[daip] Re: DBCON

[Subhashis Roy]

Hi,

I still find that there is indeed a fuzzy emission, and no point source
when I use TIMERANGE 05 0 0 0 10 0 0 0, with the DBCON output. (In my
previous write-up, the TIMER was just a typo, I indeed checked in the
MSGSRV before writing to you, the number of UV-data, which was copied by
the IMAGR). In fact, I am attaching a KNTR output (ps.gz file) of the map
(I hope that the DBCON you are running and what I am running are behaving
in the same way).

One need not go up to level of imaging to understand the problem. If one
just look with UVPLT for time vs. phase plot, it becomes clear (though,
amplitude is fine).

One more info., the error DBCON is making to the phases are antenna based,
and not baseline based. When I run CALIB with SMODEL=1,0, the phase
becomes zero.

Also, I donot understand your comment, as 360" shift is not large enough
for DBCON to handle. Do you mean that the 'w-term' becomes so large, that
the phase error for not taking care of it is ~1 radian ?  
However, it can easily be shown that for a 6" beam, due to the w-term, the
phase error is less than 0.1 radian for a shift of 5'.

Hope this will enable you to fix it.

Thanks.

					      Subhashis


On Sun, 15 Dec 2002, Amy Mioduszewski wrote:

> Dear Subhashis,
> 
> Eric asked me to look into your bug report.  I did the same test
> that you did and found there indeed seems to be something wrong
> with the data that DBCON shifts 360asec.  However not as wrong as
> you state in your message:
> 
> > This indicates that DBCON has not shifted the phase of the second data set
> > properly.  I would like to add here that the map made with TIMERANGE=5 0 0
> > 0, 7 0 0 0 does not even show a source. There is some fuzzy emission seen
> > near the centre, indicating the phase has been shifted wrongly for
> > different uv-values.
> 
> The second data set has a day number of 7 so there is no data between
> 5 0 0 0 and 7 0 0 0, if you set TIMER=5 0 0 0 9 0 0 0 you should see
> a source (I did) at the phase center.  However, the source is a bit
> messy and the concatenated dataset makes a much higher noise image than
> the non-concatenated datasets.  So there is something odd going on.
> 
> However I think this may be because a shift of 360asec is too large
> for DBCON.  When I tested it for smaller shifts (36 and 3.6 asec), things
> seemed to work fine, the noise in the images went down by ~2^0.5 etc..
> DBCON does not shift the tangent point, only the phase center.  The
> DBCON EXPLAIN file suggests that UVFIX should be used to shift the
> tangent point for large shifts. 
> 
> Cheers,
> 
> Amy
> 

-- 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: DBCON.PROB.KNTR.ps.gz
Type: application/x-gzip
Size: 204982 bytes
Desc: 
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20021217/3ba27ed7/attachment.gz>