LED Shorted??

Jim Thompson a écrit :

Yup, I have seven or three or fifteen, pi, sqrt(two) and lots of others.

How much of a short is a blob of fused silicon?

I don't know, I've never flashed an LED ;-)

Anyone with experiences where the failure was a short? ...Jim Thompson

No LED displays in your bedroom?

No, not for long, anyway.

That's my suspicion. I'm tasked with designing a driver with fault report capability. As usual, the competition's "fault" specs seem downright nonsensical. ...Jim Thompson

Why don?t you try Lumex. Contact a Field engineer, perhaps he can get you an internal contact.

Cheers

Most of the shorts I've dealt with recently were the result of gross reverse overvoltage, though current-limited, and resulted in insignificant residual resistance levels.

These were 1W emitters for lighting applications, in large strings with extensive heatsinking, where human factors and lack of reverse protection was an unfortunate combination.

Some of these parts glowed red under reverse bias before actually failing. approximately 10% of the string's elements survived without apparent degradation and no open circuit elements were formed.

In the past, I've associated end of life or infant mortality of LEDs mostly with lead bond failures or low emmissivity, neither of which resulted in a purely 'resistive' DUT.

RL

I don't know how you'd get real numbers for a general question. There are many LED manufacturing processes and they're constantly being upgraded.

I've observed some small LEDs show damage from inductive reverse voltage spikes and I've had several recalled Luxeons Rebels fail. They presented a very weak short that could be blown away with some additional damage. Depending on the drive current, they'd go out, blink erratically, or flicker. Most would flicker at 100% normal current.

The Luxeons have an ESD device that will cause flickering when driven below the recommended currents. My 21 Rebel flashlight leaves a few microamps flowing so it can be found in the dark when the brightness is turned all the way down. It sometimes twinkles like Christmas tree lights.

I've had some whites die by dimming themselves over a few months, OTOH I've got a little LED widget powered from the mains, series cap and resistor, with coloured LEDs been running for about 8 years no problem.

But I didn't keep the failed by dimming white LEDs, threw one out only recently too, as I reused the transformer with a 'proper' constant current sink with a UV LED to light the luminous clock in the bedroom.

We don't have the radiation style luminous clocks from my childhood any more :( But radioactive smoke detectors in most homes is harmless?

What sort of LEDs data you after, white ones, I guess? Or any sort?

Grant.

I thought pi was silly enough to be somewhat unreal, or was that surreal? :)

This much -->

Time for an experiment? Take a bunch of LEDs, toss 'em into a microwave oven...

Sense LED current, if above nominal range report short, if below nominal range report open?

Ed

PWM'd Constant current drive... video application :-) ...Jim Thompson

So you have the current through string and voltage across the string?

Usually a zener from top of string to limit overvoltage when LED opens, at least I put one there so the inductor doesn't kill itself.

Sure, you see a shorted LED by string V drop, but does a dim LED change resistance? Nobody said yet. Or, you driving so hard you know they going to short? Heat V drift too. Hope they pay you heaps ;)

Grant.

I bought some low-end white (InGaN) Luxeon Rebels for circuit prototyping because I expected to kill a few by accident. On the contrary. I ended up TRYING to kill some and they wouldn't die. Efficiency recovered as soon as they cooled down. They'd actually unsolder themselves and drop off the board before frying. They only ones to fail were the recalled chips.

The AlInGaP Rebels were a different story. Exceeding the maximum rating caused rapid permanent decay of efficiency.

Reliability testing under normal and abnormal conditions:

Separate sense path for forward voltage?

Yep. Comparator for short (to rail), another looking at cascode behavior when an open occurs... it dies ;-) ...Jim Thompson

So, a series R and compare LED Vf? Low V = short, normal V = ok, high V = open. Hey - it's your chip acreage I'm spending, not mine. :-)

I'm just guessing you can do it without any knowledge of typical R(ledshorted), which might turn out to be non-typical.

Ed

...

Reminds me of the old i8748 (Intel microcontroller), put it in socket reversed, the corner power lead wires would glow bright red under the quartz window, put the thing in right way around, and no harm done :) Power 5V supply was 7805 on small heatsink from 12V, so it'd be a good

I get that impression from the LED driver datasheets being so careful about taking LEDs to the limit without breaking them. The big LEDs too expensive for me yet. I'm happy with strings of 5mm LEDs at the moment.

Odd cutaway drawings don't show the anode?

Thanks, Grant.

Some things are still missing from the task description.