I have had the same problem, so I started replacing my FP300’s with the Lafaer Presence sensor. They are not as nice looking as the FP300s, but once you get them paired, they work very well, plus they use AA batteries.
Adding to this thread since it’s the more active FP300 discussion, and I want to consolidate what’s known about the underlying cause.
There’s a parallel thread (Firmware problem with Aqara - Presence Multi-Sensor FP300, 155180) where I and others have been tracking the same issue. Short version: after Aqara firmware 0.0.0_6542 (released May 26 2026), the FP300 stopped emitting autonomous Zigbee attribute reports for manuSpecificLumi unless the platform driver explicitly configures reporting. This causes the 24-hour dropout and stuck-state symptoms on Zigbee.
The diagnosis (andreas_roedl, SmartThings driver maintainer, on the Aqara forum thread 299625 post #10): platforms must bind manuSpecificLumi (cluster 0xFCC0, manufacturer code 0x115F) and configure reporting for 0x0142 (presence) and 0x014D (PIR motion). Type Uint8, Min:0/Max:300/RepChange:1.
Status across platforms:
- SmartThings Edge driver: fixed (andreas_roedl)
- Zigbee2MQTT: fix released via PR #12383 (drauziooppenheimer, merged)
- ZHA: pending quirk update
- Homey Aqara app: no update yet
Direct confirmation on the Homey side comes from @Viger_Ostensen’s June 7 post in this thread — he captured the Zigbee interview log for his FP300 showing every attribute has reportingConfiguration: NOT_FOUND, and the bindings section only contains time and ota (no manuSpecificLumi binding). That is exactly the missing configuration andreas_roedl identified.
@TedTolboom — the reference implementations are public. Is porting the same fix to the Homey Aqara app something you can look at? Happy to test a beta if that helps.
My own situation for context: 4 FP300s on Homey Pro 2026, all on firmware 6542. Even after replacing all IKEA TRETAKT plugs with ThirdReality Gen 3s (which addressed Aqara/IKEA routing incompatibility for me), 2 units still drop to a no-detection state every ~24 hours. “Try to Repair” resolves it temporarily by re-running the interview — which is consistent with the missing reporting config theory.
Have you actually used the LWR02? As it seems to become available not before August this year?
I have not used the new one. The one I was talking about working great is the LWR01. I will order the new one when it becomes available later next month. Hopefully it works even better than the first one.
I agree. My FP300’s seem to crap out about one per month so I will get a few too
Curious what your experience will be. Looks like a very nice device based on their product page 
.
If it performs like the LWR01, then it should be good. They are just so ugly. My wife hates them compared to the FP300’s
Nice
Think I’ll be putting my FP300 on sale then and grabbing these next month
