Hi,
for a project I need to know how precise the TCS3200 color sensor is.
I need it to distinguish different shades of blue that can be very similar to each other.
Does anyone know how precise it can get?
Is it only able to distinguish heavily different colors or can it get more precise?
Are there any videos where it is shown checking on similar colors and actually being able to sense the difference?
Try it out and let us know if it works. Your definition of "very similar" is probably different that mine.
Keep in mind that the measured color (spectrum of reflected light) of an object depends extremely strongly on the type of illumination.
1 Like
I was hoping that someone that already has it could tell me more about it since I have to make a uni project around it and wanted to know if it's possible beforehand.
I wish I could attach an image to this post to show exactly how close these shades are.
Thank you anyway!
I have used that sensor, and my GUESS is that it will not useful to distinguish "very similar colors".
But it would be a very good learning experience for you to try it out.
Humans see colored objects very differently than machines do, because of neural processing of the entire scene, in the eye as well as the brain.
1 Like
Our project comprehends a device that needs to distinguish various shades of blue presented as concentric bands 3 cm thick. This device is directly placed on the coloured surface so that the sensor is really close to what it needs to see, making the light conditions very controlled.
The color difference is between rgb: 38, 47, 79 and rgb: 54, 68, 94. I don't know if it helps but maybe you know if that's inside the tollerance.
How can the "light conditions" be controlled, if the light is blocked by the sensor board?
It would be a good idea to get some actual, practical, hands on experience before designing the project.
1 Like
You can get more information about this chip here
Given the nature of the output I suspect you would have great difficulty in getting the same set of numbers from the same shade of reflective card. This is because measuring colour is not a precise thing.
Anyway as a University project you should be in a position to answer your own question when you get one. Because you should already know about precision, repeatable and noise.
You will also probably have to make those measurements in a light tight box, to prevent any ambient light interfering with your measurements.
1 Like
I explained myself poorly. The sensor is placed right on top of the surface meaning that the 4 LEDs are pointing directly at the surface making the thing something like a light box, in which ambient light is blocked but there's the light from the LEDs.
I'm just starting my research and I'm trying to cope with bad organization from the university's side 
Thank you for the imput!
Thank you for the info!
We're trying to get in touch with our professors but we just started thinking of this idea so I came here to get some advice. This project is poorly organized by the professors so we are trying to get help wherever we can.
The device would act exactly like a light tight box. It will be placed on the surface covering any possible external light entrance and the surface will only be illuminated by the 4 LEDs from the sensor itself making the measurement as precise as possible.
Thank you again!
1 Like
It sounds to me like everyone is theorizing how this thing should work, with no practical experience.
This usually leads to a "learning experience".
1 Like
You're very right, I was hoping to find it but it's ok. This gives me some reassuring and some info I can work with.
How does that work then? There must be some sort of gap between the sensor, surface and the LEDs to allow the light to get from the LEDs to the sensor. Don't forget that light travels in both directions, both into and out of a light box.
For your information I was a senior lecturer at a UK university, which is the same grade as Professor in much of the rest of the world. So I know that the more evaluation you do, the better he/she will like it.
Contrary to what virtually all students think the object of a project is not about getting something working, it is to see exactly how the student copes with an investigation. If you show that it can't work and present the reasons why, there is no reason why you should not receive a good pass mark.
1 Like
There is a gap between the surface and the sensor. All the external light is blocked from around it by placing it in a hollow cube with one face missing. The sensor is facing towards the opening and with some distance from the ground.
What I haven't said is that we are not making something they asked us to do but this is something we are making from scratch and with no guidelines around it. From this comes my critic on the course and for many other reasons that don't matter right now.
Thank you for the input
Yes that is that way projects normally work. They are not trying to make you get a kit working and you are free to make your own choices. That is how it should work.
However if you are in France then I am informed that they have a much more restricted form of projects.
1 Like
Seems like a good guess.
Started a project with that sensor a few days ago,
and I am at the point that I can just reliably detect the three primary colours.
Shades of blue? Let us know how it goes.
Leo..
Yes this sensor is capable of measuring different shades of blue. Where is your schematic and pictures of the setup?
I made 2 stroke gasoline analyzer based on tsc3200 which came to the level of 1000$ hatch spectrometer products as per 3rd party testers.
1 Like
In good lighting, one assumes. 
I have a module with four 5mm white LEDs, with the sensor inside some sort of lense mount.
Reliability is only good without ambient light and a constant distance between object and sensor.
Leo..
The project is more likely to succeed with the AS7341 eight-color sensor, as at least 3 of the photodiode/filter elements would respond to bluish colors. It also has a built in white LED for sample illumination. I've experimented with it and am quite impressed.
Then there is the AS7265x, with 18 channels (quite amazing):
SparkFun Triad Spectroscopy Sensor - AS7265x (Qwiic) - SEN-15050 - SparkFun Electronics
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.