0

My concern/question: If one has a string of series RGB LEDs (w/ independent elements) and was to light green to 100% and red to approximately 75% -- in an effort to illuminate the Amber color -- how consistent will the color be from one LED to the next. More specifically, is the human eye capable of perceiving any discrepancies? Clearly this is heavily dependent on the type of LED chosen, so please allow me to clarify further.

What I’m looking to do is drive a sting of RGB LEDs (independent elements), where each string (Red / Green / Blue) is controlled via a switch-mode controller. In essence, each string is controlled via a constant current source. Thus, there will be three controllers: one for the red string; one for green string; and one for blue string.

Schematic Snippet

Luminous intensity (i.e. the mcd output of the LED) is fairly linear with respect to current (I realize it's not perfect). Therefore, for all intents and purposes, we’ll assume if we cut current in half, we cut the luminous intensity of the associated string in half.

LEDs are far from perfect. This is to say that, when comparing one LED to the next, luminous intensity across the elements (for a given current) will be different. Particularly, the LEDs I’m looking at (ASMB-TTB0-0A3A2) employ the following luminosity spectrum (@20mA).

Luminous Intensity of RGB Elements for LED ASMB-TTB0-0A3A2

To make this a bit simpler, let's assume we have only two LEDs. Let's also assume that we have determined the normalized amount of current that needs to be pumped through the Red and Green elements in order to emit a perfect amber color. The main question is, when considering the worst-case luminous variance between the two LEDs under test, can the difference in color be perceived?

I have the LEDs of interest on order, and the necessary components in the lab (to build constant current sources) to allow me to do some testing and perhaps collect some empirical data. Going back to our assumption that current linearly maps to the mcd output, one could vary current through elements to effectively mimic the luminous variance effect, no? The issue with this, however, is that I do not know where I'm starting at on the luminous spectrum. For example, let's assume I increase RED to 9.01 mA while decreasing GRN to 0.88mA to explore one of the worst-case scenarios. If the LED was already at the upper end of the RED spectrum, and here I am further increasing RED's current (mcd output), I've misrepresented the worst-case, if that makes sense.

Proposed empirical test in lab

My questions:

  1. Am I overly concerned here?
  2. Does anyone have experience with a similar paradigm and could share some tribal knowledge?
  3. How reasonable/unreasonable is my proposed lab experiment?

Sorry for rambling, as I'm new to this site (or posting to the site, anyway). I thoroughly searched for my answer, and although I found answers related to matching intensity across elements, I did not discover topics related to color consistency. I'm very appreciative of your time/effort in helping me answer this question.

cguenther0929
  • 1
  • 1
    LED binning is a thing. So this is probably going to depend on what level of consistency you pay for. – Chris Stratton Aug 04 '20 at 13:08
  • Thank you, Chris. Good point. I failed to list that as a question, as I've contemplated that, too. "Should one be concerned to the point where we pay a premium for binned LEDs?" Perhaps the answer is yes! – cguenther0929 Aug 04 '20 at 13:27
  • Probably you are over concerned, the colour differences for a full batch will be negligible. Will the LEDs be in a fixed strip, or will there be a way of replacing the ones that lights in a different hue? But bear in mind: 100% red LED + 75% green LED **won't NEVER NEVER produce a "perfect amber color"**. This is a physical reality. Red + green only **deceive** your eyes into the perception that they are seeing amber light, but they aren't. If you want perfect amber light, use amber LEDs. – mguima Aug 04 '20 at 13:34
  • Thank you, mguima. Said LEDs would be in a strip, and individual elements can be replaced if needed. Basically, we’d be performing an internal binning, if you will; which was discussed yesterday. Yes, good point, I shouldn’t have said “perfect amber color”. Thanks. – cguenther0929 Aug 04 '20 at 13:48
  • @cguenther0929 What are you trying to achieve? Do you want every RGB LED in the string to look as though they are both the exact same color as well as brightness to a human? Or are you doing something else? You use the word "luminous" which implies a human instead of "radiance" which is a physical measure independent of humans. So I'm uncertain of your goals here. Also, contrary to what mguima wrote, the color differences between LEDs from the ***exact same*** wafer will look different to a human when driven with the exact same currents. So I've no idea what planet that comment comes from. – jonk Aug 04 '20 at 17:57
  • Jonk, yes, I’d like brightness and color of all LEDs in the string to appear the same to the human. Color takes priority over brightness. I realize there will be differences, but are said differences inconspicuous enough? I’m essentially making a bar of light (12 LEDs). All LEDs are close to each other in a single file line. Ostensibly, if two LEDs operating next to one another were experiencing worst-case extremes, it would definitely be noticeable, but I’m not sure how likely such an edge case is. The row of LEDs sits about 0.5” behind a clear overlay (long/narrow) window. – cguenther0929 Aug 04 '20 at 19:21
  • @cguenther0929 I spent years with Siemens/OSRAM (before Siemens sold OSRAM) doing RGB LED calibration work. You can see one of many modules I worked on in [this answer](https://electronics.stackexchange.com/a/468743/38098), for example. The spacing is close on these modules and it was very important that they look similar to each other in the displays. Separately, I worked on other products designed to "bin" LEDs according to brightness and color (both were required, so this was a 2D binning matrix.) So I know a few things. What you are suggesting wouldn't fly. But the needs may be different. – jonk Aug 05 '20 at 00:21
  • @jonk -- Thank you for your input. It’s evident you have the industry experience and carry the accolades to back your advice. Yesterday, I had our PCB layout guy evaluate board space/etc. if we were to replace RGB LEDs with discrete elements. This was a mitigation strategy, but I’m now convinced that we need to move away from RGB LEDs. – cguenther0929 Aug 05 '20 at 12:56

0 Answers0