Question to guru Bill Roberts... phase shift, distortion

Discussion in 'Mastering' started by Alécio Costa - Brazil, Jan 2, 2003.

  1. Alécio Costa - Brazil

    Alécio Costa - Brazil Distinguished Member

    Mar 19, 2002
    Hi friends!
    I had previosly posted this issue here and at PSWbut up to now, no one had come with a very good explanation. You can talk about polos, zeroes, transfer function, delay group, phase distortion and so. I will do all that I can to understand properly this... here itgoes..
    Whenever I apply HPF´s to digital equalizer like Q10 ( Waves Q series), I sometimes face a nasty distorion no matter if I am applying up to -6d uts at some bands an that the input signal is properly under 0 db.
    some talk about distortion phase that HPF´s add to the transfer function and so..
    So, let us see if now we get a better in deep explanation
    Thanks again
    Nice 2003 to all
  2. audiowkstation

    audiowkstation Active Member

    Jun 29, 2001
    Question 1.

    Is this in 24/44.1?

    If so, I have the answers.

    If 24/higher than 44.1, I have other answers.

    If in 16 bit, another set of answers.

    Which one Alécio?
  3. Alécio Costa - Brazil

    Alécio Costa - Brazil Distinguished Member

    Mar 19, 2002
    Lots of guys tld me that this is normal, even happening with TOP digital hardware from other brands..

    The scenario you already know my dear friend: A PT TDM 5.1.1 session with Q10 inserted at a 24bit/44.1khz recording.

    Thanks Bill!
  4. audiowkstation

    audiowkstation Active Member

    Jun 29, 2001
    This is what is happening. When you EQ, you are changing the shapes of the interactions of the waves. To change the shapes properly, the upper harmonics that will be generated will be upwards of 30KhZ and 44.1 is a brick wall at 22,050HZ.

    This brick wall filtering is bad enough for one channel.

    As the wave shapes are formed with upper harmonics in mind during the render, the inability of them becoming formed causes a DC shift in the audible waves. IT will actually throw them off center and shape the zero point slightly forward of where it should be. The folowing zero point actually trys to move back and this is what causes the DC shift.

    Like I say, it is bad enough for 1 channel but get 32 of them running and this spells trouble. One way to avoid this is to use very small increases or decreases per render, usually a 1/2 octave below or above the fundamental. Doing this render several times will bunch them up again so the best thing to do is track more properly and stay away from the C-10 if possable.

    Next thing you can do is use a higher sampling freqency that will allow for more upper harmonics. You will still have the problem of adding tracks so that is one reason I do my critical equalization with the room, microphone or the source.

    I hope this helps.
    Their are other factors too as the math cannot allow even a -10 at times.
  5. Bob Olhsson

    Bob Olhsson Distinguished Member

    Feb 13, 2001
    Nashville TN
    Home Page:
    There's something even more basic going on. ALL filters ring. The sharper the filter, the greater the resonance.. You will often need to reduce the gain when you cut a frequency when you are running close to the top..
  6. audiowkstation

    audiowkstation Active Member

    Jun 29, 2001
    That is right Bob!

    This "ringing", is the "overshoot, trying to correct itself"

    That is one reason I loath 12,18 and 24dB/octave slopes on loudspeaker crossovers, passive or active.

    I even designed a 3dB per octave slope for a 2 way once because of this ringing and overshoot.. (simple 6dB audibility)and it always seems to crawl up the odd order harmonics as well. Very audible. It is odd order intermodulation distortion PLUS and VS harmonic distortion.

    I like to call it nastyness.
  7. Alécio Costa - Brazil

    Alécio Costa - Brazil Distinguished Member

    Mar 19, 2002
    Finnally, an explanation that convinces me!LOL
    In fact Bob started a very good explanation at PSW at the time I posted in there either.
    The same happens with Q4. sometimes just lowering 0.3dB is enough to achive good free distortion nice headroom.

    Nice 2003

Share This Page