Diode Green Lasers (Part 1, Wavelength and Efficiency)

Today’s blog is the first in a series about green lasers, and in particular about how wavelength, power, efficiency, and lumens relate to each other.  Also, I’m also going to write a little about the difference between and R&D announcement and what needs to be known to build a projector.

I believe that direct green lasers are the key to making very small embeddable pico projectors regardless of the display technology be it LCOS, DLP, or Laser Beam Scanning (LBS).   Unfortunately the “physics” of green lasers makes them hard to produce.

The laser makers know that somewhere in the range of 532 nanometers (nm) is a very good wavelength for  “green” in a display.   Some of the key reasons being lumens (brightness), efficiency, and color space (which will be discussed in the next part of this series).

The problem laser makers have is that going from say 510nm to the longer wavelength of 532nm is difficult and they often end up trading off the stability/lifetime/yield=cost; and/or the efficiency; and/or the power output of the laser to get a better wavelength.   Thus when I see a company announce a “breakthrough” with a better wavelength, I’m curious what they had to give up to get it.  So as the laser makers improve the wavelength, they often have to then go back and “fix” things that they “broke” to get the longer wavelength.

To make a marketable projector there are a number of important spec’s we need to know, some of the key ones being :

  1. Wavelength
  2. Efficiency (electrical power to light power)
  3. Power of the light out
  4. Lifetime (experimental lasers often degrade quickly)
  5. When it will be available and at what cost at what volume

Often the R&D announcement usually talks about 1, 2, or at most 3 of the above specs and almost never #4 and #5 above.     But it is kind of hard to build a product when you don’t know what it will cost and when it will be available.

Sometimes the prices is give in the form of a “riddle wrapped in an enigma” that almost sounds like and answer to #5, such as “the million piece price will be $X.”  But notice it didn’t say how many dollars you will spend on the first 10 and 100 thousand on your way to buying 1 million units or in what year you will be able to buy those million units.

Now on to discuss some of the technical parameters:

Light is a form of energy/power measured in Watts and Lumens or the “photopic response” is how bright a human perceives light.  It turns out that humans have different sensitivities to different wavelengths/colors of light.   At the extreme ends of the light spectrum for example, infrared and ultra violet are invisible to humans and thus produce no “lumens.”

The graph at the top shows how a “Watt” of light of a given wavelength is perceived by human in terms of lumens.   As the graph shows, a 532nm green has 603 lumens/Watt, whereas, 510nm green has only 344 lumens/Watt or about 57% the lumens/Watt of the 532 green.  In between, 525nm green with 542 lumens per Watt or about 89% that of 532nm green.

Something else to notice is that a typical blue laser is around 455nm and only produces about 33 lumens per Watt.  You need the blue to mix with the green and red to get white and the color in between, and it takes power yet produces few lumens.

A typical red laser is in the 640nm range and has 120 lm/W.  This is a deeper “red” than necessary for typical display application and is inefficient, but just like it is difficult to make longer wavelength greens, the physics of the red laser is just that it is difficult to build stable shorter wavelength red such as 625nm.

Wall Plug Efficiency (WPE) – The WPE is the simply the light output in Watts divided by the electrical power (voltage x amps) put into the laser diode.   The WPE of the laser does not include the power of the drive circuitry external to the diode which can be significant.

Note that WPE alone does not factor into it the wavelength.  So if you are comparing a 510nm WPE to a 532nm WPE in terms of lumens per power Watt, you really need to multiply the 510nm green WPE by 57% to compare equivalent lumens per Watt of electrical power.

In the next article in this series on diode green lasers, I’m going to discuss the color space problems with 510nm green (namely it can’t be used to get a good bright yellow).   For those who know about color spaces or can figure it out just from a a diagram, I have included below a CIE chart with the color space triangles for 510nm, 525nm, and 532nm green holding blue at 455nm and red at 640nm.

Credit:  I presented a version of the 2 figures in this post at SID 2011 in conjunction with our paper (co-authors of the paper where Dr. Bill Mei and Dr. Shawn Hurley)

Karl Guttag
Karl Guttag
Articles: 244


  1. For an excellent and more in depth technical information on the issues in making green laser crystals, see: “http://www.opnmagazine-digital.com/opn/201109/?pg=43#pg47″ rel=”nofollow”>Osram, Nichia, SumItomo simultaneously Demonstrate First True Green” DGL http://www.opnmagazine-digital.com/opn/201109/?pg=43#pg47.

    But please note the word “Demonstrate” in the title. With products based on material science like this, there can be (many) years between “demonstration” and production.

  2. Karl,

    This is very good information indeed, as there is a scarcity of resources available to provide any measure of understanding of the very less than clear technological challenges surrounding direct green laser diode development. I look forward to the next of your series.

  3. Karl, interesting stuff and I’ve seen people present these plots before. However, I think there is a point missing. When talking about lumens per watt, what is important is this ratio when you produce white since white is almost always the brightest part of an image. Hence what matters is how colours need to be mixed. As an extreme, hypothetical example, imagine you had a blue of 487nm (cyan really, 125 lm/W) and a red of 590nm (orange really, 517 lm/W) along with 532nm of green. These “red” and “blue” wavelengths are very efficient but I calculate that the resulting white will have an efficiency of about 240 lm/W. You would expect that changing the blue and red to the commonly available 442nm (27 lm/W) and 642nm (109 lm/W) would be worse but I calculate that this gives 258 lm/W for white. Using 430nm, 560nm and 610nm would give a poor colour gamut but a huge 480 lm/W for white. This was presented at SID in Germany this year.

    • Ian,

      Certainly what you say is true, you can’t look at just efficiency or white point. The teaser figure at the bottom for the next part in this series is a CIE chromaticity chart which I plan to discuss next. To have a good projected image you have to solve for multiple things at the same time. One point I am trying to make/show is that if you use a 510nm “green” then while you get something that looks “green,” you can’t make other colors, particularly a bright yellow.

      I’ve played with the various wavelengths of R, G, and B and like you said, it is funny how the wavelength of blue has next to no effect on the efficiency. You can sort of see this in the photopic curve since blue is on the flat part of the curve. While everyone in the industry tends to talk about the wavelength of green, it turns out that red is also pretty significant. But just like it is hard to get a longer green, they have a hard time making a shorter red. The ~640nm reds that are available are much deeper than you need for making a projector (but some color “purist” love to see them) as they are so far outside any color space that is used. But if you look at the photopic curve in the post, shorting the red even a little has a significant effect on the lumen output per Watt.

      Also in theory if you had a four color there are things that can be done to improve efficiency, but then combining the colors is an issue.

      BTW, thanks so much for a “technical question.”


Leave a Reply