Gotcha - I’ll look in the morning - but sync by way of delay is likely your culprit - lots of nondeterminism in there. Better to pick timing marks and wait for that. I can show you in detail tomorrow.
Also the ws2812 style family (including apa104/apa106/tn1809) of leds are absolute shit for POV, with only having a 400hz pwm cycle rate - some of the variation that you are seeing may even just be that. The apa102 has a 19.2kHz pwm rate - much much better.
Thank you for the tips. My eventual use of these could only just-barely be considered POV so I might be ok. As for the video, I’m mostly taking advantage of the POV effect to check my pattern reproduction and consistency – I’m not looking to actually draw any specific designs in the air. So I think 400Hz is plenty for my needs.
Dan, coming at it with fresh eyes this morning, I think there’s a good chance you’re right about the PWM limitations. If it’ only has 400Hz PWM that means there are 400 “segments” a second. With my update rate of 212Hz, each section of my pattern would be ~2 of these “segments” When I spin my pattern up to be visually stable, it looks distinctly like each section of the pattern has a subtle line down the middle – this might be the PWM cycle. The “ants marching” problem could then easily be a side effect of being roughly half the maximum rate but an uneven divisor, i.e. each section is occasionally getting 1 or 3 rather than 2 PWM segments. Shit.
Or not. Just tested the theory by splitting each segment into three and writing a pattern of colors, i.e. RGR-GBG-BRB-GBG-BRB. The fan can’t stably spin slow enough for me to get the pattern visually stable, but I can clearly see each segment being split into three colors, and it looks pretty even. So I don’t think I’ve hit the PWM limit, nor the protocol limit. Must be some other timing issue.
Using FirstLight
WS2801 - first led goes to white via red.
WS2812 - the white spot moves smothly for the fist 16 lights then stutters for the rest of the LEDS
OK WS2801 works fine when I have the strip plugged into the correct Pins! Oops!
OK there seems to be a problem with the Bluetooth stack enabled and writing out to the WS2801. I kinda expected this and I guess the stack is cauing to much of a delay to make the strip think you’ve finished sending data. Which strips would handle this delay better?
Apa102 or lpd8806
Yeah, ws2801 latches when the clk line is held low for 500us and Nordics Bluetooth soft devices have interrupt handlers that run for over 1ms. There may be options, however.
For the time being I’m going to have a button that depressed for 5s+ will enable/diable bluetooth for pattern programing It s a shame as I was going to pair with the phone and use the phone to do beat detection and control.
@Daniel_Garcia Do you have a wiki that has all the different types of chipsset ans different features/restrictions would be very useful.
https://github.com/FastLED/FastLED/wiki/Chipset-reference - and I need to update it with more info, now that I have it.
Thanks! Just tested this with an RFDuino and APA102 strip. Works perfect (with setDither(0)).