[evlatests] Some results from 3-bit test of Friday evening

[Rick Perley]

    A four-hour observation of 3C84 (29 Jy, unpolarized) was made Friday 
night.  The A1C1 and A2C2 IFs were both tuned to 12.976 -- 15.024 GHz, 
and used the 3-bit samplers, while B0D0 was tuned to 14.000 -- 15.024 
GHz , with 8 bit samplers.  The eight IFpairs for B0D0 were thus set at 
the same frequencies as the upper 8 pairs of A1C1 and A2C2. 

    The observations began when 3C84 rose, at dusk, as the goal was to 
determine what effect a large temperature change would have on the cross 
polarization.  As luck would have it, the weather was cloudy (but calm), 
so little temperature change occurred...  Nevertheless, many interesting 
effects were observed, some simple, some not so ...

    1) Two antennas were heavily shadowed at the beginning (no surprise) 
-- but AIPS cannot flag these, since the shadowing antennas are not in 
the array.  However -- Tsys measurements can very easily see when the 
antenna stops 'seeing' reflected ground radiation, so I used the time at 
which Tsys dropped to normal as a reasonable proxy. 

    2) IF 16, for both A1C1 and A2C2 is rubbish on all antennas. 

    3) Delays are very, very stable (and now quite small, thanks to Ken 
fixing these yesterday).  The only variations are of the level  of 
picoseconds, and timescales of minutes, which I presume are tropospheric 
in origin. 

    4) The switched power on all data are in general very good, but ea07 
(in both 3-bit paths, and the 8-bit path) has numerous 'glitches' (bad 
values) which I have not yet tried to correlate with scan changes.  
Editing these are easy, and the remaining data (99.9%, at least) seem 
fine. 

    5) ea07, on the 3-bit paths, is very troubled.  There is real 
trouble with subband 8 for both 3-bit paths (13872 -- 14000 MHz), which 
the fringe levels are very low, very likely due to a very large 
resonance which appears in that band -- at the same frequency for both 
paths, and for both polarizations.  Unfortunately, this subband is not 
overlapped by the 8-bit path, so I can't tell if it is in the 8-bit B 
and D IFs.  This 'resonance' is very strong and at least 10 MHz wide.  
It is present at all times on both polarizations.  It is only seen on 
ea07, and is thus not external RFI.  I can easily see it in the 
cross-power for all baselines to ea07 -- this may well be radiation from 
ea07 getting into the other antennas, hence to the correlator, and 
visible as common source. 

    6)  All antennas show, in the 3-bit paths, internal 'birdies' in 
various places in the total power spectra.  These birdies are not the 
same in RCP as LCP, nor the same between A1C1 and A2C2.  There are 
typically three to four such 'birdies' within each 2.048 GHz spectrum.  
They are not very strong -- typically less than 10 dB above the noise, 
except for ea07, where it can be worse (especially that 'killer 
resonance' in subband 8). 

    7) These 'birdies' are absent from the 8-bit path, with a single 
exception:  ea07, in subband#1 (14000 -- 14128 MHz).  However -- this 
subband is clean in A1C1 and A2C2.  The birdies in the 8-bit path is 
weak, only 2 dB above the noise, and barely visible in cross-power. 

    I ran out of time today to try calibrate these data.  I'll do that 
tomorrow.