The Saga of the Noise Floor

[Stupendous Man <richmond@a188-l009.rit.edu>]

  In TN 76, I showed that repeated measurements of bright stars
from Mark IV images exhibited a "noise floor" of about 0.006 mag.
I've been trying to figure out the major contributor to this
noise.  Arne has given some good suggestions.  Message 66 on the
TASS E-mail list for March 2001 provides some of my earlier 
results in the search.

  After some additional experiments today, I can tabulate a
number of sources and the size of each:

        quantization of magnitudes:             0.0005  mag
        noise in the master dark:               0.0005  mag
        noise in the master flat:       approx. 0.0013  mag  in V-band


  Inspired by Arne's comment that one candidate might be the software 
which extracted the magnitudes via synthetic aperture photometry, I went
back and examined the 'phot' program in the XVista suite.  Years ago,
I introduced an approximation to deal with pixels which fall partially
within a circular aperture.  I modified the code to calculate more
exactly the flux contributed by such partial pixels, and re-analyzed
the Mark IV frames on Data Disk 16.  The result -- negligible change
on a star-by-star basis, and no change in the measured noise floor.
Rats.

  I still haven't found the major contributor.  My next attempt
should be to check the sky values calculated by the synthetic 
aperture photometry routine, I think.  This is a good excuse to
crawl back into the guts of the image processing code I've written
over the years :-/

                                          Michael Richmond