I have noticed that in urban areas at night, blue lights have
a tendency to stand out. One application of this is the blue lights used
for police emergency phones in some college campuses and the like.
Until now I believed the noticeability of blue lights was a result of
"blue impact" or stimulation of the blue-sensing light sensors in human
eyes. I even made up a unit of "blue impact" that I called the "blumen" for
"blue lumen". Back when the lumen was defined in terms of the original
official photopic function (1931 CIE "Y-bar" function), blumens were equal
to lumens times the ratio of z to y chromaticity coordinates. (The 1988
redefinition of the photopic function messes this up, generaly just a
little)
So I notice that a major college campus where I did delivery work at night has LED blue lights for police emergency phones. Most have the usual 470 or so nm blue LEDs. I notice that current bright blue LEDs have peak wavelength mainly around 466-470 nm. But the blue peak of human vision is a little below 450 nm. And I wonder about the lack of blue LEDs with output concentrated towards this wavelength.
Nichia, the pioneering manufacturer of high brightness LEDs, originally produced a 450 nm blue model. Panasonic had blue LED lamps with this chip or a similar chip, available from Digi-Key for a while. But these are not available now from the sources that I know of. One reason is that Nichia's current 470 nm ones produce more "blumens" and more lumens than the older 450 nm technology. Also the older 450 nm technology was a broadband blue, whitish with actually a minority of the spectral output in wavelengths where the human eye's blue response is over 60 percent of peak.
Since the days when 466-470 nm narrowband blue LEDs superseded 450 nm broadband ones, I wondered about why all the easily available blue LEDs were 470 nm ones when 450 nm ones could yield more "blumens". I suspected that 470 nm ones were favored merely because they produce more lumens and more candela - which are specifications that people would recognize and which are useful to put on a spec sheet but do not relate to the less-known "blue color impact" or my "blumens" or blue-relevant derivable units such as "blue candela" (candela times ratio of total Z-bar to photopic total of all wavelengths produced.)
Back to 450 nm blue for the usefulness of 450 nm blue...
I have noticed that Cree makes 450 nm blue LED chips. The more interesting
one is their C450-CB290-E1000 which has peak wavelength around 445 nm and
dominant wavelength (defines color to "official standard observer") around
450 nm. They also make the C460-CB290-E1000 chip which has a peak
wavelength around 456 nm and a dominant wavelength around 460 nm.
The power output according to Cree is 3.5 mW at 20 mA, at least for the
450 nm chip. So I wonder why nobody seems to be making easily available
lamps with chips that promise to produce lots of "blumens" and with a
distinct pure deep blue color.
So I search, search, beg, search and search and ask trying to get my
grubby hands on a few LED lamps with the 450 nm Cree chips.
I got a few from someone. Maybe they just wanted to relieve the world from
me asking for such things, or maybe they actually believed my intent to
see if I could make demand for these things by finding good use for these
that many would dismiss.
So I test and compare them. My work in this area is a bit slow since I have paying applications for my time outside my fulltime job. But I screw around and at times screw around with LED lamps and at times screw around with these blue ones.
For a few weeks I just compared the 450 and 460 nm (dominant wavelength) versions of the deep blue lamps with Cree chips that I got. I expected some obvious advantage of the 450 over the 460 due to higher "blumens" - the two wavelengths were surprisingly similar in all forms of noticeability.
I view the lamps directly on-axis and off-axis in dark, moderately dark, room-light and daylight conmditions. I shine the lamps onto a white target in equal manner from short and long distances in all sorts of lighting conditions. The advantage of 450 over 460 is subtle, almost down to "prettier color". I was disappointed.
So I get a 470 nm LED lamp and tape black electrical tape around it to make its output distribution as similar as possible to that of the deeper blue 460 and 450 ones that I got. Then I measure optical output of each of these with an unusually broadband solar cell (black single crystal silicon basically satellite grade). I adjust/change the dropping resistors for each LED lamp to get solar cell current outputs (to a milliammeter) proportional to wavelength - ideally equalizing optical power output, and assumed not deviating much from that since all lamps in question are narrower band LEDs in the same basic color of the visible spectrum.)
Results of 450 vs. 460 vs. 470:
"Blumens" is not as overwhelming a factor as I thought. I did some
preliminary tests of firing these lamps through a diffuser to equalize
light output pattern (the diffuser was plain white paper) to make them as
equal as possible both in optical power output and pattern of light
distribution. With the output being illuminated spots on a sheet of white
paper, visual tests are limited to viewing of illuminated spots.
As for 450 vs. 460 vs. 470 for illuminated paper spots: In the dark visual
impact has some dependence on lumens, or maybe the scotopic equivalent of
lumens (which peaks around 507-508 nm as opposed to 555 for photopic
but still favoring green or blue-green as opposed to pure blue.)
The 470 did best in dark conditions and 460 usually did better than 450 as
opposed to the other way around. All three were nearly equal, with 450 and
460 very nearly equal and the 470 maybe 40-50 percent "brighter" than the
450. By lumen and "scotopic lumen" accounts these should be more disparate
so "blumens" count for something, just less than I thought they would.
In brighter daylight, all three were surprisingly equal. The 450 had some
advantage but it was subtle, almost down to "prettier color". At many
times the 470 had a minor but significant advantage - apparantly
noticeability of a blue light is some combination of "blumens" and
lumens and neither figure alone. I am still working on how much weight
each figure should have.
UPDATE 8/19/2000 - I put the three LEDs on a board and took it outdoors in an urban area at night with a friend to see how they stack up. We went to different areas of a large apartment complex parking lot:
Dark, dimly lit with high pressure sodium, brightly lit with high pressure sodium, lit with metal halide, and lit with incandescent.
I illuminated a piece of paper with them so that I had three "light sources" of equal size, directional characteristics, and equal optical power.
Results: The 460 and the 470 mostly came up as equally visible and noticeable, with the 450 looking dimmer - about 3/4 the brightness of the other two.
NOTE - in RGB LED displays such as large TV screens using R, G, and B LEDs, "blumens" seriously count for the blue elements and there is also desirability of best-matching the blue phosphor of the usual TV's and monitors. I think 460 is best for this. 450 has a little more blue impact than 460 but my recent tests show the advantage of this to be small and 450 has apparant color varying with lighting conditions (sometimes looks violetish) in a way that 460 avoids. So I think 460 is better than 450 or 470 for 3-color LED TV screens. But in other applications 470 seems better than 450 or 460, to my surprise.
I am still working on this.
Please read my Copyright and authorship info.
Please read my Disclaimer.