Indeed I am still wondering what you have changed from one processing to the other as the software that you are using is always the same.
the MASTER variable in the project.conf must be the full path to the splitted and orbit corrected master (processed with SNAP) in BEAM-DIMAP format.
In case you might need it, you can give me more info about the dataset and parameters so I can try to reproduce your problem. Or we can see it together
Let me know