Security

Just thought of another possible advantage. No need for an external power source, although I suppose there is no reason you couldn't use the 12V from the Idranet cable.

My understanding is that typical security PIRs are deliberately desensitised and debounced because you don't want false alarms. Whereas the Idratek PIRs are high sensitivity and then debounced in software/firmware to give rapid response for presence detection.

I thin they can be too sensitive, certainly our small cat has not problem setting them off, so I'll be changing a number to std pet PIRs.
I will take 12V from Idranet.
Another advantage is the look. std PIRs are designed to be unobstrusive and fit well in the corner of a room which is a good place often. Not so the Idrtek modules and their need of a back box.
For windows I am following the same approach of wired in parallel, I cannot think why it wouldn't work. Just got to find time to fit the sensors! As it is winnter the windows aren't being opened much, so it can wait.
JonS

Unfortunately ours are being opened (from the outside and for the wrong reasons).

With regards to PIR's I keep hearing references to pulse counts and pet alleys what exactly is this referring to?

No idea what a pet alley is but AFAIK the PIR sensor itself needs to trigger the number of times set in the "pulse count" for the PIR electronics to open the relay to say "there's a moving hot thing". THis may be a person, animal radiator (temp change) or sun. Most PIRs allow pulse count to be set.
HTH
JonS

My guess is that it has something to do with the detection pattern.

Its mainly a matter of choice. Off the shelf PIRs can be used with and powered by the IDRATEK system and can have advantages such as already mentioned eg. different detection patterns, corner location and so on. Some people find the IDRATEK ones less obtrusive when mounted in certain locations (eg. ceiling or above door frame) also the existence of multi-sensor modules sometimes makes it more cost effective and/or convenient to use an IDRATEK module. But as I say, horses for course. Sometimes its good to have more than one sensor so a mixture can be deployed (and indeed we'd encourage it since it is of paramount important to have a system that works well and suits the customer taste than for us to try and palm off our own product where its not ideal).

Pulse counting tends to mean that the PIR does not trigger until n number of transitions have been detected - typically n = 1,2,3. Pet alley, as I think Gumby is suggesting, will be a detection zone which requires larger bodies to trigger the unit - probably the fresnel lens has wider diffraction points in some areas to achieve this

Regarding sash reeds these can be wired in parallel or series or each one to an individual input. For security applications its typical when wiring multiple switch sensors to one input to do it in series and use normally closed sensors so that if sensor wire is cut it will cause an alarm, so even single sensors might be better being normally closed for same reason.

Karam
IDRATEK LTD

Pet tolerant PIRs work in one of 2 ways. The usual method is to use a special lens which blanks out the lower portion of the detection zone, so pets can move about the house without triggering sensors. Older sensors used a slotted disc which you rotated to achieve the same thing. Some newer sensors use a different method. The sensor is split into zones or curtains. And if a certain number of zones are triggered then it must be a person, and if its less then its a pet. These work very well for cats who like to jump up on to cupboards etc.

Security

the other question is, how many PIRs to have, in a room ... I'm sure Idratek could handle several, and bigger rooms probably need more, but what's the optimum ... can there be too-many .. how about one on every wall, at shoulder-height, two if the wall is over fifteen-feet long, andmaybe three if it's over thirty' ???

Can you have too many? Well my bank manager probably thinks so

I would think maybe two PIRs is sufficient for most rooms if you can't get away with just one. The problem with more maybe "leaking" into locations you don't want triggering for a particuliar room. For example a wall location next to a door would pickup movement the other side of the door if the door were open.

Security

leaking, and maybe Cortex, too - how complicated a situation can it deal with, and how well can it integrate & augment with data from other sensors - light, temp', humidity ... even CCTV ... to make sense of things ???

I'm tempted to make provision for maybe 100s of sensors, of various sorts, to ensure best-chance of catching anythings that's going-on ... such an approach being helped by Idratek's propensity to include anscilliary inputs in its modules (which helps with affordability) ...

The temptation is most definately there! how much is enough? What do you include? An approach like this could uncover many latent applications for the technology especially when the data from lots of sensors are amalgamated.

At present Cortex maintains a sense of presence but not task or person.

If you want this you need to overlay it manually via macros etc. Certainly richer sensing can support this, for example, I'd be tempted to fit reed switches to all kitchen cupboard doors and drawers (if I thought I could get away with it) to support inferences such as cooking vs lingering, for example, which might cause different lighting schemes to be engaged.

At the same time I think that integrating the information from more and more sensors will become harder programmatically ... it would seem to me that a probabilistic solution would be required, but then training becomes an issue ...

Security

>integrating the information from more and more sensors will become harder programmatically ... it would seem to me that a probabilistic solution would be required, but then training becomes an issue ...

the obvious way, perhaps, is a neural-net approach ... allowing actions based on patterns in the data, without having to laboriously identify them oneself ...

'would be good if a simple version could be added to Cortex, in some way ... not sure what hooks this would need / use ...

reed-switches (in series) ... must look to see what's available, and how to ensure reliability is good !