As of Wed Jan 17 16:44:57 2018

MBDLY: Fits multiband delays from IF phases in SN table


INNAME                             Main input file (name).
INCLASS                            Main input file (class).
INSEQ             0.0    9999.0    Main input file (seq. #).
                                      0 => high
INDISK            0.0       9.0    Disk unit #.       0 => any
INVERS                             Input  SN file version no.
OUTVERS                            Output SN file version.
BIF               0.0       16.0   First IF to use. 0=>1
EIF               0.0       16.0   Last IF to use.  0=>last
TIMERANG                           Time Range is consider.
SUBARRAY                           Subarray (default=1)
FREQID                             Frequency ID (default=-1)
OPTYPE                             'DISP' -> fit dispersion and
                                     MB delay from SB delays
                                   else   -> fit MB delay from
                                     SN table phases
PRTLEV            0.0       10.0   Print fit results
                                   >=1 fits, 2=IF resids.
APARM                              Editing parameters
                                   (1) = min SNR; def=5
                                   (2) = max rate rms (deg)
                                         def = 50d
                                   (3) = max MBD rms (deg)
                                         def = 20d
                                   (4) = min. no. IFs. 0->all
                                   (5) = MBD Search range.
                                         0 = ambiguity.
                                         <0.  Use zero.
                                   (6) > 0 remove residual phase
                                   (7) > 0 keep bad data
                                   (8) DISP only >0 -> keep only
                                       dispersion, zero delay
                                       and phase
                                   (9) DISP only > 0 -> do NOT
                                       correct delay and phase
                                       for dispersion
                                   (10) DISP only, > 0 ->
                                       average DISP over R and L


Type:  Task
Use:   MBDLY fits multiband delays and, optionally, dispersion from
       the IF phases in an SN table produced by FRING or CALIB.  The
       results are placed in another SN table.
  INNAME......Main file name (name).       Standard defaults.
  INCLASS.....Main file name (class).      Standard defaults.
  INSEQ.......Main file name (seq. #).     0 => highest
  INDISK......Disk drive # of image.       0 => any
  INVERS......Input extension file version #.
  OUTVERS.....Output extension file version #. 0=>make new
  BIF.........First IF to use.  0=> 1.  Use to determine MBDLY
              for dual frequency SN table
  EIF.........Last IF to use.   0=> last.  Use to determine
              MBDLY for dual frequency SN table.
  FREQID......Frequency ID (may need to set to 1)
  TIMERANG....Time range to consider
  OPTYPE......'DISP' -> use single-band delays to fit for a
                        single multi-band delay plus dispersion for
                        each record in the input SN table.
              else   -> use the phases in each IF to fit for a
                        multi-band delay
  PRTLEV......<= 1.0 -> show multiband fits.
              >= 2.0 -> show phase residuals for each IF
              Suggest using 2.
  APARM.......Editing control parameters:
              (1) is the minimum average SNR (WEIGHT in the SN table).
                  0 => 5.  0.01 lets everything by.
              (2) is the maximum acceptable rate scatter among the IFs
                  in units of degrees over the integration period. The
                  default is 50 degrees.
              (3) is the maximum acceptable rms residual phase of the
                  IF phase residual in degrees.  The default is 20
              (4) is the minimum number of IF's for a valid solution.
                  The default is all.
              (5) is the search range (+/- value) for the MBD.
                  Default (0) is the natural range and is equal to the
                  least-common stepping frequency, but is often chosen
                  too high.  Try APARM(5)= 10 for DELZN work.  If <0,
                  then assume MBDLY = 0.  This is useful for just
                  averaging the phases over the IF's.
              (6) If =0, keep residual phase in data.  This will leave
                  the observed phase in the data after removing the
                  multiband slope.  Use with CL2HF.
                  If =1, Remove residual phase from data.  This will
                  zero the observed phase.  The phase inserted in the
                  SN table is the average observed phase.  Use this
                  option for further imaging.
              (7) > 0 => pass "bad" data.  Set this equal to 1 if you
                  want to pass through 'bad' data previously blanked
                  out by MBDLY.  A zero or low quality code (5, 6, or
                  7) will be set by CL2HF.  Useful for multiband
                  observations when a source is detected in one band
                  but not the other.
              (8) DISP only: If > 0, then write the dispersions in the
                  output SN table but zero the phases and single-band
                  delays.  After CLCAL, you then run FRING over again
                  on the dispersion-corrected data.  This may produce
                  a more reliable set of phases and SB delays.
              (9) DISP only:  If > 0, do NOT correct the SB delays for
                  the dispersion.  Otherwise (if APARM(8) <= 0), the
                  delays are corrected for having a dispersion now
              (10) DISP only: If > 0 average R and L fit dispersions
                  before correcting phases and delays.  NOTE - THE


DOCUMENTOR:  Ed Fomalont   95.03.08, rev 06.05.21


New method added May, 2015:  OPTYPE = 'DISP'.  Use MBDLY in this mode
on an SN table written by FRING in which a separate delay is found for
each IF (spectral window).  It uses the delays found by FRING to solve
the equation
      SBdelay = MBdelay - Dispersion * lambda * lambda
by least-squares.  The task writes the MBdelay, and Disp in the
appropriate columns of the output SN table.  By default, the task will
correct the SBdelay by adding Disp*(Vlite/freq)^2 leaving the delays
as SB delays and the phases by subtracting Disp*Vlite+lambda.

