The JCMT TSSs have recently reported that the pointing with RxA has degraded:

Date: Fri, 24 Nov 2006 23:33:02 -1000
From: Jeff Cox 
Subject: [Jcmt_faults] [20061125.007]

Scans #5,6,7,16,17 show unusually large UEL pointing offsets for W3(oh). 
Seems like the more pointings I did, the more negative the values would 
become. Then when going to OMC1, the values changed by 8 or more arcsecs. 
Pointing does not seem stable at all. Time lost is the accumulation of 
extra pointings that were necessary.

              ------------------------------

Date: Sun, 26 Nov 2006 00:54:50 -1000
From: Jeff Cox 
Subject: Large UAZ and UEL

Pointing offsets seem unusually large. Tried 6 successive pointings on 
W3(oh).(I know not the best pointing source, but sometimes gotta use it). 
The UAZ/UEL from the previous source was -7.48/-0.14. By the time I 
finally got a good fit on W3(oh), values were 13.15/-5.22. See scans 
#23-28

              ------------------------------

Date: Mon, 27 Nov 2006 00:15:47 -1000
From: Jeff Cox 
Subject: [Jcmt_faults] [20061127.002] Pointing problems

The pointing seems anything but stable and repeatable. At certain AZ/EL, 
no problems. But at others, not good. Checked RxA, TMU, and SMU, no 
problems there. Scans #28-31 on L1551-irs5 were very frustrating. Tried 
all different manners of reducing, and still did not get good results. 
Went to CRL618, and had to repeat pointing 2 times to get a stable point. 
Did try all manners of reducing on that one too. On most pointing sources, 
I have had to repeat at least once to get a good result....Am I doing 
something wrong, or have others noticed problems with the pointing?

              ------------------------------

Date: Mon, 27 Nov 2006 08:27:17 -1000 (HST)
From: Jim Hoge 
Subject: Large pointing offsets

        In my opinion, pointing has been quite poor ever since the
earthquake.  I suspect our rms is close to 5 arc seconds when we (TSS) are
used to an rms closer to 2 arc seconds.

        Note that with the new system the pointing is a bit of an art: 
        some modes appear not to be very good in pulling in pointings 
        that are off. Furthermore, with all the options available you 
        have to make sure to realize what may or may not work.

        OMC1 and W3(OH) are VERY extented sources and I would not expect 
        them to work using line-mode, especially not when chopping off 
        only 60". You may be able to point on them using continuum mode 
        (of course, don't fit a baseline in that case), but even then I 
        would only accept the result if it is repeatable. Note that OMC1 
        and W3(OH) are recommended against to be used for pointing.

        Given last night's weather, there is no reason not to use more 
        compact line-sources for pointing. Using those, the pointing has 
        been very stable and repeatable over the past couple of months.

    UT dates    Rx   No. of points     CA    dS      IE    dZ   Notes
   -----------  --   -------------   -----------   -----------  -----
   20061105      A        13         +1.7 +- 1.7   -2.9 +- 2.1
   20061106      A        13         +0.6 +- 1.9   -0.7 +- 2.6
   20061107      A        11         +1.1 +- 2.8   +1.3 +- 2.3
together
   20061105-07   A        37         +1.1 +- 2.2   -1.9 +- 2.6

   20061112      A        17         +1.8 +- 3.7   -1.2 +- 3.6
   20061113      A        22         +1.9 +- 3.9   -2.3 +- 4.7
   20061125      A        16         -0.2 +- 2.6   +6.0 +- 7.8
   20061126      A        20         +3.8 +- 5.8   +2.9 +- 3.3
   20061127      A        23         -0.8 +- 4.0   +2.2 +- 4.8

The raw performance certainly looks horrible in the period since the warm-up (Nov 14-21), but it was v.poor on UT12 & 13 also. However, after removing data of w3(oh), n7538irs1 and omc1 :

    UT dates    Rx   No. of points     CA    dS      IE    dZ   Notes
   -----------  --   -------------   -----------   -----------  -----
   20061112      A         8         +0.9 +- 1.7   -3.1 +- 3.2
   20061113      A        12         -0.8 +- 2.1   -4.0 +- 3.8
   20061125      A         7         +0.4 +- 1.9   -0.3 +- 4.9
   20061126      A         8         +2.6 +- 2.6   +3.6 +- 2.4
   20061127      A        17         -0.5 +- 3.3   +1.8 +- 5.4
   20061127*     A         9         +1.0 +- 1.5   -1.5 +- 3.1 

   20061112-27   A        44         +0.7 +- 2.3   -2.0 +- 4.4

The line marked * also shows the impact of removing measures of L1551 taken at high elevation, and with very inconsistent results. The last line shows the combined data, without any attempt to equalize the nightly collimation changes. The data certainly still seem ratty (in elevation in particular). A full TPOINT 7-parameter optimization doesn't improve the residuals enough ((dS,dZ)= (2.4",3.7")) to suggest that a model change is warranted.

Conclusion so far
Even restricting analysis to exclude targets with morphological complexity leaves a marked degradation in elevation pointing not easily explained by atmospherics, nor solvable by a model change.

20061128

    UT dates    Rx   No. of points     CA    dS      IE    dZ   Notes
   -----------  --   -------------   -----------   -----------  -----
   20061128      A        12         +1.7 +- 1.9   -0.4 +- 3.1