I am planing to make "reflectometer" to test reflectivity of surfaces/mirrors.
I would like it to be highly accurate. Lets say 0% is no reflectivity(dark) and 100% is lightsource not reflecting from any reflective surfaces but directly from source to transistor. I would like 0.1% accuracy.
Do you have any recommendations what device to buy? Is any phototransistor enough, or any photodarlington transistor, or should I take some specific model that someone knows.
Also regarding source, any recommendations there, laser or regular led?
Pretty much any light sensor will do that but you will have to calibrate it yourself because
"lightsource ... directly from source to transistor" is a meaningless statement unless a luminous value is specified.
IOW, is your lightsource a birthday candle or a floodlight?
Well as I am interested in in relative value - percentage of calibrated dark and calibrated full.
As I saw few tutorials their phototransistor work in range of 2-500 which is less than i would like. My light source would be taken based on capabilities of sensor. If sensor can mesure from 0-100 lux, I would take laser, or LED that has maximum of 100lux.
Thank you, I did order that one now, but I think it will not be sensitive enough as it supports maximum of ~65000 lux so very very bright. I will try with it but open for other suggestions as well.
Seriously?
You probably should consult with ESA or NASA and learn what they are doing when imaging our Sun.
Attenuation is a very nasty problem for broadband light; filters are usually spectrum specificL; ND, neutral-density filters are available, but not inexpensive. I seriously doubt that your light sensor is wideband, so the attenuation concerns are even more complex.
I suggest obtaining some photographic reference devices - "grey cards". They are calibrated standards, I believed the norm is 60% reflection. You will need that because you can't ensure linearity with only a white and black reading. Actually you should get a "white card" at the same time. I have a kit with white, black and a series of graduated greys.
The way you used the unit "lux" doesn't seem right. I suggest you research that.
The regular Arduino has a 10-bit DAC (0-1023) so you are right-at the resolution limit if you use the full range. And that's resolution, not accuracy. (The accuracy isn't bad... I think it's usually within 2 counts as long as you have a rock-solid reference.) And that doesn't account for any analog resolution/accuracy/noise issues.
...So you may want a sensor with a higher-resolution digital output.
I would guess that the extremes (a mirror or black surface) will be difficult. And measuring a mirror probably requires a diffuse light source.
And you might need multiple ranges... Like a multimeter has several different voltage-range scales.
Thank you all very much on responses! I believe I will buy TSL2591 mentioned above as I think that is exactly I wanted and as lightsource I will use a laser. Regarding color/ wavelength aluminum coating reflectivity is very similar in visible and IR/UV light.
To test valid calibration, I will use astronomy grade mirror, that has already measured reflectivity in its certificates. Those mesures run around 93 to 98%.
How will you calibrate your project?
No light = 0%
and
Full ligh= 100%
Is okay but the response of your phototransistor is not exactly linear.
A relative 50% signal response will not equal 50% light level to the "accuracy" you require.
Hello TomGeorge Well as with my experience with scales if resolution is 0.1 g accuracy is 0. 5, so I was aiming as something like that in percentage. I see what you mean by not being linear, idealy I would have multiple calibration steps like 0 25 50 75 1000, I will see what I can do about that. as I know I have only 0, 100, 92 and 97%
@Wawa and @build_1971 Yes only one color is more then enough as there is not much discrepancy in aluminized glass.
Well regarding hardware part, other huynis doing that, so I am not sure I will leave aligment to him
Done with my part, now sending it to "hardware" guy to set it all up.
Code could have been written better, more reusable and OOP, maybe using interface for different sensor/display, but for version 0.0.1 it is fine.
It seems it might work, reflection from standard mirror was ~85% and from polished stainless steel disk 52%, but we will know after case is made.
