Second Order Referenced Pointing

                          May 1, 1996
                          K. Sowinski

  With this announcement I declare second order referenced pointing 
to be usable.  I will endeavor to explain the intricacies of its
use.  If a motto were to be attached to this memo, it should be
that it took longer to write this than it did to make it work.  

How it really works

  I first define "primary pointing offsets" as those determined with no
other pointing correction applied.  "Secondary pointing offsets" are
those determined, usually at a different band, relative to the primary
offsets.  The use of referenced pointing is controlled by a flag, the 
referenced pointing flag, on each source card of the observe file.  The
value of this flag is used to tell the system whether a particular
pointing scan is to record primary offsets, secondary offsets or nothing
at all.  For scans which are not pointing scans, this flag tells the 
system whether or not referenced pointing corrections are to be applied
or not.  The details:

Flag    Scan    Action
----    ----    ------
blank    PTG    Apply no previous offsets; any previously determined
                primary and secondary offsets will be forgotten; the
                offsets determined during this scan will be remembered.
        ~PTG    No previously determined offsets will be applied.

T        PTG    Apply previously determined offsets; the offsets
                determined during this scan will not be remembered.
        ~PTG    Apply previously determined offsets; if both primary
                and secondary offsets are available, their sum will
                be used.

S        PTG    Previously determined offsets will be applied and
                saved as "primary" offsets; the offsets determined
                during the scan will be saved as "secondary" offsets.

R        PTG    Any previously saved primary offsets will not be changed;
                previously determined secondary offsets are forgotten;
                the offsets determined during the scan will be saved
                as secondary offsets.

Neither "S" nor "R" should be specified for a scan that is not a pointing
scan.  The value of the "R" (refresh) option is that the secondary 
offsets can be redetermined at a later time, or a different band, without
having to measure the primary offsets again.  Of course this only works
if the pointing has not changed so much that the beam at the frequency 
of the refresh pointing scan misses the calibrator.  


Examples

  Suppose that you wanted to do second order referenced pointing at Q
band.  The following sequence would be adequate:

Source          Band      Ptg Flag
------          ----      --- ----
ptg calib        XX        blank
ptg calib        QQ        S
real source      QQ        T

If you wished to touch up the pointing after a while but think that the
primary pointing offsets are still good enough, then add:

ptg calib        QQ        R
real source      QQ        T

Suppose you wanted to alternate between Q band and K band:

ptg calib        XX        blank
ptg calib        QQ        S
real source      QQ        T
ptg calib        KK        R
real source      KK        T

If the primary offsets need to be remeasured, then an X band scan will be
required with the referenced pointing flag set to "blank".


Observe

  I have no idea whether Observe allows values other that " " or "T"
for the referenced pointing flag.  Perhaps Wes can "reply all" to
this message and tell us.  If Observe only allows "T" or "blank" than
post-Observe editting will be necessary.


When it fails    ("But stick your courage to the sticking-place")

  Using second order referenced pointing greatly complicates the
answer to the question, "did referenced pointing work?".  Here is an
answer to that question.  A status bit recorded on the archive tape
explains whether or not offsets were successfully applied to each
antenna.  The precise meaning of this bit is:

1   Pointing correction was requested, but the last pointing
      scan failed for this antenna or was more than 12 hours old.
0   A pointing correction was either not requested, or was requested
      and successfully applied.

If only primary referenced pointing is used, then this bit describes
success for each antenna in a straigtforward way.  If a pointing scan
is done to determine secondary offsets, then, for succeeding scans,
this bit reflects the success, or lack of it, of that pointing scan only.  
The validity of the pointing scan that determined the primary offsets
is reported in the data recorded during the pointing scan that attempts
to measure secondary offsets.  I claim this is reasonable behaviour.
While I am too lazy to defend that thesis here, I will be happy to 
discuss it if anyone disagrees.  

  FILLM could learn to something interesting with this bit.  At the 
least, it could record which antennas are uncorrected for each scan.

  It is worth noting PTG also complains during the pointing scan
when apparantly working antennas do not have valid pointing offset
solutions.  

Future work

  Someone needs to look at how well pointing really works at Q band.
How much flux is needed? How short must the integration time be?
are two obvious questions.  Currently the pointing system does
work for integration times less than ten seconds, but only 
grudgingly.  This could probably use some work.

======================================================================