[evlatests] Two Remarkable Characteristics

[Ken Sowinski]

On Thu, 25 Aug 2011, Rick Perley wrote:

>    1) Good
>
>    We are all familiar with the 'subband 0' rolloff, of still
> mysterious origin (although I understand there's a good chance this
> originates in the station board?)
>    In determining the bandpass, I reviewed the bandpass solutions.  The
> 'rolloff' is completely absent in two antennas -- ea19 and ea21, on IF
> 'B' only!  Subband '0', on those two antennas on that IF is beautifully
> sharp and flat, except for the sharp drop due to the anti-aliasing filter.
>    A number of other antennas show a dramatically smaller 'rolloff' on
> various IFs -- but none are flat like ea19 and ea21 on IF 'B'.  This
> bandpass was identical on all three runs.
>    Keith has been informed, and reviewed the T304 data for those
> antennas -- nothing unusual.  It's gotta have something to do with the
> station board.  (Can we make all the others like these two?)

This is interesting, but hard to place anywhere than the
downconverter or, perhaps, DTS module.  A simple test to
exonerate the station board would be to look at the output
of the bandpass plotting web page.

There are two choices of subband zero filter.  The one we have
been using since the prototype correlator suppreses DC.  Its
responce begins to roll of around 12 MHz and is down by more
than 50 dB at DC.  there is only one directory where all the
filter files reside and all the station board use the same set
of files.  The other choice of filter for subband zero is a low
pass filter which has a flat response all the way to DC.  Until
a month ago we had no example of such a file in any obvious place
here.  We now have a copy, but have yet made it accessible to
station boards.

>    2) Bad
>
>    The antenna stabilities were all very good ('good' defined as +/- 5%
> or so) in all subbands *except* one -- subband 16 showed huge (up to
> factors of 2) variations in fringe amplitude on timescales of minutes,
> in all four IFs, **** on some antennas only****.  The bad antennas were
> the same in all three runs (which were distributed over more than a
> month in time):  2, 7, 11, 14, 15, 17, 18, 19.  Some others were less
> variable, but still well outside the norm:  27 and 28.  All other
> antennas were as 'good as gold'.
>    There is no spatial relations of the listed antennas.  They behaved
> correctly in all other subbands.
>    Now, subband 16 does contain moderate RFI -- in the lower 64 MHz:
> 6100 to 6165 MHz.  But it's hard to see how this could cause such
> instability in some antennas, but not in others...

I think Dan has made the case that we can blame this on interference.
For the record when Rick writes subband 16 he means what is called
subband 7 in the correlator.  It is the 128 MHz subband that lies at
the edge of the 1 GHZ baseband which is away from DC.