[daip] multiband delays

[Leonia Kogan]

Mark,

You may be right if FRING has a "good" algorithm and the condition of a 
good coverage of one Fourier transform axis (at least the right 
sampling) is satisfied.
The following example elucidates the possible problem at CLCOR=> FRING
Suppose we have 4 even distributed frequencies IFs:

0, 1, 2, 3 MHz;

Lets CLCOR recalculates 1" shift to delay = 1.1 mksec

Then the phases in turns are

0, 1.1 2.2 3.3  turns (1)
which is equivalent to

0, 0.1 0.2 0.3 turns  (2)

CLCOR calculates the phase correction as COS and SIN of (1,2) and 
obviously does not make a difference between (1) and (2)

The FRING (as I read at help) fits the delay at the sequence of the 
phases using the least square. This algorithm may not be good enough
to discriminate (1) and (2)
Having fitted the straight line into the sequence of phase given by (2), 
FRINGE will find solution
for delay = 0.3/3=0.1mksec  .not. initial 1.1mksec !!!

About your address to the mapping of a source:

The gap in UV coverage (df in our case) limits the maximum size of the 
source feature (delay at our case)

Leonia



> Leonia,
> 
>    Regarding the multiband delay problem, I gave it more thought and I 
> don't think the issue is a large delay.  Assuming FRING has a good 
> algorithm, eg a Fourier transform of frequency to delay, there should be 
> no problem estimating multi-band delays in the presence of phase wraps.
> (By analogy, when mapping a source from uv-data, the Fourier transform 
> locates the source in the image plane in the presence of multiple phase 
> wraps.)
> 
> However, I looked back at my FRING parameters and I has set a delay 
> window (DPARM(2)) of 20 nsec, which was too small for the 1" shift on 
> long baselines.
> 
> Mark
> 
> 
> -------------------------------------------------------------
> Mark J. Reid                 Phone: 617-495-7470
> Harvard-Smithsonian CfA      Fax  : 617-495-7345
> 60 Garden Street             Email: reid at cfa.harvard.edu
> Cambridge, MA 02138, USA     Web  : www.cfa.harvard.edu/~reid
> -------------------------------------------------------------