Why does this inverse square calculation fail to predict actual data?

[JimLub]

TL;DR

A simple test using solar cell and LEDs - shows dramatic differences between predictions and actual data. Test results are shown in attached file.

The test data and notes are attached - showing that the inverse square calculations fail to reasonably predict the actual dimming of light over a test distance of 168 mm. Did I err in my test design or my calculations?

 

[mjc123]

A matrix of LEDs is not a point source - not even a good approximation at short distances. At 3/4" square, it is bigger than the distance to the detector at the closest distance. There is no way that this is going to give you an inverse square dependence of the intensity. I don't suppose it's even emitting light equally in all directions.

 

[sophiecentaur]

It's an easy calculation to show where the ISL fails, using two point incoherent sources, spaced at a certain distance and with a detector at some distance away. The resulting intensity will be
1/R₁² +1/R₂²
which is not 2/R²
except when R₁ = R₂
i.e. along a normal to mid point of the line of centres.
Edit: as R increases, the departure from ISL is less and less.