I do exactly the same as Gil and Roger. And if it does not match , I change the model definition , so that I have no worry about thinking about (whether there is any difference between) xl output and FP and bridge mode and fseq and importing old stuff to data layers etc etc.
Another thing I do , is have just one model come on at a time for 10 seconds and then go off, so that if there is an overlap it becomes apparent. For the faces , I do each channel for 10 secs.