• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 30 subscribers
  • Views 953 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PCM1792A output attenuation - how does it work exactly?

bowerymarc
Expert 1935 points
Other Parts Discussed in Thread: PCM1792A

the PCM1792A data sheet doesn't go into any detail about how exactly the output attenuation (volume control) works exactly as far as its signal processing functionality (the control aspect is well documented).  So I need some additional information from someone at TI.

- where in the signal chain (e.g. the functional block diagram) is the attenuator?

- is it analog or digital (obviously, it's digitally controlled)

- if it's digital, what impact on the signal resolution does it have?  (e.g. is there extended resolution math, truncation, etc.)

Look forward to some details!

Thanks,
Marc

  • 0
    TI__Mastermind 27850 points
    Hi Marc,

    The output attenuation on the PCM1792A is a digital attenuation. This will limit the resolution available the greater the attenuation, you will naturally have the headroom before the attenuation effects the signal integrity of the part performance at around 127 dB SNR to the audio source.

    Justin
  • 0 in reply to Justin Bohr
    Expert 1935 points
    Justin, your answer makes very little sense and doesn't answer some key critical questions I had asked. Again - if digital, what's the impact on the signal resolution? You're multiplying 2 numbers, how is the result treated? Is it truncated? To how many bits? Where in the block diagram does the attenuation take place? Perhaps you can get one of the designers to accurately answer the questions, thanks.
  • 0 in reply to bowerymarc
    TI__Mastermind 27850 points
    Hi Marc.

    Please look at the explanation on page 27 of the PCM1792A data sheet.

    Justin
  • 0 in reply to Justin Bohr
    Expert 1935 points
    Hi Justin,
    If the data sheet had answered my questions, I would not have asked them. As it is, none of the questions were sufficiently answered, or answered at all, by any page of the data sheet, or any other post on the e2e board, or the eval kit documentation. That is why I am seeking feedback from someone who is familiar with the internals of the design of the part. Perhaps you can contact someone on the design team to get some accurate and adequately descriptive information.
    Regards,
    Marc
  • 0 in reply to bowerymarc
    TI__Mastermind 27850 points
    Hi Marc,

    As this is digital attenuation it will happen in the digital domain. As stated before as you increase the attenuation the signals resolution will be affected, aka truncation, by eventually pushing the singal into the noise floor. This will only have an audible effect when the headroom between the resolution of the signal lets say an ENOB of 16 1/2 bits, and the ENOB of the part around 22 bits, is eliminated by the digital attenuation. This will happen in the x8 oversampling Digital Filter and Function Control section.

    Justin
  • 0 in reply to Justin Bohr
    Expert 1935 points

    Justin, I guess I'm not being clear. I'm not asking for theoretical examples. I'm asking for specific details of the part. How many bits are calculated during the digital attenuation (that is, multiplier width) and is that width reduced by truncation or rounding, is dither applied? If it's happening in the interpolation filter section, is it happening before or after the interpolation? It appears you are unable to answer these questions, based on your responses, so please contact someone within TI with knowledge of the chip internals to answer. Thank you.
    Marc

  • 0 in reply to bowerymarc
    TI__Mastermind 27850 points
    Hi Marc,

    This is the information we have:
    -The digital engine is 24 bits from start to finish
    -There is no rounding, just truncation of the LSB (if reaching into the noise floor) after multiplication of the attenuation coefficient with the audio data at 24 bits.
    -We do not have confirmation on if the attenuation and truncation happens before or after interpolation, but if the engine is 24 btis through out the entire path, then I do not believe it would matter as you would get the same result since there is no rounding.

    Justin
  • 0 in reply to Justin Bohr
    Expert 1935 points

    Hi Justin,
    Thanks this is good information. We'll have to re-characterize our product in light of this, since the THD will change with attenuation (probably won't be noticeable until it's quite attenuated, but still...)
    Best,
    Marc