>> >how many LEDs were used
>> >
>> > My guess is more than 6
>>
>> Oh, at least 8, easy
>
> And I wonder how many of them work today......

The unit in use will perhaps be something like E6 to E7.
Let's aim high.
Of the say 8E7 LEDs in use most will still be functional as 
early mortality is not a major failure mechanism.
Assuming a design that at least aims to be robust in 
practical use you can probably extract absolute-mortality 
(ie alive/dead) from other measures of operational efficacy. 
Positing a classic bathtub model is possibly a good first 
start.

As is shown here

        http://www.weibull.com/hotwire/issue21/hottopics21.htm

early failure in a bathtub model is subject to the Weibull 
distribution which I discussed recently in the context of 
LED failure prediction.
See eg figure 2. The most useful thing that you can conclude 
from this curve, which has an x axis range of 10 years / 
87,600 hours = entirely appropriate to this problem, is that 
1st day rates are going to be extremely hard to predict with 
any sort of certainty and are going to be far higher than 
the 10 year mean. The y axis here is specific to a 
particular product so is not useful here.

If we posit a 10 year absolute mortality rate of 5% (which 
I'd hope to be high for a quality product) due mainly to o/c 
of the electrical circuit due to eg bond wire failure, 
bonding and soldering/welding separations etc, then the 
day-one rate may be 3 or 4 orders of magnitude above the 10 
year mean. Thus if we have a 10 year mean rate (as above) of 
0.05/(10*365) = 14E-6/day or say about 20 per million per 
day mean 10 year rate. If we took the day one rate to be 3 
orders of magnitude higher then we'd expect about 20,000 
failures per million per day or about half the 10 year 
absolute mortality quantity. This may or may not be high but 
my feeling is that for a quality product it is.

However, using this figure then it suggests that about 2% of 
the LEDs used in the display may be dead due to absolute 
mortality / bathtub / weibull failures. In reality this is 
far higher than I have seen in practice.

This assumes not only that the reasonably arbitrary E3 
multiplier has a dickie birds chance of being something like 
correct but also that other modes are not dominant.

In well designed equipment with LEDs being operated at  no 
more than design current, and well heat sunk so die 
temperatures are not exceeded then even if lumen degradation 
was extremely rapid by normal standards it would not play a 
significant part in 1st day failures. However, if the 
equipment is built in a manner which even vaguely 
approximates the utter lack of design and attention to basic 
engineering principles which is evident in the vast majority 
of product from that area then anything might happen. It is 
entirely usual for many LEDs to be operated in hard parallel 
with no attention to current spreading. Also, current 
ratings are often not adhered to and thermal considerations 
often seem to not be a consideration.

So, I'd hazard, almost all of them.


        R






 

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