Odd LED failure

Other Hardware General Electronics
1

I say odd because although I've seen (and made) lotsa LEDs fail, I never have seen this one.

I made a device for a friend, part of which was a little status report board attached to the Arduino on which were mounted two common anode RGB LEDs.

A request to allow making them be off, just off, rather than reporting the status was made.

Lazy I was, I just threw a switch onto the board to cut off the Vcc. That worked well for months.

Now the LEDs would not turn off, was the claim. I had to see the impossible for myself, and I won't tell you how long it took me to find the fault.

Fortunately the status board was detachable. I found that if the green element of one of the LEDs was given HIGH from the Arduino board, other LED elements with LOW were being illuminated, not super bright but annoying, especially since it shouldn't be possible.

I will change the code. A switch on an input and logic to just turn off the LEDs accordianly.

Here's my finger drawn schematic. The little triangles at the bottom go to Arduino outputs, where LOW would turn on the LED on that pin and HIGH would turn it off. If the Vcc switch is closed.


image
image1312Ă—963 135 KB


I made a temporary fix that will probably be there as long as this worked in the first place by inserting a regular diode in series with the kinda broken green LED element.

I do not know the name for this mode of failure.

a7

2

Are green LEDs susceptible to reverse bias leakage? (like an avalanche diode).

3

ISTR increased leakage as a symptom of mishandling (don't recall, possibly a static discharge symptom). Don't think it was color-dependent, just given as a reason for handling precautions.
But it's a long time ago, and with LED technology that has most likely long since been improved upon. Vendors/suppliers generally still treat LEDs like other static-sensitive semis; but then again, I've received wirewound 2W resistors in antistat baggies, so...

1 Like
4

Perhaps the issue arises because the rgb elememts are on a common substrate.
Depending on its composition the green LED may need a much higher forward voltage than the red & yellow.

5

It would be useful to see a more detailed diagram to see if there is a leakage path .
I would test the led away from the board to check it .

Also does your code set the pinMode ??

Also if the leds had a power feed whilst the Arduino was off , you could have damaged one of the pin driver circuits by back feeding it .

6

I may have mishandled or abused the LED, but not to the point that it did not function to my plan in the circuit for months. the thought makes me a bit nervous as the implication is that I predisposed it to early failure. I don't think any are in such a lazy kludgy circuit, however.

In the deployment, the LEDs were always on output pins, between Vcc and the pin each with a series current limiting resistor. No pinMode tricks, just set to OUTPUT.

I didn't think to check the values of those resistors, but I am reasonably confident that I aimed for the 7 mA that makes plenty of light these days.

When I was testing, the circuit I drew is what I had in front of me.

No connection made to the point marked with the circled plus (Vcc when attached to the Arduino), .

5 volts from a USB wall wart at the far end of the resistor on one of the green elements and ground on the far end of any of the resistors on the far end of the elements of the other RGB. One by one lighting up the LEDs in the other package. My detailed notes from the experiments (cough!) fail to indicate if the other LEDs in the same package lit up.

That's drawn in thick brown ink above. + and ground symbol. I've just noticed I drew an RGY LED, the yellow should be blue. I guess I was thinking about traffic light FSM controllers. :expressionless:

I do not know over what time it went from OK to a certain kind of not OK. In any case, something is different about it now. So I call it a failure.

Or blue, haha. I measured the Vf and found red to be about 1.9 and both green and blue about 3 volts, with 9 mA from an LED tester running through.

I can't say what percent of the time the light switch was open to suppress the light, nor can I say which LED elements were on, or off off (HIGH and switch open) being stressed by what is effectively a large reverse voltage, for what kind of percentage of the time. Typical diodes have an absolute maximum of 5 volts reverse, which I never thought about and may not have even realized I was visiting upon the LEDs in some common circumstances during routine operation. I avoid maximum ratings.

I placed it back in service with the hacky fix, almost hoping that another element fails, placing for no good reason my money on the green element of the other LED.

a7

7

None of this precludes anyone from walking up to your device and discharging a polyester slack's worth of charge directly into the device, does it? I have noticed across the years that the extremes of static-discharge-prevention we went to back in the prehistoric 80's (my first boss lost his sh.. regularly about careless lab staff), have become a thing of the past in all but the most conscientious of circumstances; granted, devices are more robust, too, and we probably know better how to design for "survival in the jungle". We got a new distribution of office chairs shortly before I left, and they 'lit up your life' if you slid off of them when getting up, then touched something grounded. Quite the surprise!

But it's only a possible, I grant you. I wasn't suggesting you were careless, really.

8

Haha, no offense. I am careless and largely care free.

So yeah, that's a good thought. With the switch open, the diodes make a floating circuit between pins that were only outputs, so were in fact then just an invitation waiting to be accepted.

No one who might be able to say that happened is talking…

So the hack was just dumb and I have already copped to being lazy. One for the books, I can only hope this sticks and I spot similar accidents waiting to happen before anything is wired up.

I made the device before I went all in for smart LEDs. If that had been in the board, I def would not have solved the problem by opening their connection to Vcc.

a7

1 Like
9

Around here, the number of people in any given 1000 who would even know enough to care, remember, or think they should/shouldn't admit to it, is vanishingly small. Static is an annoyance, a reason for hairspray and dryer sheets, not something you want to keep away from your valuable electronics.

10

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.