[daip] CLCOR position shifting

[Eric Greisen]

Mark Reid wrote:
> Eric Greisen wrote:
> 
>> Do remember that 1 arc second may exceed the accuracy in the CL table 
>> for VLBI observations.
>>
> 
> Actually I'm unaware of such limitations.  Can you give the specs that 
> cause the limitations?
> 
> Mark

The CL table stores corrections in REAL*4 which means about 22 bits of 
accuracy.  The EVLA uses Earth centered coordinates so CLCOR for a 
position shift put in one for each antenna that was really enormous - 
the delay portion of some arc sec * radius of the Earth.  For the EVLA, 
the actual correction is the difference between two really close values 
but they were so large that the difference was mostly in the bits that 
got lost when the delays were stored in the CL table.  We changed CLCOR 
to use a reference antenna for these shifts so for the EVLA the result 
is accurate enough.  For VLBI, there is a  more serious limit because 
the baselines are really long.

I though we - mostly you and Leonia - went over this a month or so ago 
and at the time you found shifts of 0.1 arc sec worked fine while 1.0 
arc sec had problems.

Eric