[daip] [!4845]: aips - UVPRT is getting frequencies wrong, also SPLIT

[Michael Bietenholz]

Michael Bietenholz updated #4845
--------------------------------

UVPRT is getting frequencies wrong, also SPLIT 
-----------------------------------------------

           Ticket ID: 4845
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/4845
           Full Name: Michael Bietenholz
               Email: mbieten at yorku.ca
             Creator: User
          Department: AIPS Data Processing
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template Group: Default
             Created: 28 April 2014 10:48 AM
             Updated: 28 April 2014 10:48 AM
                 Due: 30 April 2014 10:48 AM (2d 0h 0m)
      Resolution Due: 06 May 2014 10:48 AM (8d 0h 0m)



(31DEC14; SPLIT.EXE, UVPRT.EXE from 2014 Apr 1)

For multi-IF data set where IF2 is 8.00 MHz above IF1, UVPRT will print out "chan freqs" that are in fact 16.00 MHz apart, not 8.00, so I think its adding the freq offset twice somehow.  The uv-distances that it prints out seem to scale correctly with the actual IF freq, rather than the version that UVPRT prints out.

LISTR seems to get the freq. right.

There is vaguely related bug in SPLIT: if APARM(1)=1 but if NCHAV is set to some number *larger* than the number of channels in an IF, the output channel width reflects NCHAV, not the number of channels that SPLIT actually averaged

For example, If the IFs are 16 MHz apart and have 16 x 0.5 MHz channels, then:

SPLIT w/ NCHAV=0; APARM(1)=2    produces an output file with 1 channel, width=8 MHz
SPLIT w/ NCHAV=256; APARM(1)=2 produces an output file with 1 channel, width=128 MHz

The 128 MHz channel with is of course 256 channels x 0.5 MHz, the only problem is that in reality SPLIT is still only averaging 16 channels together, so the real output width should still only be 8 MHz.   This can lead to errors in uv-distance and consequently sky-plane scaling.  

I guess this is sort of covered by SPLIT.HLP's warning about IF-dependent editing, but it would probably be better if SPLIT based its calculation on the number of channels actually averaged rather than input setting of NCHAV.

------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff