I have the above sensor at the bottom of an ultra-matt-black shroud, using the PCB's built-in LED for non-ambient constant sample illumination.
When I put a material sample on top of the shroud, no ambient light can ingress, and the program identifies the sample by comparing the values read to those stored in an array. That works surprisingly well with this low-cost device.
However:
After powering on, the LED is on all the time, which is not really desirable, because the LED heats up over time, thus changing the LED's colour temperature, thus changing the values read, thus eventually no longer making it possible for the program to identify the sample.
Also, a linear actuator moves depending on which sample was identified, and when there is no sample on top of the shroud, the sensor reads whatever illumination is present in the lab environment, thus sometimes falsely identifying a sample; particularly very pale pastel colours, and the linear actuator moves instead of not moving at all.
The question is, what could I do to 1. have the LED come on only while a sample is put on top of the shroud and 2. prevent linear actuator movements when there is no sample on top of the shroud.
Is the only solution for 1. to use a momentary switch and for 2. to put a proximity sensor next to the TCS34725?
Maybe we could give better answers if we could see a picture or 2 of your setup and the code that you have to run it. Does the shroud move up to place the sample and down to block light and read the sample? Could you put something to read the position of the shroud?
As you can hopefully see, it's an extremely ultra-matt-black shroud (tricky to photograph because it reflects hardly any light) that later will be sunk into the lab bench's surface. Like I wrote, the material samples are put on top of the shroud so that no ambient light can ingress (the sample swatches are always bigger than the shroud's opening and the samples are not translucent).
My hunch is that maybe there's simply just no programmatic way around integrating a proximity sensor onto the PCB that could then make a boolean change from false to true once the opening is blocked by a sample, thus 1. triggering the switching on/off of the LED and 2. disabling the colour sensor when no sample is present at all.
As it is now, when I put a swatch onto the shroud, the sample is lit by the LED (always on), the colour is sensed, the values compared to the ones stored, and a linear actuator moves. That works well, even reliably differentiating very close greys like RAL 7035 and 7047. But when I remove the sample, the colour sensor senses whatever is above the shroud - for example the lab ceiling, and the lighting of the room changes during the day (windows, task luminaires on or off, etc.). At the moment, I switch the whole shebang on and off with a momentary switch, but that's a clumsy way to do it.
A phototransistor next to the shroud that would derect the presents of a sample? The sample would block ambient light from falling on the phototransistor when a sample is in place.
