I am building 2 RGB LED lamps, each lamp is made from 6 strings of 25 WS2812b LEDs mounted on heatsink mini-boards driven by an Arduino Uno and based on the Fire2012 sketch. (Photos included…)
I noticed a problem while testing the 6X25 array on full white brightness as I also want the lamp to also serve for normal lighting. After less than a minute at full brightness, one of the pixels became yellow and then another one became cyan. I was surprised at the heat generated by the strips but decided to let the test continue as I thought I was just burning one of the 3 colors in some flaky WS2812b but it got gradually worse and I started to worry 2~3 minutes into my test when I had 3 LEDS completely off and half a dozen or so colored LEDs so I turned off my PSU.
Now here’s the weird part… After a cool down period of a ~1 minute, I loaded a basic test sketch into my Uno and found that some of the LEDs previously expected to have burned out were working normally ??? Although some did not and the LEDS that were physically after them was not getting any data. I switched off again and waited a bit longer (2~3 minutes) and now all but one LED was 100% operational except for one LED that was not able to display the blue component of RGB but the LEDS that were physically after that one were working normally thus receiving correctly their data.
Now another weird event, I pressed my finger on that last flaky LED and it was now working OK !?!? All 150 LEDS were now OK !?!?
My basic test sketch (MAX Brightness set at 50%) has now been running for a few hours without any problems. Note that this sketch goes through all colors and brightness variations (up to 50%) so I never have a white display at full brightness.
I noticed a similar report a short while back and wonder how many others encountered this type of problem on WS2812b LEDs driven at full brightness white for any length of time !
Looks like a great project. I’m going to guess that what’s happening is that your PSU is sagging under the load, and the end LEDs are getting undervolted and fading out. WIth the load reduced, the PSU can deliver full power all the way along the strings.
A question: are you supplying power at just one end of each strip, or at both ends?
Ah! OK, more more power thought: could you measure the actual voltage at the base of the strips, and at the end, and across the LEDs that are fading out? I’m wondering if it’s somehow too high or too low. (Not that I have a concrete reason to suspect, but I do. But as noted a few minutes ago, I am often wrong…)
@Mark_Kriegsman I did not measure at the end but I have adjusted my PSU to be exactly +5V at the beginning under full power.
Due to the short length (25cm), low power of a single string (< 1Amp) and the solid 22AWG wire used, I do not expect a significant drop !
Hi again +Mark Kriegsman, I finally did some actual +5V measurements on the full length of my strips and the variation I get is between 4.98V to 5.01V as expected.
Do you have a scope you can put on the voltage lines? (or maybe a multimeter in AC mV setting) I’ve had a switching power supply inject a bunch of noise onto the +5V line that caused flickering and weird colors.
@Tod_Kurt I do not have a scope but should add it to my XMAs wish list
I am 99.99% certain that this is not related to the PSU. The problem was only visible on 7~8 pixels out of 150 after a couple of minutes running at 100% brightness white.
However, I stopped my testing at full brightness as I was losing more and more pixels over time !
please keep us informed. I’ve got 120 of these puppies in a hat and need to power them by using 2-4 16850 lipo’s @ 3.7 with boost or 7.4 with buck. I saw that fastled has power functions. Can’t find a descent manual.
It’s unrelated, but why didn’t you use something like the 144/led/M strips, rather than wire individual modules in that way?
The mounting strip acts like a heat-sink and I have run one for 2 hours on full white to work out maximum battery life, and had no problems with colour change. I also ran Fire2012 for more than 3 hours on full brightness for the same reason, again with no change in colour, that I could see. I’m a big fan of doing as little soldering as possible
Hi @Mike_Thornbury , There are a few reasons to selecting these mini-boards instead of a regular strip:
It allows very easy irregular spacing between the pixels. Although for this project I am regularly spacing these miniboards without any wiring between them giving the equivalent of a 100LEDs/M strip.
It is actually cheaper ($$ per pixel) than the strips. But it does require a fair amount of soldering work.
The manufacturer claims that it provides its own little heatsink. Although I do not think now it is very efficient !?
Thank you very much for your report on the tests that you have conducted on a 144LED/M strip full white ! It reassures me that the device itself (WS2812bb) should take the heat normally.
I was very careful with my soldering but now believe that I might have damaged some pixels by overheating.
Hi @Kenneth_Tan , my tests revealed a surprising amount of heat from these pixels when running full white for a length of time… probably not enough to burn but still quite hot.
I would strongly suggest that you conduct full power tests with your LEDs before you begin wearing it… I would also appreciate your feedback on the results !!
