Well I need to reopen this thread to see if anyone can point me in a direction to resolve the issue.
The issue remains the same:
- If I power from both ends I get consistent color for all colors from end to end, but randomly the strip will go into a “confused” mode. You tell it fill purple and you get five or six sections some purple, some green, some rainbow (different colors in each pixel for a a foot or two). The confusion never seems to be the same, but rather totally random. I can than either disconnect the power to one end, reconnect the power, send the fill color again and usually get the desired results. Probably a 60% success rate so far. Or, as we discovered today,
when it gets confused I can fill the led with Red (255,0,0). If the entire strip responds to Red I can than resubmit my desired color (purple, gold, whatever) and get a 90% success
rate. If the Red flood does not work the first time, do it again and it usually works. I get almost 100% on two tries. Doing the same with Blue (0,0,255) or Green (0,255,0) does not seem to reset the led strip.
- If I power the WS2812B led strips from one end only I do not get the “confused” led lights, but I cannot get consistent color down the entire strip for some colors. Red, Blue and Green provide full color for all 404 leds (22 ft). But a color like Purple will be purple for about half the strip and than start transitioning in color until the far end is blood orange. Another unusual reaction is that a multicolor display can have different results depending on which color is background and which is foreground. I can set the background to Green and than have the animation play in Red to get consistent Green/Red throughout the animation. But if I set the Background to Red and the animation to Green the first half of the led strip is Green/Red, but the second half is Green/Black. On the other hand I can do Blue/White or White/Blue and get consistent results throughout the entire strip.
I have searched for solution online and so far not finding any answers.
It does not appear to be wattage. I have used 5v 10a and 5v 15a power supplies with exact same results. I even tried a 5v 5a, but I worry that that one is insufficient to run the strips.
I have tried running both front and back ends from the same power supply. Same results. I have tried separate power supplies on each strip, front and back. Same results.
It does not appear to be total number of leds. I am running two parallel strips of 404 leds. If I run only one strip or both results are the same.
I have used different WS2812B led strips with same results.
It does seem to be have some relation to number of leds. If I limit the strip to 300 leds I have not had the issue occur as of yet, but only 3/4 of the piece is lit up which is not an acceptable implementation.
I have tried resetting the software (exit and rerun). Sometimes that works sometimes it does not. (50% success).
Full power off the Raspberry Pi and the associated AllPixel is a 100% success, but this is not an acceptable means for use on a publicly accessible art piece/commercial structure.
I do have a 100uf cap on each power line. If I run with or without the 100uf makes no difference.
It’s as if the 5v coming from both ends locks up the chips and causes them to pick a random color. If you can just clear the line by disconnecting power or forcing Red it resets the line and starts working again.
Any ideas?
Harold
ps I should have some pictures and videos next week at the grand opening to share with you. It has really turned into a very cool project and has a real “Wow!” effect on everyone that sees it.
···
On Tuesday, December 12, 2017 at 4:08:29 PM UTC-6, Harold Breeden wrote:
A new issue on my project involves the power supply to power the WS2812B led strips.
I am running two (2) strips of 404 leds per strip in parallel. Of course this requires a 5v power supply so I am using a 5v 10a power supply.
The strips seem to run nicely in this arrangement for most colors, but some colors get in trouble. I can red, blue or green and have consistent color all the way down the strip on bot strips. But if I run purple I get purple about 1/3 of way ten start losing color until at the end it shows a red-orange color, apparently losing it’s blue color in the process.
I have tried running just one strip at a time instead of the two in parallel and get exactly the same results.
I had read that I could power the strip from both ends. So I mounted a 5v 10a power supply on both ends of the strip. That resolved the color integrity issue.
But it introduced a new issue. Transitioning from one pattern to another or doing a reconfigure of the AllPixel driver (such as change number of leds) will result in really weird patterns.
Example: set strip to all red and I get all red. Than set it to all purple. Instead of getting all purple I get sporadic colors along maybe 50% of the one strip and 30% of the other. I set it to all purple again and both strips go all purple.
If I disconnect the power supply at the far end of the strips than when I transition from all red to all purple I get 100% of all leds, but I also get the purple on one end to red-orange on the other end. But never the partial set of leds.
Is this due to some result of the AllPixel?
My first reaction is no, it cannot be the AllPixel because the AllPixel is using Data/Ground not the 5v power supply.
Any ideas as to what I am dealing with?
