Have been running this clock for almost 9 months now:

Archive FastLED Archive
1

Have been running this clock for almost 9 months now:
http://www.instructables.com/id/Infinity-Mirror-Clock/
But it started getting stuck erratically a couple of weeks back. The clock would run for a few hours and then stop - the LEDs would get stuck. I tried everything to troubleshoot them - going so far as to replace all the components (except the LED strips one by one). But to no avail.

Long story short - I have narrowed down to the following:

  1. The first LED in the code remains ON at full brightness at all times
  2. This caused the first LED to overheat and stop transmitting data
  3. The solution is to change the white colour to (50,50,50)
    It’s working now - but will update if there are any issues.

Thought I’d post this if anyone else is having similar problems with WS2812B strips.
http://www.instructables.com/id/Infinity-Mirror-Clock

2

Wow. You may have the first documented case of this, but I am certain it’s happened before, but maybe just not as clearly diagnosed.

Also, we’ll call this one more reason why folks should consider using something lower-power than 255,255,255 for white. If you want a similar apparent brightness to other full-brightness pure hues (eg red=255,0,0), I’d choose a value between 85,85,85 and 100,100,100, depending on the application. If you want your white noticeably brighter than the pure hues, you might want to go as high as 128,128,128 or so. It’s all subjective and up to you as designer and artist to choose the best option.

But it’s super interesting to learn that (1) an overheated pixel will stop passing data to the pixels beyond it and (2) a pixel can be made to overheat just by 255,255,255 for an extended period in an enclosure!

But let me also clarify one other thing: after you let it cool, and reprogrammed it not to run so hot, the troublesome pixel fixed itself? Or did you also have to do some hardware repair/replacement?

Thank for the update on this project, and for sharing what you’re learning.

3

I didn’t change the troublesome pixel. The clock is built in a way that replacing the led strip is extremely difficult.

I guess there wasn’t any permanent damage. It’s been working for the last 3-4 days - so fingers crossed.

4

Yeah- replacing the internals there would be quite nasty. Well, keep us posted!

Now that I think about it, I did one much smaller installation piece in a (sealed) shadow box and after about …uh… three years (!?!) I started seeing some of the LEDs and driver chips start flaking out.

I guess all those ventilation holes in electronic device enclosures are there for a reason – and the reason applies to high-power LEDs as well as to everything else.

5

First law of designing anything:
It will break.

First rule of designing anything:
How the frack will I repair this damn thing?

6

True. You live and learn.

7

Hi @Dushyant_Ahuja ,

This is very similar to what happened to me with WS2812b LEDs individually mounted on small PCBs. However, you problem occurred after months of good operation while I experienced this immediately.

My problem also disappeared when I simply reduced the MAX brightness specially on the white colors.

I concluded then that I must have speed up this effect with heat application while soldering but remained convinced that the WS2812b package itself was fragile to heat build-up.

I asked around on this forum if anyone experienced similar behavior with 100% brightness white color settings for a significant time but never really got any confirmation !

Your experience simply strengthens my conclusion about the heat sensitivity of the WS2812b.

By the way, excellent project !!!

My best regards, JP

8

Thanks @JP_Roy

9

I’ve had a few issues like that with some bad LEDs on WS2812B strips after a while of usage one of my channels would lock up and only the first LED would work. I attributed it a bad LED that is on the verge of dying it shows after it heats up for a little bit.