[evlatests] L-band 1 GHz-wide spectra

[Rick Perley]

    Michael Rupen supplied me with a prototype WIDAR database consisting 
of a few minutes' observation of 3C273 at L-band.  There are 8 adjacent 
subbands (IFs), each 128 MHz wide, each with 1024 channels.  The data 
averaging was a sprightly 0.1 seconds. 

    The temporal amplitude and phase stability is very good.  There is a 
large delay error, which was removed via the AIPS program FRING -- the 
value of 136 nsec was the same (to within 1 nsec) for each IF (except 
for IF#5, centered at 1560, where the solution was destroyed by the very 
strong satellite RFI in that subband), and at all times throughout the 
scan.  (The solution interval used was 10 seconds). 

    With the removal of the delay-induced fringe wind, we can peruse the 
phase structure over the entire passband.  The results are very 
satisfying:  All subbands, and all channels have the same phase (25 
degrees) to within +/- 5 degrees.   Other than near RFI lines (and at 
1465 MHz where the interim polarizers give the well-known 'suckouts'), 
there are no significant phase slopes in the passband. 
   
    The RFI also gives us an idea of the linearity.  The peak signal 
(inthe INMARSAT bands) is 500 units.  The ampscalar averaged noise is 
about 0.5 milliunits -- a factor of 60 db below the peak.  There is no 
sign of saturation of the signal, nor of any 'splattered' components in 
the noise outside the satellite band. 

    This is all very nice.