[evlatests] New 3-bit test data

[Rick Perley]

    As noted by Ken, power levels were readjusted this afternoon, and a 
new test run.  The file alternated between 3C286 and a blank field, for 
one hour.  As before, the 8-bit and 3-bit paths were exercised in 
parallel, with 3-bit running through AC and 8-bits through BD.   The 
frequencies were lowered by 1 GHz, to avoid the strong RFI seen at 6.0 
and 6.25 GHz.  The new test covered 4.5 to 5.5 GHz.

    Unfortunately, the 3-bit data are useless, as clearly there is 
something not set up correctly.  There is no signal at all visible on 
the odd-numbered subbands, in any polarization.  There are weak signals 
present in the even-numbered subbands -- but the spectral shapes within 
those subbands are all wrong.  The amplitudes are about a factor of 20 
too low.  

    The 8-band data look fine. 

    What happened?