Color Sensor using a Single photodiode and a white LED

Status
Not open for further replies.
I

israel_Y

Member level 1
Joined
Feb 8, 2010
Messages
34
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Location
The Netherlands
Visit site
Activity points
1,534
Please, I need opinion on my project as follows:

I am claiming a color sensor using a single photodiode (OPT301) and super bright white LED by using a simple interval mapping of the photovoltage into a color based on the datasheet response curve. (please see the attached "opt301 photodiode responsivity"). Without using a microcontroller/any sort of programming.

Many color sensors i have researched so far, however, use color filters, or RGB light sources to shine the photodiode sequentially and use microcontroller to determine the color.

Please let me know if there is anything wrong with my claim. Thank you
 

Attachments

  • OPT301 spectral responivity.bmp
    1 MB · Views: 127

A white LED is actually a blue LED with a yellowish phosphor so it is not really white. It produces very little amounts of red colors.
Its color response must be added to the color response of the photodiode.
 

Audioguru said:
A white LED is actually a blue LED with a yellowish phosphor so it is not really white. It produces very little amounts of red colors.
Its color response must be added to the color response of the photodiode.
Click to expand...

Thank you for your reply,

So if I am able to control the light source somehow, i can use a single photodiode as color sensor?
 

What do you mean with interval mapping? I don't see at all how a single scalar measurement value can represent colour information. I don't know what kind of objects you consider, but to ask a simple thing, how do you distinguish reflectivity from colour with a single signal?
 

@ FvM, by interval mapping, i mean, red color is o.47V (with some tolerance interval could be 0.45-0.5 V) .. i make same interval for the other colors aswell.. up to 4-8 intervals. i m thinking to just measure the photovoltage output and say it is 'this' color based on the intervals.
 

Unless the photodiode is responsive to a single photon, and you can sample quickly enough to capture that event, then that won't work. It is not, and you cannot.

If you are only ever going to show the photodiode single, pure wavelengths, then it might be worth considering. Like by using a filter , for example...

Also think: what about the intensity of the light? That will effect the photodiode output level too. How will you differentiate output level due to intensity, from output level due to wavelength?

There is a reason for the way things are done already.
 

@ FvM
Lets say for reflectance application, one color (smooth surface color) will be reflected off and i ll be measuring that with the photodiode.. which means i m shining one colored light at a time. About intensity, i thinking that can be controlled by controlling the current of the light source.

I really appreciate your help
 

israel_Y said:
About intensity, i thinking that can be controlled by controlling the current of the light source.
Click to expand...

Yes it can, but how will you know how to modulate the intensity of the light source in order to remove that as a variable in the output, unless you already know what colour the light is being reflected off?
 

FoxyRick said:
Yes it can, but how will you know how to modulate the intensity of the light source in order to remove that as a variable in the output, unless you already know what colour the light is being reflected off?
Click to expand...

I didnt think of that. Cant i keep constant intensity?
 

Even if you know the precise spectral output of your light source, imagine this loose example:

Wavelength A gives a reflected power of 3mW per unit area, and the photodiode is sensitive to this wavelength at 2V per mW.
Wavelength B gives a reflected power of 2mW per unit area, and the photodiode is sensitive to this wavelength at 3V per mW.

The output is 6V. Which wavelength has caused the output? There is no way to know, you cannot work backwards from a single scalar value to multiple input variables. And make no mistake: you have input variables that you cannot eliminate from this system.

In a laboratory setup, with carefully measured equipment and specific, pre-determined colour samples, you might be able to differentiate between them in this manner. But in the real world, as soon as a new variation is encountered, the system will fail.
 

I may hv to go with pre-determined colored samples.. Thank you very much for your input FoxyRick
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…