Hello! I know everyone’s probably busy with BM projects, but I am having trouble with another project I am working on. We have a bar-top table running ~600 pixels, injecting power every 200 pixels. For almost every pattern we run, every few seconds the pixels glitch and show different colors. Here are some examples of patterns that glitch (some more than others):
Any ideas of what it might be? I’ll post some photos of the wiring, we are using an inverter to power each set of LEDs and the PB, with one data line (with a 470 ohm resistor) running through the whole set. We’ve tried lowering the brightness to 50%, increasing the processing power, but no luck. We pack up to leave Tuesday, so any help is appreciated!
Glitches can be caused by many many things. I would guess that in your case its not a bad LED. The animations seem to glitch around the same place in an animation cycle. What this means is that something about the data and/or power draw is causing that first pixel to misinterpret the data in some cases that seems to be reproducible.
Stuff to try:
Remove the resistor. You shouldn’t need a 470 ohm resistor. Pixelblaze has a 100 ohm resistor already. Adding a lot of resistance weakens the data signal. A weak data signal is subject to more EMF.
Avoid making coils (loops) of data line. Coils will add inductance to the wire and can pick up more EMF. Shorter wire is always better, so trim the excess if possible.
Try moving PB close to the first LED, and share the same power lines. This reduces any potential difference in voltages, EMF, etc.
If close isn’t possible, try running a GND wire (black wire) from PB along the same path of the data line to the first LED. Keep the data (green) and black wire as close together as possible. If possible twist them around each other a little. This ensures that any EMF that affects the data line is also on the reference GND and helps cancel them out a little.
Try thicker wire, especially to the first LED. Wire will drop a surprisingly large amount of voltage under heavy load, and that can cause data interpretation issues. You can math it out with ohms law if you know the gauge and the current. or just measure voltage at the first LED with a static pattern. This can be mostly mitigated by moving PB closer to the first LED and powering it from the same power, but wildly changing power from voltage drop can also impact wireless and stability.
Thanks @wizard for the help- tried removing the resistor, cutting out all the slack in the data line, and we have it next to the ground wire, but still happening. Will try the thicker wire next. Can you think of anything else it might be?
Also, when you say share the same power line- do you mean connecting the 5V from the first LED directly to the PB, instead of splitting it further down the wire like we have it? Would we need to do the same for ground? And could we use the second set of smaller ports on the PB to do that?
Also double check that you have ground connected directly between every led segment, eg where you inject power. Double check the connections between any two segments.
If your power supply sucks, it might be spitting out dirty power. Could try adding some capacitors near the LEDs.
Could try a different power supply and/or PB in case one of them is faulty.
If it is a bad led, try to find which led is the first to flicker, and replace it and the led before that.
I’m guessing that points to some kind of power/connection issue with the injection. @wizard, would a capacitor here help? How would we do that? What would you recommend?
Sounds like maybe there’s a connection issue at that joint from one pixel to the next. My guess is GND isn’t connected there solidly between the 200th and 201st pixel. Without GND, the data signal can be way off since its all relative to GND.
Sometimes soldering can damage an LED too, but I would suspect GND/data issues since a bad pixel is usually more randomly glitchy.
