SOLVE HOLOGRAPHY

 

        CLIC\SOLVE HOLOGRAPHY [MO nmodes] [NP npix] [IT niter gain] [FREE r1
    r2  ...]  [/OFFSET  x  y  z] [/PLOT [AMP amin amax astep] [PHA pmin pmax
    pstep] [ERRORS emin emax estep] [NUMBER]

    (Keywords like MO, NP, IT, FREE
     may be given in any order, each must be followed by its arguments; they
    are all optional)


    This command computes an antenna surface map from a  set  of  holography
    measurements.  The  set of scans (procedure HOLO) should have first been
    calibrated in phase, amplitude and RF passband relative  to  interspaced
    correlation  scans  in  the  direction  of  the source.  The map will be
    computed from the first band and subband sets chosen with  commands  SET
    BAND  and  SET  SUBBAND.  SET  BAND AVERAGE is recommended for continuum
    measurements;  only  continuum  subbands  should  be  used.   For   line
    measurements,  the continuum width of one of the correlator units should
    match the actual line width for better sensitivity.  The antenna  to  be
    studied  should  be  selected  by  command SET ANTENNA i.  Data from the
    baselines linking this antenna  (scanned)  to  other  (fixed)  available
    antennas are averaged.

    The amplitude and phase maps are obtained by Fast Fourier  Transform  of
    the  observed  beam  map. The maps will be square (npix by npix pixels).
    The default for npix is 64, it should be  greater  than  the  number  of
    observed holography scans (usually 16 or 32).

    After FFT a gaussian illumination function is fitted in the  amplitudes,
    giving  the  offset  from  the center (in meters) and the edge taper (in
    dB).  If /PLOT is given, the amplitude map will be shown (in  decibels),
    from  -15  to  0 dB, with contours in steps of 3db (these may be changed
    using amin, amax and astep).

    A lest square fit is used to correct the phases from a  remaining  phase
    offset,  pointing  errors,  and focus offsets. The panel rings following
    FREE are not used for this fit. If /PLOT is given,  the  antenna  normal
    surface errors will be shown, in micrometers, from -500 to 500 mum, with
    contours in steps of 100 mum (these may be changed using keyword  ERRORS
    followed  by  emin,  emax  and  estep).   If  /PLOT  PHASE is given, the
    residual phase map will be plotted instead of surface errors, in radians
    from -pi to pi, with contours in steps of 0.2 radian.

    If this map shows remaining 2*pi discontinuities, or  if  focus  offsets
    larger  that  1mm  are  found,  you  should  try using option /OFFSET to
    correct the phases  for  an  offset  (x,y,z  in  meters)  in  the  focus
    coordinates, before fitting. This should lead to better rms values.

    /PLOT gives also the amplitude illumination pattern, in dB; default plot
    limits are -15dB to 0dB by step of 3dB; use /PLOT AMP amin amax astep to
    change these settings.  If NUMBER is given  as  a  /PLOT  argument,  the
    panels numbers are drawn.


    The rms values for the phase and the normal surface errors  (in  radians
    and  micrometers)  are given, both with and without amplitude weighting.
    The contribution of the illumination amplitude distribution and  of  the
    observed phase errors to the antenna efficiency are given.

    Finally, if nmodes is larger than 0, a listing of  panels  displacements
    is  computed, and written in a file like "panels-an1.dat". This uses the
    parameter nmodes which is the number of modes used for each panel: 1  is
    only  the translation mode (normal to the antenna surface); 3 (the usual
    setting), adds both tilt modes, radial  and  tangential,  but  no  panel
    deformation;  4  adds  a torsion mode and 5 a motion of the panel center
    relative to its edges (there are only 5 screws for each panel, thus only
    5  possible  modes).  In the file "panels-an1.dat" a line for each panel
    is printed. The first two numbers are the panel numbers, followed by  up
    to  five  screw  settings  (three  only  for  the inner ring). All screw
    settings are equal if nmodes was set to  1,  only  one  number  is  then
    printed.

    If niter is not forced to zero, the fit  is  obtained  iteratively:  the
    panel orthogonal deformation modes are computed from the aperture phase,
    then the phase change that  these  deformations  would  have  caused  is
    computed  (by  doing  a  FFT  to  the  beam  map, doing a cut-off at the
    observed map size, followed by a FFT back to the  aperture  plane),  and
    subtracted  from the aperture phase; second order panel deformations are
    computed from these residuals, and so on. The number of iterations niter
    and  a  gain  to  this  iterative  procedure may be specified ([IT niter
    gain]); their default values are 5 and 1.0 . Use IT 0 for  no  iterative
    procedure  at  all.  At  each step the phase residual rms and the rms of
    panel deformations fitted are given (weighted by  the  fitted  amplitude
    illumination and counted perpendicularly to the surface).

    A gildas image file of the results is kept (e.g. "jj-mmm-yyyy-an1.map"),
    in  which  plane  3  in  the amplitude in dB, plane 4 the raw phases and
    plane 2 the residual phase in  radians  (plane  1  contains  the  fitted
    amplitude, i.e.  a Gaussian).