[evlatests] Spectral Line Noise, cont.

[Rick Perley]

    Some clear patterns are emerging now. 

    For the impatient, the bottom line appears to be:

    We have troubles with 12.5 MHz BW, in spectral line modes.  All 
other tested modes and BWs appear correct.  (But, not all modes and BWs 
have been checked yet). 


    Details:

    This morning's test was done at X-band, on a strong calibrator 
(1800+784), and a nearby blank field. 
    Modes checked were:

    1) Continuum, BW = 50 MHz.
    2) PA, BW = 12.5, 3.125, 0.78 MHz
    3) PB, BW = 12.5, 3.125, 0.78 MHz
    4) 1A, BW = 50, 12.5 MHz. 

    The data are all of exceptional quality.  No flagging was required 
(except for the 50 MHz BW data in mode 1A, where the VLA and EVLA were 
incoherent).  The data were loaded as correlation coefficients, and 
calibrated using the known flux density of the calibrator.  All 
statistics and images were made on the noise field data (except for mode 
PA and BW = 12.5 MHz, where the problem reported earlier caused the 
blank field data to appear as if it were mode PB.  It's still not 
certain what mode that scan was at). 

    For all tests done at BW other than 12.5 MHz, the derived 
sensitivites -- as reported via histograms of the real and imaginary 
parts of the visibilities, and as images made in all four Stokes 
parameters -- gave and SEFD between 290 and 320 Jy -- exactly the 
correct value. 

    For the 12.5 MHz spectral line modes, the results are:

    Mode = PA:  The X-hand visibilities (RL, LR) gave normal 
sensitivity.  The parallel hand visibilities (RR,LL), gave an SEFD = 730 
-- at least two times too high.  (Important note:  because I didn't have 
the noise field to work with here, the SEFD values are less reliable, as 
they were derived from the strong source.  And since PCAL cannot be run 
on a single observation, the X-hand histograms and images are 
contaminated with leakage flux.  However, the factor-of-two difference 
between X-hand and parallel-hand results is secure. 

    Mode = PB:  All four correlations provide sensitivities a bit more 
than twice what they should be -- SEFD = 600 Jy. 

    Mode = 1A:  The RR corrrelation provides a sensitivity about twice 
what it should be:  SEFD = 530 Jy. 

    What next?