[evlatests] Update on Strange R-L phase behavior

[George Moellenbrock]

Sanjay-

You're saying that if the vp is registered stably to a tracked point on 
the sky offset from the source, the source will travel through the 
voltage pattern in a systematic way, by the parallactic angle rate (and 
this is also why off-center sources travel through beam "annuli").   
Yep, and that will create some sort of effect like this, if the pointing 
error is large enough, and the vp R-L phase interesting enough in the 
right way on that scale (but freq-dep...).  Reference pointing is the 
exercise by which we try to keep the source stationary in the vp to 
avoid this.   Unclear if reference pointing can be effective near the 
zenith, though.....  R/L Amps can say a lot, I think, if pointing is 
going way bad near zenith.

I'm saying (in my 2nd paragraph), that on top of this, the source itself 
rotates (spins in place) as it also possibly travels (as above), and 
this rotation (at a stationary point in the vp, if the pointing is good 
and stable), all by itself, causes the R-L phase to evolve (think 
rotating helices).   This is the parallactic angle rotation itself, not 
its effect on a source's offset location in the vp, and which changes R 
and L phase in opposite directions. Now, because the antennas are all 
nominally mounted in parallel, the gross parallactic angle phase (what 
we normally think about) differences to zero in the /parallel/ hands, so 
doesn't matter in ordinary gain solutions (see VLBA, where parang phase 
/will/ show up in gain solutions if you don't pre-apply it).  However, 
the mechanical errors in antenna structure and alignment mean that each 
antenna has a peculiar residual in its driven rotation for the observed 
source and which is modeled nowhere /for phase cal purposes/.  The 
differencing in the parallel hands isn't perfect, and this blows up the 
R-L phase near the zenith.   Can it do enough?  Still TBD...

These are distinct things, related through dependence on parallactic 
angle, I think.   Rotation must occur.   Travel through the vp /may/ 
occur, but won't (significantly) if pointing is good enough.

-George



On 3/30/22 13:32, Sanjay Bhatnagar wrote:
> George:
>
> 1. The effect I am thinking of is more like in the first few sentences 
> of you second paragraph.  Source moving _systematically_ in the R and 
> L voltage patterns.  The precise track can be written down as an 
> expression (as also suggested by Steve).  It is not a source wander.  
> In general it also not a pure rotation (as you seem to imply).  Also, 
> with significant pointing offsets antenna polarization squint matters 
> for the kind of investigations done here (variation of R-L phase with 
> time).
>
> 2. CASA imaging *does* account for geometric effects (e.g. antenna 
> offsets, squint, effects of non ideal aperture illumination as 
> measured with holography, all of these as a function of time, etc.).  
> Some of these are needed, and even used for VLASS imaging.  Also these 
> corrections, in general, can only be done during imaging (i.e., can't 
> be done in the traditional pre imagining calibration step).  For 
> compact sources one can approximate, I _think_, these corrections via 
> transitional calibration (as in AIPS or CASA calibration modules) but 
> only *after* eliminating pointing offsets
>
> On Mar 30, 2022 12:57 PM, George Moellenbrock via evlatests 
> <evlatests at listmgr.nrao.edu> wrote:
>
>     Sanjay-
>
>     I think you are describing phase variation within/across the
>     voltage pattern, and the source wandering around in that. 
>     Wouldn't that be quite band-dependent?   I think Rick was going to
>     look for R/L amplitude effects which might be evidence of that
>     sort of thing.  And we might expect that wander to be less
>     systematic/symmetric, probably.   Still, wander around the beam,
>     especially near zenith, is likely at least a confusing factor, indeed.
>
>     The geometric effects I've been trying to describe will operate
>     even if the source is strictly stationary (in direction) in the
>     voltage pattern.  But it is still /rotating/, or more to the
>     point, the antenna (and thus feed) is rotating about the direction
>     to the source in a manner that is a function of mechanical
>     imperfections described by the pointing model (and related
>     effects).  This rotation causes differential advance/retard of R
>     and L phases, relative to whatever phase the vp introduces at the
>     point the source pierces it (assuming stable pointing).    And to
>     be clear, CASA (nor AIPS, to my knowledge) incorporates
>     geometrical info from the pointing model to correct the
>     differential rotation of the antennas (which gets interestingly
>     large near zenith).  And this would be via the parallactic angle
>     correction, which I suspect Rick hasn't been applying, else we'd
>     probably see more interesting things, like more odd symmetry
>     effect, if AIPS is still using geocentric latitude for the
>     calculation (alas CASA does, too, because the overall impact is
>     still fairly small for most observations, compared to likely
>     posang errors from other causes).
>
>     As for solving for the effects as Steve suggests, we may already
>     be doing so, e.g., in the pointing model; i.e., existing terms can
>     suffice, at least qualitatively if not to scale, and maybe some
>     new term is needed... My point is that we are not doing the
>     /peculiar/ feed rotation calibration explicitly /anywhere/**, and
>     so the effects thereof must show up at some level in solved-for
>     phases in the manner Rick has shown (possibly, or probably,
>     confused a bit by what Sanjay describes, but not so much as to
>     obliterate an otherwise very geometric-looking systematic effect),
>     and may, in fact, be the actual explanation---if the required
>     mechanical errors are significant enough to do it.
>
>     (**is the correlator at all aware of the pointing model?  for
>     reasons/other than /net path length, if even that?)
>
>     Cheers,
>
>     George
>
>
>
>
>     On 3/30/22 11:42, Sanjay Bhatnagar via evlatests wrote:
>
>         A simpler way to achieve the same would be:
>
>         1. For deriving R-L phases, use source model that includes
>         known effects of antenna pointing offsets (from pointing
>         measurements) and measured antenna aperture illumination
>         patterns.  This can be done in CASA.
>
>         2. I am less sure here, but since the celestial source is
>         compact, I _think_ if the data is pointing offset-corrected
>         before deriving R-L phases, it will effectively achieve almost
>         the same as above.
>
>         sanjay
>
>         On 3/30/22 10:47 AM, Steven Myers via evlatests wrote:
>
>             If the explanation is geometric, then can we write an
>             equation mapping (AZ,EL) of the antenna and (HA,DEC) of
>             the source, including the various physical offsets, to the
>             observed R-L phase, and then solve for these offsets using
>             the data in hand?
>
>
>
>             _______________________________________________
>             evlatests mailing list
>             evlatests at listmgr.nrao.edu
>             https://listmgr.nrao.edu/mailman/listinfo/evlatests
>
>
>         _______________________________________________
>         evlatests mailing list
>         evlatests at listmgr.nrao.edu
>         https://listmgr.nrao.edu/mailman/listinfo/evlatests
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20220330/64b3727f/attachment.html>