[daip] DOBAND=1 looses channels 1, N

[Eric Greisen]

Michael Bietenholz writes:
 >    LISTR, UVPLT, SPLAT and presumably many others seem to loose the
 > first and last channels in an IF when DOBAND=1, ie. for 8 channels/IF,
 > setting DOBAND=1 causes channels 1 and 8 to be flagged.  As far as I
 > can tell, my BP table has useful values for all 8 channels, and there
 > doesn't seem to be a weight per channel in the BP table so I don't
 > think the BP table should be flagging anything (same thing happens
 > with a 16-channel data set, ie. channels 1 and 16 dissappear).
 > 
 >    Now I know the end channels are usually a bit flakey, but is this
 > supposed to happen?
 > 
 > (31DEC03 AIPS; LINUX and SUL)
 > 

Is this VLBI data?  The bandpasses have to be shifted as a function of
time and antenna due to the way VLBI correlators work.  This usually
causes one or both of the outer channels to be flagged.

Eric Greisen