PIr motion sensors

Yes they are inherently AC coupled.

I would guess that "hot" and "cold" polarity signals will be treated the same.

Yes. They detect the changes, so the can also be seen in the signal.

-- Mikko

Your average porch light sensor uses a segmented Fresnel lens that casts about a dozen images of everything in its field of view, and a split sensor with the two halves wired in inverse parallel and the cut line running vertically.

Any basically vertical thing that moves through the FOV thus produces about a dozen cycles of AC, which is buffered by a JFET follower and detected by a rectifier. (It's a very clever design, really.)

A cold vertical object will do the same thing with a 180 degree phase shift, so it will trigger fine.

Things that are too close, or that occupy too large a field of view, don't work as well.

Cheers

Phil Hobbs

Ah, I didn't realize the sensor was split! I had thought the sensor's output would have a rather slow filter on it and the "live feed" compared against that (counting on the mechanical chopping of a "moving heat source" to generate a detectable "difference")

Or, that move too slowly, apparently (hence the reason I assumed the implementation that I did). Or, along the "wrong" axis.

I had worked on a custom chip RFQ that worked along the lines you mention. An additional channel was to sense a dummy PIR. Seemed like overkill, but as they say, not my job to ask why.