Its aways been a grey area with power supplies to fire detection, despite the BS 5839 offering advice and reccomendations, and BS 7671 stating "see BS 5839 for further spec". As always drawings can't lie or be misenterpreted as can written text,.and I think drawings should be used more often to offer support to what is written down, and reduce the never ending ambiguity associated with regs. When I used to get fire alarm work, I would always install a Wylex 106 switched fuse at the mains or similar bearing the do not switch off engraved label.
About 11 years ago I installed an economy fire alarm in some bedsits in York, as the job was partly funded by a grant from the council. They offered a spec that I didn't fully agree with, they hadn't allowed for remote indication for when the rooms were locked, an interface to shut off the fans in the kitchen, along with appliances using a heat source, and the 20db+ loss rule through a fire door wasn't adhered to. I mentioned this to them before the work commenced at I was told that spec complied with BS 5839? Despite my concerns that I found whilst carrying out a fire risk assessment on the property to decide which category and type of detection to install.
Anyway I decided to do the work and listed them along with other departures they hadn't allowed for on the certificates. After the work was completed, the jerk of works who sounded like he was ex forces came round for the handover, and asked me why I had put a dp key switch next to the panel? I replied "isn't that obvious"? He then stated that if the key wasn't available no isolation could take place at that point, or there could be more than one key available to more than one person and that would contravene the procedure with locking and tagging out (food for thought). And in the event of a switched fuse spur it could be accidently switched off and the fault buzzer not taken any notice of, and an unswitched fuse spur only offered single pole isolation, and not double as required with BS5839.
He also tried to snag the fact that the pvc trunking was supporiting fire alarm cables and they could drop in the event of a fire. When I showed him that "P" clips had been used inside the trunkig as well, he backed off.
He then went on to try lecture me about isolation in general, but then said he was satisified as I had provided a supply via a lockable switched fuse at the mains in take, and the only thing that could interupt the supply to the panel was a power cut, a fault on the circuit, or someone switching off the supply to it. And so there was no need for local isolation at a single panel, as it wasn't a rotaing machine that could potentially cause amputation or worse.
Anyway I let him go on, and the job was handed over and I noted the departures on the certificates and appended their letters and spec to them, and I was paid. I mentioned it to my niccy area inspector, and he said if there had been more than one panel or other ancilliary devices or equipment connected to it, then local isolation would be required for functional switching off. So as to reduce the interuption of the supply to the panel, and rely solely on battery back up and potentially reducing the standby time afforded by the batteries. And as the power supply could be isolated and locked off at the source, it would comply to the EWAR, despite the requirement of BS 5839, and BS7671.
So it would seem from my experience with the above scenario, that local isolation isn't always necessary, except in certain circumstances?
Despite our better judgement and experience?