[daip] Ringing from BPASS

[lincoln greenhill]

Hi Eric,

I hope not to take much more of your time on this matter.

> I do not understand the origins of the data - I had thought that the one 
> source was just a UVCOP extract from the multi-source - but in fact the 
> two had different BP tables.  The history file of the one-source seemed 
> to contain several runs of BPASS - but according to local VLB folks, 
> using BPASS at all is the first conceptual problem.  For astrometry 
> anyway, it is not needed.  

I will ask Ingyin to review what she has done in generating the data
and in use of BPASS; we will review inputs.

Concerning the conceptual error - there are multiple ways in which to
process the astrometry data.  I prefer to run FRING on one scan to 
estimate/remove instrumental delay and a DC term, then run BPASS on the
same scan to remove high-order terms in the bandpass, and FRING again to obtain 
delay measurements on the full time series of calibrators.  By evening out the 
bandpass phases between FRING runs, the second run is not affected by
loss of S/N for stations with large swings in phase across the band and
the possibility that those phase swings might affect the measurement of
a linear phase gradient w/ frequency.  The last should not occur, I admit,
but I am being better safe than sorry.

> I do not know why the 2 bandpasses are 
> different - they are very similar except at the edges but I noticed a 
> column in the bP table that is not used for much contains the apparent 
> shifts by antenna and time and they were different which I do not 
> understand at all.  If I thoiught that bPASS were broken I would spend 
> time on this but I have already spent 2 days or more with time from Amy 
> and Vivek and I conclude that the issues are not with AIPS (at least 
> this time).
> 
> CPASS should not be used for the gross BP shape although it has use 
> according to some for slight shape changes with time in e.g. the VLA.

This comment in particular interests me.  Claire and we have been using CPASS 
successfully in a 43 GHz VLBI time-series program, where we adopt the 
autocorrelation spectrum for amplitude via BPASS and adopt the CPASS polynomial 
for phase response.  I had not thought about using it to deal w/ VLA time 
variation.  Makes sense.  In the EVLA era, I presume this will be behind us.

Best, Lincoln




> 
> ERic Greisen

-- 
Lincoln J. Greenhill               Harvard-Smithsonian Center for Astrophysics
Office:     1 617-495-7194         60 Garden St, Cambridge, MA 02138
Cell:       1 617 851 8918         E-mail:  greenhill at cfa.harvard.edu
FAX:        1 617-495-7345         http://www.cfa.harvard.edu/~lincoln