Teensy v3.2 + octoWS2811 + WS2812b flickering

Status
Not open for further replies.
C

Cyanide

New member
Dear all,

Recently I installed 2 WS2812b ledstrips (60leds/meter) behind my television to simulate some sort of amiblight. Each led string consists of 120 LEDS and both strings were cut from the same 5m strip.

I use a 5V 20A power supply to power the leds. Each strip has the +5V lead soldered on the beginning of each strip as well as the ground connection.

The ledstrips are controlled through a Teensy 3.2 and octoWS2811 combo. As I have only two strips, only 2 out of 8 possible channels are used.

I used the standard code on the teensy: https://github.com/PaulStoffregen/OctoWS2811/blob/master/examples/Adalight/Adalight.ino

The adalight is fed through a program called hyperion [https://github.com/hyperion-project/hyperion/wiki] which runs on my pi2.

This setup has been working flawlessly for 1/2 year; but now one of the two strips starts flickering and shows wrong coloring.

This does not happen all the time and seems random. I checked every connection on the strip and resoldered every wire; but this did not solve the problem.

The strange thing is that the other led strip is doing fine and works perfectly. The strips are being driven by the same Power Supply; so I don't think this is the problem. The output of my Power Supply is about 5.25 volt, and I don't have a massive voltage drop at the end of the strip. The voltage values seem good.

I use thick wiring for the ground and +5V leads, but the signal lead is quite thin, as is the case in a normal cat6e network cable, which I use to carry the signal lead.

What could be causing these issues?

Many thanks for any help.
 
The LEDs each have a controller, and the cheaper strips have a history of failing with age due to heat damage to the die which corrupts data to all downstream LEDs. Quick but boring option is to try swapping out the strip.

If you want to test in situ an all white load test and then scrolling a dark pixel through may help identify if there is one or more problem children in the strip. If that's the case then you can cut out the offender and splice the remainder back on, depending on failure mode planting it at the end of the strip is do-able but possibly not worth the effort.
 
Indeed, the reality is most LED strips are made with low quality manufacturing. The heat from the LEDs also tends to thermally expand the flexible strip, which stresses the (usually low quality) solder joints between the LEDs and strip. These problems are so prevalent that they made a WS2813 chip which automatically routes data around defective LEDs, so one failure doesn't take out the rest of the strip.
 
Thank you for your comments. I replaced the led strip and the problem has been resolved. It was indeed a faulty led somewhere in the strip. I have yet to identify the culprit, because i need some more time to design a test pattern with the processing app. I'm not yet that familiar with that code. It would be great to have a worst case- scenario test pattern script so we can use that to check our led strips. I ordered the strip in China and paid 25€ / 5 meter (60 led/m). In Europa they are far more expensive, but maybe the quality is also better; or is that not the case?

Concerning the WS2813, that one seems promising, but it has four connections instead of 3. Is that strip also compatible with OctoWS2811?

Thank you.
 
Quality can be better but finding them is always tricky since as you found out the problems may not show up till 6 months later. As far as I know the WS2813 use the same commands, just have a jump over in and out signal that is used if a given chip doesn't receive anything from it's primary port it can fall over to the bypass from pixel n-1.

You just wire the bypass to your normal drive pin and leave them to it. This will not fix the dead unit so that will stay black and won't help if you have corrupt data from some more obscure failure mode that you seem to have met.

Possibly overkill for your use case since replacing the strip and/or doing surgery isn't a big hassle. If you are building a jumbotron then paying extra for a device that turns a dead string into a dead pixel may start making a lot of sense.
 
Status
Not open for further replies.
Back
Top