Playing With UV LEDs

[Mike O'Dell]

the current leading-edge crop of ultra-high-output white LEDs are,
in general, blue LEDs with phosphors which are primarly yellow but
are a mix of phospors to produce the desired color temperature.

at the moment, Cree holds the record for the highest lumens-per-watt
efficiency as well as the highest lumens per single chip.  these
records slosh back and forth with considerable frequency, however,
as this is an extremely active development area. the EU has funded
an R&D program aimed at going to LEDs for general lighting
applications ASAP and it's catalyzed a lot of work.

what is now clear is that the design of LED-based lighting is
a systems design problem, not a component design problem.
the ultra-bright LEDs indeed generate considerable heat but
they are much, much more heat-sensitive than xenon-tungsten
(the primary competitor). the "ballast" and housing and heat
sink all have to be designed together with a specific choice
of LED(s).

one alternative not widely considered for terrestrial applications
but which sees a lot of use in large boats and yachts is 
"centralized" lighting. in this model, a single
high-intensity-discharge tube of several hundred watts
generates a metric-shitload of light, which is then
distributed to where it's wanted with glass or plastic
fiber bundles. dimming can be as simple as a mechanical
shutter and as sophisticated as an LCD panel. individual
luminaires can filter for color temp as require, too.

the net is that it saves a dramatic amount of electricity
and considerable weight as well. the speed and ease with  which 
one can install 2000 pounds of copper in a boat is truly scary.
(remember - it's all 12 or 24VDC, so I^2 * R losses get to be
non-trivial quite quickly)

one other "issue" with LED luminaires is that LEDs are,
by comparison with xenon-tungsten incandescent lamps,
highly directional. historical lighting models contain
assumptions about uniformity of the light field from 
a source (an "isotropic point source") and it is both
expensive and likely unnecessary to simulate an IPS
with LEDs. so lighting design models are being revised
to accomodate the physics of the "emission cone" which
better characterizes the flux from an LED. 

of course this all interacts with the design of
luminaires, too, so it's been interesting to
watch the evolution of luminaire design over
the last 4-5 years as designers get a handle
on the characteristics of these new sources
all the while the sources were evolving rapidly
as well.

there's probably a decent PhD topic in the history
of science here somewhere if you know someone looking
for one.

    -mo