<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<p>Regarding geometrical explanations, some more musings...</p>
<p>* An important objection to my suggestion that antenna-dependent
collimation offsets might be pushing the effective pointing of
each antenna's primary away from the target source (such that the
drives effectively rotate them for a direction different from that
of the source, and differentially among antennas) is that we might
have expected this to show a more interesting band-dependence.
But what if the whole feed cabin, as a unit, were not quite
centered properly at the vertex (probably by rotation, not
translation, about the primary's focus point), such that all of
the feeds nominally point to a convergence point not on the
symmetry axis of the primary (i.e., the boresight line). In this
case, I think all of the feeds (per antenna) would suffer the same
collimation rotation (w.r.t. the primary's axis) relative to
nominal, and thus force the same net boresight pointing offset.
Would it be interesting to compare the scale of inter-antenna
collimation offsets with the scale of (per-antenna) inter-band
collimation offsets? If the latter are smaller than the former,
then a common (per antenna) collimation offset could be relevant.
Do we have any idea what the (band-independent) per antenna
collimation offsets are <i>absolutely</i>? I.e., how the feed
cabin is oriented w.r.t. the primary? What is the spec on
placement of the whole EVLA feed cabin (I seem to recall it was a
tight fit....)? I don't know how this is easily pinned down now
if degrees of freedom in pointing will compensate for it.
(Remember, what we are seeing are net differential effects between
antennas, in their peculiar departures from the simple spherical
geometry that governs how Az/El motion <i>is driven </i>to track
a point on the sky.)<br>
</p>
<p>* I think maybe displacement of the subrefector mount (w.r.t. the
primary's symmetry axis), so as to similarly asymmetrize the
primary optics (and possibly done deliberately to accommodate the
feed cabin positioning?), could also do something like this, but
again one must think carefully about what band-dependence should
be expected. (I haven't managed to conclude anything on that.)<br>
</p>
<p>* And what about perpendicularity of the elevation axes w.r.t.
the Az axis? For a reasonably correct Az axis orientation
(vertical w.r.t. array center location coords), I think the Az
rotation compensation required near the zenith to keep up with the
source would yield the even symmetry effect. I.e., at transit, (I
think) a tilt in the elevation axis in a vertical E-W plane would
look like a longitude offset in the antenna position (which causes
even). Is the effect correlated with the elevation axes'
perpendicularity? This strikes me as a promising possibility if
the even symmetry is the dominant one. Residuals to this are then
due to Az tilt (which must have <i>some</i> effect) and other
similar things at lower levels, including the hysteresis evident
in most of Rick's vs-Elevation plots of antennas with large even
symmetry effects (i.e., falling el actually doesn't quite match
rising el)...</p>
<p>-George</p>
<p><br>
</p>
<div class="moz-cite-prefix">On 3/30/22 10:26, Rick Perley via
evlatests wrote:<br>
</div>
<blockquote type="cite"
cite="mid:cf6cb9a3-58a8-b589-c92a-688c2fb67199@nrao.edu">An
update, and a suggestion...
<br>
<br>
Eric cleared up some AIPS software problems, and I can now more
quickly and confidently make various plots.
<br>
<br>
Attached is my best and clearest example of what is going on: This
is a plot of the R-L phase (NOT RL phase) for four carefully
chosen antennas -- ea01 , ea05, ea06, and ea22, at C-band, as a
function of elevation.
<br>
<br>
The reference antenna is ea09 -- chosen because when using this
one as reference, the great majority of the other antennas display
the cleanest 'even' signature (when plotted versus time or HA).
Using one of these displayed antennas as reference would merely
subtract what you see from all the others, so that ea09 (for
example) would show the same effect, but with the phase declining
with elevation.
<br>
<br>
Key points are:
<br>
<br>
1) The four different sources (color coded) all follow the same
curve very closely, arguing strongly that the underlying cause is
a function of elevation, and not HA or parallactic angle. (When
plotting the data against these, much messier plots are
generated).
<br>
<br>
2) The same plots are seen at every other spectral window within
this band,(!!) and in every spectral window at L and S bands.
(!!!). Not only the same shape, but the same magnitude. (!!!!)
The effect (as seen by these antennas, using ea09 as reference) is
solely a function of elevation, and is independent of observing
frequency.
<br>
<br>
3) I've 'cherry-picked' the antennas to show. About half the
remaining antennas show the same relation as those shown here, but
not as tightly as shown in these. It's clear that the reason is
that there is an 'odd' factor which causes a different (R-L) phase
difference between the east and west sides of transit. And for a
few antennas, other factors, unrelated to 'odd' or 'even'
symmetries have caused large phase differences.
<br>
<br>
Barry has opined for an antenna-based problem (something within
the electronics which is strongly elevation-sensitive). But, in
an experiment run by Paul two days ago, no elevation-dependency on
the 'auto-cross' phase was seen. (This monitors the phase
difference in the injected noise-diode signal -- and so is not an
astronomical observation). Arguments based on a temperature
effect in my data are hard to sustain, as the outside temperatures
on the night of my observations were exceptionally uniform
throughout the period -- and it was quite breeezy as well. These
results argue for an origin preceding the injection of the noise
diode signal.
<br>
<br>
So -- what to do next to isolate the cause(s)?
<br>
<br>
I'd like to try the 'over-the-top' observation. If the effect is
truly due to an elevation-dependent effect within the antenna,
then it should continue to increase as the antenna is tilted 'over
backwards' -- the antenna elevation is then greater than 90
degrees. This should cleanly separate effects due to elevation
from those due to HA or parallactic angle. Observing OTT also
reverses the orientation of the R and L 'squint' beams, so should
be definitive in eliminating that origin.
<br>
<br>
I suggest we do this with sources which transit both to the north,
and to the south of the zenith. All my current examples are from
sources which transit on the south side.
<br>
<br>
Doing this with the current 'A' configuration might also
illuminate any dependencies on antenna placement -- despite all
our antennas nominally having parallel azimuth axes, sources will
transit at slightly different times. I don't think this is an
issue -- but who knows? We might be surprised ...
<br>
<br>
This is a fair investment of time -- a few hours. But I think we
need to do something like this to make any progress in isolating
the origin(s).
<br>
<br>
Rick
<br>
<br>
<br>
<fieldset class="moz-mime-attachment-header"></fieldset>
<pre class="moz-quote-pre" wrap="">_______________________________________________
evlatests mailing list
<a class="moz-txt-link-abbreviated" href="mailto:evlatests@listmgr.nrao.edu">evlatests@listmgr.nrao.edu</a>
<a class="moz-txt-link-freetext" href="https://listmgr.nrao.edu/mailman/listinfo/evlatests">https://listmgr.nrao.edu/mailman/listinfo/evlatests</a>
</pre>
</blockquote>
</body>
</html>