[daip] Re: bandpass problems

[Eric Greisen]

Phil,

I checked the shifting (since I agree about the symptoms), but that is
identical.  The reason I asked about the version was that early
31DEC01 versions still had the error that shifted cross-correlations
before the fit using the shift of one of the antennas.  The correct
fit is the instantaneous current observation followed by a shift of
each antenna solution.  Then an inverse shift as needed before the
fit.  Those shifts were all identical between the two data sets as
they should be.

When I run BPASS with identical parameters on the two data sets I get
essentially identical results.  I just repeated the test with
normalization 3 instead.  The 4 shapes are nearly identical, the 4
sets of phases are nearly identical, the amplitudes are offset between
those with type 1 and type 3 normalization as one would expect.  Type
1 normalization is the correct usage for VLBI since the gains are set
by all channels (unlike the VLA).

I cannot see how you got the two different BP solutions that cause the
odd offset/slope at the right hand side of some of the spectra.  Could
you have improved the CL table in the interim?  DOCAL 1 is important
in your data set to remove the delay phase slope.

See you Sunday.

Eric