Linux/PROCESSOR-SDK-AM335X: Multiple McASP I2S Audio Input Serializers

Jarrod,

Our expert on this is out today, but we should be able to give you some feedback early next week.

Could you please share what version of the SDK you are working with?

Thanks.

Ron,

Alright, that sounds great!  Thank you.  I started with the latest (05_01_00_11), so the kernel version 4.14.

Thanks,

Jarrod

Peter,

Thank you very much for the reply!  Sorry for the delay i was caught up into some other things.  Anyways the information you gave me has pointed me in the correct direction I think.  I didn't realize the channels were organized in the slots in that way, which is why i though everything was all mixed up!  I recorded from the raw PCM and opened the file and everything was on the correct channel for the 6 channel configuration you presented above, i will just have to redo my asound.conf to map them correctly. 

The other problems i am having that prevented me from realizing the channels are simply ordered different was a manufacturing issue with the microphones and something with the WM8580 ADC settings.  I thought the slots were messed up causing the garbled audio with the 2 channels for the ADC of the WM8580 but turns out something must be wrong in the WM8580 setup because it is not putting the data on the I2S bus properly, I verified by tieing the I2S data line from the ADC directly to a DAC output data line and sure enough the same garbled up audio, so nothing with the McASP setup.  I haven't been able to figure out the problem quite yet with the WM8580 setup, but i will keep working at it.  As for the microphones they were accidentally washed during manufacture and it damaged them, they just needed replaced.  I have multiple devices on the I2S bus so my setup is rather weird, as you pointed out the WM8580 only has 1 stereo capture, i actually have a couple I2S Mics and a Synthesizer chip (generates I2S audio data from MIDI input) providing the other inputs.  Here is a very simplified block diagram of what I have going on:

The Synthesizer chip and Microphones need no configuration and they are working properly now, the synthesizer chip has to be the master so my setup is rather strange, pretending the WM8580 has 3 stereo captures so i can expose as 6 channels. 

Anyway i think you have given me the information to point me in the right direction.  I will keep working at the WM8580 ADC setup issue and post back my final asound.conf changes once i get everything working fully!  Thanks Peter!

Thanks,

Jarrod

Krunal,

Thanks for the help!  Here is the corrected asound.conf in case anyone stumbling across was curious:

#Our output mixer/controller
pcm.dmixed {
    type dmix
    ipc_key 2048
    ipc_perm 0666
    slave {
        pcm "hw:0,0"
        format S32_LE
        channels 2
        rate 48000
        period_size 682
        buffer_size 10912
    }
    bindings {
       0 0
       1 1
    }
}

#Our main 6 channel input mixer/controller
pcm.imixed {
    type dsnoop
    ipc_key 3025
    ipc_perm 0666
    slave {
        pcm "hw:0,1"
        channels 6
        format S32_LE
        rate 48000
    }
    bindings {
        0 0
        1 1
        2 2
        3 3
        4 4
        5 5
    }
}

#This puts stereo out on just the "front"(VOUT1) jack
pcm.frontx {
    type plug
    slave {
        pcm "dmixed"
        channels 2
    }
    ttable.0.0 1
    ttable.1.1 1
}

#Define the 2 channel line-in
pcm.lineinx {
    type plug
    slave {
        pcm "imixed"
        channels 6
    }
    ttable.0.0 1
    ttable.1.3 1
}

#Define the 2 channel mic-in
pcm.micinx {
    type plug
    slave {
        pcm "imixed"
        channels 6
    }
    ttable.0.1 1
    ttable.1.4 1
}

#Define the 2 channel dream-in
pcm.dreaminx {
    type plug
    slave {
        pcm "imixed"
        channels 6
    }
    ttable.0.2 1
    ttable.1.5 1
}

The channel alignment was my big problem with using the multiple serializers and there were just other issues confusing it.  I still have an issue with my WM8580 codec ADC not behaving properly as a slave, it seems to be an MCLK issue, but since the master is not the AM335x either but another chip entirely this is not an issue for this forum, so yeah all solved from the AM335x side of things : ).

Thanks!

Jarrod