Wozzer is right. The facility of Discovery to deal with the loss of the protocol and still initiate a fire signal is down to a circuitry that monitors that the device is receiving protocol pulses and if it sees none for 180 seconds it can generate it's own series of special pulses, on sensing a fire condition, which panels which have extra circuitry to detect these special pulses can respond to. I understand that apart from a couple of special panels specifically designed for the 'marine' market there are no UK panels that use this facility i.e most UK panelsdon't have the circuitry to deal with the special pulses, so the 'benefit' of it has no real use in the UK at present.
The liklihood of a panel processor failing and still having the normal loop voltage available is very possible. The processor will control the protocol pulses but is unlikely to control the loop power requirements.
Discovery protocol is just an extended version of the XP95 protocol (extra bits of information) and has been around for years, and most control panel manufacturers have models that support it.
Wiz;
In my understanding, there is no need for extra circuitry for the panel side to make a discovery trigger as a conventional detector in case of lost of protocol, the processor can monitor the impedance at the loop in both ways, if it sees drop of impedance to a level say i.e. 470Ohms, it trigger an alarm signal...etc
This can be done by software rather then additional hardware, by incorporating a simple If statement algorithm...
If (presence of protocol) then
Statement 1: Device behave as analogue addressable device
Else
Statement 2: Device behave as conventional device
End If
Statement 1: can take place as long as the processor is receiving analogue values from all devices in the loop.
Statement 2: takes place if not receiving any analogue value.
Now, the major concern is that, would this algorithm works for each device separately or only when the whole protocol is lost, by means a 'Partial loss' ( device no reply ) or 'General loss' ( no reply from all devices ), this is actually down to the programming to setup either way...etc
However do the existing manufactured panels have such algorithm incorporated, the answer is: I don't know! 
This is down to substantial evidence…etc.
Sorry Benz but you are wrong. It never operates as a non-addressable device, in the common sense of the description, therefore special panel circuitry is required to react to it's special pulse facility to warn of operation when the normal protocol system has failed.
You don't have to believe me (you rarely do!) just give the technical department at Apollo a call (ask to speak to The Oracle) and you will find, for some strange reason, that you'll get exactly the same answer that I have given you.
p.s. 'conventional' is now an obsolete term in the fire alarm industry. All systems that aren't addressable should now be called 'non-addressable'
Wiz;
In what I am wrong? And in what I don't believe you? I believe into the technical evidence and explanation from who ever the come from, once I am convinced I will say thank you for that, as usually I do, but please allow a delay time compatible to my slow assimilation timer
As for the philosophy of the appellation of either conventional and its new fashion non-addressable appellation, even the latest guides of Apollo still mentioning conventional rather than non-addressable and many people in the industry still use conventional name... I think things will take time to be standardized into one general single appellation; there is no rush... etc.
In fact conventional system is addressable too, it’s addressed by zones, while the analogue addressable one is more specific (it's addressed by devices within the zones/loops) and so it is addressable too or rather it’s analogue addressable, thus both of them are addressable.
Conventional name was a good appellation in the old days, and used to make sense, before analogue addressable was developed, conventional appellation was compared to the previous standalone smoke alarms where there was no conventions, standards and norms, therefore the conventional was the first one to be designed to certain conventions and therefore the name of conventional system came out...etc
Now conventional name is compared to its successor system rather than the predecessor one, the successor to conventional is more efficient in terms of conventions and therefore they didn’t call it the ‘more conventional’ system, because probably this will not make sense any more, they have called it analogue addressable system, which is more descriptive name, and the new name for conventional which is non-addressable came out by comparing the conventional to its successor one instead the predecessor one, so they preferred to use ‘the new latest name minus one’ which is normally in more descriptive word the ‘less addressable’ but they preferred the name of ‘non-addressable’ which is in my opinion is NOT the right name...etc
Because in fact things come back into a closed vicious circle the conventional system is addressable too and they call it 'non-addressable' just because it's ‘less addressable’ than analogue addessable one... or probably the most descriptive name making the distinction between them, is the name of 'non-analogue' instead of non-addressable, or rather 'non-Digital' as for the analogue addressable is 'Digital' this may make more sense to the appelations...etc but just not 'non-addressable'
Never mind.
Better of finding another name rather than 'non-addressable' or 'conventional'!
Waiting for what the future reveal
