Review on CQRobot Ocean: TSL25911FN Ambient Light Sensor for Raspberry Pi, Arduino, and STM32. Wide 0-88000Lux Detection Range, 600M:1 for Industries, Mobile Phones, Computers, Lighting, GPS, and More. by Sebastian Erik

Most ambient light sensors take care of what the human eye pays attention to. That's fair as it's related to what they're being used for. For a penny you can get a Light Sensing Resistor (LDR) and for a little money you can get an LDR breakout board that includes a voltage divider circuit and whatnot. I tried a few of these and they didn't give me what I needed for my weird use case. This board is relatively expensive (but not by much unless you buy a lot of them), but the TSL2591 sensor has several advantages. First, it offers a digital readout over an I2C interface, meaning the ADC circuitry and all the necessary settings are already included. Second, it includes two different de facto lords. One sensor is more sensitive to visible light and the other is more sensitive to IR. Depending on the application, you can read one or both channels. Third, this particular breakout board uses a JST connection and includes a cable with Dupont connectors. This gives you a very secure and fairly compact connection at the end of the sensor and makes it very easy to connect to the microprocessor board. The sensors' digital output needs to be in "lux" but you should be aware that you may need to go through a calibration process if you really want it to be true "lux". This is not a sensor or board error. It's just a fact of life when it comes to physical things. On the other hand, if you just want to emphasize "more light means more numbers", you don't need a lab calibration. You can easily experiment with the programmable gain settings to get all the interesting levels in the ranges you find appropriate, even if they're not quite "luxury". My application for this is a pretty narrow niche. I have a gas water heater that sometimes turns off the pilot light. I wanted the sensor to be able to notify me when this happens so I can fix it instead of being surprised when hot water is needed. I can position the sensor to face the small window you're looking through when you turn the test light back on. With cheaper sensors, the differences between a full gas flame, a pilot flame, and nothing at all were indistinguishable. I think that's because these cheaper sensors don't pay much attention to the part of the natural gas flame that you actually see. With this sensor experimentally programmed to an appropriate amount of gain and using the sum of both sensor channels, full throttle is 50-60k, pilot light is 150-160 and nothing is 0. This is perfect for my use. case, although I don't know what those units represent in absolute engineering terms. When full throttle shuts off I don't get a reading for a few minutes, but I think that's because the sensor is being affected by the heat coming through the tiny window. After a few minutes, the readings will return to the 150-160 range.