ADAU1401A - Connecting external slave mode codecs

Hello all,

I am considering the possibility of using the ADAU1401A in my project.

I wish to use several external codecs with I2S bus. the codecs are in slave mode.
I've noticed the ADAU1401A serial input port works only in slave mode, while the output serial port can be configured as master or slave:

I believe this configuration is addressed on the ADAU1401A data sheet, page 45:


I would like to verify if I correctly understood the way the codecs should be connected to the DSP in this configuration:


Is this the right way to do it?
Second, do I have to use a clock buffers for this configuration, or the output BCLK, LRCLK can safely drive the load of several codecs inputs?
Many thanks!

Nir
Parents
  • +1
    •  Analog Employees 
    on Apr 20, 2020 6:08 PM

    Hello Nir,

    Yes, this is the way to clock this configuration. The DSP clocks the codecs and its own serial input port. 

    The buffers is a more difficult question. The propagation delay of the buffers are added to the time it takes for the clocks to reach the other part. For the data going to the codecs this should not be a problem. The delay is quite small. However, for the serial data coming back to the DSP, this delay is added to the time of flight, the capacitance of the line, the internal delay of the codec to output the next bit after receiving the next bit clock bit and the time it takes for the signal to return. The time of flight back. If you are going at a really high sampling rate like 192kHz then all of this might be a problem. For lower rates this should not be an issue.

    Then I like that you can dampen the output of each buffer separately to control the transmission line reflections. So the buffers can be handy for that reason.

    The part should be able to drive three inputs as long as the traces are not super long. But a buffer may have some advantages. So I am a little on the fence about it in your case. Hopefully I gave you enough info to make the decision based on your exact plans.

    Dave T

Reply
  • +1
    •  Analog Employees 
    on Apr 20, 2020 6:08 PM

    Hello Nir,

    Yes, this is the way to clock this configuration. The DSP clocks the codecs and its own serial input port. 

    The buffers is a more difficult question. The propagation delay of the buffers are added to the time it takes for the clocks to reach the other part. For the data going to the codecs this should not be a problem. The delay is quite small. However, for the serial data coming back to the DSP, this delay is added to the time of flight, the capacitance of the line, the internal delay of the codec to output the next bit after receiving the next bit clock bit and the time it takes for the signal to return. The time of flight back. If you are going at a really high sampling rate like 192kHz then all of this might be a problem. For lower rates this should not be an issue.

    Then I like that you can dampen the output of each buffer separately to control the transmission line reflections. So the buffers can be handy for that reason.

    The part should be able to drive three inputs as long as the traces are not super long. But a buffer may have some advantages. So I am a little on the fence about it in your case. Hopefully I gave you enough info to make the decision based on your exact plans.

    Dave T

Children
  • Hello Dave,

    I plan to use 44.1KHz sample rate, so as you wrote, it's not super fast sample rate so I believe the delay won't be significant.

    I will try to test both configuration, with/without the buffers...I intend to keep all traces short, and to use control impedance so not using the buffers may be a sufficient solution.

    Thanks again for your advice, you have helped me greatly.

    Nir