[evlatests] Interferometer Data from 27 Sept 2005.

[Rick Perley]

    I've reviewed the data taken by Ken and me late yesterday afternoon. 

    The observations were taken in spectral line mode, with 6.6 seconds 
averaging,
mode '4' (providing AA, CC, BB and DD combinations, but no cross 
polarization
combinations), with 3.125 MHz bandwidth, giving 64 channels in each -- 
49 kHz
spectral resolution. 

    We observed at 6cm, both a strong source, and nearby blank sky. 

    a) As reported by Ken, IFs 16A and 16C were dead. 

    b) Also as reported by Ken, 14B was weak -- by a factor of 1.5 to 
2.0 in noise. 

    c) Bandpass shape:  The residual phase curl exceeds 1 radian only at 
the lowest
500 kHz in all 6 IFs that I got solutions for.  It is flat and smooth 
over the rest of
the bandpass.  The amplitude, over this bottom 500 kHz is rolls off, 
quite different
than the sharp drop seen in VLA bandpasses -- the amplitude for these 
drops only
in the bottom 100 kHz.  In fact, this roll-off will reduce the closure 
errors associated
with the phase residual -- so I'm inclined to ignore it. 
       The bandpass of 14B is special -- with a kind of double-hump.  
Perhaps related
to the poor SNR?

    d) Temporal behavior.  The 6.6 second amplitude and phase solutions 
for antenna
14A, C, and D
are comparable to VLA antennas in their scatter.  Not better, not 
worse.  (We do not
expect to see SNR differences in these, due to the strength of the 
calibrator).  The
amplitude solution scatter for 14B is much higher than the others -- by 
a factor of
3 to 4. 
       The solutions for ant/IFs 16B and 16D are bizarre.  In the 2nd 
scan, the fringe
power drops by a factor of 3 to 4 every 20 seconds.  The other 2 
solutions (remember,
the data are sampled every 6.6 seconds) are normal.  The amplitude 
scatter of these
good solutions is like other VLA antennas. 
    The phase behavior of 16B and 16D is even more bizarre.  It's too 
complicated for
a short description!  Removing the weak amplitude data did *not* result 
in stable
phase solutions -- these phenomena appear to be at least partially 
decoupled. 

    e) Sensitivity. 

    I formed rms matrix listings of a single 48 kHz channel, giving the 
ampscalar rms over
a 5 minute period, on both the blank field and the strong calibrator.  
The same picture
emerges from both:
    14 A, C and D look like a VLA antenna.  The expected reduction in 
rms noise, from
the lower system temperature (by a factor of roughly 40%) is not seen. 
    14B is 1.5 to 2.0 times too noisy. 
    16 B and D look like a VLA antenna. 

    I also looked at the rms amplitude and phase across the bandpass, 
after applying the
calibration and bandpass corrections, discarding the edge channels.  
These gave the
same results as the temporal rms, given above. 

    In short -- 14 and 16 (or at least those channels which are behavior 
`normally') are
about as good as a decent VLA antenna, but not as good as they should 
be.    The
discrepancy is by 25 to 40%. 

    f) Baseband Total Power Spectra.  The `spike', located 250 kHz from 
the baseband
edge, is now barely visible in these high-resolution spectra.  It is 
clearly seen
in 16D, and 14B, where it contributes less than 1% of the total power.