[evlatests] Bizarre Pointing Errors

[Rick Perley]

     My recent observations of the nucleus of Cygnus A at K, Ka, and Q 
bands has revealed some bizarre phenomena, clearly due to pointing issues.

     These observations were taken on three days, the first in late 
September, the others over the past week.  Each has ~10 minute long 
observations of Cygnus A at K, Ka and Q bands, which were preceded by a 
referenced-pointing solution on the very nearby calibrator J2007+4029.

     Following the usual calibration, I used the central nucleus of 
Cygnus A (~ 1.2 Jy) as a calibrator source.  At these high frequencies 
and long baselines, the lobes are completely resolved out.  There is 
some residual visibility from the compact hotspots, mostly noticeable in 
K-band, but at Q-band, the hotspots lie right on the antenna null, so 
all the VLA 'sees' is the nucleus (and some weak surrounding structure, 
which is what we are interested in scientifically).

     To summarize the above paragraph:  To all intents and purposes, 
Cygnus A is a 1.1 Jy calibrator source, very close to the standard 
calibrator J2007+4029.

     The results from performing a calibration on the nuclear source 
were quite a (disturbing) surprise.  I have attached four gain plots to 
illustrate the weird phenomena.

     There are two phenomena to mention:

     1) Oscillating off-source pointing.

     Plot ea08+25-Q-Day3.png  shows the Cygnus A nucleus gain solutions 
from ea08 and ea25, at Q-band.  A level of ~1000 is what we hope to 
see.  (At Q-band, we're delighted if the values are between 900 and 
1100).  Values greater than 1100 indicate the antenna is off-source.  
Roughly, the offset in arcseconds is given by:  77*sqrt(log G), where G 
is the gain value on the vertical axis).

     There are six observations, each 10 minutes long.  The gain 
solution time is 6 seconds.  The elevations of the six are 83, 79, 72, 
66, 60, and 53 degrees.

     Immediately apparent are sinusoidal-like gain variations at some, 
but not all, of the observations.  The period is a few minutes, and the 
amplitude of the gain corresponds to offsets of 70 arcseconds (for ea25) 
and 30 arcsecond (for ea08).  We know these are pointing, since the 
amplitude scales with frequency in the way expected, and the same 
patterns are seen on the preceding observation at K and Ka bands, with 
the same offsets (in arcseconds).

     These oscillating patterns are quite common, especially with these 
two antennas.  A more spectacular is shown in the next plot:

     ea08+25-Qbig.png

     This one is from a different observation day than the preceding.  
The elevations are 58, 64, and 70 degrees.  The larger offset is almost 
1 arcminute.

     The third plot shows how this effect can 'come out of the blue'.

     ea25-K.png

     This one, from K-band, shows how ea25, in the middle of the 
observation scan, suddenly decided to head way off the source -- by 
*more than two arcminutes*, then begin the same oscillatory pattern.  
Again, we know this is a pointing issue, since the amplitude of the 
effect is much higher with increasing frequency.

     2) Antennas which decide to point elsewhere.

     Similar to the plot noted above, is the phenomenon of antennas 
moving off the source, as shown in the last plot

     ea15+21-Q.png

     Here, the antennas start at the correct on-source position (since 
the gains are ~1000), but after a few minutes, clearly start moving 
offsource, ending up, after a minute or two, pointing off by an 
arcminute or more.  In this last plot, ea15 did this on the first scan, 
and ea21 did it on the second scan.  The short time-scale over which 
this large error occurs is what is surprising (to me).  Since the Q-band 
observation is the last in the sequence (a calibration, and pointing 
calibration follows), I can't say if the offset carried over to the next 
observation.

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

       Ken and I discussed these results.  He notes that the oscillatory 
pointing offsets are likely due to encoder calibration errors -- these 
are normally found by applying the procedure 'PN3dB'.  He ran this last 
night, and found a small correction for ea08, but none for ea25.  So if 
an encoder issue, it's time-variable...

     In my opinion, this is a serious problem -- high fidelity 
observations of large sources are impossible with this kind of pointing 
behavior.  And it is much more common than we thought.




-------------- next part --------------
A non-text attachment was scrubbed...
Name: ea08+25-Q-Day3.png
Type: image/png
Size: 26058 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20171027/0ee80c55/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ea08+25-Qbig.png
Type: image/png
Size: 24383 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20171027/0ee80c55/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ea25-K.png
Type: image/png
Size: 19442 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20171027/0ee80c55/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ea15+21-Q.png
Type: image/png
Size: 19673 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20171027/0ee80c55/attachment-0007.png>