[daip] Problem with VLBA data tape

[Eric Greisen]

Alan Marscher writes:

      daip is the best address for aips problems - Ulvestad is
subscribed to that list.  I have asked the analysts to prepare a tape
of this run to give to me and I will play with it.

 > We are having trouble loading the files (using FITLD in AIPS, version
 > 31DEC99) from the data tape for observation BG099, 29 June 2000.  The files
 > have a lot of new tables, presumably for automatic calibration,

      There are a variety of new tables for gain transfer but they are
not that new by now.  We have been doing that nearly 2 years.

 > but some of these must be incompatible since 'DOCONCAT' doesn't

      Are you simply trying to concatenate the multiple files from
this run in one run of fITLD?  Or are you re-running FITLD on one tape
or multiple tapes or are you trying to concatenate to some earlier
data?   (I.e. I do not know what you mean by DOCONCAT does not work)

 > work.  So, we get a separate file for each UT section.  One anomaly
 > is that the UT table (which I've never seen before) has a
different

      Something is very wrong.  The UT tables are the uv data and they
are supposed to be translated into a UV main file and deleted.

 > version number for each section of data on the tape.  Furthermore,
 > the frequency numbers for the different sections correspond to
 > different frequencies.

     What frequencies did you observe, correlate, ...?

 > 
 > Jose Luis reports that he encountered this error once before for

      reports to you but not to us.

 > another data set.  So, either it occurs occasionally or occurred
 > over some finite period of time.  We don't have the problem for
 > a data tape written in late 2000 for an observation on 1 Oct 2000.
 > So, it's not a matter of having too old a version of AIPS.

      31DEC99 is unfortunately not a well-defined version.  There is a
frozen 31DEC00 which is 31DEC99 grown up - see Aipsletter on www pages.

 > 
 > Of course, it's possible to deal with this with a lot of patience
 > (running DBCON, etc., on each file) and a lot of disk space.  But

      DBCON does not renumber antennas and may have other troubles as
well.

 > it seems like a waste.  I suspect that it was caused by an error
 > at NRAO when writing the data tape.  If so, then it would be best

      If the frequencies vary, that is set at the time of correlation.
The writing of the tape is very automatic and there is no way that the
analysts can alter the content of the files written by the correlator
when they extract and copy a particular experiment.  That is why I
asked for a copy of the tape rather than simply sending you another,
probably identical, tape.

 > from our viewpoint if NRAO would write and send us another data tape.
 > (Please send it to Alan Marscher at Boston University rather than
 > to Jose Luis Gomez in Spain.)
 > 
 > Please advise.
 > 
 > Thanks very much,
 > Al Marscher, Jose Luis Gomez, and Svetlana Jorstad
 > 

Rather than sending vague "does not work" messages it would help if
you provided details, e.g. the inputs to and messages from FITLD and
anything else that tells you things did not work - e.g. IMHEADERs of
the output file(s), PRTAB on FQ tables, etc.

Eric Greisen