WM8805 doesn't work with TDA1543 in hardware mode?

[xiedidan]

Hi, I've set the WM8805 to hardware mode, with 16bit master mode I2S output.
When 192k/16bit spdif data is feed in, I measured the I2S output rates of WM8805:
BCLK=12.28M
LRCLCK=192K
Since 12.28M=2*192K*32, meaning the output is 32bits?

I read from the datasheet that in hardware mode, WM8805 generates 128fs MCLK with 192K input. And BCLK is 1/2 of the MCLK.
However, 12.28M is far more than TDA1543 could handle(9.2M).

Any way to make WM8805 work with TDA1543 in hardware mode?

 

[abraxalito]

If you read the datasheet you'll find that 192k only works in software mode.

<edit> At least that's the way I read the datasheet but reading again it says in hardware mode that all input rates are supported. Which is confusing to say the least...

PPS When I was working out how to use this chip, I recall reading that in hardware mode the BCLK is always 64X the sample rate - i.e. 32bits are clocked out per sample. The only way to get 16bits per sample (and hence 32X BCLK) was to use software mode.

 

[xiedidan]

If you read the datasheet you'll find that 192k only works in software mode.

<edit> At least that's the way I read the datasheet but reading again it says in hardware mode that all input rates are supported. Which is confusing to say the least...

PPS When I was working out how to use this chip, I recall reading that in hardware mode the BCLK is always 64X the sample rate - i.e. 32bits are clocked out per sample. The only way to get 16bits per sample (and hence 32X BCLK) was to use software mode.

Thanks abraxalito

Yes BCLK is 64X the sample rate.

I'm still confused by the I2S format - datasheet says - 'Word length of 16, 20 and 24 bits are supported'. So 32 bits per sample is not supported, but we do get such output...

If I feed 192k/24bit into WM8805, will it still output 32bits per sample? (guess it adds 8 zeros at the end of each sample)

 

[abraxalito]

I'm still confused by the I2S format - datasheet says - 'Word length of 16, 20 and 24 bits are supported'. So 32 bits per sample is not supported, but we do get such output...

I think that must be referring to the formats of the SPDIF input - it can support up to 24bits (unofficially, I think that's part of the AES/EBU standard). 32bits per sample is not possible over either of these formats - there's not enough space

If I feed 192k/24bit into WM8805, will it still output 32bits per sample? (guess it adds 8 zeros at the end of each sample)

Yes, that's my understanding - the 32bit frame will get zero padded.

 

[xiedidan]

I think that must be referring to the formats of the SPDIF input - it can support up to 24bits (unofficially, I think that's part of the AES/EBU standard). 32bits per sample is not possible over either of these formats - there's not enough space



Yes, that's my understanding - the 32bit frame will get zero padded.

Looks like I have to find an MCU for software mode then.

 

[abraxalito]

I can recommend LPC1114 - it has hardware I2C and comes in a DIY-friendly DIL28 package. Or for a finished one on a PCB you can't go far wrong with this one for $8 : http://www.mouser.com/ProductDetail/STMicroelectronics/STM32F0DISCOVERY/?qs=sGAEpiMZZMvFPGEOwQcrYysX8Y3LPqzB%2f58XePqj%2fvE%3d

 

[xiedidan]

I can recommend LPC1114 - it has hardware I2C and comes in a DIY-friendly DIL28 package. Or for a finished one on a PCB you can't go far wrong with this one for $8 : STM32F0DISCOVERY STMicroelectronics | Mouser

That's nice.