[evlatests] More on full 8GHz Bandwidth

[Ken Sowinski]

On Fri, 16 Mar 2012, Rick Perley wrote:

>    1)  All seven fully-equipped antennas fringed very well on all eight
> IFs.
>
>    2)  As expected by Ken, the odd subband issues seen yesterday
> disappeared in last night's tests -- these were evidently caused by
> something left over from the prior phased array tests.   But not all
> subbands are at the right levels, however, with subbands 0 and 15 (or 1
> and 16 in AIPS-language) usually far off -- this must be a requantizer
> issue.    A general pattern is easily seen, although not all antennas
> show this equally.
>                For the BD IFs (B1, B2, D1 and D2), it is always subband
> 0 which is low, typically by ~5 to 10 dB.
>                For the AC IFs (A1,C1,A2 and C2), it is always subband
> 15 which is low, typically by 5 to 10 dB.
>    This looks like the 'DC offset' problem.

This is caused by the power in DC dominating the power in the
subband so that the requantizer cranks down the gain.  We need
to discuss whether we want to address this by removingthe offset
arithmetically in the subband adjacent to DC, or use a FIR which
suppresses DC as we do for the 8-bit samplers.

>    3) Six of the seven antennas had correct power levels.  But ea25 did
> not.  The A2C2 and B2D2 IFs were low by about 4 dB.  Furthermore, A2 was
> clearly much noisier than all others -- most easily explained if the
> system temperature is much too high.  Bad attenuator setting?  Strange
> FE bandpass?  (This is unlikely, as A1 -- at an adjacent frequency --
> looked fine.

If A2 is noisier thah A1 it is unlikely to be high T_sys.
Perhaps a T304 problem, or software mis-setting the attenuators
by a lot.  We should see if this is repeatable.  Something to
look into next week.