Is Audio MCLK always = 256 * Fs

[Amninder]

I am new to Audio Systems and want to know if MCLK in audio systems is always equal to 256* MCLK . If Yes, who so (Is it due to some oversampling requirement in CODECs). If not, how to determine this multiplication factor?

 

[Dan Mills]

256 is common, but many parts can be configured for 128/256/512/1024 as well as (sometimes) some odball ones like 384.

Delta sigma modulators like a clock at a reasonably high multiple of the sample rate, and the common serial protocols for off board use (AES3 and SPDIF) are biphase with 64 bits per frame, so 128 times LRclk is required to generate the biphase stream.

Regards, Dan.

 

[Amninder]

256 is common, but many parts can be configured for 128/256/512/1024 as well as (sometimes) some odball ones like 384.

Delta sigma modulators like a clock at a reasonably high multiple of the sample rate, and the common serial protocols for off board use (AES3 and SPDIF) are biphase with 64 bits per frame, so 128 times LRclk is required to generate the biphase stream.

Regards, Dan.

Thanks for the reply, Don.
So, is there a criteria to choose one multiple over other. For example, what would be pros/cons if I choose say 512 over 256 * Fs for a I2S based audio system?
Regards
Amninder

 

[Dan Mills]

Lots of the stuff out there just has an internal divider that can be configured to produce 128 or 256 times for internal use from something higher, so really it is pick your oscillator having drawn up a table of what all your parts can accomodate.

Usually somewhere in the 24Mhz sort or region is good as you can get sensable dividers for single (512), dual (256) and quad (128) rate from that. allowing you to for example run 48K, 96K and 192K from the same crystal, two rocks in that region can give you every rate that is in common use.

Regards, Dan.