The sensor does not know what a pet is
A passive infrared sensor does not detect animals or people. It detects change. Specifically, it watches the infrared pattern arriving from its field of view, through a faceted lens that slices the scene into zones, and it fires when the pattern across those zones changes in a way that matches moving warm bodies. A human, a dog, and a radiator with a curtain blowing across it all produce pattern change. What the sensor's logic does with that change is where every promise about pets lives and dies.
Pet immunity, when implemented properly, is a combination of geometry and signal processing. The geometry part is installed height and downward angle, which together determine how high above the floor a warm object must be before it fills enough of the sensor's zones to look like a person. The signal processing part is a set of thresholds on the size, shape, and duration of the moving signature. A small warm blob that never rises above a certain apparent height is classified as a floor-level animal and ignored. A taller or faster signature is classified as a person and reported.
The weight number on the box, the common claims of immunity up to 20 or 25 or 40 kilograms, is a translation of that geometry and those thresholds into a number a shopper can compare. It is a guess about how big a floor-level animal looks from the stated mounting height. Which is why the number is only valid under the conditions it was measured in, and why the same sensor mounted higher or lower has effectively been given a different pet-immunity rating that the box knows nothing about.
Where the guess breaks
The first failure mode is furniture. Pet immunity classifies by apparent height. A cat on the floor is a small signature. The same cat on the back of a sofa, on a dining table, on a bed, is a small signature at human height, and the sensor has no way to know there is furniture involved. The blob is now up where people are, moving like a person, and the alarm logic does exactly what it was designed to do. Homes with a sensor covering a room where pets have elevated perches will produce alarms the owner finds inexplicable, because everyone involved, the owner, the pet, and the sensor, is behaving correctly within their own understanding of the situation.
The second failure mode is size. A large dog is not far from a small person in apparent signature. A sensor tuned to ignore a 15-kilogram dog has very little margin before it starts ignoring a person walking at the far edge of the room, and a sensor tuned to catch that person will start seeing the dog. The weight rating is not a wall; it is a balancing point, and big dogs sit close to it from both directions.
The third failure mode is stairs. A cat or dog climbing a staircase within the sensor's view presents a continuously rising signature, passing through the entire height range as it climbs. There is no mounting height that makes stairs safe for pet immunity. Stairs also happen to be where animals spend energetic time, which makes them a natural place for false alarms that the owner cannot reproduce on demand.
The fourth is proximity. A pet that approaches the sensor itself gets physically closer, and apparent size grows with proximity. A small cat at one metre from the sensor can present a larger signature than an adult at the far wall. The zone directly under a ceiling-mounted sensor, roughly a metre and a half of blind-and-blooms directly beneath it, is a classic false alarm source in homes where the animal's route passes the mounting point.
What installation can actually fix
Mounting height is the master variable, and the manufacturer's stated height is not a suggestion; it is the calibration condition for the pet immunity rating. A sensor specified at 2.2 to 2.4 metres and installed at 1.8 metres has had its immunity rating effectively reduced, because the downward angle now sees floor-level animals as taller signatures. An installer who drops the sensor to keep it out of a doorway has quietly converted a pet-immune sensor into an always-alarm sensor.
Position selection beats sensitivity adjustment. The productive rule is to place the sensor so that the protected zone contains floor level only, with no sightlines to furniture tops, counters, beds, or staircases. That usually means aiming the sensor across the room at the entry line rather than down into the room's interior, and it means accepting that a PIR in a room full of cat highway is the wrong tool regardless of settings.
Sensitivity settings exist on most units and are worth understanding before touching. Turning sensitivity down reduces pet alarms and also reduces detection of people at range, which is the entire purpose of the device. The trade is direct and inescapable. A sensitivity setting that reliably ignores a dog on a sofa is a setting that will eventually ignore a person walking softly on a far carpet.
Some commercial-grade units offer zone masking, physical or logical shutters that blind the lower portion of the field of view. This is a genuine solution for the furniture problem, because it enforces the height classification optically rather than algorithmically. It is worth asking about specifically rather than hoping it is included.
The layered approach that actually works
A single sensor asked to distinguish an intruder from a resident animal in a furnished home is being asked a hard question with one small piece of evidence. Systems that work reliably use more pieces.
Door and window contact sensors are completely immune to pets. A magnet opening is a magnet opening. In homes with animals, the perimeter, the doors and windows, carries the primary intrusion detection, and the interior motion sensor is either omitted or configured as a second layer that is only armed when the layout permits it. This is not a compromise; it is the standard professional pattern, and it removes the false alarm problem entirely from the perimeter line.
For the interior layer, placement discipline, the correct mounting height, sightlines away from furniture and stairs, does most of the remaining work, and zone masking handles the rest where the hardware supports it.
The verification step that most installations skip is the soak test. Leave the system armed in pet-tolerant mode for two or three days of normal animal life and read the event log. Then walk the room at the far end like a person with somewhere to be, and confirm the sensor catches it. A configuration that passes both tests, quiet through animal chaos, loud for a human crossing the room, is done. A configuration that passes one is not.
Reading the specification honestly
When comparing sensors for a pet household, four specifications carry real information.
The weight threshold should always be paired with the mounting height it was validated at, because a number without its height is a marketing figure, not an engineering one. The field of view pattern, the number and arrangement of detection zones, describes how finely the sensor can classify signatures, and more zones generally means better discrimination. The availability of optical or logical masking describes whether the furniture problem has a hardware answer. And the alarm logic, whether the unit requires one zone break or several simultaneous breaks to fire, determines how well it can tell a rising signature on stairs from a walking human.
None of these appear reliably on consumer packaging, and all of them appear in the engineering documentation that a supplier can provide on request. That documentation is also the honest signal of whether the manufacturer engineered the pet behaviour or described it.
On the manufacturing side, we build motion and contact sensor programmes together, which reflects how they are actually deployed, as complementary layers rather than substitutes, and the pet behaviour documentation is part of the specification rather than a footnote. The sensors and alarms range covers the standard builds, and the door and window sensor manufacturing programme is documented on the sensor manufacturing page. For volume and OEM enquiries, the manufacturer site and the contact page are the direct route.
Top comments (0)