Dear Francesca,
Am I wrong?
No, you are right!
Decaying radioisotopes as source terms in FLUKA are treated in so-called semi-analogue mode. This implies that decay products (e-,e+,gamma,alpha) are explicitly sampled and transported, but they are sampled from inclusive database information: on average you get the right number, kind, and energy of decay products, but one loses possible correlations among them (hence the attribute “semi” above) as turned out to be the case in your example: the Ka and Kb lines should be indeed exclusive, but they are actually sampled independently.
Sorry that I missed pointing this out in last evening’s reply.
Now, in practical terms, a viable solution is indeed to use the source.f routine as your intuition suggests, manually introducing the energy and branching ratios of the lines you care about and steering the sampling accordingly.
With kind regards,
Cesc