>>> Needless to say, to have any chance of achieving rated
>>> lifetimes LEDs need to be driven in a constant current 
>>> mode
>>> at rated current or less.

> They did a study over on candlepowerformums to check 
> brightness of
> the "extra bright" white LEDs available on eBay (from 
> China),
> compared to brand-name Nichia white LEDs.  Some of the 
> imports
> really were as bright as claimed, but they deteriorated 
> much
> more quickly than the Nichias, and VERY quickly indeed 
> when
> "overdriven" as common in flashlights (The chinese LEDs
> deteriorated much more quickly when overdriven than the
> Nichias.)

If you look through those refs I provided you'll find a 
report on some BIG brand name LEDs that had very very bad 
lifetimes. As well as thermal and dye issues there are 
encapsulant performance and die wire longevity issues to 
consider.

Standard market practice in cheap retail devices (and even 
in dear retail devices) is, as I said, to hard parallel 
LEDs. This can lead to in excess of 2:1 +  current 
distributions - depending on how well binned the LEDs are 
for Vf at operating current.

The products that I'm assisting with the development of 
initially had parallel LEDs with spreading resistors. 
Current was moderately well controlled but not well defined 
across battery lifetime. All the products will now use 
current driven LEDs in series strings with well defined 
operating current at all battery states.

The manufacturer involved has consistently used a single LED 
supplier who is as about as "name brand" as any Asian maker 
(but who I'll not name here for reasons which seem good to 
me at present) and whose products have performed well for 
them. I will be doing ongoing tests of my own.

Lumens per Watt are indeed "right up there" with the best 
available in anything like the price range. You can buy 
binned LEDs at very high prices that double the l/W figure. 
ie we are getting 50+ l/W typical figures. Cree (I think it 
was ) will sell you a 100 l/W LED commercially but to get 
this they spec it at a current which is less than its rated 
maximum - which will also help its lifetime. People are 
getting 130+ l/W in lab samples but AFAIK these have not 
appeared on the market in other than samples yet. For white 
LEDs 100% is a bit over 300 l/W (varies with chromatic 
spectrum due to the definition of a lumen relating to eye 
response) so there is quite a lot of improvement to look 
forwards to yet.

People are getting upwards of 70% Quantum efficiency at SOME 
wavelengths but much poorer at others and so far this 
doesn't translate overall into anything really close to 
theoretical maximum.  However, it's impressive to realise 
that a 100+ l/W LED requires only 2/3 the cooling of a 
resistor dissipating the same power. 1/3 of the cooling is 
done by optical radiation :-).




                Russell

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist