[daip] VLBAEOPS version TST spoils phases

[Eric Greisen]

Eduardo Ros wrote:
> Dear Eric,
> 
> In the last days I was analyzing old archival data, and some weeks 
> earlier my own data from CTA102 and I noticed that the TST version of 
> VLBAEOPS spoiled the phases (to be seen in POSSM after running it).  
> This does not happen in the NEW and OLD versions of aips in the MPIfR 
> installation.
> 
> Maybe you want to have a look to it.  If needed I can send you some 
> plots with the NEW and the TST versions.
>
This is curious.  I have looked at the code and there is a comment in it 
that suggests to me that there is an error - the pernicious sign of the 
Y antenna coordinate.  Unfortunately that does not explain your remarks 
at all since the code in at least NEW and TST in CLCOR and in the 
procedure file VLBAUTIL in NEW and TST is identical.  If I had to guess 
- they should all be bad for your data.  The versions of AIPS in Bonn 
are all current - so 31DEC09 has in it the change that made all antenna 
files be on a right-handed coordinate system.  If you run PRTAN on your 
data set you will see a keyword stating XYZHAND='RIGHT' and all 3 
versions agree.

I have the author of the 3 places in CLCOR affected 'SUND', 'EOPS', and 
'ATMO' examining the code.  Unfortunately, he leaves on vacation after 
tomorrow but I hope he can provide an answer by then on whether the code 
as it now stands - with RH By in the antenna table - is wrong or not. 
The fix is easy.

This does not explain the apparent functioning in NEW and OLD.  Did 
those versions copy the data file the same way TST did?  Are the CQ 
files the same or is this the same data set simply run with version = 
'new'?  What do the Cl tables look like before and after and what do you 
mean by spoiled the phases using POSSM?  POSSM makes a spectral plot - 
did EOPS add a large delay component causing a big phase ramp?  That is 
what POSSM should be used to display.

Eric Greisen