Interesting pixel failure mode - what was it?

[k6ccc]

Background, I have been running WS2811 pixels since 2012 so I have a few failure modes.  I had a new one and wanted to point it out so if you ever see this, you might save a little time troubleshooting.

My pixel tree is 26 strings of 100 WS2811 pixels running at 5 volts.  These pixels have REALLY small wires - this is important.  In my case the controller is at the top (behind the star).  One of my strings was working properly for the top 40 pixels and then the next 15 or so faded out dimmer and dimmer until the bottom 40 or so pixels were completely dark.  My suspicion was a voltage issue so I measured the string at the bottom and found essentially zero volts. I next focused on a possible short circuit in one of the lower pixels that was shorting power to return.  To confirm this, I cut the positive wire between pixels 48 & 49.  As soon as I did that, the pixels above that (which had been dim) came up to full brightness.  Unfortunately about the only way to locate the bad pixel is to repeatedly cut the positive wire somewhere and see if the the problem clears (problem farther along the string) or is not affected (closer to the controller).  I finally narrowed it down to one of 4 pixels, but I had five cuts that either had been temp spliced or were still cut.  Since I have several hundred spare pixels for this prop in 50 pixel strings, I just replaced the section that had a bunch of cuts and what I removed will become spares.

The reason that the small wire size in important is that since the short circuit was about 80 pixels from the controller and the very small wire resulted in not enough current draw to blow the fuse.  Also note that if I had opened up the controller box and measured the current for that string, it likely would have been higher than the rest.  But that was more trouble than it was worth (unless it had blown the fuse).

Here is a hint for this troubleshooting, make your cuts starting near the end of the string.  The reason is that if you make cuts closer to the controller, you have to repair those in order to do the next test.  If you cut near the end of the string, you can continue to make cuts without having to repair the string first.

 

[dgrant]

Interesting idea on the troubleshooting. If it had been me, I probably would have started in the middle first, therefore trying to eliminate which part of the string the loss was in, then move on from there. But, you have already considered that and found that your method is faster and less invasive. I'm wondering if its possible to visually detect the failure point to the led prior to the problem-child led?

[k6ccc]

Nothing I could see.

 

[dgrant]

Just thinking out loud here, suddenly crossed my neurons, what if a DC sensing probe could detect the change along the path? I'm thinking it would have to be really extra sensitive but hey, it might work...

[TheDucks]

5 hours ago, dgrant said:

Just thinking out loud here, suddenly crossed my neurons, what if a DC sensing probe could detect the change along the path? I'm thinking it would have to be really extra sensitive but hey, it might work...

A quality IR imager might show a change in current (heat)

[k6ccc]

A DC clamp on amp meter would work, but remember that has to clamp on one wire, so you would have to split the wires apart at each place you want to test.  Doable.

A Thermal imaging camera would likely also work, but I don't have one of those.

 

[dgrant]

Most of the time, a failure point is obvious but not always as per this failure. What we need is something inexpensive that we could possibly pass over the strip or nodes and get an indication if the voltage/current is getting past that point or not. Same would be nice for data and or clock signals. It shouldn't be that hard to develop. There's already line testers for Cat5-6,7 that can determine how far along a cable where there's a fail point. Why not similar but appropriate for pixels? Toggle a known good pattern or constant along the length and have the testing device look for deviations. Someone with more brains than I should be able to figure it out. I mean, I can probably build you a space shuttle in your backyard, but putting together a designer custom circuit for detecting tiny current measurements is beyond my knowledge base. Give me a schematic and I can track things down...lol.