I’ve been asked to resurrect an old project that uses a bunch of LEDs and drivers. Since I do some fast updates, the WS28x series are out of the picture. And since I’d like to get rid of the drivers, that leaves me with the APA102 … or SK9822? Is there a difference between them? My vendor says they can supply the APA102s if I really want that, but they claim the SK9822 are, and I’m quoting here, “more stable”.
So what’s the skinny? @Daniel_Garcia ?
Hmm - not sure on that one - though to be fair I haven’t stress tested sk9822 strips - all I’ve ever gotten was a board with a few hundred on it which historically has always worked better than the strips.
I just received a 1 meter strip of SK9822 LEDS 144/m. Tested them with a Wemos D1 mini at a data rate of 36MHz with absolutely no issues. Couldn’t believe how awesome they were working with that speed. Tested with Mark’s discostrobe code so things were blinking like crazy. But in a good way.
What happens when you go longer than one meter though? Because I think part of the apa102 issues had to do with the physical length of the strip.
I only have 1 meter to test with right now. I’ll buy a couple more and check them out. I know @Leon_Yuhanov has done some pretty extensive testing with them and thinks highly of them. Daniel if you want I can send a couple meters to you as a thanks for everything you’ve done for this community. Just let me know and we can chat privately.
Yeah, the vendor I talked to said something similar about the APA102 regarding a very long run, and then proceeded to say the SK9822s are “more stable” for that. I guess I can always try and if they don’t work out, I can repurpose them for something else.
If the sk9822’s are regenerating both clock and data signals - I suspect that could go a long way towards helping things. I really should just go and order a handful of 9822 strips for laughs/work here. (I’ve also got a set of ws2813’s - which may have solved the interrupt problem enough for me to allow interrupts on AVR platforms, even - need to test the timings)
If you have the choice between the 2 then I’d definitely go with the SK9822. I have a 2 meter strip of APA102 that can’t handle anything faster than 4Mhz. They are terrible quality (had to cut out the first 15 LEDs because the copper pads were shorted out between each other) but I’d still go with the SK9822.
My last order of apa102’s had a bunch of bad pixels on them - I spent about six hours one night testing nearly 100 meters of strips and cutting out/soldering over individual bad pixels. And by bad, I mean they failed after about 10 hours of burn testing (which I started doing after the first few meters of random failures)
@Daniel_Garcia Not sure how to send you a private message via G+. I’ll order 2 meters of Sk9822 for you tonight. If you let me know of a good way to contact you I can get them sent your way.
I have a 5 meter strip of SK9822 chips. I’m testing them to see if it is really a valid alternative to APA102. I’ve read somewhere that they use more power when leds are off.
I’m building my own led product so I have to take EMC in account as well. Especially on high clock speeds this can be an issue. So it’s nice that you can run them at super high speeds, but that makes them also more sensitive for radio waves…
Hey guys, I have been using the SK9822 for a while now. They are a good APA102 replacement. But apart from having a 30Mhz Maximum drive speed, they are pretty much identical to APA102. @Brian_Lewis Very impressed that you got 36Mhz on your stip, i have not been able to get above 30mhz on 1 meter . Id say the variability may be in ht manufacture of the strip not the pixel. @Daniel_Garcia I can drive 500 pixels at 10-12Mhz with a 1 meter Cat5 cable. And and 22Mhz with a very short lead from an RPI.
@Kasper_Kamperman I can sort-of confirm the “the use more power” although I’m not sure what the cause is. I have purchased from 2 separate manufacturers of STRIP and both were identical. On average a 1 Meter SK9822 trip used about 10% more power. But i was testing with one of those pass-through USB things so its not really acurate
I’ve been doing some very non-scientific testing of my 1M SK9822 strip consisting of 144 LEDs today. I am currently driving it with the data rate set to 44MHz and using Mark’s discostrobe code. I added a fps counter and removed the delay and an average fps of 615. I am driving everything with a Wemos D1 mini at 160MHz and not using any sort of level shifting. I have absolutely no hiccups or random flickering happening. Very pleased with the quality and performance of this strip.
FYI Daniel. Your strips are ordered. Takes Ray Wu about 7 days to get them shipped then another 4-5 days to get to me.
Unless you’re using another library - you’re only driving the leds at about 2.8mhz on the d1 mini, as I don’t support hardware spi on the esp8266 yet (also - doing the math - 615fps for 144 leds at 32 bits per led is ~2.8mhz.
Also - sub 1m lengths is not where the problem rapidly devolves - it’s when you get to multiple meters.
@Brian_Lewis thank you! You can ping me on hangouts with danielgarcia at gmail.
@Daniel_Garcia Well that makes more sense. I’ll grab one of my teensy boards and mess with it since it supports hardware spi. Not that it matters with only having 1M of strip to work with but it would be interesting to see when things start to get funky.
@Brian_Lewis Thanks for your test results, and @Daniel_Garcia thanks for clarification about maximum speed.
@Leon_Yuhanov did a quick test with a teensy 3.2. Test results were less impressive. Got the speed up to 12MHz before I ran into issues but I’m wondering if adding a level shifter would make a difference. I have some good ones from Texas instruments that I will try tomorrow.
@Brian_Lewis Do add a level shifter or a NULL pixel very close to the Teensy, you should definitely get at least 20Mhz on the SK9822. Any less and id be worried.