I have a presence sensor switch in a very small bathroom.
I noticed that the detection angle range is not really wide.
The coordinate positions I tested that did not trigger motion (or rather eventually lost presence) : 60 cm in front of the switch and 100 cm either on the left or right of the switch.
From my calculations this means that in either of these two positions I’m at a direct linear distance of 115 cm from the switch and at a 30° angle with the switch and the wall the switch is on.
This also means the switch view range would then be less than 120°. That is strange because the switch is supposed to have a 150° angle range.
The switch is set with -600 and 600 for width range. Depth range is set to 0-145. Sensitivity was tested at high and medium.
In my testing of the switches, its horizontal angle is surprisingly wide. Like nearly 150°. But I also have a very small water closet (toilet only room). The room is less than 100cm deep in front of the switch, and I sit on the toilet with my head at about a 10-15° angle to the right of it, maybe 30cm away, and at eye level. It was still losing track of me at times. I’ve been able to get the Report Target Info option working, and I was surprised at where it was putting me sometimes - ~105-110cm out. I suspect it’s getting a radar reflection off the far wall, and so reading a much further distance. Once I adjusted for that, it’s been working much better. Unless you’re savvy enough to compile and deploy your own modified version of Zigbee2MQTT, you’ll just have to wait until the option to do that gets merged and released in order to get that sort of feedback. In the meantime you could try extending your max depth to twice what you would expect, and see what happens.
I believe the “reflection” hypothesis has further evidence because in two other bathrooms that face into a tub/shower combo with a tile wall about 2m out, I sometimes see a Y coordinate of ~2.5-3m.