- Thu Aug 20, 2015 12:21 pm
#26598
In my implementation, the DMA is the limiting factor, unless you start chaining extra DMAs together. A quick calculator says around 450 with this implementation. Though I guess if you chain them it could become 900 without too much effort.
- Thu Sep 17, 2015 12:02 pm
#29268
Do you have any insight into what I2S_BCK_DIV vs I2S_BCK_DIV, they are both 4 bit (16 value) registers that clearly affect the I2S speed but it's not clear to me the difference and there are many possible combinations of values that make I2S just stop.
I'm currently working on bi-directional I2S data transfer with DMA which I'm having pretty good luck with and will be posting shortly.
I'm currently working on bi-directional I2S data transfer with DMA which I'm having pretty good luck with and will be posting shortly.
Last edited by DanielC on Mon Sep 21, 2015 4:07 pm, edited 1 time in total.
- Thu Sep 17, 2015 3:36 pm
#29285
Very nice 
I do want to do the same with APA102 Ledstrip... and update my Android App : https://www.youtube.com/watch?t=1&v=zr7ZrnF-bfM
But seems difficult under arduino firmware ? and SPI ?
I do want to do the same with APA102 Ledstrip... and update my Android App : https://www.youtube.com/watch?t=1&v=zr7ZrnF-bfM
But seems difficult under arduino firmware ? and SPI ?
- Thu Aug 18, 2016 12:09 pm
#53231
Apparently the DMA has a max buffer size of 4kB, but by rotating buffers I've been able to get more.
Right now, I'm controlling SK6812 RGBW, with 32 bit per led instead of 24, in 4 I2S bit per LED data bit period, up to 367 LEDs. As a 5M strip is 300 leds, I'm good with that.
Also on SK6812RGBW, the tolerances on the timing are awesome. I could reliably get the clock to 1333Khz instead of the specified 800Khz and all is fine (given 5V power is adequately distributed at various points in the strip).
sfranzyshen wrote:What is the maximum number of LEDs that can be driven by this method? Since the ram (memory) is the limiting factor ...
Apparently the DMA has a max buffer size of 4kB, but by rotating buffers I've been able to get more.
Right now, I'm controlling SK6812 RGBW, with 32 bit per led instead of 24, in 4 I2S bit per LED data bit period, up to 367 LEDs. As a 5M strip is 300 leds, I'm good with that.
Also on SK6812RGBW, the tolerances on the timing are awesome. I could reliably get the clock to 1333Khz instead of the specified 800Khz and all is fine (given 5V power is adequately distributed at various points in the strip).
