The Jupiter map 20161209s101 allows to estimate the level of the "plateau". The figures below shows the signal of all arrays in % of the Jupiter peak.
Alessandro. Interesting. In particular to see that you confirm that the footprint is the whole array and not a single readout line. This is (again) a bit surprising to me. 1000 pixels seeing 1% each makes 10 times higher than the source itself. If it was the optics so we would have to admit that we absorb in the main beam less than 10% of the total. I would have preferred to see a slice of the array, and attribute this to a change in the readout line impedance.