192kHz sampling rate on the Soundbite board

[petr_sh]

Hi, I need to enable analog inputs with 192kHz sampling rate on the Soundbite board. I use the the standard Freescale template application configures AK4584 to generate the sampling clock for all components. In the template code I have found options for 96, 64, 48 kHz only. In my understanding I need to reconfigure the Clock Mode Control register on AK4584. Could somenody advise the proper setting for 192 kHz?

[JohnnyChen]

Please note that AK4584 does not support 192K ADC but it do can generate
LRCLK/BCLK clock for AK4556.

From the features list of AK4584 data sheet(
http://datasheet.digchip.com/025/025-00607-0-AK4584.pdf) we can see that
this codec
1. Support 24bit 2ch  96K ADC, 192K DIR
2. Support 24bit 2ch 192K DAC, 192K DIT

In order to support 192K, we should use 128fs mode and quad speed  mode as
well. In the section "System clock", that mentioned "The ADC is
powered down
during quad speed mode".

AK4556 clock has been configured as slave move with CKS3:0 = 0100. It should
support fs=192K.

I also have tried that on my two Soundbite boards. 

One is ok to generate 192k clock and get proper output for I/Os 3 to 8 with I/Os 1 and 2 'muted'.

Another one only generate 44.1K clock. I think the second board must have something wrong with codec. I didn't do further test on that. Perhaps we can read back all register setting from codec to double check it.

[petr_sh · Accepted solution]

Hi,

The configuration settings Freescale Support team provided for AK4584 correctly  generate clocks for AK4556 (the changed part in the configuration is: REG#>9=$A6; REG#>2=$E). I can see LRCLK is 192KHz. I use the audio processing chain: J3->U2(AK4556)->U1(DSP56371)->U2->J4. And the assembly template application works well with them, it passes through the 1KHz sine wave pretty clearly.

But, the problem is with the C-template application. It corrupts the input sine wave (applied to J3 input for the instance) even sampling at 96KHz, and in the more severe way at 192KHZ. So, I think there is a problem in the Soundbite C-template application in the way it configures the HW or processes the interrupts.