WLED Matrix glitching beyond 'max' brightness when sending http pixel updates

[DaverDavids]

What happened?

I have a python script that updates pixels on my WLED matrix display by http json calls. I was having reliability issues and realized that sometimes, when sending pixel updates to the display, it will flash random pixels super bright for a split second... if the USB power supply isn't extra powerful, the esp32 crashes due to low voltage.

This is a four output set using parallel i2s, 1024 LEDs total, target 42fps. Each output has a 330-350mA limiter set and 'brightness factor' is set down to 10%. I can turn the display all white and turn the brightness to max, and it won't crash, I have the brightness properly limited. But the glitch seems to disregard the hard brightness limit and if enough pixels are randomly part of the glitch, it will crash due to power draw.

I'm running an alpha build because at the time I made this display, this build corrected another bug issue I was having. WLED 0.16.0-alpha (2511210).

I've attached pictures as well as the code I'm using to update it, in case it's relevant. Thanks for your help.

Is there any log I can include to see what WLED is doing when it updates? Network capture? Not sure what would help.

To Reproduce Bug

I'm not sure this bug is reproducible in other environments. It happens when my code sends pixel updates and it doesn't happen every time.

Expected Behavior

Display shouldn't show pixels that weren't drawn to it, and hard brightness limit should be respected in all updates.

Install Method

Binary from WLED.me

What version of WLED?

WLED 0.16.0-alpha (2511210)

Which microcontroller/board are you seeing the problem on?

ESP32

Relevant log/trace output

Anything else?

matrix_server.py

wledflash.mp4

Code of Conduct

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

[DedeHai]

these glitches look like "corrupt hardware signal" which can happen due to interrupt starvation or just plain hardware issues and are outside of WLED code.
Since you are using parallel I2S which is not prone to such glitches it's a bit puzzling on where the glitches are picked up.
Do you ever see these glitches when not bombarding the ESP with data i.e. using internal effects?

[DaverDavids]

As far as I can tell, it only happens when it's being updated with new pixels.

[DedeHai]

did you test with latest nightly?

[DaverDavids]

I've just updated to latest nightly and I will have to test with it for awhile to see if it appears. I'll put it back on a good-enough but not exceptional power supply to see if it browns out.

I guess hardware glitches/noise would make more sense that it exceeds the allowed brightness level. I'll report back. Thanks.

[DaverDavids]

Haven't had a glitch since I flashed to latest nightly..... but then I flashed back to my nightly from a few months ago and still am NOT having the glitch anymore, even though I was when I first posted this. So I've not got it pinned down and can't reproduce it anymore, I'll return if I figure out anything concrete that can be software related but for now we can assume it's a hardware problem.

Thanks.