Observers may forget that some frames lost due to missing synchronisation. There is no synchronisation between the readout of the detector and the satellite or wheels motions. As a result, at least one frame is lost at the beginning or at the end of the integration in a given configuration/position. In the manual we mentionned that, for security purposes, the logic would add two frames, one at the beginning and one at the end, to each integration in a given configuration. It does not do that, and so you have to do it. Thus if you require N frames to be taken either per raster point, or per CVF step, or per beamswitching cycle, you must actually enter in the PGA the following N(Total):
Note that these formulae are conservative and slightly overestimate the number of synchronisation frames necessary. Note also that ISOCAM's simulator takes them into account to compute the observing time ( axis on the graph), but as you are required to enter N in the PGA, and not T, you have to include these synchronisation frames.