Data Line signal suddenly gets noisy - Neopixel

Community PCB Design
1

Hi,

I am building a PCB with an 11 x 10 matrix of neopixel LEDs, WS2812B.

  • I have a 1000uF cap on the VCC line coming in from a 9V adapter. It is then regulated to 4.95 volts.

  • Using ATMEGA 328_TQFP to run the LEDs. 5V data signal.

  • Data line has a 431 ohm resistor before the first LED

  • 5 V powering the LEDs

  • 0.1uF Capacitors at each LED

The first 10 LEDs light and respond fine. From there on, the string does not work. I replaced the 11th LED thinking it was damaged, but did not help.

I replaced LED 10. Did not help.

I then monitored the data signal with an oscilloscope and see that the Data Out signal from LED 10 does not have a nice flat top at 5 volts. (how can I attach an image without a URL?)

So it looks like LED 11 is not working due to a bad data signal from LED 10. Installed a new LED at LED 10 and it still had the same bad shape for Data Out.

LEDs 1 - 9 have signals that all look the same, clean flat top.

I compared my PCB data signal to the 8x8 Neopixel Matrix Adafruit sells. LEDs 1 - 9 have data out signals that looks like Adafruit’s.

Any ideas on why my data out signal from LED number 10 suddenly looks different from other 9 LEDs?

Some sort of PCB design issue?? Have images of oscilloscope to attach, but do not know how.

Thanks for any advice.

Rich

2

Use controls in “Full Editor” to attach pictures.

Since your problem appears instantaneously as opposed to gradual degradation it is probably a layout issue with either the signal line leaving led10, the power to led10 or the ground at led10.

You mentioned the cap before the regulator. How big a cap is there after the regulator?

  • Chip

I do not work for SparkFun

3

The 1000uF cap at the barrel jack.

4

PCB Layout and PCB itself

5

Thanks for replying Chip - I have attached the schematic of the regulator.

I have 10uF capacitors on both sides of the 7805. In addition I have a 1000uF cap at the 9V barrel jack. This is recommended by Adafruit to protect the neopixel LEDs.

6

Chip - What you say makes sense: it is sudden at LED 10, so must be something wrong with the PCB there. What should I look for? Damaged trace? Bad trace to ground…? I will look closely with a magnifying glass at LED 10 and LED 11 traces on board.

7

You can also check voltages at led10 supply and ground pins. And/or check traces with an ohm meter while power is off. But from your scope trace it does not look like the signal is degraded enough to be the source of the problem. Check the supplies at the dead led. Check the signal at the input of the dead led.

  • Chip