HLK-LD2450 ceiling vs wall mount - why is ceiling so much worse?

I have a Athom PS02C3MZ presence sensor that uses the HLK LD2450 mmWave sensor. This sensor has a 6.0m range.

I got this with the Athom ceiling mount - however, I’ve now found out that apparently the LD2450 is recommended to be wall-mounted…not ceiling mounted.

Apparently with the ceiling mount, the range is around 2.0 meters - but if you wall mount it, you get the full 6.0m range. I was curious why this is - possibly I’m missing something very obvious about angles here?

I taped these temporarily to the ceiling, and used the HLKRadarTool app to do some tests. The ceilings are around 2.3m high - and yes, you basically do only get the 2.0m range, for some weird reason. It can’t detect people who are further out - it just loses tracking on them.

But if you put the sensor around chest heigh facing outwards - you do get 4-5 metres of range - and it does seem to nearly get 180 degree coverage as well.

Does anybody know why the ceiling mount position is so much worse, or why the angle of coverage seems less?

I believe the LD2450 is built more for wall mounting as the radar signal is more targeted towards the front of the sensor so it has a much better range for distance. Its not great at the wider angles and above and below. So when mounted on the ceiling, the range of the sensor becomes how wide the 360 coverage of the sensor is (ie. left/right/above/below) hence the poor range.
Think if it like a torch light. It lights up straight in front of you but around you, its fairly dim. Think of the radar as similar to the light from a torch.

Consider the image a radar would see from persons from

  • the side
  • the top
    both are some much different in shape, that the software (firmware) needs to be competely rewritten for those profiles.
    This is also the reason that a top located sensor is very good in ‘fall detection’ as the shape of a dot changes to person shape.
    Top scanning is just hard to make presence detection precise enough

@athua @checking12 Thanks both for your explanations.

@athua I know for wifi APs (and other RF transmitters etc), the datasheet will typically give you a radiation diagram or polar plot, so you can see the signal propagation through free space:

(Source, more info, eevblog post)

Does anybody know if there’s something similar for the LD2450, or other mmWave sensors? (Or I guess, whatever the equivalent to a RF radiation diagram here is).

The closest I can find is on page 14 of the user manual - but I’m not sure what the diagram means?

Are the concentric circles the tracking range? How are up/down represented in the diagram?

@checking12 Yeah - fair enough - I guess if you see it from a profile view - top of head is harder to find vs getting a proper view of the whole person side on =).

It’s a bit annoying, as now I have a bunch of Athom presence sensors with the ceiling bracket - and it seems the ceiling bracket might be useful :expressionless:.

And Apollo also make a deal of having the ceiling mount:

It’s weird though - why would they sell them, if the mmWave sensors aren’t designed to be ceiling mounted at all?

Or is there some tweak to at least make the ceiling positioning slightly better?

Use the ld2450 to detect movement in a really small zone to turn on the lights (let’s say kitchen or living room entry), then use the ld2412 (which has a much better range) for presence to keep the lights on.

You can also tilt the 2450 sensor a little bit so it doesn’t look straight down. It requires a lot of tweaking.

A lot of questions so here’s at least some answers I know.

You’re right, the user manual there doesn’t have a radiation plot at all, only an approximation based on tracking range. It doesn’t explain anything about up and down but data from other parts indicate the functional pitch angle to be +/- 35 degrees. The “Up” and “Down” labels on the left image are necessary to define the positive and negative azimuth values.

Yes, the concentric circles are tracking range in two-meter increments. Up and down (pitch) is not represented and would require figuring out the proximate blind spot. The offset of the range values doesn’t help with seeing dissymmetry, but there is a slightly better range at an azimuth of -30 and -45 degrees.

Visualize it as the sensor stuck to a flat wall crossing from side to side at 90 degrees, with the PCB antenna pads facing toward 0 degrees at the top and the ‘down’ as shown in the left image toeward the ground. It can do its tracking out to 8m for anthing directly in front of it, a bit over 7m for things 15 degrees to either side, just over 6m for things 30 degrees to either side, and about 5m for things 45 degrees to either side. Then only about a meter for things 60 degrees to either side. So while it has a maximum range of 8m, the ‘true tracking area’ is only about 5m out in a 90 degree cone.

The whole ceiling mount thing is honestly a tricky thing. The LD2450 is made to map position, speed, and vector in a 2D plane across the azimuth of the sensor. Think of it as a flat fan of sensing down from the ceiling that reports where up to three distinct objects are located and how fast they are moving in which direction on the plane of that fan. But most humans will walk through that plane, not move up and down and left and right inside it. ‘Shining’ down from above means it can only see what is directly below it in a narrow zone.

So the final two questions… They sell them because fall detection and narrow zoning requirements can make use of the ceiling mounts. For example, hospitals, nursing homes, and retirement homes. Places where knowing the information about something moving up and down is important.

As for tweaks to make the ceiling position slightly better? It’s good for a specific purpose that is probably not the purpose you want it for. To have view of the whole room, you need numerous ceiling units with slightly overlapping views - think of spotlights in recessed ceiling fixtures and you want even coverage of light around shoulder level. It requires a lot of light fixtures.

Long story short(ish):
Ceiling mounting is meant to monitor a small region and focus on how high the moving object is. 45 degrees means that at the ground, the furthest to the sides it can see further than a meter is the same distance as the height of the ceiling. Useful for fall detection and tracking a small area but ignoring things at specific heights, like a pet near the ground.

Scanning an entire room requires wall mounting.