Sometimes on .show() the lights are all set wrong?

[mpeddicord]

This is a great project. I'm using it as the light controller for LEDs around my bed.
One thing I've seen though that I haven't been able to solve is sometime when I update the lights. They seem to get the wrong data. Usually it's like the data is offset. Often it manifests as the entire strand goes to full white for just the one frame.

When I use the examples I don't see the problem. It only seems to be when I'm doing a lot of manual setting of pixels using slices. One reason why it hasn't been such a big deal is because I'm using Home Assistant to send MQTT commands to the Pico, and that's usually just one update at a time. But it's become clear that the problem happens like 1 in 200 updates, roughly. So, when I'm updating manually I almost never see it. But when I'm trying to use a custom animation with a fair amount of complexity, I see lots of wrong updates.

I suppose it's possible it may not be a software thing. But I'd like to know if you've seen something like this before. (Also, everything is sharing a ground correctly)

Sorry if this isn't the right place for this.

[blaz-r]

Hi,
I haven't encountered this issue before, but it is possible it is due to software if updates are done too rapidly, though I'm not sure.
Could you please send the animation implementation so I can test it to see if I encounter similar thing.

[Davis8483]

I'm having a similar issue. In my setup I have a pi pico w connected to my computer over serial. The part of the pico's code that handles the commands is in a separate thread from the one that updates the leds.

LED Bug Video

There are two led strips side by side. The first one is set to display one pixel of blue and the rest red, although it has random flashes of blue. This bug only happens when its connected over serial.

[mpeddicord]

That's actually how mine also usually manifests with a shorter LED count. Longer stands got worse for me. I kept thinking it was a power supply or grounding issue. Because you definitely do see weird flickering like that if you're not sharing a ground between the chip and the LED strip. (I found this out from firsthand experience and subsequently burnt out one of the data pins of my first PICO W before realizing all the things I was doing wrong)

Unfortunately, I didn't ever resolve this though, I put the project on hold instead.

I just wanted to chime in with additional similarities to you. I too was running my main code and the light updates on a separate thread. Though, I think I remember ruling that out. I almost had a small sample project to show the weirdness, but I had a baby and now there's no time.

Good luck! I'd still love to get to the bottom of this one. Whether it's software, the hardware itself, or some weird interaction between the two.

Last little data point, I'm using these LEDs exactly:
https://a.co/d/7uFL16u

[blaz-r]

Thank you both for this. As I said in previous comment, it'd be great if you could give a minimal code that can reproduce this issue. Another question I have is if you are both using Pico W or Pico?

I think this can be software, hardware or even mix of both issue. It seems like the timing might be a bit off every once in a while. This might mean that the PIO wait intervals are incorrect, or PIO frequency or even something else.

Neopixles do have some tolerance for timing, about +- 25%, that's why I don't think that the problem comes from timing, because then it'd be more frequent. Could be something with PIO buffer.

It could be problem with "latching" at the end, where strip needs some time for colors to stick. But if you kept default delay of 0.0001 seconds, which equals 100 microseconds, it should be enough as strip should require 50 microseconds. It could as well be something about how this is achieved with time.sleep and threading that you are using.

With this many possibilities, I can only help if you can send me the code where this problem appears.

[Davis8483]

I was finnaly able to reproduce it with simplified code. Btw, I'm using a pi pico w.

Upload this to the pico as main.py with the neopixel library in a folder titled pi_pico_neopixel

from pi_pico_neopixel.neopixel import Neopixel
import _thread
import sys
import time

strip_length = 8
strip_pin = 14
strip_mode = "GRB"

led_strip = Neopixel(strip_length, 0, pin=strip_pin, mode=strip_mode)

def serial_communication_thread():
    while True:
        serial = sys.stdin.readline()

        sys.stdout.write(serial)

# start the serial communication handling thread
_thread.start_new_thread(serial_communication_thread, ())

while True:
    time.sleep(0.05)

    # set all pixels red
    led_strip.fill((255, 0, 0))

    # set last pixel blue
    led_strip.set_pixel((strip_length - 1), (0, 0, 255))

    led_strip.show()

Then run this on your computer.

import subprocess
import time

try:
    import serial

except:
    subprocess.run(["pip", "install", "pyserial"])
    import serial

port = "COM4"
baudrate = 9600

ser = serial.Serial(port=port, baudrate=baudrate, timeout=1)
time.sleep(1)

print("starting data tansfer")

while True:
    # send alot of data
    for index in range(50):
        ser.write("meaningless data".encode())
    
    # delay
    time.sleep(1)

[blaz-r]

Thank you. I'll look into it when I get some time. I suspect it has something to do with threading and timing.

[Davis8483]

Any update?

[blaz-r]

I'm really busy right now and don't have access to my pico. I'll try to test this as soon as possible, but might still take a few days.

[Davis8483]

Ok, thanks!

[blaz-r]

Hello.
I managed to reproduce your problem on my pico.
I'm really not sure how this can be solved, but it seems that PIO doesn't work that well when serial is actually working is thread.

I rewrote the entire PIO assembler code for leds, to really get the timing down to 10ns precision, but it still happens.
I even changed the "latching delay" from 100 micro seconds (required should be 50) to 500 and it still occurs.

Tried to add flush to serial, or call gc every here and then and still no success.

At this point I don't know what else to try or what could cause the problems, but it seems that serial reading and writing somehow messes with PIO, or the code that loads data to PIO buffer.

[blaz-r]

I think I might have solved it, but I'm unsure.
I checked the docs and I see that state machine put can accept entire array. So I replaced the loop with just one call inside show(), and the flickering is gone for me on ws2812b.

Can you try replacing show with the following and tell me if it solves the issue:

    def show(self):
        """
        Send data to led-strip, making all changes on leds have an effect.
        This method should be used after every method that changes the state of leds or after a chain of changes.
        :return: None
        """
        # If mode is RGB, we cut 8 bits of, otherwise we keep all 32
        cut = 8
        if self.W_in_mode:
            cut = 0

        self.sm.put(self.pixels, cut)
        
        time.sleep(self.delay)```

[Davis8483]

It still seems to have the random glitching. I might just rewrite my program to exchange less data.
Edit: I'm also using ws2812b leds

[blaz-r]

I'm sorry to hear that. I'm really not sure what to do at this point. I will update the code to remove the for loop nonetheless as this seems a better approach.

As for this issue, I'm really not sure how to solve. Even when debugging I had a lot of problems with serial. I've also seen a lot of people having issues with threading online, so it might be a bigger issue.

If you by any chance find a solution, please share it here.