[daip] TVFLG reports INCORRECT PHASES (OLD/NEW/TST)

Michael Rupen mrupen at aoc.nrao.edu
Sun Apr 27 12:52:32 EDT 2003


Dear DAIP,

  I have a very odd problem with TVFLG.  I have a typical VLA continuum
uv-data set, 3.3s integrations, at 6cm.  This is one of the transient
monitoring runs, so I've reduced these guys in the same fashion zillions of
times without any trouble.   On this particular occasion, I happily went
through initial flagging & calibration (TVFLG, SETJY, CALIB, GETJY, CLCAL),
then went back into TVFLG to check that the calibrated amp. & phase look OK.
This is where things got weird.  TVFLG claims phases up to 180d -- it looks
like a few antennas are not calibrated for a few sources.  The amp. & the
phase diffs. look fine.  Now I try to check the phases in UVPLT, one IF at a
time -- they look spiffy!  OK, back to TVFLG, but now instead of just
setting CALCODE '*' I explicitly list the calibrators with SOURCES.  Now all
the phases look wonderful!!!  This is true in OLD, NEW, & TST.  I tried
re-creating the NX table with INDXR, but got the same result.  I would have
thought this had something to do with how TVFLG interpolates the CL table
entries, but that doesn't make any sense:
  * there are CL entries every 10s (yes, I checked this with LISTR), and I
    have minute-ish scans, and the CL entries themselves change slowly and
    smoothly (and not through 180d on the offending antennas)
  * the phases shown in TVFLG are consistent for whole scans and even
    across scan boundaries for a given source
So I am very confused!

I've put the offending data set in

  /home/verdi/ar508_26apr03c.uvfits

if you'd like to have a look.  I run TVFLG as

  default tvflg
  getn  <DATASET>
  docal 2 ; gainuse 2 ; flagver 1 ; sour ''; docat -1
  dparm 0 ; dparm(6) 3.3333333 ; dparm(3) 1
  calcode '*'

which fails, and then add
  sour '1331+305','17009-26109','17519-25240','1824+107',''
which works -- despite the fact that it's 17009 & 17519 which give problems
when I just rely on calcode '*'.

Thanks very much --

               Michael

p.s. I'm running on verdi, a Linux box.



More information about the Daip mailing list