Index: trunk/MagicSoft/Mars/Changelog
===================================================================
--- trunk/MagicSoft/Mars/Changelog	(revision 3205)
+++ trunk/MagicSoft/Mars/Changelog	(revision 3206)
@@ -5,4 +5,8 @@
                                                  -*-*- END OF LINE -*-*-
  2004/02/16: Markus Gaug
+
+   * mcalib/MCalibrationPix.cc
+     - documentation from David's email added
+     - EffectiveQE decreased from 0.2 to 0.18 (see documentation)
 
    * mcalib/Makefile, mcalib/CalibLinkDef.h:
Index: trunk/MagicSoft/Mars/mcalib/MCalibrationPix.cc
===================================================================
--- trunk/MagicSoft/Mars/mcalib/MCalibrationPix.cc	(revision 3205)
+++ trunk/MagicSoft/Mars/mcalib/MCalibrationPix.cc	(revision 3206)
@@ -40,4 +40,48 @@
 //   Error F-Factor = 0.02
 //
+// - Average QE: (email David Paneque, 14.2.04):
+//
+//  The conversion factor that comes purely from QE folded to a Cherenkov
+//  spectrum has to be multiplied by:
+//  * Plexiglass window -->> 0.96 X 0.96
+//  * PMT photoelectron collection efficiency -->> 0.9
+//  * Light guides efficiency -->> 0.94
+//
+//  Concerning the light guides effiency estimation... Daniel Ferenc 
+//  is preparing some work (simulations) to estimate it. Yet so far, he has 
+//  been busy with other stuff, and this work is still UNfinished.
+//
+//  The estimation I did comes from:
+//  1) Reflectivity of light guide walls is 85 % (aluminum)
+//  2) At ZERO degree light incidence, 37% of the light hits such walls 
+//    (0.15X37%= 5.6% of light lost)
+//  3) When increasing the light incidence angle, more and more light hits 
+//     the walls.
+//
+//  However, the loses due to larger amount of photons hitting the walls is more 
+//  or less counteracted by the fact that more and more photon trajectories cross 
+//  the PMT photocathode twice, increasing the effective sensitivity of the PMT.
+//
+//  Jurgen Gebauer did some quick measurements about this issue. I attach a 
+//  plot. You can see that the angular dependence is (more or less) in agreement 
+//  with a CosTheta function (below 20-25 degrees),
+//  which is the variation of teh entrance window cross section. So, in 
+//  first approximation, no loses when increasing light incidence angle; 
+//  and therefore, the factor 0.94.
+//
+//  So, summarizing... I would propose the following conversion factors 
+//  (while working with CT1 cal box) in order to get the final number of photons 
+//  from the detected measured size in ADC counts.
+// 
+//  Nph = ADC * FmethodConversionFactor * ConvPhe-PhFactor
+// 
+//  FmethodConversionFactor ; measured for individual pmts
+// 
+//  ConvPhe-PhFactor = 0.98 * 0.23 * 0.90 * 0.94 * 0.96 * 0.96 = 0.18
+// 
+//  I would not apply any smearing of this factor (which we have in nature), 
+//  since we might be applying it to PMTs in the totally wrong direction.
+// 
+// 
 /////////////////////////////////////////////////////////////////////////////
 #include "MCalibrationPix.h"
@@ -62,5 +106,5 @@
 const Float_t MCalibrationPix::gkConvFFactorRelErrorLimit = 0.1;
 
-const Float_t MCalibrationPix::gkAverageQE                = 0.20;     
+const Float_t MCalibrationPix::gkAverageQE                = 0.18;     
 const Float_t MCalibrationPix::gkAverageQEErr             = 0.02;  
 const Float_t MCalibrationPix::gkConversionHiLo           = 10.;