The incandescents lasted forever (well at least 4-10 years) until they were
turned off.

I believe I have the supplier mixed up. It wasn't OSRAM but another supplier
with similiar type name???. OMRON or something..Been awhile now. These lED
indicators were all crap and we tried many different styles and many
different current levels. When run at their rated current (I think about
20mA) they all went up in smoke after a few years, anyway. The main (130vdc)
ballast resistors were mounted elsewhere so they weren't a problem. The
problem, as I saw it were they were designed as a 24v bulb with 24vdc worth
of ballast in a miniature bulb....that's a no..no and did them in from
localized heat. Finally, after about 15 years of experimenting with them and
different breeds, the Engineering department decided to ignore the
manufacturer's advice, went back to incandescents and replace the bulbs
every few years when the device was de-enrgized, basically.

As I stated, the LED units are back without any diffusion. LEDs just don't
put out enough light to make them look like incandescents with diffusion and
still be visible with bright lighting. The red and yellow ones were never a
problem, only the green, other than being short lived.

"Don Klipstein" wrote in message
...
In article , Josepi wrote:
As some of the articles point out LED testing may be done unfairly, is
many
cases. The manufactures show lumen output for bare elements and then add
the
reflectors, lenses and other external parts later.

The ballast in not usually included in the efficiency testing, either.

Are these the white phosphour screen based LEDs, you refer too?

As a side note our company put in hundreds of OSRAM indicator pilot lamps
on
electrical control panels. After 10-15 years of replacing bulbs, burnout,
sock melting, changing ballast current limiters, lenses and filters,

Due to someone not knowing how to implement the LEDs properly, though 15
years ago efficiency of LEDs was a lot less and maybe they could not have
been implemented properly.

we changed them all back and retrofitted them to incandescent bulbs.

Certain colours, green especially, could not be dicerned, when
illuminated,
if there was any windows with sunlight entering into the buildings. If we
put a similar green pilot lamp with a lime green filter in it (unlit)
beside a normal green illuminated unit, no difference could be detected.

This problem is very easy to avoid with the green LEDs that are
available nowadays, not too hard to avoid with green LEDs that have been
available since about 2000-2001 or so.

When we increased the drive current, the bulbs only lasted a month or so
(at
a cost of about $5 per bulb). These were very tiny LED segments with about
9
elements in each bulb. The ballast resistor dropped the current from a
130vdc battery bank and was a burn hazard for humans.

Have a look at what just one modern good InGaN green LED can do with
5-10 mA now, or what one made by Nichia in 2001 can do.

Inverter technology
was a much better proposition but too expensive a retrofit for so many
bulbs. They spent tens of thousands of dollars trying all of OSRAM's
tehnologies they had availble for about 10 years and finally went back to
incandecent bulbs with low current supplies (less than the LEDs) and the
bulbs last about 10-15 years (or until your turn them off, after a few
years
of usage...LOL).

Did you run controlled tests? I have heard of testing showing that most
incandescents do not lose much life to cold starts. They do become unable
to survive a cold start before they become unable to survive continuous
operation, but not by a lot. The usual incandescent failure is from a hot
thin spot in the filament, prone to temperature overshoot beyond its
already-excessive temperature when a cold start is imposed upon it. This
bad condition of an aging filament accelerates worse than exponentially,
and an aging filament that cannot survive a cold start will kick the
bucket soon no matter what.

- Don Klipstein )

In the last few years the pilot lamps got smarter and went to a
non-filtered
LED holder, so the area of illumination decreased and the LED elements
were
now visible. This made the LEDs visible and workable but the whole thing
dazzled the eyes like a Christmas tree.