[daip] Re: A question regarding flux-calib after flagging

[Eric Greisen]

papadop at phys.ethz.ch writes:
 > Hi Eric, here I am again, that 300 sigma observer....Padelis Papadopoulos.
 > 
 > 
 > I think now I have a good feel of how flux calib works, namely, the GCAL
 > in the CALIB program, goes ahead (when I run it for both primary and
 > secondary) and solves the gains for the primary calibratory, per antenna
 > as it does for the 2ndary.  Then GETJY  goes ahead and scale the SNs
 > of the 2ndary using the gain solutions found for the primary (I guess it 
 > will
 > be a scaling of the amplitudes  only?), the latter having the correct 
 > *absolute*
 > values since we know the flux of the primary.

        Yes only amplitudes

 > 
 > 1) If one is forced to flag an antenna only for the primary (I had to do 
 > that a few
 >     times) but not the secondary,  then naturally no gain solution and 
 > flux-scaling factor will
 >     be present for that antenna for the primary, so what scaling factor 
 > is then used for
 >     that antenna to scale the SNs of the 2ndary in the GETJY procedure
 >     (an average of all other ones, a standard 0-th order value)???

        The flagged antenna is ignored and the average taken over the
remaining antennas.

 > 
 > 2) After GETJY I get consistent flux values for my secondary, between 
 > the 2 IFs
 >     as well as for the two freq IDs (they are all very close in 
 > frequency), in several
 >     consecutive days. That value is 3.55 Jy (with a variation of at most 
 > 0.05 Jy among
 >      days and freq IDs). In *one* day though I get
 >      S=3.21+/-0.05 Jy for both IFs, and Freq ID=1
 >      S=3.55+/-0.05 Jy for both IFs and Freq ID=2.
 >      Unlike other cases the disagreement stays no matter what plausible 
 > flagging I do
 >      (on other occasions, flagging a bad antenna or two always brought 
 > the estimates
 >       of S between the two freq IDs into agreement in all days).
 >      any suggestion on this?

        I have no ideas about this other than to be rather certain
that there is a bad solution in the one FQ for that day.  Examine the
SN table for that FQ with SNPLT and/or PRTAB or LISTR will do gains.

Eric Greisen