[evlatests] Missing Scans in wideband test data

[Rick Perley]

    We ran a 1 hour test on Monday night, in 'wideband' mode, 
alternating between X and C bands, while also including C-band 
referenced pointing in narrowband mode. 

    Large fractions of the expected resulting data are missing.  I 
filled the database with no flags applied.  FITLD reported a huge 
fraction of integer zeros: 

    24% for the narrowband K-band data
    22% of the narrowband X-band referenced pointing data
    47% of the wideband X-band data
    13% of the wideband K-band data. 

    The OPT set up the following:

    a) Four short scans on 3C286 in narrowband mode
    b) A single short scan on 3C286 in wideband mode (K-band). 
    c) Observation of J2007 at K-band (wide) (referenced pointing not 
applied)
    d) Referenced pointing on J2007 at Cband
    e) A short observation of J2007 at K-band (referenced pointing applied)
    Then, two loops around the following:

    f) Reference pointing on J2007 at C-band (narrowband)
    g) Observation of J2007 at Xband (wide)
    h) Observation of Cyg A at Xband (wide)
    i) Observation of J2007 at Kband (wide)
    j) Observations  of Cyg A in three position (E, Nucleus, W) at 
K-band (wide). 

    What we got is rather different:

    a) The initial narrowband observations of 3C286 are present. 
    b) The short wideband scan of 3C286 at K-band is present.
    c) There is then a 6 minute gap with no data at all.  This should 
have been both the slew, and the wideband K-band observation (without 
referenced pointing).  The duration requested was 6 minutes, which is 
more than enough for the slew and observations -- probably the problem 
is the 3-minute limit for any given scan -- nor matter what that scan is 
doing.  (i.e. -- slewing counts!)
    d) Referenced pointing at c-band was present.
    e) The wideband observation of J2007 at K-band, with referenced 
pointing applied is missing -- and was apparently *replaced* with a 
C-band observation in narrowband mode!
   
    More gaps appear with the loops:

    f) Referenced pointing on J2007 at Cband was present.
    g) The observation of J2007, in wideband at Xband was present.
    h) The following observation of Cygnus A at X-band is missing (a 
3-minute gap)
    i) The K-band observation of J2007 was present.
    j) Only one of the three observations of Cygnus A (the first one, 
Cyg-E) was present.  There is a 6-minute hole corresponding to where the 
other two observations should have been.

    On the second time around, the calibrator observations and the 
referenced pointing were present, as above, but
Cygnus A at X-band was again missing, and the observation of Cygnus A at 
the west end was also missing. 

    The 47% integer zeros reported for the X-band wideband data closely 
matches the fraction of time that should have been spent on Cyg A at 
that band.  But the 13% zeros for the wideband K-band data is quite a 
bit less than that expected from the missing scans -- 38%. 

    So -- what went wrong here?