sky spectrograph for LSST

The concern is spatial variation in OH sky structure. 

Monocam data have 300 electrons per pixel in NIR per minute in 0.4*0.4 arcsec^2 pixels. So broadband integrated flux is (300/(60*0.16)) = 31 electrons per sec per square (10 microns) * (4/f)^2. Say we go to f/2, that's 4 times as much flux. So at f/2 we get around 1 detected electron per square micron per second, at nominal QE with two aluminum reflections. Assume a factor of two loss for spectrograph performance, and we're back to 0.5 electrons in y band per square micron per second. A 1 mm fiber is diameter of 1000 microns and an area of 7.8E5 square microns. That would give us around 2E5 electrons per second broadband. 

The focal length of the lens determines the FOV on the sky. Let's say it's f/2 and focal length of 135mm. The fiber then spans an angle of 1/135 radians, or half a degree. 

If we spread the light over 1000 pixels, we get around 500 electrons per pixel per second.