Yes, I think engineers try not to specify where the power comes from because they aren't sure where it should come from. I'm never sure either.
BTW, the code requires monitoring of this power.
I would not take it from the machine room lights/receptacles. I think that's supposed to be a dedicated circuit. I do like the idea of deriving it from the elevator power.
I'm never sure how to handle making sure the elevator makes it to the recall landing before we trip the power.
This is something engineers have to design themselves, when, in my opinion, this is something that should be standardized. I'm refering to the entire elevator/fire alarm/shunt trip power/shunt trip monitoring ball of wax. We draw something up on paper without really knowing any details about the elevator control panel, or the fire alarm panel. If the shunt trip is tested at all, it is tested once at startup and that it.
It just seems like all the functions that are needed for this should be part of the fire alarm system, or maybe part of the elevator control panel. It should part of a tested and listed package. Not something that we force every electrical engineer to design from scratch.
Steve