Timescale for scanning over filter transmission curves
                            Michael Richmond
                              Feb 27, 2008

  I've tried to do a very rough estimate -- good to a factor of two or so --
for the time it might take us to check the filter transmission curves
on orbit.  The basic idea is the following:

       - start with an incandescent bulb to make light of all wavelengths
         
       - use a monochromator to send light within a narrow range
             of wavelengths into an integrating sphere

       - light from sphere goes into fiber, then to Ring of Fire
         
       - light eventually falls on focal plane
         
       - calculate the number of photons striking each pixel per second

         
What I want to find out is, "how long will each exposure last?"  This will
give us some idea for the time to scan over a large span of wavelengths
covering the range of SNAP detectors.

  I grabbed parameters for the optical system and properties
of the Ring of Fire from the document

     http://spiff.rit.edu/richmond/snap/flatfield/aug18_2005/aug18_2005.html

That document provides both "optimistic" and "pessimistic" values
for the parameters.  In this short note, I pick values in the middle
for each parameter.  I can do a more detailed report later.

  My light source is a 100-Watt light bulb with a filament
running at T=3000 Kelvin, which miraculously transforms all
input power into light.  I figure this can be scaled to more
realistic light sources easily.

  If we pick bins in wavelength which are 10 nm wide, then 
within each bin, we find the number of photons striking each
pixel on the focal plane per second to be about ...

      at   400 nm     about     310  photons/sec      exptime 32 sec

      at  1000 nm     about  10,400  photons/sec               1 sec
  
      at  1700 nm     about   9,300  photons/sec               1 sec


... and if we require 10,000 photons per pixel for each exposure,
then the exposure time is shown in the rightmost column above.

  If we wish to sample the entire range from 400 nm to 1700 nm 
at 10 nm intervals, we need 130 exposures.  Again, in a longer report
I will calculate the total time properly, but for rough purposes, 
we can place a _strong upper_ limit of 130*32 = 4160 seconds.
That's really a big overestimate, since the exposure time will
drop very rapidly as we move from 400 nm to 500 nm to 600 nm.

  A simple conclusion is that the exposure times required under
these circumstances are typically much less than the readout times.
But remember, everything is based on a 100-Watt light source
running at 3000 K, and the optical setup described in my earlier document.

  I will be travelling for the next few weeks, so I may not be able
to respond in detail to technical questions.  But I may be able
to provide simple clarifications.

                          Michael Richmond