[evlatests] Pointing Troubles

[Rick Perley]

    I ran referenced pointing check run Saturday morning, with excellent 
weather conditions. 
(The overnight winds we experienced in Socorro early that AM had little 
effect on the site). 

    Observations were made in coupled pairs:  A pointing raster 
(hereafter called a 'wiggle')
on a strong source, followed by a 2nd wiggle, at the same band, with the 
referenced pointing turned on.
The goal was to see if the pointing offset determined in the first 
wiggle was removed in the second,
as it should be if the program is working correctly. 

    I did this on three sources, 1800+784, 2202+422, and 0319+415 -- all 
of them unresolved and
with well more than enough flux density -- at C, X, K and Q bands. 

    The results for the EVLA antennas are completely chaotic, whereas 
for VLA antennas, the
results are exactly as expected. 

    I reviewed carefully the results at C and X bands only.  These were 
sufficient to allow me
to conclude that there is no discernible pattern to the results.  I find:

    a) Cases where large offsets are excellently reproduced with and 
without the referenced
pointing applied. 
    b) Cases were large offsets are excellently removed, so the 2nd 
wiggle shows no residual
pointing error.
    c) Cases were a completely new solution was obtained on the 2nd 
wiggle, with no obvious
connection to the first solution. 

    These 'random' events are seen in both Az and  El solutions, 
independently. 

    One separate conclusion can also be discerned:  Antenna 14, and 
C-band, has a large
(2.50 arcminutes) offset in azimuth, constant with elevation.
    The reaction of referenced pointing to this large offset may be 
useful.  Below is a
short table:

             Antenna 14 Az error at C-band
    Elevation     Initial Error      Secondary Error
---------------------------------------------------------------
    81 deg.            2.50                       2.25
    74                    2.55                       1.90
    46                    2.50                       0.63
    33                    2.55                       0.57
--------------------------------------------------------------

    The consistency of the initial (non-referenced) error suggests the 
'wiggle' pattern
is working correctly.  (Ken has looked into the database, and has found 
the reason
for this error, and has already corrected it.  He can explain how this 
happened, if he
wishes).  The 'secondary' error (offset after the initial error was 
purportedly applied)
has an elevation dependence, suggesting a mis-application of the error 
(and not a
random event, as hinted above!) 
    There is evidence of a similar dependence for antenna 16.