search

bk1285 / rpi_wordclock

Software to create a Raspberry Pi based wordclock
GNU General Public License v3.0
214 stars 107 forks source link

Problems with Raspberry Pi Zero W and the W2812B LEDstrip #151

Closed BowenDragonheart closed 4 years ago

BowenDragonheart commented 4 years ago

Hi WordClock makers!,

I am trying to build my own Word Clock and am in the final stage of testing the LED strips. Unfortunately, I ran into problems here.

I'm using a Raspberry Pi Zero W, combined with a Mean Well power supply (5V, 50W) and the a W2812B LEDstrip.

Using both the strandtest.py and the test script of the source directory, I run into a strange problem: I need to specifiy four more LEDs than are actually in the LEDstring to have all light up.

To investigate more, I made a simple program based on the strandtest.py and an issue raised in the rpi_ws281x github (https://github.com/jgarff/rpi_ws281x/issues/314) .

from neopixel import *

LED strip configuration: LED_COUNT = 10 # Number of LED pixels. LED_PIN = 18 # GPIO pin connected to the pixels (18 uses PWM!).

LED_PIN = 10 # GPIO pin connected to the pixels (10 uses SPI /dev/spidev0.0).

LED_FREQ_HZ = 800000 # LED signal frequency in hertz (usually 800khz) LED_DMA = 10 # DMA channel to use for generating signal (try 10) LED_BRIGHTNESS = 255 # Set to 0 for darkest and 255 for brightest LED_INVERT = False # True to invert the signal (when using NPN transistor level shift) LED_CHANNEL = 0 # set to '1' for GPIOs 13, 19, 41, 45 or 53

strip = Adafruit_NeoPixel(LED_COUNT, LED_PIN, LED_FREQ_HZ, LED_DMA, LED_INVERT, LED_BRIGHTNESS, LED_CHANNEL) strip.begin()

strip.setPixelColorRGB(0, 127, 127, 127)

strip.show()

If I run this, the first LED become white. So far so good. If also the second LED is activated,

strip.setPixelColorRGB(0, 127, 127, 127) strip.setPixelColorRGB(1, 127, 127, 127)

then the first remains white, but the second becomes yellowish green. If also the third is given a signal

strip.setPixelColorRGB(0, 127, 127, 127) strip.setPixelColorRGB(1, 127, 127, 127) strip.setPixelColorRGB(2, 127, 127, 127)

Then the first and second become white and the third bright green.

To conclude

strip.setPixelColorRGB(0, 127, 127, 127) strip.setPixelColorRGB(1, 127, 127, 127) strip.setPixelColorRGB(2, 127, 127, 127) strip.setPixelColorRGB(3, 127, 127, 127) strip.setPixelColorRGB(4, 127, 127, 127) strip.setPixelColorRGB(5, 127, 127, 127) strip.setPixelColorRGB(6, 127, 127, 127) strip.setPixelColorRGB(7, 127, 127, 127) strip.setPixelColorRGB(8, 127, 127, 127) strip.setPixelColorRGB(9, 127, 127, 127)

Results in that the first seven are white, the 8th is yellow and LED9 and LED10 are not giving light at all.

I tried with several LED strips of 10 and all gave the same result.

Who can help me to solve this? I also made an issue at the rpi_ws281x github, but there people appear not to be that active. . You will have my eternal gratitude if you can help me with this.

Best regards, Bart

cmaxl commented 4 years ago

Hello @BowenDragonheart,

welcome to the Wordclock builders. I have three things you might consider to get rid of that behaviour:

  1. Are you using this commit of rpi_ws281x as described in Software Setup? grafik
  2. Remove the level converter. (I am also using a Rpi Zero W and the LEDs light up fine witout one.)
  3. Swap the Meanwell with an USB wall adapter. (Maybe this is a better and safer solution.)

Greetings, plotaBot

BowenDragonheart commented 4 years ago

Thank you for your input!

  1. Yes, but I will do a clean install right now just to make sure that everything.
  2. I don't use a level converter.
  3. I will give this a try.

Furthermore, I have read that it is important that the LED strip and the PI do not share the 5V+ (https://tutorials-raspberrypi.com/connect-control-raspberry-pi-ws2812-rgb-led-strips/)

This is currently the case for my clock. Why is this important? Can this explain the wrong behaviour?

cmaxl commented 4 years ago

Ok. Have you connected both Ground (Power Pi AND Power LEDs) to the Pi? I am not experiencing any kind of problems using one Powersupply. My wiring looks like this: grafik

Maybe another thing causing this problem could be related to the beefy powersupply iterfering with the signals. btw: 10-15W should be enough and an ordinary phone charger is capable to deliver such power. Since you are not running all 114 LEDs at the same time on white with maximum brightness you should be fine. (I once measured this value and Pi + all LEDs were using <2A peak)

Let us know if anything I suggested works or not ;)

BowenDragonheart commented 4 years ago

I reinstalled the pi and the behaviour is exactly the same. So it is at least reproducible.

Next thing I tried is not using the same power supply for the LEDs and the PI. No difference.

The third test is exclude the beefy power supply. So now I power the Pi through the USB and the LEDs with GPIO's. Exactly the same result.

So now can conclude that it is not the power supply.

I realise now that I am not using exactly the commit to which you are referring, but the latest version. Do you think there is actually a problem with a newer version of the rpi lib?

There are three suspects left:

Thanks for helping me plotaBot

BowenDragonheart commented 4 years ago

Is that not the problem? It is not an RGB strip, but a RGBW strip. Is this not supported?

Oke, I can get the standard rpi_ws281x test to work now by specifing in the main.c that I'm using a SK6812 RGBW strip. Now actually 10 LEDS start to shine when I specify a Width of 10.

However, the simple python script still doesn't work. I think it only supports the RGB strip? So does that mean I have to actually buy a new LED strip? :-/

cmaxl commented 4 years ago

Apparently rpi_ws281x supports RGBW but implementing it requires more effort and a strip at hand. Sorry if you bought the wrong LED strip. Unfortunately I am currently not able to develop a rpi_wordclock RGBW support (time and no LEDs). Or you, Bowen, are passionate enough to dig into that?

@bk1285: Would you add the enhancment label so @BowenDragonheart is able to use his LEDs (somewhere in the future).

bk1285 commented 4 years ago

Hi @BowenDragonheart

to address the LEDs, jgarff's library is currently used. As mentioned here, it's probably an issue, that you need to provide another byte to setup the fourth color channel. As soon, as jgarff's library supports this, I expect the wordclock to work smoothly with th RGBW-type of LEDs.

Meanwhile, you'll have to choose standard RGB-strip.

Best, Bernd