It sees infrared
The sensor is keenest at 850 nm, in the near infrared you cannot see, and your eye is keenest in the green. So two lights that look equally bright can read very differently: a filament bulb or daylight reads several times higher than a white LED, and a TV remote you cannot see at all makes the reading jump.
Two different ideas of bright
Your eye is most sensitive to yellow-green light, around 555 nm, and hardly sees deep red at all. Much past 700 nm it sees almost nothing: that is infrared. The sensor on this board is most sensitive at 850 nm, already in the infrared, and its datasheet gives a range of 400 to 1100 nm. It is a silicon part, and silicon sees further into the red than you do.
The dashed curve is your eye and the green one is the sensor. The shaded shape is the light itself. A white LED is almost all visible light, so its shape sits under your eye's curve and hardly reaches the sensor's peak. A filament bulb is the opposite: most of what it gives out is infrared.
The bars underneath are the point. For light that looks equally bright to you, each bar is how much the sensor reads, with the white LED counted as 1. They are worked out from textbook shapes, not measured, so read them as rough sizes: a fluorescent tube somewhat higher, daylight several times, a filament or halogen bulb something like ten times.
What that does to a reading
A number is only comparable with another number taken under the same kind of light. Change the bulb and every threshold moves. Bring the block to the window and daylight reads much brighter than the room looked. The datasheet quotes its currents in lux, but lux is defined by the eye's curve, and the datasheet does not say what light it measured under. The lux figures in this book are the datasheet's, and they hold for that light.
That is why this book calibrates rather than converting to lux. A night light reads the room it is in, under the lamps it has, and sets its own lines.
A remote you cannot see
A TV remote flashes an infrared LED, usually at around 940 nm. You see nothing. The sensor sees a good part of it, so pointing a remote at the block while pressing a button makes the reading jump in bursts. It is a quick test that the sensor is alive in a dark room. It is also a reason a night light near a television can flick on and off.
Seeing a remote is not the same as reading one. The IR receiver block is built to decode what the remote is saying; this block only sees that it is shining.
When it does not work
Different lamps give out different amounts of infrared, and the sensor counts all of it. A filament or halogen bulb reads far higher than an LED that looks the same. Calibrate again under the new light; the night light sketch does it every time it starts.
A remote flashes an infrared LED, and this sensor sees infrared well. Point the block away from where remotes are used, or average over a longer window so short bursts matter less. Infrared lamps on some security cameras can do the same, all night.
Not on its own. A lux meter is filtered to follow the eye's response; this sensor follows its own, which runs well into the infrared. Its datasheet's lux figures hold for its own test light, which it does not name. Under one kind of lamp you can calibrate it against a meter, and that calibration holds for that lamp.
An infrared-blocking filter would bring its response closer to the eye's, which is what many purpose-made ambient light sensor chips build in. This block has none. For deciding whether it is dark yet, it does not need one.
Why the last digits wobble under some lamps, and the 50 ms average that settles them.
The flicker under room lights →Edit this page — content/books/ambient-light-sensor/it-sees-infrared.mdx
Questions about this product
See what other owners have asked, and read their solutions.
Ambient Light Sensor
Loading discussions…
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.