[evlatests] Stability Tests on Sept 29, 2005

[Rick Perley]

    The last attempt at determining baselines for antennas 14 and 16 
failed, as the phases
seemed non-sensical.  This encouraged a test, involving cycling around a 
few sources to
see if the EVLA antennas phases 'connect' from source to source. 

    The short answer is:     They do not. 

    Details:

    The test ran for 30 minutes, starting at 1PM.  I used C-band, as the 
weather was
cloudy with large T-storms passing over at the end of the run.  A 1.6 
second averaging
was used, and 50 MHz BW, continuum mode. 

    Antenna 16 was dead at all IFs -- I think it was stowed throughout 
with Rob doing
some system work. 

    Antenna 14 worked in only the C IF channel.  Channel A was dead, 
channel B
provided correlation values at least 8 times too high (reason unknown), and
channel D gave values 100 times too high, with phase = -180 and steady.  
Clearly
not working. 

    It should be noted that no attempt was taken to set the powers at 
the appropriate
levels.  The only two gurus who know how to do this (Barry, Ken) were 
both out. 

    Seemingly useful data were obtained from  channel 14C. 

    A) Amplitude behavior.   Dropouts, last a single integration record 
(1.3 seconds),
where the correlated amplitude drops by 10 to 40 %, were noted 
throughout.  When
they occur, they are always separated by 10 seconds.  But they do not 
happen *every*
10 seconds (in fact, perhaps only 5% of the data are so affected), and 
when they occur,
they do not necessarily fall on 10-second 'ticks'.  Each scan was 
different -- some had
no dropouts, some had many.   There is no corresponding phase effect for 
these
amplitude dropouts. 
       Other than the dropouts, the amplitude scatter in the solutions 
is about the same
as a VLA antenna. 

    B) Phase behavior. 
          i) Within any one scan (of a single source), the phase 
behavior is the same as
an adjacent VLA antenna. 
          ii)  There is no connectivity in phase from scan to scan.   
However, there is
good evidence that the same phase returns for a given source.  (The 
weather became
very poor at the end, so this conclusion is a bit tentative).   Hence, 
there could be a
Bz error in the baseline.  However, if so, it is *very* large, for a 
change in declination
by 1.7 degrees changed the mean phase by over one radian -- I figure a 
40 cm error
in Bz is needed. 

    More data are needed (in better weather) to confirm this.