For those who work with nCounter rcc files from two or more similar codesets (rlf) and have no calibrator sample:
the program finds the common genes between codesets as it should, but then calculates normalization factors separately for the samples in the different codesets. So you cannot use those factors together downstream in the analysis.