Basic HTML Version
NORDIC LIGHT & COLOUR
131
thing about the ‘naturalness’ of a color, but not really anything
about the ‘vividness’ or the ability to distinguish between subtle
differences in hue. Further, which is probably the most prob-
lematic issue in relation to information and general evaluation
of the color-rendering capabilities, CRI cannot be compared
across different color temperatures (K). It is problematic, as
incandescent light bulbs, who are the basis for the develop-
ment of the CRI, have a CRI 100, but at 2700K, whereas some
(mostly early) CFL’s and LED’s have a higher color temperature
(>2700 K).
Further, good color rendering is difficult to obtain, if the level
of luminous flux per unit area (lux) is lower than 100 (Rea and
Freyssinier-Nova, 2008). This is also a problem in terms of
being able to assess this based on the information on the light
source’ package, not least because the level of light is indicated
through luminous flux (measured and indicated in Lumen),
which requires a conversion of values just to get to know the
level of illumination, measured in Lux. The difference between
lux and lumen is that lux takes into account the area over
which the luminous flux is spread. 1000 lumen, concentrated
into an area of one square meter, lights up that square meter
with an illuminance of 1000 lux. The same 1000 lumen, spread
out over ten square meters, produce a dimmer illuminance of
only 100 lux. If a light source is placed central in a room, with
the intention of illuminating most of the room, it would require
a rather high luminous flux, in order to obtain a lux-value above
100. As many LED’s still struggle with obtaining a luminous
flux above 1000 lumen (corresponds to 50 watt incandescent
light bulb), using a LED for general lighting would give very
poor color rendering. However, this would also happen if using
a 50 watt incandescent light bulb – but that discussion seems
absent when talking about incandescent light bulbs, and is not
dealt with in the pamphlets either.
Getting back to measuring the Ra value, obtaining a Ra value of
100 merely implies that each of 8 pale test color samples (and
sometimes an additional 6 ones) looks exactly like they would
under a black-body radiator (incandescent light) at 2700K.
However, this only applies if the tested source has a color tem-
perature
corresponding
to 2700 K as well. Secondly, but equally
important, the evaluation of the color rendering based on the
test color samples is based on an average across the samples,
making it difficult to determine in what color areas of the spec-
tral distribution that the color rendering is better or worse.
Thus, what is deemed a ‘good quality’ color rendering seems
very much associated with how colors look like under incan-
descent light (a black body radiator). As the incandescent light
bulbs’ spectral distribution is more intense within the red and
yellow colors of light than for instance the fluorescent bulb and
to some extent the LED’s, it may cause problems for particu-
larly LED’s as they often consist of narrow-band light sources,
such as red-blue-green LED’s used to produce white light.
That the CRI is not being without problems for other light
sources than traditional incandescent light bulbs as well as
traditional fluorescent light bulbs is also documented by Corell
et al (2009). They show that a particular LED that is designed
without the blue part of the spectrum (a ‘no-blue’ LED) gets
a lower CRI score than a yellow fluorescent tube, although
when comparing the color rendering through the Gretagmac-
beth color rendition chart (a visual measurement), the colors
on the chart are rendered much better by the LED than the
fluorescent tube. It is however important to note that in this
case, neither the tested LED nor fluorescent light bulb are de-
signed to have a good color rendering as the light sources are
designed not to have any light below a certain wavelength, and
thus would not be regular white-light sources. Therefore the
mismatch between the sources CRI light sources and the visual
comparison would presumably not be as evident for regular
light sources that would often be white-light light sources.
However, it does indicate that the CRI has limitations in relation
LED light (Rea and Freyssinier-Nova, 2008). This issue is in fact
dealt with in a recent paper presented at International Com-
mission on Illumination (CIE) conference 2010, where Smet
et al (2010) discusses alternative ways of using samples for
testing color rendering capabilities, particularly in relation to
testing LED’s.
No matter the complexity of even getting the evaluation of color
rendering right, the information about both color rendering
capabilities and color temperature on the packages is quite in-
consistent. One of the reasons why there is no information about
this on the incandescent bulbs may be connected to the incan-
descent light bulb having been the only, or at least main source
for domestic light for a very long time, and that no one has paid
much attention to explicating the various functionalities of this
bulb, as no (few) other alternatives were available. Either the
package presented in Figure 7 is old, and therefore holds no in-
formation, or the lack of information is a reminiscence of the in-
candescent light bulbs ‘era of monarchy’. But it does not explain
the lack of information available for the newer technologies.
This may partly be explained by the nature of the requirements
in the European Eco-design Directive for (non-directional) light
sources (244/2009, part 3), as only some information is required
to be on the package itself, whereas other information (including
color rendering) is only required to be publicly available on the
internet. What is further interesting is that this only applies for
non-directional bulbs, however with some exemptions, whereas
directional bulbs, such as LEDs, are not included, and therefore
remain unaccounted for in this sense.
ies,