WS2811 Pixels Keep Failing - What's Going On?

[AlexTheCat123]

This is my first post here, so bear with me if I leave out any important information that you guys might need to know!

This is my second year doing a light show and I'm having some really strange problems with my WS2811 pixels. They just keep failing at a very high rate of several per day. Most of the failures involve a pixel completely dying and not passing any data on to any of the rest of the pixels in the strand. A few of the failures have been where the pixel still passes data on perfectly and the red and green LEDs work fine, but the blue LED in that particular pixel is stuck on (it's always the blue one for some reason). The first problem is much more severe than the second one, but it would be nice to fix them both. I keep on splicing out the bad pixel (and the one before it) and replacing them with new ones, but other pixels just keep on failing and I've probably replaced at least 50 of them so far. I'm not really sure what's happening here because everything worked perfectly last year without a single pixel ever dying. The only thing that I've changed from last year to this year is upgrading the wire from my terrible solution of twisting three 24 gauge hookup wires together from last year (don't judge me too hard for that; I'm a broke college student and I was just using what I had) to some much better 18 gauge three conductor wire, but I can't think of any reason why that would be causing the issues. And by the way, I'm using an F16V3 controller in case that matters. I'm getting pretty concerned that I'm just going to have to cancel my show for this year if things keep on dying, so I would really appreciate some help with this!

Here are the things I've tried so far (with no success):

• Adding a 330 ohm resistor to the data line of each channel right as it goes into the start of the pixel strand.
• Turning down the power supply voltage from 12V to 8V (the terrible wire from last year was dropping the voltage to 8-ish volts by the time it reached the lights, so I wanted to replicate that with the new wire to see if that was somehow the problem).
• Checking the pixels for signs of water damage. They're sealed very well in a block of epoxy, so I don't think that water or humidity is the culprit.
• Checking the power supply for excessive ripple with my oscilloscope. Things look totally fine there.
• Probing the data signal with the oscilloscope. The voltage looks a little high (the scope says that the max is around 5.6-5.8V, depending on the channel) but I don't see why this would have changed from last year given that I'm using the exact same controller.
• I've tried to see if there's a pattern when it comes to the pixels failing (like it always being the first one in a strand or something). It seems to be completely random.

And here are some links to the parts I've used that I think could be relevant to the issue:

• WS2811 Pixels
• F16V3 Controller
• Power Supply - I don't think that this is the exact one that I used (I just used one that I had laying around from another electronics project), but they look the same and have the same current rating.
• New Wire

Thanks for reading and please let me know if you guys have any suggestions!
Alex

[MikeKrebs]

50 out of how many?

My first thought was that output voltage may have been out of spec but you did check for that. 11.8 v should be fine. Actually the led you linked say they are regulated in the q&a so any voltage should be good. the description says they are silicone filled and not epoxy. that makes me wonder if you have leakage of moisture around the wires. I have some silicone filled from ray that are less silicone spatula and more silicone caulk. Very suspect on their ability to to withstand movement.

Have you played with cutting any of the dead ones open? would be curious if power is the issue? in another thread someplace someone mentioned the blue and V- (ground) are physically close together on the chip. if you had some kind of leakage there maybe?

Looks like the led come with preterminated ends so no chance of connecting power reverse?

[angus40]

This is my first post here, so bear with me if I leave out any important information that you guys might need to know!

This is my second year doing a light show and I'm having some really strange problems with my WS2811 pixels. They just keep failing at a very high rate of several per day. Most of the failures involve a pixel completely dying and not passing any data on to any of the rest of the pixels in the strand. A few of the failures have been where the pixel still passes data on perfectly and the red and green LEDs work fine, but the blue LED in that particular pixel is stuck on (it's always the blue one for some reason). The first problem is much more severe than the second one, but it would be nice to fix them both. I keep on splicing out the bad pixel (and the one before it) and replacing them with new ones, but other pixels just keep on failing and I've probably replaced at least 50 of them so far. I'm not really sure what's happening here because everything worked perfectly last year without a single pixel ever dying. The only thing that I've changed from last year to this year is upgrading the wire from my terrible solution of twisting three 24 gauge hookup wires together from last year (don't judge me too hard for that; I'm a broke college student and I was just using what I had) to some much better 18 gauge three conductor wire, but I can't think of any reason why that would be causing the issues. And by the way, I'm using an F16V3 controller in case that matters. I'm getting pretty concerned that I'm just going to have to cancel my show for this year if things keep on dying, so I would really appreciate some help with this!

