[daip] CLCOR
Eric Greisen
egreisen at nrao.edu
Thu Nov 26 11:12:21 EST 2009
Andreas Brunthaler wrote:
> Hello,
>
> I noticed a bug in CLCOR, version 'TST' using OPCODE 'ANTP' and 'ANTC'.
> When I shift a source, it changes the delay for all antennas, expect the
> first one (see attached plots). I tried this now with two different
> datasets, with the same result. I looked at some older data (reduced in
> April 2009). There CLCOR worked fine for all antennas. Also the Version
> 'NEW' works fine.
>
> Other changes applied by CLCOR (e.g. ATMO, PANG, EPOS) work for all
> antennas.
This is not an error. Delays and phases are all relative anyway. The
problem with the old version of the code was that it put in an absolute
change in delay. With the antenna at an earth radius, an serious
position shift would make the phase change extremely large. For a
nearby antenna, the phase change would also be extremely large. The
relevant thing to that baseline would be the small difference in these
exceedingly large numbers - a difference that could not be accurately
represented in single-precision floating point numbers as stored in the
CL table. I changed the code to make everything relative to the first
antenna so that EVLA and even VLBA changes would be more accurately
represented.
The result on your data is the same assuming that the old method did not
lose accuracy. On the EVLA for normal position changes the phases were
not accurate within a full 360 ambiguous lobe.
Eric Greisen
More information about the Daip
mailing list