l> To be "white" is one required to include all wavelengths of light from
> infrared to ultraviolet ?

No.

> Or does one just hit the high points of the cone cells ?

No.

The following is far easier to understand and to get a feel for than
will probably at first be apparent. Stare at variations of a CIE1931
chromaticity diagram * (see below) while invoking colour temperatures,
black body locus ** and the like and before long it actually SEEMS to
make sense. YMMV E&OE etc.

So;

White is a brain perception which occurs when the "vector sum" of the
amplitudes of all wavelengths present on the CIE1931 chromaticity
diagram * fall on the black body locus** in a specific range near the
centre of the diagram.

Vector summings near this locus appear whitish. My "favouriite" white
is at coordinates 0.31, 0.31 on this chart. This is a "daylight white"
 which is a little on the bluish side for some. Some people prefer
"white's" which are off to the right along the locus - known as "warm
whites" and with a greater "yellow" content.

The modern "white" phosphor" LED, invented by Nichia,  mixes only two
colours in its vector sum. The LED proper generates a deep blue, and a
proportion of the blue is absorbed by a phosphor and re-emitted in the
yellow part of the spectrum so that the vector sum of the two lies
somewhere in the above region. Note that while the yellow is a single
re-emission wavelength it in turn can be thought of in terms of a mix
of the RBG primary colours of this chart. In this case it is NOT made
that way, but could be arrived at in that manner - with possibly
different second order effects - see below.

* The 1931 in CIE1931 is the year 1931 AD and simply refers to the
year when this version and its associated arcanery was promulgated.
There are later version which are far less used and arguably of
potentially greater value, and there are other ways of color energy
mapping. This is an extremely black art indeed and delving beyond
CIE1931 is seldom necessary.

** The black body or "Plankian" locus is the colour emission line of a
black body radiator as it is heated. A reasonable approximation is a
very small hole in a box whose interior is held at a given temperature
and a nice expensive and convenient alternative is a Platinum-Black
surface or even sheet Platinum (such as foil), given appropriate
mechanical arrangements.

The temperature involved is the "colo[u]r temperature" so beloved of
LED sellers.
* ** CIE1931 chart here with Plannkian or black body curve shown.


http://upload.wikimedia.org/wikipedia/commons/b/ba/PlanckianLocus.png

You can find my favoured 0.31. 0.31 (dimensionless axis) here.

Time to stop before we get too 'Boxian'***  in our 'model', but note
that the traditional RGB colours are at the extreme corners of this
model and that all colours shown in a triangle between these colours
can be obtained by vector summing. Some colours along the blue green
edge can NOT be obtained in this way - and that leads to a vast new
area. The Adobe RGB colour space uses colours outside this space to
obtain a broader "colour gamut". As gamuts get more extensive you
start to find that extreme regions are printed in white, as printing
inks are not available which extend far enough towards the area of
interest.

This leads us back to Gus's original "bluier than blue" question and
we see that colours "outside" the standard RGB can bve very useful
indeed.

Better still, having FOUR "primary" colours available can be of
immense use. If we had eg a greenier than green at 0.1, 0.9 and
another at say 0, 0.9 we could fill the top left corner with color
that would otherwise be unobtainable. Note that human eye response
will aso start to play a part at extreme 'edges'. Some recent displays
are using 4 emitters to allow better color mappings.

Eye response also plays a part in lumen output for a given energy
level, as the definition of lumens includes standard eye response.
Quite apart from the efficiency with which a light can produce a given
wavelength, or mix of wavelengths, this leads to "warm white" LEDs or
CFLs being of lower relative luminous efficiency than 'daily white'
emitters.

The white from a white phosphor LED (usually) contains only two
predominant wavelengths, as noted above. A significant minority of
people find such 'white' visually unacceptable, without necessarily
being able to say why. In some cases people draw conclusions about why
a light source is unnacceptable to them, and express their objections
in those incorrect terms. This can complicate remediation.
eg "flicker" is seen as a problem by some, even at drive frequencies
of 100's of KHz - well above the rate of human optic nerve response.
There may conceivably be secondary effects due to frequency, so such
complaints should be treated with due care, in case there is an as yet
unknown mechanism at work. However, even LEDs driven with the purest
of DC may elicit complaints of 'flicker' from some people, indicating
that while the effect may be entirely real, the perceived symptom is
not well correlated with the actual cause.


              Russell McMahon


*** "All models are wrong, some models are useful".






* ** CIE1931 chart here with Plannkian or black body curve shown.
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist