[evlatests] A strange S-band spectral feature ...

[Rick Perley]

    In reviewing my 'book of plots', Michael noticed a curious 
replication which had gone unnoticed by me. 

    In 5 of the 8 subbands between 2.0 and 3.0 GHz, a 10-MHz-wide 
'block' feature appears in exactly the same place in each spectrum, 
relative to the subband boundaries.  (That is, it's in the same spectral 
channels).   It's clear from the shape and location that all of these 
are 'echos' of the peak of the DSR signal at 2330 MHz.  The spectral 
windows in which these echos appear are:  2, 3, 4, 5 (very weak), and 
7.  (I count these from 1 to 8).  Spectral window 3 contains the original. 

    Perhaps even more curious is the distribution of this echo amongst 
the baselines.  There were 14 antennas in the array observing at S-band 
-- giving 98 baselines.  Of these, only 26 showed any hint at all of the 
'echo' on subband 7 (where the averaged 'echo' is quite strong -- about 
1 dBabove the noise). 

    Getting even more curious:  The presence or absence of the 'echo' 
certainly doesn't close.  I could find only one tendency -- antennas 
closer to the CB had more baselines affected.  Antennas 6 7 8 9 and 10 
-- all at or near the ends of the north and east arms, had only a single 
baseline affected, while antennas 14, 15, 22, 24 and 26 -- all on the 
west arm or center of the north arm, had about half their baselines with 
the signal. 

    But nothing is simple in this game.  The baseline with the strongest 
'echo' was 7 x 27 -- the pair at the end of the east arm.  So I suspect 
this proximity correlation is spurious. 

    It is more likely that this echo is somehow related to the 
'lumpy-bumpy' phenomenon -- subband 3 clearly saturated for many, 
perhaps all, of these baselines showing the 'echo' -- but why would a 
saturated subband cause replicants in the other subbands?  The 'echoed' 
signal is about -40 dB below the original.  I thought the anti-aliasing 
was better than this?