[evlatests] Pointing Offset Results

[Rick Perley]

    More information from the elevation-gain test. 

    The observations were taken on a single source (J2202+4216 -- about 
10 Jy) with referenced pointing for each band, applied to itself.  This 
means:

    For X-band:  A  normal referenced pointing was done, and applied.
    For Ku-band:  An differential referenced pointing, using X-band as 
the reference, was done and applied.
    For K-band:  The Ku-band solution was applied, and the difference 
measured and applied.
    For Ka-band:  Ditto above, using K-band as reference.
    For Q-band:  Ditto above, using Ka band as reference. 

    The observation started with the target source at transit, and ended 
at elevation of 10 degrees.  Weather conditions were ideal -- calm, clear. 

    Attached are plots of all the results.  For each antenna is plotted 
the solutions -- when reviewing them, note the order given above.  (Also 
note that the plots claim the source was 3C84.  It wasn't -- my bad).  
    A few global comments:

    1) Many antennas show a strong elevation dependency in the X-band 
results.  This can only be due to a model error.  Ken, on the basis of 
last night's pointing run, has found the origin, and inserted corrections. 
    2) If a receiver is missing (such as at Ku-band for a number of 
antennas), it seems clear that the subsequent solution (i.e., at K-band, 
if the Ku-band receiver is out) has forgotten the preceding solution 
(X-band, in the case noted here), so that the new solution is an 
'original' -- with no preceding solution remembered.  I don't think this 
is good behavior (but others may argue otherwise). 
    3) For most antennas, the residual pointing corrections are very 
small, with little evidence for notable collimation errors.   And there 
is little evidence for a time (or space) dependency in the collimation 
offsets.  However, there are exceptions.  We should run this program 
once every few months to determine how important these exceptions are. 
    4) There is an interesting anti-correlation between Ku offsets and K 
offsets, evident in many antennas.  The secondary solution at Ku-band 
often    makes a ~5 -- 10" change -- which is promptly removed in the 
subsequent K-band secondary solution (which has applied the Ku-band 
offset).  Is this an erroneous Ku-band solution?  Or is there something 
odd with Ku-band pointing?  (I don't know the answer). 
    5) The 'rms' error in the pointing residuals, for most antennas, -- 
especially at Ka and Q bands -- is very small -- probably 3 or 4 
arcseconds, and maybe better.  This is in excellent agreement with the 
post-fit scatter in the elevation gain curves at the higher frequencies, 
determined from the same data.   Basic conclusion:  referenced pointing 
works very well under these observing conditions