[evlatests] Aliasing

[Barry Clark]

Finally the penny dropped.

The aliased power adds to the straight through power as 
1 + v(f)exp(4*pi*f*Del)
where f is the frequency in the bandpass, and v(f) the aliasing 
function of the digital filter, as provided by Mike, and Del is the
differential delay to the two antennas involved introduced by the 
delay lines in the correlator.

I have been assuming that the delay Del was just the adjustable part
of the delay due to geometry, the w of (u,v,w).  It is not.  It also
contains the compensation for the different fiber delays to the two
antennas involved.

The current fiber delays are given below
ea01   W10     5214
ea04   W18    12633
ea11   E10     7129
ea13   N16    12290
ea14   E16    12004
ea16   W6      3120
ea17   W14     9144
ea18   N12     9288
ea19   W4      3943
ea21   E2      5661
ea23   E18    13493
ea24   W12     6613
ea25   N2      4941
ea26   N6      6203

So, in the observation Gustaaf recently took with w near zero, baselines
in which these fiber delays are nearly the same will have the classic, 
rising to low frequency, profile - baseline 17-18 in particular, but also
1-25 and 13-14.  Baselines where the difference is near a microsecond will 
still be plunging down to low frequency, with a turnaround below the usable 
bandpass - 21-24, 11-26, 19-25 for instance.  Baselines where the difference 
is large will look all squiggly - (4, 13, 14)-(16, 19 25).