On Sun, 30 Oct 2005, wrote:

LED bulbs SHOULD NOT generate sufficient heat to explain significant
temperature rise, but one still active from the batch that contains the
failed bulbs warms up enough to be felt by touch of the glass bulb.

It is a common misunderstanding that LEDs produce trivial heat. In
fact,
they produce quite a bit of heat per lumen. The primary reasons for the
misunderstanding a

1) This heat is emitted to the rear of the junction, not "out the
front" in
the same direction as the light as is the case with most other light
sources, and

2) Most people's only LED experience is with the very small, relatively
low
output items used as indicator lights on e.g. electronic equipment and
dashboards. A great deal more light is required for illumination than
for
indication.

The backside heat produced by LEDs powerful and numerous enough for
illumination is *not* trivial, and it creates two problems: There is
the
issue of thermal resistance in the materials used to build the lighting
device, its housing and surroundings -- this doesn't really differ from
the
same problem with any other light source, with the exception of the
location/direction of the heat. This is throwing obstacles into the
development, for instance, of LED vehicle headlamps. Not only is the
significant backside heat creating thermal management problems
requiring
elaborate and expensive solutions (fan-cooled headlamps!) to avoid
exceeding
temperature limits in the lamp housings themselves, but the lack of
*frontside* heat means the lens doesn't defog or thaw when the lamp is
turned on. Too much heat in back, not enough in front!

LEDs are also highly temperature sensitive with respect to their
output.
It's common to see a swing on the order of +60% at -30°C to -40% at
+40°C,
relative to nominal output at 20°C. Most other light sources in
current use
are minimally temperature sensitive in this manner, if at all. Filament
and
arc lights generally don't know or care what the ambient temperature
is.
Fluorescent lamps tend to have low output at low temperatures (and
unreliable starting, in extremely low temperatures), but their internal
heat
tends to counteract low ambients such that this tendency amounts to
little
more than extended ramp-up time when started from cold. While thermal
direction and management issues are causing problems for headlamp
engineers,
the temperature sensitivity of LEDs' output makes problems at the other
end
of the vehicle: How do you design a brake lamp that is guaranteed to be
sufficiently intense at 40°C when its emitters are producing only 60%
of
their rated output, while not being overly intense at -40°C when the
emitters are producing 160% of their rated output? It's not as easy as
it
sounds.


I am using the 120V LED bulb to light my computer keyboard. Since the bulb
is fed by the UPS, I prefer the LED bulb in order to reduce the load on
the UPS during power failure. Since I did notice a voltage peak in the UPS
when it turns on, I wait for it to stabilize before turning anything else
on, including the light. Still, two bulbs did not survive a month each. I
believe this is due to poor workmanship. Trial and error at the prices of
these bulbs is not good enough.

Unfortunately, it really is your only reliable option.

DS