According to the documentation, "An illuminance of 0lux indicates that the sensor is saturated and the configuration should be modified".

However, in total darkness, the Bricklet also reports 0 Lux (I checked).

Q: How do I distinguish between A: sensor overloaded by light (too much light), and B: complete darkness (no light)? In both cases I read 0 Lux.

0 lx always indicates an invalid measurement. Unfortunately, the datasheet of the sensor doesn't document under which conditions this could happen. With testing we found that this happens if you saturate the sensor. For example, by selecting an illuminance range that is way too small or an integration time that is way too long for the current light conditions. Hence the documentation suggesting to adapt the configuration in that case.

But another user already found that this could also happen under very dim light conditions. I could reproduce this by covering the sensor with two sheets of printer paper and a piece of duct tape. I was going to update the documentation about this at the time. But it seems this never happend. If updated the documentation now.

Photron,

Thanks for answering.

I can not think of a way to correct this in software; at unlimited range and 50ms integration time, it is rare but possible to have saturation and a reading of 0 Lux. When this happens and the light is then reduced and stable, the reading appears to decrease over several seconds, e.g. the sensor takes time to recover from overload. Interesting.

One way to solve this might be to have a second sensor with a 10-stop ND filter in front.