Has anyone run into an issue with WS28xx strip where they become glitchy when running at full white? Basically anytime a fully white pixel is shown that downstream pixels begin to exhibit issues and stop working until a power cycle is performed. I am currently in a fight with our vendor on quality issues (TBD if this is the problem).
However, by decreasing the global brightness (leds.setBrightness or leds.setMaxPower) down to a certain point seems to remedy the problem although at a drastically reduced perceived brightness (on the order of 60%). This is definitely a workable solution though I’m bummed I can’t ever run the LEDs at max brightness with white without exhibiting issues.
Hoping someone else has run into this! I’m open to testing suggestions and can provide more data/photos/videos if that would help.
how many leds and what is your power source and where are you injecting power? my guess is that when you are driving the pixels at full white you are causing enough of a voltage drop that leds further down the line are getting starved, not just in their brightness, but messing with the timing of receiving incoming signals as well.
When I measured voltage drop across the strip the max I was seeing at the end of each strip was around 4.7V. I can build a test rig tomorrow and do some more in depth testing to see if power injection would help with this.
And part of the issue is just voltage drop across the strip and the max amps that the strips themselves can carry - so there’s a point where it doesn’t matter how large your power supply is, you’ll still have to inject powe
Interesting +Daniel. I’ll have to do some testing on running full white and dropping power in at various points to see where I notice failures. I’ll post the results of my testing here for future folks to keep in mind.
I’m about to start a project running high-density LEDs (144/m) across almost 12 m so hopefully, my designed power drops will be sufficient right now they are every meter give or take.
@JP_Roy these are WS2811 but I’m making the switch to ws2812b strip with a different vendor so I’ll do some A/B testing and see if there are any differences.
Ran (and still running) into similar. In my case, I’m convinced it’s just cheap (eBay) LEDs. Usually though, once it glitches, it stays glitched in my case. Had to rethink an entire exterior lighting project that had entire strips silicon encased strips siliconed to brackets. This year, trying to make modular 10-LED pieces that can be swapped out if necessary. If that doesn’t work… just going to buy more expensive ready-to-go outdoor-suitable something or others.
On the plus side, I’ve become very good at snipping out LEDs and soldering in replacements.
What is the voltage output of your microcontrollers GPIO pins? I’ve had several projects using a 3.3V microcontroller like the ESP8266 that worked fine with a small quantity of LEDs or at low-brightness, but #FFFFFF would glitch them out. Using a level-shifter solved the problem.
The other people are right, you don’t have enough amps, either total amps, or you have too much of a drop on your chain, and re-feeding power in the middle and at the end should fix your problem.
I am curious if you made any progress with your issue !?
From your description it does sound like you may have some power supply issues but I kinda latched in to the part about not being able to drive your ws2812 pixels at full brightness white which is something that I (and AllanGee by the looks of it) have been struggling with to this day !
On the power issue, it’s usually fairly easy to detect by
putting an amp meter in line with your power supply and see if it maxes out with a brightness less than 255/all white. If so, there is your problem indeed
as others have also mentioned, feeding power into multiple places on your chain, helps. You can measure voltage in multiple points and see if it dips. If it dips too much, there again is the problem.
I personally have a 12x12 matrix which I powered from both ends, and it still gets glitchy on full white in the middle (away from both ends with power)
So an update for everyone here. I’ve done some testing with the limited LED strip I have in-house - which is in a questionable state so I’ll post further conclusions when my spares arrive back in-house from Burning Man. I also need my better multimeter to come back…
Test 1 - Color vs Voltage at end of strip. 120 WS2811 LEDs.
There were two behaviors that I noticed that I’ll try and classify further once I have more LEDs available.
Category 1.
Single Pixel failure. Data is not passed correctly downstream.
At low enough brightness 0 - 30%, the strip will work most of the time at this low brightness.
Any higher than 30% and the strip repeated fails with the data stream not being cleanly passed (observed with o-scope) to the next pixel. Power injection at the multiple points does not help (points tried with DC bench supply were before/after failed pixel and 25/50/75/100% of strip.
Category 2.
Erratic strip behavior (flashing, stuck colors, etc) at brightness levels greater than 30%.
Power injection into the strip (points tried are same as category one) fixes the erratic behavior.
Category 1 failures seem to me to be pixel failures since the data stream gets corrupted repeatedly by the same pixel. These I like since they’re repeatable and I can fight with my manufacturer over quality issues on. Category Two failures definitely seem to be related to power injection. I plan on doing current readings once I get more LEDs and crates back in-house next week. I’ll follow up with more results then!
