I'm designing an Arduino Nano circuit that has the ability to sense when an external LED fails, and I need to test it. Obviously, the easy solution here is to substitute short and open circuits where the LED is supposed to be. I would assume, however, that testing the circuit with an actual failed LED would give better data.
So how can I break an LED without catching stuff on fire? Most internet searches turn up results on how NOT to break stuff. Typical.
The component is a low voltage, 15 mA UV LED, looks similar to the ones that come with Arduino kits.
Linking to the datasheet for a similar LED, as I don't have access to the original.
Maybe OP is designing / building a LED tester? In that case you want to test the tester with known components; good ones (often known as golden samples that are kept for reference) and faulty ones.
@sterretje
That's more or less the case. Sensing component failure is the entire point of the project, and I don't think I can properly test it without a faulty LED.
My first instinct was to hook it up to a power source without a limiting resistor, but every time I've accidentally done that in the past, something starts smoking...
A failed LED doesn't emit. So it's dark. Put a black heatshrink sock over the led, or black paint, or black tape, and if you sense a change, you're done.
If you really have to have a dead LED in a production environment test, cut the trace to the LED in a sample unit, and randomly insert that unit in the production line. Or, remove the current limiting resistor for the LED. But I vote for the black heatshrink option, because you don't have to have a 'designated test unit', you can simply put the hood over a random item in the test line.
C
Page 3 of the datasheet says that you can jab at your UV photodiode, if colder than 20C, with up to 80mA for 1/10 of the time. Consider the time constant to get average heating within the micrometer or so nearest to the active junction inside the device (sometimes you can see a little square thing set in the middle of the LED clear encapsulation). To a thing so small, a millisecond is much too long a time. You should aim for about 11 or 12 microseconds off (to cool down) 1 microseconds ON 80mA. Custom circuits would be needed for that, and how to make a capacitor-limited on time 5 Volt flasher is beyond the scope of this forum, but I'd do it with a 2N7000 nfet driven with the rising edges of fastest acheivable audio out from the arduino. That might get more UV light out than 15 mA continuous due to amplified spontaneous emission. The surest way to find out is to test it. Oh you wanted to break something instead ?
The most common failure of an overdriven LED, as could be found by giving this one 40 mA steady for one second, is that the light emitting junction overheats and rather boringly makes a lot less light next time. Don't waste your 375 nm good diode trying that. Get a bag of 20 cheaper visible LED and overdrive one of them instead. Amber ones turn a redder shade of amber whilst very damagingly hot, and if you are lucky and good at pulling the plug very quickly on spotting the very slight colour change then you might even get away with it and it will work as amber after it has cooled down for a second or two.
Using a resistor and as common a Voltage adjustment as a rheostat or potentiometer, you can set up to measure forward current vs forward voltage (middle left of page 2 of datasheet) at two analog input pins such as a0 a1 as quickly as you can sweep from 0 to enough Voltage to give you 15 mA. It would be sensible to set up a light detector and op amp circuit to give you about 0 to +5 Volts at double the expected light output. That uses a3. A similar op amp circuit with probably a resistor value changed and a more expensive UV photodiode for 375nm can go to a4 so that you never need to fiddle with disconnecting the silicon photodiode for visible light. Arduino has enough pins that you can just do that.
Break a few ambers by somewhat overdriving them to test them this way, before vs after seeing how orange you can get it before the junction diffuses. You should find that the "V I" chart goes not-quite-right while hot and worse after an irrepairable overdrive accident. The "V I" chart of an ideal charred lump of charcoal is an uninteresting straight line through (0,0).
Now you will have a stash of a few odd failed LED's which don't light up properly, and can be neither fully open circuit nor fully short circuit. What would you need to distinguish between those ones and undamaged pristine originals ?
Breaking the connection (or removing the current limiting resistor) will make an open circuit.
And with the current-limiting resistor in-place you can usually short the LED and the resistor will still limit the current but if you're driving it directly from an Arduino you should make sure the resistor limits the current to 40mA or less,
... I don't think I've ever killed an LED but most semiconductors will short when they "burn up" but if you connect it to super-high current source like a car battery or an AC power outlet it will probably "blow open" but it might also literally explode so I wouldn't do that without eye protection and other precautions.
I have seen some LEDs DOA from the factory and they are usually open or internally reversed (so they work backwards). I've never seen one shorted from the factory,