Is it possible for the sensor to take into account a sudden increase in value, e.g. PM2.5? The goal is to detect a person smoking cigarettes on the terrace next door. The PM2.5 background is not constant and the window closing scene closes it constantly. Or maybe just set a higher threshold?
The Sonoff Air Guard SAWF‑07P will detect cigarette smoke, but only through its TVOC sensor, since it doesn’t have a PM2.5 optical module. Cigarette smoke contains a huge amount of volatile organic compounds, so TVOC spikes within 1–3 seconds, and eCO₂ rises shortly after because it’s calculated from VOC levels. Outdoors (e.g., on a terrace) it works the same way, but the reaction depends on wind direction and the smoke disperses much faster. The device won’t show PM2.5, PM10 or CO, but a sharp TVOC jump is a reliable indicator that someone is smoking nearby.
Use a camera with AI 
PM2.5 maybe calculated based on TVOC.
PM2.5 and PM10 values are shown directly. I didn’t know they were calculated. TVOC is not shown and cannot be included in the scene. If I set the PM2.5 scene trigger threshold to 52, it detects a smoker on the terrace. But it happens that the background exceeds this value. Firmware modification regarding PM jump would be great if possible.
When I wrote that PM2.5 may be calculated based on TVOC, I only meant that some sensors on the market estimate PM2.5 from other parameters because they don’t have a real optical laser module. I didn’t mean that you ought to calculate it yourself
A battery‑powered sensor definitely does not have a laser or optical PM measurement system, because those modules require active airflow and draw far too much power to operate on batteries.
As for the “PM jump”: that’s really something that should be handled by automation rather than firmware. The firmware’s job is simply to report the raw readings as the sensor measures them. Logic such as “detect a sudden PM2.5 spike” or “ignore background levels above X” should be implemented on the platform where the user creates their rules or scenes, because everyone has different environments and different thresholds.
SAWF-07P has constant power supply. Are we talking about the same sensor?
Yes, we’re talking about the same thing. The mention of battery‑powered devices was meant in a general sense, not as a specific reference to anything in particular.
Sonoff, in its official documentation, does not explicitly mention a “laser optical sensor” or “optical PM sensor” in the user manual or on the product page. However, the displayed values – PM measurements in µg/m³ – suggest the use of a standard laser optical sensor, as only such a sensor can measure particle concentrations in real time at a home level.
In other words, the documentation refers to the PM measurement function but not to the sensor technology.
I also noticed that the SAWF-07P does not trigger a scene when setting more than one internal criterion. Only one choice, e.g. PM, makes the scene work. If I also add temperature, the scene will not run. Isn’t that a mistake?
Possible, but it’s related to the scene logic in the eWeLink app, not to the device itself.
Now I can select the criteria and save the scene that opens the window. But it doesn’t work. It’s strange that I manage to save such a scene to perform. There are two separate scenes operating.
This is an issue that should be reported to the eWeLink dev team.