from:	 Young, Ken 
to:	 "Primiani, Rurik" 
cc:	 Mark Gurwell ,
         Jonathan Weintroub ,
         Eric Keto ,
         Nimesh Patel ,
         Chunhua Qi ,
         Lars Kristensen ,
         David Wilner ,
         Qizhou Zhang ,
         Sridharan Tirupati Kumara ,
         Raymond Blundell ,
         Jun-Hui Zhao 
date:	 Mon, Dec 2, 2013 at 1:31 PM
subject:	 Re: SWARM Phase Closure
mailed-by:	 cfa.harvard.edu
Hi Rurik,

   No, I did not reject the guard bands. I've attached plots without 
the gaurd band channels to this message. The first set of plots shows 
the Legacy Correlator data again. This time the pseudo-continuum was 
built from passband-calibrated data in the overlapped region. Applying 
the passband calibration had virtually no effect. The second three plots 
show the SWARM data after passband calibration, with the first and last 
1000 channels rejected from the summation used to produce the pseudo-continuum 
channel. One could now claim that triangle 2-5-7 closes to 0, I think, 
but the highest signal/noise triangle, 2-5-6, is clearly not closing to 0, 
and 2-6-7 isn't either. The last plot included shows triangle 2-5-6 with 
the first and last 4000 channels rejected. That plot looks nearly the same 
as the 1000 channel rejection one does. If I trim the guard band channels, 
but do not apply passband calibration, I get plots that still look like the 
ones I sent originally. So it is really the passband calibration that 
improved things, not the rejection of the guard band channels.



On Mon, Dec 2, 2013 at 12:59 PM, Primiani, Rurik  wrote:

    Hi Taco,

    Very interesting/confusing.

    When you calculated the pseudo-continuum phase for SWARM did you use
    the entire band (i.e. all 16k channels)? Or did you excise the guard
    band?

    Thanks,
    Rurik

    On Mon, Dec 2, 2013 at 11:40 AM, Young, Ken  wrote:
    >     I've looked at the phase closure on BL Lac for the SWARM data we took
    > the week before last.    Attached are plots of all phase closure triangles
    > which include antenna 2.   One set of plots shows the phase closure when the
    > pseudo-continuum was calculated using the portion of the Legacy Correlator
    > which overlapped in frequency with the SWARM Correlator (Legacy Correlator
    > chunks s37 through s48).   The second set of plots shows the phase closure
    > when the pseudo-continuum was calculated from the SWARM data.    As you can
    > see, all the triangles close to 0 when calculated from the Legacy Correlator
    > data, and none of them do when calculated from the SWARM data.   I think
    > this means that there is a baseline-based phase error in the SWARM data.
    > The signal/noise on the SWARM plots is also much higher.
    >
    > Taco
    >