mimxrt10xx: Verify and possibly fix neopixel_write

[arturo182]

Copied from stm32, seems a bit buggy, only works sometimes (sets the wrong color when not working).

[hierophect]

I'd like to take a look at this as part of a revamp I want to do to the STM32 version.

[arturo182]

I only tested by looking at the neopixel and checking if the colors change correctly, I would like to actually connect a scope on the line to see what is really happening there, but not sure when I'll get time to do this.

[hierophect]

@arturo182 when it goes wrong, what happens? You said wrong color, does that mean "white" or "off" or actually literally the wrong color but with normal brightness and stuff?

If it's a timing issue it should either be off or a blinding white and is probably a boring processor cycle counting problem. If it's actually a different color else your data is being garbled which is more interesting.

[arturo182]

What I've been seeing is that if I try to set pure red I get sometimes orange, sometimes yellow, sometimes white, and if I keep retrying, eventually it will be red.

I think putting a scope on it will be the easiest way to see what's up.

[hierophect]

Do you have a Saleae? Hands down the easiest tool for this kind of monitoring that I've used. Scope will also show it but the Saleae is just so easy to set up and read with.

[arturo182]

I have a Chinese clone that works with the Saleae UI and I really like it, I was just worried if it's fast enough, also it's digital-only so if there's some weirdness on the line, I would miss it.