Just curious what everyone sees with current sensors.
It seems to me that the sensors do a poor job detecting the difference in illumination in anything less that rather bright light.
In other words, even in a room where I can see perfectly fine--the sensors may show "0 lux". I know the human eye is extremely adaptable over a massive range of light--just wondering how others deal with illumination?
In general, I haven't done much with them (somewhat for that reason) and focus on which lights are on, sunrise/sunset, etc. instead. But, there are some things that could be done better if the sensor was more sensitive in the lower ranges (tbh, I tend to think they should adjust the lux with something like a logarithmic curve--because they could use more sensitivity at lower levels and less at the other end).
I think (or maybe guess is the better word) the problem with devices like hue is that the the light sensor is under the diffuser dome so I suspect that this will alter its response characteristics.
If you look at the datasheets of commonly available light sensors the spectral responses are pretty much always claimed to mimic the human eye (of course ymmv). But this assumes the sensor is directly exposed to ambient light.
In any case the larger issue (for me at least in an indoor setting) is that the update frequency is pretty low to extend battery life. Also, using indoor sensors for sunrise/sunset adjustment isn't the best way to go, as the low lux sensitivity is pretty poor. Better to use an outdoor sensor as the light level range is much greater.
I wasn't thinking that so much--as trying to decide how light a room was so I might temper my automatic lighting accordingly. But, there's not a lot of difference between "normal room light" and "really dim room light". With the Inovelli ones (on USB power), I get 2 lux in an area that's very dim, 26 lux in the room I'm currently in (not bright, but certainly lit), vs 48 with a sensor pointed fairly directly at a couple bulbs that aren't that far apart.
So, for "indoor" type light, it seems the range is about 0-50 lux (and I'm not so sure how consistent it is--since these are rather new).
They'd be so much more useful if the range was about 10x that much.
I am using 4 Xiaomi Zigbee light sensors and 2 Fibaro Ms with light
sensing capabilities. 3 Zigbee sensors are basically external facing
out through the windows. The response is relatively fast so I have
to create a Light Filtering rule in RM. Now all my curtains and lights in
kitchen and living room are controlled by real light levels.
I found that the sensors did not seem to agree on the level with the coolcam recording in the 1000s and the hue and aotec recording differently but in the hundreds at most, all the outdoor sensors seem to agree being in the 1000s and even up to 10000s.
I've never had a sensor read 0 unless it was very dark.
I used hubigraphs to monitor the levels for weeks to get to know each sensor and finally use the outdoor sensors to replace sunset and sunrise. Which works out we as the levels at the back and front of the house vary due to the time of day.
As for the indoor sensors i am starting to use them and have bought a cheap non smart lux meter to compare them.
The out come of the comparison is they are generally accurate and surprisingly it may be 5000lux outside and only 200 indoors with an overcast day being defined as 1045.
With the exception of thee coolcam you levels seem the same as mine.
I decided to go with my weather station's illuminance vs indoor sensors. With a bit of trail and error I have a range low to high of a dead zone to keep the rule machine from flipping my dark outside variable. With this solution it global for the whole house vs. room by room but the values I chose works well for the rooms that are normally occupied that needs light when illuminance drops seems to work well:
I have a Aeotec Multi-Sensor6 (USB powered, facing out through glass) and a Xiaomi Mijia Light Sensor. Both report pretty frequently. My rules are using < 20 lux to turn on lights and > 30 lux to turn off lights for both sensors. This is what the readings were yesterday (Dallas, sunny day).
The "green" is the Aeotec sensor that is in a room facing east and the "blue" is the Xiaomi sensor in a room facing west. There is some "noise" due to cloud cover, but this seems to work for me.
I have 2 windows in the living room and sliding balcony door in a bedroom.
All windows and door are facing West so Sun is gradually gets in.
I am using individual light sensor for each location.
Plus I have to create RM based Light Filter.
Only doing this I was able (finally after trying many simpler solutions)
to control blinds and lights very reliably and got 99% wife approval.
EDIT:
I forgot to mention, there are two internal light sensors in addition
to 3 external.
There are many different ways how to attack this problem.
Recently I moved in the apartment complex and live on 9th floor.
There are many high raised building around.
View from the windows is excellent but simple sun tracking will not work.
As well as passing clouds. Some of them are fast moving but some of
them too slow.
Bottom line:
What could be better than real time illuminance tracking?
This took me a while experimenting with different light sensors,
sensor's locations, and algorithms. But as a result everything finally
works next to ideal (no matter what is done it never will be ideal).
