What is a large number of APA102c LEDs?

[clfaye]

Hi all,
I've done a lot of work with WS2811 and variants in fastLED. When I do a large number, I run them in parallel. I am just starting to experiment with the APA102c, and I am thinking of a pretty large project with about 3000 LEDs. I'll be using a teensy 3.6. Can 3000 LEDs be controlled over a single data and clock line? Or - would I need to figure something out since there is no parallel solution for fastLED for the APA102c?
Thanks!
cf

 

[PaulStoffregen]

Can 3000 LEDs be controlled over a single data and clock line?

No, sadly this isn't usually feasible.

APA102 has a problem where the clock signal isn't perfectly regenerated. Each pixel shortens the low time slightly. It's sort of like a photocopier making a copy of a copy of a copy, each time getting a little worse. After about 100-150 pixels, 24 MHz no longer works. The longer you go, the worse the timing margin becomes. Just yesterday someone was reporting problems with a 2000 LED chain only working with very slow clocks, like ~1 MHz.

Unless you're doing a persistence of vision project, or you're using a non-Teensy board which lacks non-blocking libraries for WS2812, I would advise against APA102 LEDs. I know many people feel like APA102 is superior, but in fact the poor quality clock regeneration circuitry is a huge problem as you try to scale up to a larger project.

 

[clfaye]

hmmm - that is very good info. I was looking forward to the better low brightness performance. What do you think of having multiple arrays of APA102c LEDs on different pins - so the max number on a set of clock and data pins would be 256? Although I shudder at the wiring. I think you wouldn't get a parallel writing benefit, but it seems like it might solve the clock signal problem. Or - would adding buffer like the 74HCT245 every 256 pixels do the trick?

 

[PaulStoffregen]

Here's a writeup where I investigated the APA102 clock problem.

https://www.pjrc.com/why-apa102-leds-have-trouble-at-24-mhz/

My general recommendation is to only use APA102 for POV.

 

[clfaye]

yup - it's kind of a lot of work to perform an experiment on with a LOT of LEDs on the line. I'll do some bench work on the APA102s for fun, but plan on WS2812s for the project. Thanks for the write up!