[daip] new antenna ref. system

[Eric Greisen]

Luca Moscadelli wrote:
> Dear Eric,
> 
> Just a question to undertand better the backward compatibility of the DEC10
> AIPS version when working with the old  reference system for the antennae.
> 
> Running 31DEC10 CVEL on VLBA 2008 January data, I observe a large 
> average shift
> (up to more than 1 channel) with a large difference (plus minus one 
> channel)
> among different antennae.
> 
> Running 31DEC07 CVEL, the average shift is only a small fraction of a 
> channel,
> and very similar among the antennae.
> 
> It is not clear to me why using a different antenna reference system 
> should cause  different shifts.
> 
> If you need a better description with reported CVEL output, please tell me.
>

I can think of no reason why there should be any change - except one. 
That one is user error.  If a data set is first looked at in any way by 
31DEC09 or 31DEC10, the antenna file will be changed to have all Bx, By, 
Bz systems right handed.  31DEC07 assumes that the VLA is right handed 
and the VLBA is left handed.  The only difference in the code itself on 
this point is 1 line which is changed from

       IF (VLA) BY = -BY

to

        BY = -BY

since the code that does the Doppler shifting uses internally a left 
handed system.  Thus, if the antenna file has seen 31DEC09 or 31DEC10, 
then 31DEC07 will make the wrong **assumption** about handedness and 
will use a very different By.  There were also changes in modern CVEL 
for arrays that do not use the Earth's center as their center but VLBI 
systems always use the Earth's center.

If you really want me to examine these data, I need the data themselves.
I have 32-bit computers at my disposal so I can run 31DEC08 vs 31DEC10 
in debug modes to try to see what is going on.  I will need access to 
the FITS file (preferably not modified by AIPS 31DEC10) and the inputs 
that you have used.

Eric Greisen