LED fails after a couple of minutes

I have a string of LEDs in series, in parallel with 3 other equal strings, and it fails (goes dark) after a couple of minutes (the other strings are fine)!
I watched the current with a meter in series, and it looked normal until just going to zero as the string went out.
I tried 1/2 the current with an additional resistor (from 20 to 10ma), and, interestingly the LEDs still looked about the same...until, again they went out!

Anybody have any idea what might cause such behavior? I'm thinking perhaps I have one defective LED in the string, that maybe gradually warms up and fails?

Thank you!

Each string of LEDs should have a resistor to limit the current for that string; it's not a good practise to parallel LEDs without the resistor as they won't necessarily share current equally.

Normally - for me anyway - if an LED fails, it doesn't come back on, even after cooling down.

What leds are you using, what voltage are you using.?

Is the voltage DC or AC, is the voltage filtered.

A led will only handle about 5 volt reverse voltage, and if the supply has ripple greater than this it can kill the leds.

A meter will often not show the peak voltage and the peaks can be high enough to kill the leds.

I went through this some years ago with led garden lighting as they kept blowing the leds, the suppliers answer was ...quote...led technology is not good enough yet...what rubbish!

I took their power supplies and put a CRO on the outputs and a 12 volt supply had ripple as high as 35 volt, with adding some pre filtering this was reduced to a smooth 15 volt, and end of problem, but the meter only ever showed 12 volt before filtering.

The morral is the problem might be caused by what you can not see, and more information on your setup will help us find the problem and offer solutions.

Pete.

Wow, thanks for all the replies, folks!

@dougy3,

Yes, I am indeed using resistors on each string to limit current. And, yeah, I think it's really weird that the string fails after a couple minutes, yet next time I plug it in (well, after waiting several minutes at least), it repeats the behavior, i.e. the led isn't (yet) "blown"!

@SABorn

Okay, the LEDs I'm using specify 3.2V typical, 100ma max, and 5V reverse, described as "1/2 watt".
I have six in each string, powered by a supply from some Altec Lansing computer speakers, "class 2", rated 15VDC @800ma, though without a load it measures 19V. I'm using 150 ohm resistors, which keep the current in a low range of around 15ma, and the LEDs appear perfectly bright.

Again, the other three strings, wired identically, work just fine. I'll probably just disconnect each LED in turn in this string to test separately, probably just powered by a battery.

(sorry if this is a repost--I got impatient awaiting "moderation approval" some 12 hours later--despite having no links or email addresses)
Yes each string has current limiting resistors.

The LEDS are rated as follows:

Forward Current - 100mA max
Forward Voltage - 3.2 typ, 3.4 max
Reverse Voltage - 5v

There are six in the string. The power supply is from an Altec speaker system, rated at 15V DC at 800ma, second class. It measures 19V with no load. I've inserted 150 ohm resistors, resulting in currents typically only of around 15ma (the LEDs appear perfectly bright with that).

The failure "recovers" after a few minutes rest unpowered, then it recurs.

Funny, no?

(Hey! Just realized that I had used the "at" symbol before when talking about current rating. Maybe that flagged my message as containing an email address!)

Use your meter to check the volts on the leds, when they quit see where in the string you are losing the power.

Try a 300R or 1K resistor and see if the string still works and if it drop out with the higher resistor, if not then it would imply a current problem coming from a over voltage your meter can not detect.

As i said earlier its the things you cant see that causes the biggest problems, and meters dont all ways tell the truth.

Pete.

You might try resoldering LEDs, resistors, connectors ets.

Hey, just wanted to thank all of you for pointing me in the right direction!
I did isolate the problem LED, and I think it was just a dicey solder connection there. So thanks, RonSimpson, in particular!