On Monday, March 24, 2014 12:37:34 AM UTC+8, Louis Boyd wrote:   
   > sgpopticsguru@gmail.com wrote:   
   >    
   > > On Saturday, March 22, 2014 11:40:21 PM UTC+8, Mikko OH2HVJ wrote:   
   >    
   > >>   
   >    
   > >>   
   >    
   > >>> I repeat the test for various exposure durations under the same   
   >    
   > >>> light   
   >    
   > >>   
   >    
   > >>> and do a linear regression on pixel values. The linearity of the   
   >    
   > >>   
   >    
   > >>> pixels are very good but they do not cross from zero, the   
   >    
   > >>> regression   
   >    
   > >>   
   >    
   > >>> line has a DC offset.   
   >    
   > >>   
   >    
   > >>   
   >    
   > >>   
   >    
   > >> There's a bias voltage in the sensor output signal by both the   
   >    
   > >>   
   >    
   > >> chip amplifiers and purposefully added so that AD converter   
   >    
   > >>   
   >    
   > >> does not have to start from zero, where it may be non-linear.   
   >    
   > >>   
   >    
   > >>   
   >    
   > >>   
   >    
   > >> This is removed in astronomy by taking 'bias images' in dark with   
   >    
   > >>   
   >    
   > >> zero exposure time and substracting this from the image.   
   >    
   > >>   
   >    
   > >>   
   >    
   > >>   
   >    
   > >> --   
   >    
   > >>   
   >    
   > >> Mikko OH2HVJ   
   >    
   > >   
   >    
   > > Mikko, I do this. If you look at my description, I first measure FPN   
   >    
   > > and take it out. Despite this I see a DC offset later which makes me   
   >    
   > > wonder.   
   >    
   > >   
   >    
   >    
   >    
   > What is the magnitude of the effect your seeing?    (measured A/D counts   
   >    
   > out of full range A/D counts) at the zero crossing?    What kind of    
   >    
   > sensor and camera?  Manufacturer & model of both?   
   >    
   >    
   >    
   > How are you  determining that the exposure is uniform over the nominal   
   >    
   > exposure time?  What is the light source?   What method are you using to    
   >    
   > start and stop the exposure.  Switching the light source on and off? A    
   >    
   > mechanical shutter in the camera?  Interline transfer on the sensor?   
   >    
   >    
   >    
   > This is just a guess based on limited information, but you may be    
   >    
   > measuring the ramp of the opening  and closing of the camera's shutter    
   >    
   > if it's a Scientific grade camera with a mechanical shutter.   That can    
   >    
   >   be removed in the reduction process.a   It can be a simple constant    
   >    
   > subtracted from the exposure  if the shutter is always allowed to open    
   >    
   > fully  before closing.   It's not an error.  It's just a fact that    
   >    
   > doubling the time between opening the shutter and closing it allows a    
   >    
   > little more than twice the light to pass though as opening it and    
   >    
   > closing it twice.   
   >    
   >    
   >    
   > You may also care if the exposure  at the center of the sensor is equal    
   >    
   > to the exposure off axis.  Most scientific cameras put mechanical    
   >    
   > shutters right in front of the sensor and have two leaves which open    
   >    
   > from the center which  gives a longer exposure in the center of the    
   >    
   > sensor.  That can be more  important to precision photometry than    
   >    
   > linearity vs exposure time.   
   >    
   >    
   >    
   > What is your specific application?   
      
   Dear Luis,   
   Thanks for the detailed answers. The magnitude of the effect is very small but   
   it is there nonetheless. Camera is a global shutter cmos from Aptina.   
   (AR0134). Light source is a point LED with diffuser in front. Since camera   
   shutter is electronic, most of    
   your theory may not be applicable. I do not have an integrating sphere handy   
   so I use a point source and change the current on the LED to change overall   
   light level.    
      
   I am essentially calibrate the colors of the camera perfectly, hence my effort   
   to take out the FPN/Dark Noise/Row Col Noise/PRNU etc. so that I am left with   
   temporal noise and real light.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)