LED Noise

Status
Not open for further replies.
M

mlathrop

Member
Hello,
I wanted to see if anyone else has had issues with LEDs flickering odd colors when using the OctoWS2811 adapter. I just started seeing this issue after rewiring my panel. Power runs up the back of the panel on both sides and the center (photos below). The flickering occurs all over the panels (it isn't isolated to one data line etc). Anyone have any ideas? I want to get this panel perfect before I do the other 49!

IMG_4507.jpg

-Matt
 
Unsure how much you have already done with it but in order:
Can you turn each pixel on one at a time without trouble? Idea is to at minimal power consumption look for interference type noise issues or anything string related
If single pixels are reliable then problems or most likely power related and it becomes a case of chasing down if problem is peak power capacity, or the slew rate of the power supple (how fast can it reach it's rated capacity). Be careful adding random electrolytic caps in this sort of quest since you can end up with resonant circuits but the idea is to provide enough capacity to keep 5V stable as the pixels change the current draw.
If single pixels (or all pixels at RGB 3,3,3) still flicker than it's time to hunt noise on the data lines. Including, does it flicker if you write once and then leave the pixels cached at the current level?

First up, is anything else connected to the Teensy, and do the code examples work? I ask since Octo uses 4 pins for it's own purposes and I got caught when I added a last 'fail' LED on pin 4 since even declaring that as an output mucks with the DMA colour quality, but interestingly doesn't prevent it working completely.

A handy option for any software side doubts is to use the neopixel library working through each of the eight strands to bitbang a test pattern.

If you have a scope then it's time to go checking wave shapes and levels looking for glitches or out of spec leading edges.
If your using a meter the things to check for are the grounding of the Teensy to your power supply, and if Teensy ground is actually the same level as the pixel ground level, in the same line confirm that the Octo 5V rail to the level converters is the same level as the strand 5V rails.

GW
 
SO I figured it out and as a computer scientist it turned out to be one of the more interesting bugs I have seen.

It turns out to be a bug in the processing 3.0 alpha! The following line was seemingly randomly changing the output values on a few pixels.

ledImage.pixels[j] = color(red(current), green(current), blue(current));

I reimplemented it using bit shifting and everything works fine! Very odd bug and would love to know what goes on!
 
It took me forever to find! It wasn't until I loaded other programs onto the teensy that I noticed they didn't have noise.
 
Unfortunately this is my own code. I submitted a bug to processing and I will be posting my full code when it is done. It's a pretty nice re-build of movie2serial. It uses a screen capturer to capture any part of the screen then processes it and sends it out. I have it working with 25 teensys simulating the work of 20k pixels (I don't have all the pixels in panels yet). I ended up having to use threading to make it work and I am getting around 54fps.
 
Status
Not open for further replies.
Back
Top