[evlatests] Partial Results from tests, 20 and 22 December

[Rick Perley]

    I obtained time on Tuesday evening and this afternoon to check 
sensitivity, etc. 

    I used the now-standard methods:

    a) fast continuum, with 0.4 sec. averaging.

    b) spectral line.  In this case, I used 16 channels over 50 MHz, 1.3 
second
integration, each frequency and each IF separately. 

    This abbreviated report concerns only the continuum data from L, C, 
and X bands. 

    A)  None of the data from Tuesday evening could be used.  Except for 
the first
scan, the antennas were never on source, apparently preferring to head into
the elevation limit instead.  I haven't seen a definitive explanation 
for this bad
behavior yet, but think it very important that one be found. 

    B) The Thursday afternoon data are better, somewhat. 

    i) L-band.  There is much RFI which has greatly degraded the quality.  I
don't believe it is the Sun (we're in D configuration, so are very 
sensitivity to
solar emission at low frequencies), as the RFI is very strong at 1465 
MHz, but
much weaker on the BD side (1385 MHz). 
       IF 16B was bad in a very odd way:  There are four discrete phase 
states,
and the system seems to cycle through them on timescale of ~10 sec.  
These four states have four
unique amplitude states to match -- one of them being zero, the others 
similar to each other.  Weird.
    Because of these problems, I did no further reduction of L-band data.

    ii) C-band.  No visible RFI, so I could proceed.  Unhappily, 16B and 
16 D were both dead. 
Ken claims they worked earlier today. 
    Data quality was good.  The phase jumps between scans (since I 
cycled through the
bands) reported by others were noted. 
    The big 4-second periodicity noted last month has disappeared. 
    But, we have a residual ~8% closure error in amplitude only, on all 
four IFs for
the 14-16 baseline.  The sense of the error is that when the data are 
normally calibrated,
the apparent flux density on 14-16 is 1.08 times that of the other 
baselines.    There is
no corresponding phase effect, nor phase closure. 
    The sensitivity for antennas 14 and 16 appears to be the same as VLA 
antennas.  Thus,
the degradation noted before is still present. 

    iii) X-band.  All IFs for both 14 and 16 worked.  Editing and 
calibration were
straightforwards.  Results:
    Sensitivity seems close to normal. 
    Baseline 14 x 16 has an 8% amplitude closure error, exactly the same 
as at C-band. 
    IF 16C has an extraordinary transient loss in amplitude sensitivity 
which has to be seen to
be believed.  The effect starts about 8 seconds after the beginning of 
the scan, causes the
amplitudes to drop by nearly 50% over ~5 seconds, then recovers 
asymptotically to
full strength over about 20 seconds.  This occured on both observations 
of the source, in
identical manners. 

    A closing note on the closure problem:  This could be due to 
mismatched bandpasses.
I'll be able to discriminate between this and other, less pleasant 
alternatives, by analyzing
the spectral line data.  Tomorrow.