Sensor to distinguish between daylight and artificial light in a room?

[crutschow]

I think a UV light detector might give you information. Maybe a far IR detector.

I have doubts that either of those can reliably differentiate between all forms of artificial light and natural light through a window.

 

[ronsimpson]

I have doubts

Me to. The sun is very broad band and light bulbs have no need to be. When you need to detect all types of bulbs.... All is a very big word.

Also; just got new windows for the house. They are UV and IR blocking. Very strange to stand in front of a window and not feel the IR coming in.

 

[Pommie]

They are UV and IR blocking.

Do they reflect the IR or absorb it? I'm guessing if they absorb it they will get very warm.

Mike.

 

[OBW0549]

Regarding what I wrote in post #17 above, I got curious and decided to refresh my memory by actually measuring the light output ripple of a few different light bulbs. Light output and ripple were monitored by a PIN photodiode feeding a fast (Tr ≈ 20 ns) transimpedance amplifier, and observed on my o'scope.

Bulb #1 was the 50 watt incandescent floodlight bulb in my workbench lamp; peak-to-peak light output ripple amounted to 11% of the average light output.

Bulb #2 was a 4 watt incandescent "night light" bulb. Ripple was 32% p-p of the average light output.

Bulb #3 was a Phillips 100W-equivalent (14 watt actual consumption) LED bulb. Ripple amounted to about 25% p-p of the average light output, and was primarily at a frequency of 78 kHz. There was very little 60 Hz or 120 Hz content in the ripple, although the 78 kHz ripple showed some 120 Hz amplitude modulation.

Bulb #4 was a 4W-equivalent LED night light bulb. Ripple was 100%, with the light output dropping to zero twice during each AC cycle. As best I could tell from looking at the bulb's innards, incoming AC passed through a capacitor (presumably for current limiting), then into a full-wave bridge rectifier which in turn fed a series stack of 6 warm-white LEDs.

Anyway, FWIW I would encourage the TS to adopt the suggestion Pommie made: to distinguish between incandescent light and sunlight by detecting the presence of AC ripple in the light output. The only caveat would be to watch out for higher-wattage LED bulbs, as they have radically different ripple characteristics.

 

[ronsimpson]

Do they reflect the IR or absorb it? I'm guessing if they absorb it they will get very warm.

The glass is not warm to the touch, inside and outside. (reflect)

 

[DrG]

I also thought UV would be a key characteristic (assuming you don't have too many mercury-vapor lamps in your living room). If I am to believe these kinds of spectral response graphs, I still think it might but I would like to see them further into UVA (and even UVB, in case you are erasing EEPROMS [edit EPROMS]- j/k). Of course, when there are multiple light sources, it gets complicated. But, if the furthest UV portion (shortest wavelength) is only, or mostly, present in sunlight, you might be able to predict the intensity of sunlight and, thus, the contribution of sunlight to total illumination.

(from **broken link removed** and a lot of other places)

**broken link removed**

(from **broken link removed**)

This guy's patent attempts to do what is being discussed. The rationale for the "why" part is for energy conservation.

 

[dknguyen]

I also thought UV would be a key characteristic (assuming you don't have too many mercury-vapor lamps in your living room). If I am to believe these kinds of spectral response graphs, I still think it might but I would like to see them further into UVA (and even UVB, in case you are erasing EEPROMS - j/k). Of course, when there are multiple light sources, it gets complicated. But, if the furthest UV portion (shortest wavelength) is only, or mostly, present in sunlight, you might be able to predict the intensity of sunlight and, thus, the contribution of sunlight to total illumination.

(from **broken link removed** and a lot of other places)

**broken link removed**

(from **broken link removed**)

This guy's patent attempts to do what is being discussed. The rationale for the "why" part is for energy conservation.

The problem with UV is a lot of UV is blocked by windows and the fact that most UV photodiodes/sensors easily cost an upwards of $80 per component, especially for the low levels of UV you are trying to detect in these cases (sunlight outside counts as low level). There are a very couple of cheap ones, but not sure how well they work.

 

[DrG]

The problem with UV is a lot of UV is blocked by windows and the fact that most UV photodiodes/sensors easily cost an upwards of $80 per component, especially for the low levels of UV you are trying to detect in these cases (sunlight outside counts as low level). There are a very couple of cheap ones, but not sure how well they work.

Yeah, that is an issue (UV blockage from windows). I did not think about that...

 

[unclejed613]

I don't think frequency detection will work for incandescent bulbs.

actually they do have a slight flicker that's easy to detect. my son has a slow motion camera app on his phone, and he wanted to know why an incandescent lamp was repetitively dimming on a video he shot.

 

[JonSea]

Deregarding 7 posts telling me I was wrong about incandescent bulbs, has anyone noticed that the OP is absent from this discussion?

 

[gophert]

Deregarding 7 posts telling me I was wrong about incandescent bulbs, has anyone noticed that the OP is absent from this discussion?

How long have you been a member here? If you don't understand how this site works, here goes...
OPs leave over 50% of posts after the thread is created and the rest of us knit pick each other's answers for the next 30 to 300 posts.

 

[unclejed613]

Deregarding 7 posts telling me I was wrong about incandescent bulbs, has anyone noticed that the OP is absent from this discussion?

you aren't completely wrong, they don't go entirely dark between AC peaks, they just dim a little.

looks like OP joined just to ask the question, and then vanished...

 

[ronsimpson]

the rest of us knit pick each other's answers for the next 30 to 300 posts.

No way. Never happens. Again you are wrong.
It 29 to 303 posts.
Considered your self knit picked.

 

[DrG]

No way. Never happens. Again you are wrong.
It 29 to 303 posts.
Considered your self knit picked.

Do you mean nitpicked?