LEDs losig brithness at a different rate, am I damaging them somehow?

byteofthat:
Could a voltage spike situation when being plugged in be the problem? Would it damage the LEDs in such a fashion?

No...

Is it only one bank of LEDs that is failing? The one connected to power directly (although via a 220 Ohm resistor)? If so is it connected to 5V or possibly higher?

How many LEDs are you running in total, and do they all go through the voltage regulator on the Arduino?

ChilliTronix:
No...

Is it only one bank of LEDs that is failing? The one connected to power directly (although via a 220 Ohm resistor)? If so is it connected to 5V or possibly higher?

How many LEDs are you running in total, and do they all go through the voltage regulator on the Arduino?

It just seems to be the LEDs that make up "KLUKKAN ER". I have one LED for each letter in that phrase. Each LED is connected to the power source directly, and each has its own 220Ohm resistor.

None of the LEDs go through the voltage regulator on the arduino. I just use the arduino to drive the shift registers and transistors arrays. I use one pin per word on the transistors array to sink an entire word to ground.

The power source is a 5 volt power supply, and I have measured it a few times, it's 5 volts. I have the positive and negative leads soldered to my prototype board, and then I run my connections from there. I have a decoupling capacitor at this point.

Does this make sense?

It does, but the only issue I can think of then is a dry joint in the circuit for that word.

ChilliTronix:
It does, but the only issue I can think of then is a dry joint in the circuit for that word.

Hmm, well I will take a look, That would be quite odd if that was the case. I will come back with info as I have it!

Re melt the joints just in case. I can't think of another explanation at the moment.

Hmm, tried using alligator jumpers to see if it was a connection problem and no change at all. I am going to try swapping one of the bulbs and see if that changes anything. If I notice a difference, I am not sure what I will do from there, because I still won't understand the initial issue.

Yeah... I don't know what's going on, but LEDs don't normally wear out or deteriorate.

I work in electronics and I've seen bad LEDs from the factory, but I don't think I've EVER seen an LED die in the field or go dim after some time. And of course, resistors don't go bad or wear out either. (If a resistor is burned-up, something else failed first or it was a bad circuit design.)

Where did you source your LEDs from?
Cheap eBay stuff from the far east often comes out of the reject bin. It could be a badly doped batch you have.

Very nice work by the way.

I thought your clock looked familiar! it is, I saw it on hackaday (I think).

if it's not a voltage spike, and it is only several rather than all, then maybe they were bad to begin with and by that I mean they're being driven too hard because they can't handle the current supplied...

Hi,

Can you please post a copy of your circuit, in CAD or a picture of a hand drawn circuit in jpg, png or pdf?
And spec of the LED's.

Thanks.. Tom....... :slight_smile:

Simple troubleshooting step would be to swap around your darlington arrays and shift registers. Make sure they're seated properly as well; perhaps your girlfriend set the clock down too hard or the heat/cold from the sunny window caused them to work themselves out.

Regarding the RGB LED strip (WS2812B) in the clock, this link is a good reference - look at the warning section 1/2 way down the page.

@Chagrin: The shift registers/transistor arrays wouldn't have anything to do with it, considering the effected LED's are not connected through them, but are connected directly to the power and ground.

Maybe @cjdelphi is correct. I replaced one LED, and it was back to normal brightness. I had then just changed them all out for the effected letters and it seems to be working. I am going to see what happens after it runs for a while. Hopefully it was just bad LEDs. Time well tell!

Thanks for all the advice, everyone.

Did you re melt the joints as a test first? If not then the problem could still have been a dry joint issue... though it would be odd if it affected all in one run of LEDs and not randomly across the board.

White LEDs fade with time, the datasheet for the particular device should be consulted.

White LEDs change colour balance too - the basic issue is that the fluorescent yellow
gel in front of the LED changes with time. The LED is typically blue/indigo/UV and the
other colours are generated by fluorescence. Its hard to engineer a fully stable fluorescent
dye - note that ambient daylight could be aging the dye more than the blue from the LED
itself even - only the manufacturer will have data on these processes as they are a property
of the dye.

The actual LED emitter itself can age too, but this is likely to be a smaller effect due to
crystalline semiconductors being more robust than organic dyestuffs.

ChilliTronix:
Did you re melt the joints as a test first? If not then the problem could still have been a dry joint issue... though it would be odd if it affected all in one run of LEDs and not randomly across the board.

I did with a couple of them, and nothing happened. I figured as much when I tried using alligator clips to see if a bad connection was the problem.

MarkT:
White LEDs fade with time, the datasheet for the particular device should be consulted.

White LEDs change colour balance too - the basic issue is that the fluorescent yellow
gel in front of the LED changes with time. The LED is typically blue/indigo/UV and the
other colours are generated by fluorescence. Its hard to engineer a fully stable fluorescent
dye - note that ambient daylight could be aging the dye more than the blue from the LED
itself even - only the manufacturer will have data on these processes as they are a property
of the dye.

The actual LED emitter itself can age too, but this is likely to be a smaller effect due to
crystalline semiconductors being more robust than organic dyestuffs.

I understand this, but it doesn't explain why they faded so fast in comparison to other lights on the clock that were on almost the same amount of time. The word "MINUTUR" was on 83% of the time "KLUKKAN ER" was. Considering the clock has been in operation for less than a month, after waiting for a week and not seeing the same behaviour in the word "MINUTUR", I found it unlikely that it was just a "wear and tear" issue.

I would be interested to see how it is in another month.

byteofthat:
@Chagrin: The shift registers/transistor arrays wouldn't have anything to do with it, considering the effected LED's are not connected through them, but are connected directly to the power and ground.

And a resistor I hope.

It needs to be a higher value resistor than that connected through the transistor because there is more voltage across those than the transistor ones.

Grumpy_Mike:
And a resistor I hope.

It needs to be a higher value resistor than that connected through the transistor because there is more voltage across those than the transistor ones.

Each LED has it's own 220Ohm resistor.

ChilliTronix:
I would be interested to see how it is in another month.

The word 'MINÚTUR' gave out, it has now sunk in brightness. I wonder if I am simply running too my current through the LEDs? I would figure 220Ohms would be enough. At 5 volts that would be ~22mA, but assuming there has to be some voltage drop, it must be even less than that.

Any more ideas now that the effect seems to be spreading?

Regarding reply#14:


Right at the power input of the led strip, what value of capacitor did you connect?
I would recommend anything from 220µF to 470µF.


Right at the led strip and connected in series with the data line, what value of resistor are you using?
I would recommend 220Ω, 270Ω or 330Ω.