[evlatests] New RFI Sweep -- interesting new results

[Rick Perley]

    Ken did an 'RFI Sweep' yesterday at noon.  Only the S-band equipped 
antennas were included:  6, 7, 8, 9, 10, 12, 14, 15, 16, 22, 24, 26, 27, 
and 28.  Some important new results are noted:

    1) He arranged that the script actually provide us the high end of 
Q-band, so I can now verify the presence of a 128 MHz comb at the upper 
end of that band.  Many, but not all, antennas show this comb.  *** By 
far the strongest*** are from antennas 10, 14, and 22 -- antennas that 
are nowhere near to each other, so responsible the mechanism is 
certainly an internal one.  The second tier of self-polluting antennas 
include 7, 8, 12, 14, 27, and 28. 

    2) The 32-MHz replicating comb is now greatly diminished.   
According to Ken, this was due to an incorrect fshift which was used for 
the older observations.  The new value has reduced this resonance (1/4 
of the subband width = 32 MHz for this experiment) from the level seen 
last week of -27 dB to a level of -42 dB.  Probably good enough? 
 
    3) The 29440 MHz 'super-RFI' line is present 'in spades' -- but now 
without the 32 and 64 MHz echoes.  This birdie is on all antennas, with 
about the same strength -- typically 30 dB above the noise.   Ken has 
proposed a mechanism -- I think it is important to verify his suggestion. 

    4) The 'lumpy-bumpy' phenomenon is present in a major way.  To 
review:  'lumpy-bumpy' means the spectrum in a single subband is 
elevated above the noise by typically 40 dB (this is HUGE) and has a 
'bumpy' spectral structure with typical scale of ~5 MHz.  In last week's 
final test, 6 baselines showed this problem, all of which were 
associated with antennas 22, 26, 27 and 28.  Yesterday's test shows the 
phenomenon on *most* of the baselines.  The root cause of this problem 
is clear -- the single subband in which this occurs is that containing 
the digital satellite radio signals (Sirius and XM).  This signals are 
*strong*!!!  The total power within the RFI zone (2310 through 2350 MHz) 
is more than 100 times the noise power within that subband.  (This also 
means this power is about 6 times the entire bandwidth noise power!)  
Not all baselines have been affected, but those that are are definitely 
associated with specific antennas.  Although we should probably not be 
surprised that this subband is messed up by these signals, I think we 
need to understand a bit better just what is happening, and learn if 
there is a way to prevent this.  I presume that the internal 7-bit mode 
should help here. 
    I'll note here that the other subbands seem completely immune to 
this super-signal, with normal sensitivity and performance.