[evlatests] EVLA Pointing

[Rick Perley]

    Another test done was to check our 'blind' pointing, and the 
repeatability of referenced pointing. 

    I did a pointing check on a far northern source (dec 80, elevation 
about 40 degrees) at 2AM.  the weather was clear and calm.  The 
solutions were then applied, and another pointing solution made.  This 
was done at X-band.  The latter solutions represent the best we could 
expect from referenced pointing. 

    This pair of observations were repeated at noon, at nearly the same 
az and el.  The weather was now clear, but slightly breezy (10 - 15 
mph).  Solar heating had been in play since dawn.  The results clearly 
show how much we're affected by solar heating of the structures. 

    I include four histograms, showing the pointing error distribution 
for Az (top) and El (bottom) for the four situations (Night-Blind, 
Night-Ref, Day-Blind, Day-Ref). 

    Basic conclusions are:

    1) Blind, night-time, ideal weather:  the required 6 arcseconds 
'rms' error is close to being achieved.   As always, azimuth is better 
than elevation.  There is a notable mean offset in elevation of about -5 
arcseconds.  (Possible explanation below). 

    2) Night-time referenced pointing is really good -- rms is 3 to 4 
arcseconds.  There is now an elevation offset of +2 arcseconds. 

    3) Blind, day-time:  A remarkably wider distribution -- pk-pk is 45 
arcseconds in Azimuth, and over 60 arcseconds in Elevation.  There is 
now a large elevation mean offset of about 25 arcseconds!  Clearly the 
effect of solar heating of the yoke.  But the large spread and offset in 
azimuth (10 arcseconds) has no easy explanation. 

    4) Referenced day-time:  A huge improvement -- but not as good as 
night-time.  The spread is about twice the night-time.  Perhaps an 
effect of the wind?  Note the offset in elevation -- about -5 arcseconds. 

    From this, it seems that the referenced pointing algorithm is 
*overcorrecting* for the elevation offsets:
           - at nighttime, the blind offset is about -5 arcsecond, and 
the corrected offset is +2
           - in daytime, the blind offset was +25 arcseconds, while the 
corrected offset is -5. 

    As for the origin of the night-time offset:  My theory is that the 
pointing model 'freezes' in a value of the offset depending on the 
left-over temperature gradient from the previous day.  If (for example) 
the pointing data were taken in early evening following a warm sunny 
day, it would be possible for some of the differential heating (causing 
a global pointing error) to put into the model.  Later night-time 
observing, when the antennas are in better thermal equilibrium will be 
off by a small negative value. 
    Of course, with the sun bearing down in the morning, one side of the 
antennas is preferentially heated, causing the large noon-time offset, 
as observed.  No surprise here. 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: OnNoon.png
Type: image/png
Size: 12477 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130228/5f944abc/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OffNoon.png
Type: image/png
Size: 13054 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130228/5f944abc/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OffNight.png
Type: image/png
Size: 13168 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130228/5f944abc/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OnNight.png
Type: image/png
Size: 11617 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130228/5f944abc/attachment-0003.png>