[daip] New Ticket - [!JNA-711883]: ANTAB and cal transfer problems

[Eric Greisen]

New Ticket: ANTAB and cal transfer problems

I have found that the problems with the calibration transfer FG and PC
tables that I pointed out in 2007 are still with us.  There are excessive
"System idle" flags that cause spurious loss of scans, in this case the
first scan after a switch from a geodetic (DELZN) segment to main data or
visa-versa.  These are caused by an assumption in the code, that hasn't
been fixed, that the flags are on at the correlator job start time.  The
state doesn't change until the next monitor data flag appears.  If the
flags are down (good data) at the start time, then the data are spuriously
flagged for the rest of the scan.  As if this weren't bad enough, the TY,
and PC data are not passed during that time.  Meanwhile some of the PC
data (pulse cal) is assigned to the wrong antenna (separate problem).   So
I am trying to replace my FG, TY, and PC tables (also GC because I know
the gain curves have been updated) and ran into trouble.  This is BW092A
in Jan, 2009 processed on my Mac at home (not the notebook).  AIPS was
updated within the last 2 days.

I ran VLOG on the cal file with the gain table concatenated.  I know from
earlier that VLOG creates exactly the same output files for the geodetic
file, the main file (7mm), and a small 1cm file.  It doesn't separate out
the data according to which AIPS file is used as the VLOG input.  That
leaves the cal file with mixed modes.  What I'm most suspicious of is that
the geodetic section has 8 single polarization IFs and hence 8 Ts values
in each record in the cal file.  The main and 1cm files have 2 dual pol
IFs with 4 Ts values in each record.  The first time that ANTAB gets a
transition from 8 to 4 values per record, the following error appears:
ANTAB1: TXTCAL: BAD TIME:   240.30  242.50 after   0 entries: SC
ANTAB1: Purports to die of UNNATURAL causes
(the time numbers can change).
First, this happened about half an hour into the run at the first
transition (I start with a geodetic section).  I then edited out all the
geodetic section data for the start and tried again.  ANTAB again died at
the second transition from geodetic data to normal data, a couple of hours
later in the file.

Note that VLOG has not put any CONTROL records with INDEX in the file.

I presume the work-around is to split the file into the 8 channel set and
the 4 channel set.  Is it expected that you cannot mix the two?  If so, it
would be nice if VLOG would do the job.  But the need may change with
DiFX.  Meanwhile, if this is an issue, it should be made clear in the HLP
file.

I can provide more information and files if needed, but only from home.

Given my on-going problems with the cal transfer data and in my current
sour mood (I had done a lot of work before finding the table problems),
I'm tempted to advise users not to use cal transfer data and maybe even
get it purged from the archive.

Cheers,

Craig


Ticket Details
===================
Ticket ID: JNA-711883
Department: Data Processing (AIPS)
Priority: Normal
Status: Open
Link:  http://nrao.helpserve.com/staff/index.php?_m=tickets&_a=viewticket&ticketid=34