Again about APA102 @ 24 MHz. It was written here often that it depends on the lever shifters, on the wiring, on the APAs but mostly on the led strip itself how many APAs can be driven at 24 MHz without glitching towards the end of the strip.
I made an interesting observation: I have a setup here which mostly works fine @ 24 MHz. BUT: Without any change of the setup sometimes the last part of the led chain flickers. Exactely the same code, same setup, same power supply, same cables… everything 100% untouched.
I checked room humidity, room temperature, led temperature, atmospheric pressure… but I found no coherent context yet which explains the reason for that.
Any ideas which physical reason or black magic could be relevant for the different behaviour of the setup?
Gremlins? 
I’m about to try adding additional level shifters in the middle of a glitching set of them…
How long is your strip length? Because that matters too. Longer strip lengths get corrupted a lot quicker
Have you looked at the signal with an oscilloscope? Two factors that are hard to control: the quality of power going in to your rig – unless you’re able to use chemical batteries; and the electromagnetic environment. Those long wires on your rig are antennas.
Unfortunately I´ve just an old 20 MHz scope and a 20 MHz logic analyzer. I´m aware about the influence of strip length, strip quality and em environment. The question is how can it be, that it works sometimes and sometimes not. And yes, for tests I removed all Wifi and cellphone stuff, any pulsed DC converters, really everything from the room (and I have no neighbors within 500 meters). Often it works perfect, sometimes under exactely the same conditions not. Maybe it´s my own body that has a different electrical potential / reflection / deflection capabilities at different times?! 256 leds by the way. Power supply is a high quality laboratory model. I ask for not more than 25% of the current it is rated for.
@Stefan_Petrick well you just answered what would have been my suggestion IE: you have some external source that occasionally generates EM that interferes with your ‘sensitive’ setup.
If it is easy, you may try to tack on some distributed small 0.01 uF caps along your strips power lines ! There are none on the APA102 strips to my understanding and I only guess that may help to make your setup a bit less susceptible to noise.
Seems to me you are being very methodical already about it, still, keep in mind that there is always a logical explanation for ‘ghosts’ ! Ideally get your hands on a decent scope if you want to ‘see’ the problem in action. Good luck !!
A couple of others come to mind: software/hardware intermittent failures, e.g. race condition that is only visible with low probability. I ran into a bug like this with a Teensy driving an APA102 strip via SPI while also communicating over USB serial. Or bad solder joints. This can make the rig sensitive to vibration.
Not to sound crazy, but check the Kp Index when you get flicker. It just might be a solar storm.
- officially the APA102 can not be driven past 20MHZ so what you are seeings is perfectly normal. I actually challenge you to get it to work at 20mhz or for arguments sake 15mhz!
- how sure are you that you have REAL APA102 pixels and not SK9822. The reason i ask is that for about a full year the APA102 was impossible to source because the company that made it was bought out. The SK9822 took over… then We started seeing APA102 pop up. I have a set of recently purchased APA102 and they are 100% SK9822. How do i know? They look identical. The original APA102 pixels had a HUGE IC compared to the SK9822.
- having said that, the SK9822 CAN be driven to a maximum of 30mhz, but getting close to that 30mhx you will start to see issues unless you have a clean room environment. ALL issues i have had with strips relate to the quality of the strip not the pixel. Is a case of how well were they soldered on to the strip and how good is the strip itself. If you look at some of @Daniel_Garcia posts, he has a PCB with APA102 on them and they can be driven to full speed.
As a rule of thumb, i never push them above the 15mhz mark because at that speed, you have so many issues with external noise its not really work it.
@John_Corbett gave an important hint: “only visible with low probability” My animations don´t run with a constant framerate. Multi layer noise and plasma stuff. So, I narrowed down the problem a little: The flicker is most likely to happen when I update the leds with 1350-1400 fps. Happend very seldom with my animation, most of the time it´s slower. FPS rate below is fine, above is fine. Some resonance effect? Further answers: Yes I´m pretty sure that I´ve APAs. No problems yet with 16 MHz SPI speed. And yes I should get a decent scope.
@Stefan_Petrick Where did you purchase them from? If they are real APAs id love to get a hold of them to test for myself. Im fairly certain they don’t actually exist anymore
@Scott_Schipper Effects of solar radiation on terrestrial electronics are rare. Huge memory arrays can be used as particles detectors, but they need to be huge to see statistically significant effect rates.
@Leon_Yuhanov , I’ve bought APA102s and SK9822s from Ray Wu and they’re visibly different. I can only assume that the APA102s are real…
@Leon_Yuhanov They are inside a GameFrame https://ledseq.com/product/game-frame/
@Jeremy_Williams might be able to answer the question where he bought the led panels.
The panels are custom made for Game Frame. They are real APA102. I was sent a test panel of SK9822 that I had issues with, and in turn sent it to Daniel Garcia, which led to the discovery of the different end frame interpretation and official FastLED support. But the takeaway is Game Frame panels were produced in the first half of 2016 and use real APA102.