Exceeding LED ratings

I wonder if you've not been to India :)

NT

On Nov 7, 12:06=A0pm, "Pimpom" wrote: =A0And if

I believe light output decreases with increase in temperature.

Dan

I haven't been anywhere that requires more than 4 hours on an airplane. My loss.

John S

snipped-for-privacy@krl.org Inscribed thus:

I agree :-) That is why I suggested the longer "off" period.

--
Best Regards:
                          Baron.

were=20

=20

None it is way too small. A decent approximation of a geometric point. (far smaller than a gluon,

A while ago, we used some Nichia white 3mm LEDs in a product, driven from a programmable current source.

Unfortunately the current source design was such that, when the set point was over ~80% of full-scale, the driver would latch-up and connect the LED directly across the supply rails (there was a small current- sense resistor in series, 10R IIRC). The duty cycle sounds similar to your application, pulsed on and off with a period of about 5 seconds.

Current measured in this condition was about 75 mA, abs max for this part is 30 mA. We had a couple of dozen units in use for several months before we noticed. Field failures : 2.

If you are able to replace any failed LEDs on the spot, I'd say you should be OK.

Yeah I assumed that temperature was responsible for both the 1/f noise and decrease in light.

George H.

Some quick data, Cheap red LED shoved into photodiode (PD). PD gain =3D 100k ohm.

(again at ~DC)

I(mA) V PD (Volts)

2.9 0.266 6.3 0.472 9.93 0.623 14.9 0.776 19.8 0.871 (and drooping with time) 30 1.03 (fast data) 40 1.07 "

The last two point were taking fast... current ramped with knob and first voltage reading recorded.

George H.

Thanks, and thanks also to everyone else who posted meaningful replies. I read them all but didn't have time to reply individually to each post. We ran trials today and everything's OK so far. The actual event starts tomorrow. I'll report the results later.

One would think that the proper culling method to obtain that desired end would be to use a thermal imager and static current flows. Develop a table (graph). The plotted points would show ANY and ALL LEDs that would suffer from an earlier death than the others. AND it would show which ones would likely last or perform beyond all the others in the batch/lot.

The IR emitted right on the LED die segment would be directly proportional to light output as well. Up to a point near the extreme ends. So no need to attempt to measure the light output. The IR will cull the best units out just fine.

Using a fixture, one could do this with a mere IR thermometer without imaging capacity as well.

Then tell your retarded mommy that I am not interested in her hamper.

You are as dumb as it gets.