Summary
The elevation encoder was replaced with the spare on 24 April. Transit tracking over the next couple of days showed a variety of 's-curve' profiles. However, the rather extreme nature of these data have provided the enginners with a better understanding of the encoder support structures. The thin (1/8") invar anti-rotation rods were replaced by stouter (3/8") rods of steel on 27 April. Transit tracking on 27 Apr should provide ultimate proof of the repair.

The Replacement
The encoder was replaced by Vernon DeMattos and team. The friction/stiction in the encoder was measured to be 3x that of the new encoder, which may explain some of the recent transit measures. Binary encoder positions were recorded before the removal and after the replacement by Neal Masuda. These were converted to decimal encoder units by Jim Hoge, and the difference used to update the encoder zero-point in the EPICS software by Nick Rees. (I could have eliminated the binary conversion had I used the 'cau: get mount:el:enc.RVAL' method Nick had recommeded - but I forgot !).

A change of more than 5 degrees was made and we were able to acquire a bright source (n2071ir) immediately with ~20" residual error.

Click here for PHOTOS of the daywork events by Jim Hoge.

Transit tracking - I
Transit tracking was then performed using the map16 method, and with the empirical correction disabled :

• from (az,el) = (147,67) at HST 15:07 to (220,65) at 16:59.
• transit occurred at HST 15:56 at an elevation of 71 degrees.

The pointing residuals are displayed below :

The elevation residuals show the familiar s-curve :

• of (semi-)amplitude 3.5",
• of (semi-)width 5.0 degrees,
• and centred at azimuth 181 degrees.

A new empirical correction was installed (and the zero-point adjusted in the THI.tide file) and the observers were informed that observing was possible at about 17:45 HST.

First shift observers used RxB, but immediately reported a 11" elevation pointing error. They persevered, adopting a strict local pointing tactic to minimize the unknown errors remaining. However, it seems likely that their data quality was compromised.

Transit Tracking - II
With the sky opacity too poor for their scheduled work, the second shift observers gave the telescope back to the observatory to resolve the pointing problems. Mars was then tracked :

The azimuth residuals have a raw rms scatter of 1.0", which is only a little larger than the seeing. The elevation residuals show a very large transit step :

• of (semi-)amplitude 9.5",
• of (semi-)width 1.0,
• and centred at azimuth 180.5 degrees.

The abrupt change in behaviour explains why earlier and subsequent all-sky pointing were not amenable to producing a new pointing model of suitable quality. This sudden change is extreme enough to have prompted further thought on the engineering difficulties faced in mounting the encoder. The tension in the anti-rotation rods must be sufficient to prevent the encoder rotating, but also sufficient to prevent the encoder from 'floating' within the bearing. It appears that the current tension is insufficient to meet the second criterion. A further adjustment to the support of the encoder was made on 25 April.

3 new transit runs
Transit trackings of 3 sources were performed during the night of UT 20010426 with the empirical correction disabled :

• from HST 19-21, on irc+10216, which transits at 83 degrees
• from HST 22-23, on 3c273, which transits at 72 degrees
• from HST 03-05, on Mars, whcih transits at 46 degrees

The first two plots show s-curves with characteristic friction-dependent spread and amplitude. The Mars data shows a sudden step, indicating no friction. The encoder seems to have become loose before the Mars data was taken. Further daytime tests Apr 26 were made to measure the motion in the anti-rotation bars during elevation slews, and new stouter bars will be manufactured and installed tomorrow that should reduce the amplitude of the step by a factor of x10, but there will be no further adjustment today - the telescope will show elevation pointing 'flop' whenever the direction of elevation motion is reversed. For tonight at least, IT IS IMPERATIVE THAT LOCAL POINTING BE DONE IN THE QUADRANT IN WHICH OBSERVATIONS ARE TO BE MADE, and observations through transit unaccompanied by local pointing should be avoided.

3 transit runs repeated
The 3 sources above were followed through transit again :

The lack of any discernible step in the first run is quite a surprise, and it still seems possible that the behaviour changes with either time or elevation - or (breaking news) with temperature : combined with the cautious loading, a 10-degree change in temperature during the night, as experienced these past 3 nights, will decrease the loading during the Mars experiments, resulting in the 'flop'.

The 1/8" invar anti-rotation bars were replaced by 3/8" steel rods during daywork Friday (27 April), and the loading on the bars was substantially increased. Transit tracking is scheduled for this evening, with the transparency forecast to be less than perfect and scheduled observing unlikely to occur.

3 transit runs repeated again
3 sources were followed through transit again :

The expected removal of the step at transit has not occurred in all cases. If there was a thermal component to the behaviour of the encoder previously, then it persists. Alternately, an elevation dependence is still a viable conclusion to draw from these transit data.

Allsky pointing (38 measures) was performed in between these transit trackings - all with the empirical correction disabled. Analysis of the change in (elevation) pointing between consecutive measures shows that for those pairs on the same side of the meridian (there are 21 pairs in the East, 8 in the West) their mean difference and s.d. are 0.0 +- 0.5", while for those pairs crossing the meridian (5 E-W and 4 W-E) the steps were 10.3 +- 1.2 and -8.1 +- 3.8, respectively, giving an 'absolute' result of 9.5 +- 1.6. Among the 9 trans-meridianol pairs there appears to be no dependence of the size of the elevation pointing change with either time (temperature) or elevation; but the dataset is rather small.

Correcting all pointing results in the west by -9.5" produces the dataset shown below :

The elevation residuals seem well behaved between elevations 30 and 75 degrees - there is possibly an atmospheric effect in play in these data below elevation 30.

The elevation encoder zero-point has been adjusted (10am HST 28 Apr), and the empirical correction lookup table has been replaced by a bunch of zeroes. The pointing should be OK provided that

• the 9.5" adjustment is made each time the meridian is crossed : +ve when going E-W, -ve when going W-E, and
• allowance for atmospheric effects are made below el=30.

Further deliberation and action will ensue.