[daip] CALIB and solution intervals

[Eric Greisen]

Olaf Wucknitz wrote:
> Hi there,
> 
> sorry for adding another aspect to this difficult issue.
> 
> In the past I always used <<10sec (e.g. 1sec) solution intervals when I 
> wanted solutions at the intervals of the integration time (typicall 10sec 
> for VLA data), because that worked better than a solint of 10sec.
> 
> Now this approach causes a small problem. Sometimes it happens that the 
> last solution of a scan is assigned an incorrect source number in the SN 
> table. I noticed this only because GETJY does not rescale those incorrect 
> entries which thus pop up in subsequent SNPLT as being a factor 3 too 
> small. With 31DEC07 this problem does not occur. It also disappears if I 
> go back to 10sec solints.
> 
> What is the recommended solint to produce solutions at the intervals given 
> by the integration time?
> 
> I am using the latest MNJ versions of 31DEC07-09 (last entry in 
> CHANGE.DOC is 12787. January 6, 2009 CALIB). The data set is just 
> about a week old, downloaded today and read in with the latest FILLM.

This should not happen of course so

(1) Can you do a PRTAB on a section of the NX file and on a section of 
the SN table covering a couple of scans to illustrate the problem?

But

(2) Why are you doing individual record solutions in a multi-source 
file?  This will have lot S/N and the standard interpolation methods of 
CLCAL will only combine the last sample of one cal scan with the first 
sample of the next from such SN tables.

Eric Greisen