[daip] aips_gripe

[Lincoln Greenhill]

{27-MAY-2000 10:55:39}
{rgalp8       UN-REGISTERED!!!!!!!}
{   24  31DEC99  DEC}
{lincoln}
{cfa, ms42}
{617 495 7194}
{For multi-src files, CALIB does not appear to make consistent choices
 of the times tagging entries in the solution table.  My tests involve a
 phase referenced VLA observation.  Scan lengths are 27 sec, with 3 sec
 deadtime between scans. (This is what is shown in the NX table.) The time
 entries for a scan that starts at 13:21:13 are 13:21:43.3 (length 0 sec)
 and 13:21:58.3 (length 23.3 seconds) and 13:22:43.3 (length 0 sec) and
 13:22:58.3 (length 23.3 sec) etc.
  
 In the past I have found that for a different multi-source file, one that
 features longer scans, I obtain better self-cal results with CALIB is I
 delete the NX table before running CALIB (this is a multi source file with
 only one source in it).  At that time Ketan mentioned that CALIB does not
 divide up time into SOLINTS properly.  I believe that I obtained better
 results w/o the NX table because it prevented the introduction of irregularly
 spaced time tags.
  
 In the case detailed in paragraph 1, the phase solutions for the 0 length
 records are substantially different than their closest neighbors, which is
 a worry.  My remedy in this case will be to use TABED to delete them.
  
 Perhaps FRING also suffers this problem.  I have not checked but expect
 the same machinery to be used as in CALIB.
  
 Final note:  There are occasional instances in the files discussed in para. 1
 where the 0 length records are absent. For example:  present at 13:20:4x,
 13:21:4x, 13:22:4x, missed at 13:23:4x, present 13:24:4x, etc
  
  }
{CALIB does not compute time tags correctly in multi-source files w/ NX tbls.}
{greenhill at cfa.harvard.edu}
{ }