1
\$\begingroup\$

I've made a really primitive spectrophotometer using an RGB LED and a transimpedance amplifier, and I'd like to fine tune it a bit. Most importantly, I was wondering if it was possible to map RGB values to a specific wavelength so that I could graph output voltage values against wavelength values. I found this online tool: Link here, but I'm not sure if it's accurate for an LED. Also, I found that most colors appear uniform with the LED, but mixing equal red and blue is always distinctly red and blue; i.e., it doesn't appear purple. Any suggestions/input?

\$\endgroup\$
  • \$\begingroup\$ What do you mean by "appear...distinctly red and blue"? Are you talking about how your eye perceives the color, or what you measure with the spectrophotometer? \$\endgroup\$ – The Photon Jul 5 '17 at 1:33
  • \$\begingroup\$ @ThePhoton eye perception; all the other colors appear as they're mapped, but when you look at the LED when it's supposed to be emitting "purple" it looks like two LEDs \$\endgroup\$ – Alex Liu Jul 5 '17 at 1:40
  • \$\begingroup\$ You probably need a decent diffuser in front of the LED to 'mix' the colors better. \$\endgroup\$ – brhans Jul 5 '17 at 1:44
  • \$\begingroup\$ What is your goal? Are you making a spectrometer? Or something different? (I can't tell for sure, but from what little I can form from your writing the online tool you linked has no value to you.) \$\endgroup\$ – jonk Jul 5 '17 at 4:24
  • \$\begingroup\$ @jonk yeah I've already built a spectrophotometer which uses an rgb LED, but right now it's purely qualitative; I want to know if there's a way to determine/control the specific wavelength emitted by the LED \$\endgroup\$ – Alex Liu Jul 5 '17 at 4:34
2
\$\begingroup\$

enter image description here

Figure 1. Apparent colour mixing when adjusting intensities of RGB LEDs.

We tend to think that by mixing RGB intensities that we can generate any colour or wavelength. This is, in fact, not true. What happens is that due to the way the eye is designed that we have the perception of the new wavelength.

The human eye can distinguish three broad wavelength bands - short, medium and long - with the three different cone types. These sensors correspond roughly to red green and blue.

enter image description here

Figure 2.

The perception of colour is due to the relative signal strength detected by each cone.

Most importantly, I was wondering if it was possible to map RGB values to a specific wavelength so that I could graph output voltage values against wavelength values.

No. The wavelengths emitted by your RGB LED won't change - only their relative intensities. (They might shift slightly due to temperature at higher brightnesses but not as much as your question suggests.)

\$\endgroup\$
  • \$\begingroup\$ In that case, what kind of white LED should I be looking for? Or are high CRI incandescents more well suited for this application? \$\endgroup\$ – Alex Liu Jul 6 '17 at 3:07
  • \$\begingroup\$ White LEDs are either RGB as you've got or are blue with phosphors which fluoresce in a broad spectrum. You can't control their wavelength - just the amplitude. Incandescent bulbs are broad spectrum. "... what kind of white LED should I be looking for?" What are you trying to do? \$\endgroup\$ – Transistor Jul 6 '17 at 6:48
  • \$\begingroup\$ I'm building a spectrometer and I need a light source for a monochromator. \$\endgroup\$ – Alex Liu Jul 6 '17 at 13:57
  • \$\begingroup\$ Well LED's aren't suitable for that. They'll give you peaks at three points in the spectrum and nothing in between. I can't suggest an alternative for a tuneable light frequency other than a broad spectrum light source such as an incandescent lamp and a prism to allow you to select a region of interest. \$\endgroup\$ – Transistor Jul 6 '17 at 16:25
0
\$\begingroup\$

An RGB LED produces only three frequencies, (in different mixtures) so with an RGB LED as source you can measure at only those three (rather broad) frequency peaks.

So you have a spectrophotometer that produces only three data points per assay, I think "rather primitive" is an understatement,

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.