Andy shows that some channels have super had linearity (15% under at low flux) - would be interesting to see how the linearity code deals with these extreme cases.