1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

S/PDIF bit scope

Discussion in 'Recording' started by clintrubber, Dec 1, 2003.

  1. clintrubber

    clintrubber Guest

    Hi,

    Having posted this before at the TechTalk-dept I figured this would be a better place.

    What would be an easy way to determine if
    an S/PDIF (or AES/EBU) signal is 16, 20
    or 24 bits ?

    Any kind of solution would be welcome,
    for instance in the form of a few CMOS
    chips and a few LEDs, or as a check
    in a .wav-editor of the recorded signal
    or who knows there's even a dedicated
    VST-plugin out there.

    There's an article here:
    http://www.digido.com/index/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=34
    but it uses a scope. No problem, but something
    more compact would be nice.

    Thanks,

    Peter
     
  2. mjones4th

    mjones4th Active Member

    That's a good question Peter.

    I don't know of any dedicated tools to do the function. I assume you already tried to find tech specs on the SPDIF device?

    mitz
     
  3. clintrubber

    clintrubber Guest

    Thanks for the reaction.

    Yep, the info on the S/PDIF format is available on various places. It's just that it could be convenient to know what a certain device is sending - so how many relevant (active) bits are there. The standard allows for up to 24, but if a wrong assumption is made (receiving/recording it as a 16-bit word for instance while it's in fact 20 wide), the wordlength would end up being truncated.

    Bye,

    Peter
     
  4. mjones4th

    mjones4th Active Member

    True indeed.

    Try this:

    Play a file at full volume through the spdif device in question. Something limited to hell so the signal level is pretty high and relatively constant (or maybe a sine wave at full volume). Record into a SPDIF device at 24 bits.

    The result, if I'm not totally off-base here, should be a lower signal level for lower bit depths. Does this make sense, or am I totally in left field?

    or you could listen for the truncation artfacts.

    Both methods are extremely inexact, but its the best I could devise.

    What is the device you are uncertain about?

    Mitz
     
  5. clintrubber

    clintrubber Guest

    This all might (1) be all too much 'academic' or (2) really important and easily overlooked. I guess it's often clearly one of the two, but not always the same one ! :)

    Could say that

    I think it's the other way around. So the MSBs (most signif. bits) always reach 0 dB FullScale, but deep down there you have that sixteenth bit (LSB) working OK there - or you have a lonely soul that just saw his little brothers & sisters getting lost & trown away (the 17th-24th bit) !

    So it'll probably matter only for reverb tails etc, but still - could imagine some detail gets lost, even during louder passages.


    No device in particular I'm 'worried' about. it's just that I have stuff around that likely gives
    16, 18, 20 or 24-bit - so it be great to have a simple indication ! It won't always be obvious to hear (except in extreme cases) so better to see it easily than to find out afterwards.

    Bye,

    Peter
     
  6. mjones4th

    mjones4th Active Member

    Duh, you're right Peter, when going from a lower bitrate to a higher one, the msb bits would contain the signal. What was I thinking.

    Maybe try opening the recordings in a hex editor to look at the contents of the file?

    Sorry i can't help you any more than that

    mitz
     
  7. clintrubber

    clintrubber Guest

    It's OK, there's always he possibility to amplify the signal by 16 times ~6.02 dB and see what's still unclipped. If so, then there were more bits than 16.

    Maybe I'll build me something based on the link above.

    Thanks for following this,

    Peter
     
  8. mjones4th

    mjones4th Active Member

    Glad I could (umm not) help Peter.

    sdevino might be able to help you though.
     
  9. clintrubber

    clintrubber Guest

    Hey please don't ! :)
    Your reactions appreciated !

    Peter
     

Share This Page