LEDs occasionally flash white on ESP32 C3 SuperMini

[ibaon]

What happened?

Hello, I have noticed that sometimes the SK6812 LEDs on my controller will just flash white. I am not sure if it is pure white, but whiteish. Sometimes it is the whole strip, sometimes just part of it, and sometimes it starts in the middle of the strip, not even at the beginning. I don't think it happens in a fixed intervall, but roughly every 30 seconds, sometimes a bit less. It does not seem to matter if it is a static, solid color or if there are any effects playing.

I am using an ESP32 C3 SuperMini combined with a 74LV1T34GVH Logic Level Shifter. They are connected to a good quality power supply which can supply more than enough current. Also I have a big 1000uF decoupling capacitor at the power supply and some 10uF capacitors on the controller itself.

I am using 514 LEDs, injecting power both at the beginning and at the end of the strip and also limited the current to the strip in WLED. The problem occurs both at high or low brightness, so I don't think it is related to voltage drop or similar.

At first I had the data line connected to GPIO3 but later switched to GPIO2, which didn't make any difference.
Furthermore I added a 62 Ohm resistor to the data line, which didn't help either.

I also tested a different controller and it seems to be fine with the other controller, so I don't think the strip itself is damaged. The other controller I tried is a Seeed Xiao ESP32 (which also uses the C3 Module). Both times I used the same PCB with the same logic level shifter etc, I just swapped the microcontroller and changed the data pin to GPIO8 for the Xiao. So I don't think it is even related to the C3, but just to the SuperMini? I have experienced some LEDs glitching with the Xiao ESP32 before as well, but not nearly this bad.

What else could be causing such a problem?

To Reproduce Bug

So far it seems like I can only reproduce it when I am using an ESP32 C3 Supermini. It happens both with GPIO 2 and 3.

Expected Behavior

The LEDs will occasionally flash white. Not the whole strip, often just part of it, not even starting from the beginning, sometimes it will only be the second half or less of the strip.

Install Method

Binary from WLED.me

What version of WLED?

0.15.3

Which microcontroller/board are you seeing the problem on?

ESP32-C3

Relevant log/trace output

Anything else?

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[DedeHai]

this is a known issue on the C3. The wifi driver interferes with RMT LED driver output. To check:

• disable wifi sleep
• set it to AP mode

if the flashing persists in AP mode it is a hardware issue with your setup.

[ibaon]

this is a known issue on the C3. The wifi driver interferes with RMT LED driver output. To check:

* disable wifi sleep

* set it to AP mode

if the flashing persists in AP mode it is a hardware issue with your setup.

In AP mode I can not see any flashing lights ... Is there an actual fix for this problem? While it is in AP Mode and not connected to my Wifi, there isn't really a way of controlling it without manually connecting directly to it, right?

[DedeHai]

there is currently no fix. well there is one but it makes it crash every few hours. the workaround is to not access it using HA or other software.

[ibaon]

Thanks. Swapping the Microcontroller to an ESP S3 SuperMini should solve it as well, right? While the footprint/pinout is a bit different, I think it should fit as well.

But I hope there will be a proper fix in the future, since I already have a few Seeed Xiao ESP32 C3 in use. While they also glitch sometimes, not nearly as bad as the SuperMini ESP32 C3.

[softhack007]

Well the "proper fix" was actually to completely rewrite the RMT driver in assembly language 😅

We have done that on classic esp32 and S3, but currently we have nothing similar for risc-v chips like the -C3.

So yes, with an S3 board - which is a good choice any way - there should be no glitches.

[DedeHai]

Well the "proper fix" was actually to completely rewrite the RMT driver in assembly language

thats what I thought too, but risc-v does not need that assembly bridge it works without it. The code works on a C3 but the increased RMT interrupt priority is not appreciated by the RTOS - and it crashes once in a while (on average about every 12 hours)
There is another hack-fix that gets rid of 99.9% of the flashes and its changing a 1 to a 4 in NPB and reduce the number of outputs to a single one.