Here are the things I've tried so far (with no success):

• Adding a 330 ohm resistor to the data line of each channel right as it goes into the start of the pixel strand.
• Turning down the power supply voltage from 12V to 8V (the terrible wire from last year was dropping the voltage to 8-ish volts by the time it reached the lights, so I wanted to replicate that with the new wire to see if that was somehow the problem).
• Checking the pixels for signs of water damage. They're sealed very well in a block of epoxy, so I don't think that water or humidity is the culprit.
• Checking the power supply for excessive ripple with my oscilloscope. Things look totally fine there.
• Probing the data signal with the oscilloscope. The voltage looks a little high (the scope says that the max is around 5.6-5.8V, depending on the channel) but I don't see why this would have changed from last year given that I'm using the exact same controller.
• I've tried to see if there's a pattern when it comes to the pixels failing (like it always being the first one in a strand or something). It seems to be completely random.

And here are some links to the parts I've used that I think could be relevant to the issue:

• WS2811 Pixels
• F16V3 Controller
• Power Supply - I don't think that this is the exact one that I used (I just used one that I had laying around from another electronics project), but they look the same and have the same current rating.
• New Wire

Thanks for reading and please let me know if you guys have any suggestions!
Alex

I would contact the vendor and request a refund .

[1pet2_9]

I'm the guy who documented the blue and ground shorting, and I, too, am getting a lot of failures. My mega tree is driving me nuts, and today I just fixed a bad pixel that was US-made, different vendor from the others, not on the mega tree, was never exposed to hot weather, 18 AWG, 12V, has seemingly good sealant, xconnect, and is almost 200 pixels away from the nearest supply. It makes no sense. I don't see why always-blue would happen, but not always-red, when the red diode has a lower threshold voltage. The data-failure-to-pass problem is easy:. the DI and DO pins are right next to each other on the chip. If either gets an open OR short, they're gone.

I would love to get to the bottom of this myself. The one thing I see is, don't let pixels point downward in the rain. That's what my American pixel did (but not the mega tree pixels).

[MrDudeMan]

Last few years I also have extreme failure on bullets...
It's nothing we are doing wrong.
First year of newly bought pixels will have normal failure rate, but subsequent years with those same pixels are are off the charts failure rate.
I have pixel batches that been running 7 years and only had to replace 3 pixels out of 7 years. These ones though were talking guaranteed 10-15 die each year. Also seems to be increasing the failure rate of them.
These new ones are a lot brighter, im guessing it has something to do with it.

[MartinMueller2003]

Blue pixels actually use more current and dissipate more power than the other colors. That may be a contributing factor to the blue dying rate. As for general pixel failure, I just threw 2000 of the latest pixels from Ray Wu out of my show. They had a huge failure rate and were super sensitive to PSU issues.

[1pet2_9]

Has anyone come up with a good, quick way to peel the broken ones open? I can look. If failure is associated with high power--not low threshold voltage--then my theory would be electromigration/hot electron. Blue stuck-at 1 suggests a blue/vdd short (and not short to gmd, too), but I don't see blue near vdd at the PCB.level. Mainland Chinese IC foundries have very high defect rates. Maybe the failure rates can be traced to Worldsemi vs. the other foundries they outsourced to. Or even a particular chip design.

[joeytx]

I bought some of those LEDs and did a review, but they were the 100 ct strings from ALITOVE. Out of the 5 3 had no problems but 2 of the strands you could see cold solder joints on the wire connections. I ended up cutting out about 20 to get a string that worked. I couldn't trust them so I tossed the strands that failed, and put the 3 that did on a tester an ran for 3 days before using them. We will see if temp and humidity effect, but I think it was just poor soldering.

[1pet2_9]

We would be able to recognize poor soldering by ripping the broken pixels open.

I believe the alitove pixels are a different IC.

[AlexTheCat123]

After reading all of these replies and seeing that so many people are having the same problem, what should I do? Should I just try switching to WS2812s or something to see if I have better luck with the manufacturing quality there? WS2813s and WS2815s look pretty tempting too because of the redundancy of the data lines in case a pixel fails.

And by the way, I dissected a few of my dead pixels to see if there was any visible damage. It took a long time to remove all of the epoxy or silicone or whatever the stuff is, but there was no visible damage after I got it all off. No bad/cold solder joints or anything, so it's definitely some sort of failure internal to the chip.