Weirdness! I haven't used NeoPixels since discovering APA102 but I picked up this great

Weirdness! I haven’t used NeoPixels since discovering APA102 but I picked up this great 32x32 NeoPixel panel that I wanted to put to use. It tested fine with the Adafruit NeoPixel library, but after switching my code over to FastLED I get this random glitch effect.

I’m running the data line through a 74HCT245 and a 480 Ohm resistor, and I’m using a Teensy 3.6 SPI pin (Pin 11). (The same glitch occurs when bypassing the buffer and resistor.)

This test animation simply moves a pixel along the string of 1024 LEDs. If I change the NUM_LEDS to 256, it moves much faster but exhibits the same glitch problem. I’ve noticed that the glitch always appears before the active LED in the sequence. Any thoughts?

Do you usw esp8266? then it might be

@Stefan_stelb_Le_Bret Thanks for the suggestion, but I’m not using anything other than the Teensy 3.6.

Also, to be clear, the glitch isn’t exactly a flicker. It moves the entire frame buffer backwards in the strip. The dots you see flickering are supposed to be a part of the consecutive sequence but they’re being rendered on the wrong LED (strip zigzags top to bottom). This is more apparent when displaying full screen animations.

Which version of FastLED are you using?

Well, I’m at 3.1.5.

But I just ran one of the examples and it works fine, so there’s a conflict with something else in my code. Odd, because the simple running dot animation above blocks everything else. Maybe an interrupt, but I don’t think I’m using any. I’ll tinker and report back.

Ah, indeed I was using an interrupt – inside the Entropy library. Removing that solved the problem. Who needs random numbers anyway.

Why do I feel the need to go play Centipede now? :wink:

@Jeremy_Williams - Nice NeoPixel Panel. Can you please share were you purchased it?

Years ago from a plant I used for manufacturing in China. They had a few spares left over from another client.

@Jeremy_Williams - Thank you for your response. Any chance that they still are making them or do you know of anyone making a large panel like that with APA 102?

@Ken_White I’m afraid the answer to both is, no.

Ah - the frame being moved backwards in the strip means that the interrupt allowing code is allowing a bit too much of a time gap when an interrupt occurs. Making this more easily tunable is one of the first things on the list when I get back to working on the library.

The biggest matrix i found so far is 2828 like: sk9822 1X Addressable SK9822 RGB led matrix flexible dot matrix display with fiber plate 2828LEDs /3020LEDs/1414LEDs free shipping
(from AliExpress Android)

I have the same issue some time ago. As far as I remember, I have fixed it by replacing the noisy power supply with a better one.