Comments below apply to all other values of OPTYPE:

FRING and CALIB determine the value of the antenna-based phase
for each IF for a specified solution interval.  The results
are placed in an SN table.  The program MBDLY then reads this SN
table (INVERS) to determine the best phase slope through the
residual phases of the specified IF's.  The results are put in a
new SN table (OUTVERS).  This delay over all of the IF's is
called the Multi-band delay or the Group delay and it is often
used in geodetic/astrometric experiments.  The advantage is that
the Group delay is not subject to lobe ambiguities associated
with the phase.  The group delays obtained from MBDLY can be
put into DELZN to determine the residual atmosphere and clock
residuals for each antenna, as a function of time.

The peculiar phase constants for each IF must be determined and
removed before MBDLY can calculate the phase slope.  The
peculiar phases can be obtained by using PCCOR and choosing an
appropriate time range for a good calibrator.  Another method
is called the 'manual phase calibration'.  Here you run FRING
on a short segment of data (less than a minute) for a strong
source with a complete set of data to obtain the Single-band
delay, rate and phase.  It is recommended to zero the rates
(using SNCOR) unless the rates are persistently large for an
antenna.  Apply the solution to the entire data base using CLCAL.
This will line-up the IF phases and define the multiband delay
to zero for all antennas during the period of the short segment
of data used in FRING.  The relative IF phases should then be
relatively stable so that the residual IF phases for any other
source (strong and without too much structure) should be linear
with frequency.  MBDLY finds this phase/frequency slope.

The Multiband delays can also obtained directly from FRING with
APARM(5)=2.0.  However, this procedure has been found somewhat
unreliable so it is best to use FRING with APARM(5)=0
and then MBDLY.

Since the phases are sampled at a few discrete IF frequencies,
the best slope is ambiguous by an amount (1000/DF) nsec, where
DF is the basic IF frequency spacing (MHz).  For example, if the
IF spacings are 0, 80, 140, 400 MHz (above a reference
value), then DF is 20 MHz.  The ambiguity spacing of the
MBDLY is then 50 nsec and this is what is used if APARM(5)=0.
However, it is best to use as small a MBD spacing as possible,
and no more than 10 nsec is recommended for most DELZN

For astrometric/geodetic experiments, the results in the SN
table produced by MBDLY are interpolated into the CL table
using CLCAL.  Then, the task CL2HF calculates relevant
astrometric quantities needed for the Calc/Solve software and
places these results in an HF extension file.  Finally the task
HF2SV translates the HF data into something that can be read
into Calc/Solve.   See instructions associated with these two

  BIF........The begining and
  EIF........The ending IF to use for the solution.  Experiments with
             simultaneous S/X bands are common.  It is recommended
             to separate the two frequencies and reduce separately.
  TIMERANG...The time range of the input SN table to process.  0 means
             take everything.
  PRTLEV.....Use 1 for most executions.  The output will list some
             information about the fits, including the rms phase
             residual.  If PRTLEV=2, the residual phase error for each
             IF is listed underneath each line.  This offset should be
             less than a few degress for a reasonable fit.
  APARM..... Editting control parameters.  Any SN table entry that
             exceeds one or more of the following criteria, will have
             a blanked multiband delay written in the output.
             (1) Minimum SNR.  The default is an SNR of >5.  This is
               reasonable. To include all data, let APARM(1) = 0.01
             (2) Maximum rate rms. This is the maximum phase scatter
               over the integration time which is caused by the
               scatter in the rate solutions for the IF's.  The
               default is 50 degrees.  This could be set to 10 degrees
               for good data.  This option removes source solutions
               which have significantly different phase rates amongst
               the processed IF's.  It is a good indication of bad
               data or too weak of a source.
             (3) Maximum multi-band delay rms.  The maximum phase
               scatter in the IF's after the removal of the best MBD.
               The default is 20 degrees.  This could be set to 7
               degrees for good data.  Use PRTLEV=2 if there are
               problems here.
             (4) Minimum number of IFs.  This is defaulted to all IFs
               otherwise no solution is given.  This can be decreased
               even to 2 IF's as long as the phase difference between
               the two IF's is less than 180 deg.
             (5) The MBD Search Range.  The default is the natural
               ambituity range which is equal to the inverse of the
               'unit' spacing in the frequency set.  For example, if
               the four IF relative frequencies are 0, 70, 385, 490
               MHz, then the unit frequency sampling is 35 MHz, or
               28.6 nsec which is the default.  For frequencies which
               are not commensurate, APARM(5) should be set to a
               reasonable range.  For example for most DELZN runs,
               APARM(5)=10 should be sufficiently wide.
                  If APARM(5)<0, then assume MBDLY is
               zero and just average phases in IF's.  This is useful
               for getting the average phase over all of the IF's for
               a weak source.
             (6) Unused now: formerly controlled whether residual
               phases were zeroed or not.  They are now passed
             (7) > 0 => pass "bad" data.  Set this equal to 1 if you
               want to pass through 'bad' data previously blanked out
               by MBDLY. A zero or low quality code (5, 6, or 7) will
               be set by CL2HF.  Useful for multiband observations,
               when a source is detected in one band but not the other.