[daip] bug in FITLD ?

[Eric Greisen]

Floris van der Tak writes:

 > There may be a bug in the way FITLD writes GC tables for VLBA 
 > data sets with different bandwidths in different IFs.
 > 
 > My VLBA experiment BT079, observed last November 14, has three IFs
 > with 4 MHz and one IF with 2 MHz bandwidth. Each IF has 128 spectral
 > channels. The data were put on tape in seven files: three for the 
 > 1-IF and four for the 3-IF data.
 > 
 > Reading these data into AIPS all at once, FITLD wrote two files. The
 > amplitude calibration with APCAL fails on the file with 3 IFs, with
 > the error message "missing gain". But APCAL works for the 1-IF file.
 > 
 > If FITLD just reads the 3-IF data (skipping the first three files) it
 > writes one file, which APCAL can handle.
 > 
 > The corrupted GC table has 40 rows where the others have 10. It looks
 > like the bad one just says the same thing four times.
 > 

This is a well known feature of FITLD.  After running FITLD one hase
to run something like MERGECAL.  Perhaps you should be using the
VLBAUTIL package described in Appendix C of the CookBook.

Eric Greisen