs. elevation is broken along the elevation axis to reveal more of the transit details, which has some of the characteristics of hysteresis.

The results of the brainstorming on this problem on Friday, and previous and subsequent suggestions, yielded several further tests, and a possible cause : Ian's (& Tim's) favoured candidate is "wind up of the structure", which it is suggested, will produce hysteresis, as is seen above.

Suggested tests and other unsorted thoughts include :

1. Careful monitoring of the elevation drive currents during manual slow elevation slews. Done - by Per on 2000 04 10 (?) and reported by Ian Pain : "no apparent glitches" in going from 70 to 40 and 40 to 90, thus eliminating such causes as "repeatable bearing defect or a very large loose component moving by a large amount".

2. Internal encoder errors - ruled out by thought experiment (IP & TCC). (Details awaiting a thorough examination of the brains involved.)

3. If there are problems at the particular elevations at which 3c273 and 3c279 transit, then tracking should be done through one of these elevations in circumstances that do not include a transit.

   In fact the data for this already exists. The data of 19991228 shows 3c273 passing through elevation 64.38 (where 3c279 transits) - at azimuth 131.6 - without any similar problem.

4. If the problem lies with the map16 method, then regular POINTING data should be taken.

   Done : either of the current datasets. The problem persists, and differences (not that we think there are any) between the centroid analysis mechanisms within map16 or POINTING are eliminated.

5. One transit following another, to examine the degree of hysteresis.

   Done - the current data. Analysis pending.

6. Tracking through the north - is this phenomenon limited to azimuth 180 ?

7. Use elevation chopping/nodding - is the effect disturbed by the need to exercise the elevation bearing/encoder in a way that does not occur with the azimuth chopping/nodding styles used above and by default ?

8. In discussing the observing plan with an observer working almost exclusively below 45 degrees elevation I suggested that his data could be adjusted for any pointing errors suffered by observing through transit knowing the functional form of the errors above. However, our data come only from two sources at fairly high elevation and it would be useful to repeat this experiment on a source transitting at 45 degrees (ie decl.=-25) This will take a couple of hours of course . . .

9. If the problem affects not just transits but all reversals of elevation motion, then analysis of pointing data may reveal this. Such data could be divided into 8 groups of sources depending upon the direction (in elevation space) of the approach to the source and the previous and subsequent tracking directions.

   Done - at least for the case of our last dedicated pointing run : the 50 data of 19991115. The 49 data pairs are assigned a 3-letter code representing (the elevation tracking direction of the first source of the pair, the elevation slew direction, and the elevation tracking direction of the second source), with A=ascending, D=descending or down, and U=up. The column headed 'del' in the table below is the mean change in elevation pointing between the two sources. The values of del for these 49 data pairs group as follows :

    
       No.  key     del   +-
       12   AUA    -0.1   0.6
        1   AUD     1.8   0.0
       13   ADA    -0.1   0.4
        2   ADD     4.7   0.8
        1   DUA    -2.8   0.0
        9   DUD    -0.1   0.4
        3   DDA    -1.5   0.8
        8   DDD     0.3   0.6
   

   The elevation pointing residuals between sources headed in the same direction (AUA,ADA,DUD,DDD) are essentially zero : whilst those involving a change of direction (AUD, ADD, DUA, DDA) show, despite the small number statistics, larger changes of consistent sign. In the case of a reversal of direction as experienced at transit i.e. ADD (although the central D is debateable) we would expect a +4.7" change in the elevation pointing residual. This is to be compared with the 4" change reported above.

   This seems fairly conclusive evidence that our 'transit' phenomenon permates all pointing, and is dependent upon the circumstances of recent tracking, the slew direction, and subsequent tracking of the sources involved. The misnomer 'transit' will be replaced by something more accurate as soon as an appropriate acronym has been found.