Antenna and Feed Geometry

[Jan-Willem]

I am still a little bit confused about the antenna and feed geometry we should record in the MSv4 antenna_xds. To help figure things out, I created a Google drawing and inserted two images of a telescope (one for the MSv2 and ASDM): https://docs.google.com/drawings/d/1HqqzhOZaJ-PYNxBdkt-bJViimhO1ngHT8GJrs3FvSjA/edit?usp=sharing The idea is to label the image to show the relationship between the quantities in the MSv2 and ASDM.

Key questions to answer:

1. What are the data variables with their dimensions we need to describe the antenna and feed geometry that will be used in offline processing?
2. How should we calculate these data variables from what is in the current MSv2?

[gmoellen]

I think we need to consider just what we mean by the net position for the reception of radiation for purposes of correlation with others, and then unpack the various components of this which have occurred historically in antennas and formats. I was about to propose that the net focus point, with the optics unwrapped, would make sense (that's the point to which light has traveled for reception: the focus point is where we put the receptor, by construction, literally), but the asymmetric optics of antennas like GBT and ngVLA introduce additional subtlety: the effective location must also lie on the centroid axis of the bundle rays collected by the primary (I think). I think we have escaped this and similar kind of subtlety in many cases (including w/ symmetirc antennas) when the antennas in an array are uniform. Just what happens in VLBI correlators would be instructive, since they are more likely to have inhomogeneous antennas (even VLBA, eg when GBT is added). We should ask Walter, et al.?

NB: Mark and Kumar on on vacation now, I think.

[kgolap]

This is my understanding of the MS V2.0 antenna info. The nominal antenna position is a position fixed on earth where the observatory correct all the phases due to geometry, direction, atmospheric etc before correlating. It need not be at the bottom of the pedestal. The nominal feed position is measured at the neutral position (For Az-EL most probably it is 0,90 deg azel angles) of the antenna.
The offset of the feed is then the vector between these 2 nominal positions is the offset (so it will be in units of position difference). For data reduction I think only the antenna position is important as the correlator model would have accounted for all other phase/slash delays...thus baseline vector is just the difference of the antenna positions in the baseline. I suspect the feed offset may only be needed if it is known that the correlator model missed some of the antenna model. As far as i know we have never used it.

[kettenis]

It is still a bit unclear to me what the purpose of this information in the MS (v2 or v4) is or would be. But indeed for VLBI the correllator model takes this all into account. There the input parameters are taken from the VEX file[1] that specifies the observation. The relevant parameters in this context are site_position (vector), site_velocity (vector), axis_offset (scalar) and axis_type. Unfortunately the exact definition of site_position and axis_offset isn't clear, but since only these parameters are used we can infer that site_position isn't what Kumar calls "nominal antenna position" in the drawing above, but more something like the "yoke antenna position".

[1] https://safe.nrao.edu/wiki/bin/view/VLBA/Vex2doc