[evlatests] Modcomp-Free tests, Friday afternoon (25 may)

[Rick Perley]

    The first hour of Dynamic time, following Software Time, was used to 
check data quality.  Three 20-minute blocks were set up, for L, C, and X 
band observations of the strong calibrator 3C147.  Continuum mode, with 
12.5 MHz bandwidth (this to reduce losses due to the remaining delay 
errors).  I ran this with 0.418 seconds averaging, to check short-term 
behavior. 

    General Results:

    Overall, data quality was very good, with very few short-term 
problems on either VLA or EVLA antennas.  Many antennas show long-term 
drifts in amplitude, which are greater at the higher frequencies.  I'm 
presuming these are due to pointing errors.  Many antennas show poor 
sensitivity at some bands -- I'm presuming these are due to incorrect 
delays.  I expect these elementary issues will be resolved over the next 
few days. 
    Other than the (remarkable!) exceptions noted below, phase stability 
was excellent, with the variations seen clearly due to atmospheric 
irregularities. 
    Especially notable is that there was (almost) a complete absence of 
the 'short shallow drops' that have been noted for a long time on EVLA 
antennas. 

    Problems Noted, which are common to all bands:

    1) The EVLA x VLA incoherence in IF2, noted at L-band in prior tests 
is shown here to be present at all frequencies.  Ken has shown this to 
be an incorrect frequency setup for the VLA antennas. 

    2) The phase in IF 'D' (LCP in IF 2) *only* is slowly rotating at a 
rate, typically one turn in a minute, which is clearly proportional to 
frequency and to baseline length.  It looks like a miscalculation of the 
fringe rate.  But also note that this slip is not seen in any other IF. 

    3) Antenna 14, in IF 2 (IFs B and D) has two different phase states, 
separated by about 50 degrees.   The antenna spends about 75% of the 
time in one, 25% in the other.  The transition between them takes very 
nearly exactly 0.8 seconds, as a single intermediate phase value is 
*always* seen midway between the two states.  The transition is always 
identical in both polarizations. 

    4) Antenna 24, all IFs, show a phase which is 'all over the place'.  
The antenna stays in any one phase state for only a few seconds, then 
abruptly transitions to another.  The transition seems to take 0.8 
seconds, as an intermediate phase point is always seen between the 
adjacent stable states.  Unlike antenna 14 (above), the phase changes 
appear random.  However, all four IFs 'jump' at the same time with an 
identical phase change. 

    5) There were no fringes from IF#2 in Antenna 19.  (I think this is 
a known problem). 

    Band Dependent Problems Noted:

    1) X-Band:   Antennas 3, 20 and 22 gave no discernible fringes.  
Delays?  Pointing?

    2) C-Band:  No fringes from antenna 19, 21 and 23.  No receiver, 
presumably. 

    3) L-Band:  The system did some very odd things at the beginning of 
the observations at this band.  This band was the first to be observed 
of the three -- the transitions from L to C, and from C to X, were 
uneventful.  I don't recall what run preceded this test, but the 
transition was far from smooth:

          a) The bandwidth changed from 12.5 to 50, then back to 12.5 
MHz, over the first ten or so minutes.  The frequencies changed, both 
between the 12.5 and 50 Mhz scans (as might be expected), but also 
between the two 12.5 MHz BW scans. 
          b) After settling down at the right frequency and bandwidth, 
remarkable changes took place:  The integration time was incorrect -- at 
3.33 seconds, not the 0.418 seconds requested.  (In the observe file, 
the only difference between these is the presence of an 'S' in the 
appropriate column.  The observe file had the 'S', but apparently the 
system didn't get it).  After spending 5.5 minutes with the wrong 
integration time, the system abruptly switched to the correct value, 
with large phase jumps for all antennas.   The conversion was nearly 
instantaneous -- a single 3.33 second record is missing, then the 0.418 
second data begins.   The confusion may be related to an earlier test (a 
few hours earlier!) which observed the same source, but in spectral 
line, with 3 second averaging.  The operators log noted the change in 
program, with no change in source. 
          c) The most remarkable event occurred 12.0 seconds after the 
abrupt change to the correct integration time:  All phases, on all 
antennas and IFs, abruptly switched sign.  This clearly means the sign 
of the 'SINE' correlation changed sign.   But why? 
          d) Forty-eight seconds later, there was a ten-second gap in 
the data, after which the phases and amplitudes changed.  Some 
amplitudes went up, others went down, typically by ~10 percent. 
          e)  Following this event, stability was restored on all 
antennas, the the remaining ~5 minutes of observation passed 
uneventfully, other than the phase issues on antennas 14 and 24, noted 
above. 

     Other than these issues, the only other significant problem noted 
was the presence of a modest number of dropouts on antenna 4.  All IFs 
drop together, for single (0.418 second) integrations.  No other antenna 
showed anything like this.