[daip] 31DEC07 LINUX midnite job (dave, 20070322.044406)

[Walter Brisken]

> The CLCOR problem has been there for some time - arising due to the
> change in which the source position is updated in the SU table (before
> the phase change is computed).  It is 2nd order in that it shifts the
> phase in the right direction but from the output position to the
> output + another delta.  It should have been and is now output
> position - delta to output position.

> The UVFIX problem was a frequency error of REFCHN channels in
> uncompressed data only.  Typically REFCHN=1 so it is 1 channel
> separation over the center frequency * the shift.  In Amy's data this
> was 1 part in a 1000.  Prior to about 2002, UVFIX did not include the
> W * Delta Z term, so the modern code really should be better in that
> respect.
>
> It would be good if you informed us of problems so that they could be
> addressed sooner.  Amy was kind enough - through her student - to
> document this issue and enable us to find the problems.

Probably should have, but my hints of these problems were very subtle 
(changes in parallax fit residuals on 8 epochs of data before adding CLCOR 
to my reduction code) and my attempts to pin the problem on CLCOR never 
were successful.  From your descriptions above, errors from both problems 
are well below my detection threshholds!

-W