[evlatests] Bizarre behavior in S-band

[Rick Perley]

    ... just when I thought I'd seen it all, up comes this one ...

    On the 17th of my 20 S-band databases (each about 1 hour long, taken 
between early November and early January), the observation was of a 
source near dec = 10.  The safe zone, or so I thought ...

    The RCP visibilities for the calibrator look fine, and the 
calibration proceeded without apparent incident.  But when I compared 
these gain solutions to those in LCP, a dramatic difference appears:  
While the RCP solutions are all quite constant over time, the LCP 
solutions -- for all antennas -- varied by factors of a few.  And the 
variation appeared to be exactly same for all 16 subbands! 

    The switched power tables give us a fairly clear view of what the 
system was doing (although not the cause).  Attached are four .png plots 
from ea02:  The R and L powers for subband #8, and the R and L powers 
for subband #12.  I picked these two for being clear of incidental 
occasional RFI -- all others behave in identical ways.  And all antennas 
behave very similarly. 
    Each plot has three frames:  The top is the synchronous switched 
power (PDif), the middle is the total power (PSum), the bottom is the 
computed Tsys (ratio of the preceding two times a constant). 

    1) EA02-8-RCP.png

    Antenna 2, subband 8, RCP -- IF 'A'.  There is little to suggest 
alarm.  The fluctuations in the total power (middle) are odd, but small 
-- of order 1%. 

    2) EA02-8-LCP.png

    IF 'C'.  Totally crazy!  Both the PSum and PDif values are smoothly 
changing by factors of nearly two.  The values when on the calibrator 
(the short dashes) follow a very different function than of the source 
(the long track).  Yet these are only a couple degrees apart in the 
sky.  We know this is purely a gain change, since the system temperature 
(bottom panel) is affected to a much smaller degree (and may be 
artificial).  The cause cannot be internal, since the calibrator and 
source values are so very different.   Note in particular that the 
highest value of system temperature coincides to when the total power 
was a *minimum*! 

    Note also that the mean total power (6 counts in RCP, 3 counts in 
LCP) are much lower than the target value of 14. 

    3) EA02-12-RCP.png

    Same antenna, RCP, but subband 12 -- IF 'B'.  Again, there is little 
here to cause concern, other than a rather peculiar, but small, 
variation in total power. 

    4) EA02-12-LCP.png

    Same craziness if this IF 'D', as in 'C'.  Note here that there is 
no variation at all in computed Tsys, other than a small bump just after 
05:00 (also seen in RCP) which is likely due to other causes. 
    The total powers in this IF pair are higher than in the other, but 
still lower than the target. 

    Comments:

    1) The cause is *not* in the visible spectrum between 2.0 and 4.0 
GHz.  The spectral plots are smooth, with only the usual RFI.  Nothing 
outstanding. 
    2) I think it very unlikely this is due to a moving satellite -- the 
run lasts well over an  hour, and the behavior remains qualitatively the 
same throughout. 
    3) Another argument that the source is not a narrow-band emission 
within the passband is that if it were, it would only affect one of the 
IF pairs -- the one the emission lies in.  But this phenomenon affects 
both the C and D IFs almost equally in all subbands.  And the effect is 
visible -- although at a much reduced level -- in A and B (RCP). 
    4) The source is not local to the VLA site -- all antennas show 
exactly the same effect, to about the same degree. 
    5) I have seen behavior like this before -- in a similar set of 
observations (of the same sources in fact) taken a year ago.  I can't 
yet tell if this is restricted to a particular declination range -- I 
think this unlikely, but since I have no clue of a possible explanation, 
all crazy ideas are welcome. 

    Now for the Good News: 

    These gain variations are *almost perfectly* corrected by applying 
the switched power.  

    Speculations:

    1) How about a variation in LO power?  It would have affect LCP much 
more than RCP (no idea how). 
    2) Can the origin be RFI lying outside the 2 -- 4 GHz passband?  
Same objection as that above applies here -- the far-sidelobes of the 
antennas are highly elliptically polarized, and there is no reason to 
think such RFI would be polarized in exactly the same state as the 
sidelobe in order to dump significant power only into the LCP side.  If 
this is the cause, it would have to be fixed in space -- either on the 
ground or on a geostationary satellite. 

    All ideas are welcome!


-------------- next part --------------
A non-text attachment was scrubbed...
Name: EA02-8-RCP.png
Type: image/png
Size: 66764 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140116/434816b0/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EA02-8-LCP.png
Type: image/png
Size: 45923 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140116/434816b0/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EA02-12-RCP.png
Type: image/png
Size: 64146 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140116/434816b0/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EA02-12-LCP.png
Type: image/png
Size: 52869 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140116/434816b0/attachment-0003.png>