Loading ...
Sorry, an error occurred while loading the content.

Re: [regsaudioforum] low frequencies on vinyl

Expand Messages
  • Jose Antonio Almagro Pastor
    Of course there s no limitation below 20Hz in vynil nor CD (I can see my woofers moving) but why do you want a speaker with no phase shifts if you have an
    Message 1 of 28 , Jun 1, 2008
    • 0 Attachment
          Of course there's no limitation below 20Hz in vynil nor CD (I can see my woofers moving) but why do you want a speaker with no phase shifts if you have an oscillating system rotating phase (a room with eigen-modes).
       
          Phase is very important to feel dry bass, but that important phase linearity it's not in LF but in the mid region. You can try using Smaartlive to get a flat phase in LF and spend a lot of hours trying to solve a natural problem that happens, and our ears-brain know well.
       
          Best regards.

       
      2008/6/1, Ted Rook <rooknrol@...>:

      Thank you Robert, it would also be interesting to hear the views of a cutting engineer on
      extreme LF limitations of vinyl, Goran?

      Yes I agree about the phase linearity effect but how would you quantify it subjectively I
      wonder? I am inclined to believe it is a secondary effect, coming after the importance of the
      amplitude extension that allows the primary signal to be heard. It was interesting to discover
      the principal behind the effect of LF phase linearity being greater than HF phase linearity, I
      seem to recall coming across it in some AES papers on high pass filters in amplifiers. If I
      recall correctly the mathematics of the phase shift effect have opposite signs at the HF and
      LF ends of things, causing a much greater extension of the LF phase shift up the band than
      the HF shift does coming down the band, which is what you said earlier.

      If what we were discussing was a CD issue I would be concerned but the effect with vinyl of a
      subsonic filter on the system resonance is a measureable and audible improvement.

      Long term I have a project in preparation to make some measurements at extreme LF using
      my system as a test setup.

      Ted



      On 1 Jun 2008 at 0:13, Robert Greene wrote:

      >
      > It is quite true that many of the old recordings were bass shy--RCAs
      > in particular. I think everyone who has listened to them carefully
      > realizes this! They are very midrangey.
      >
      > But that does not mean that vinyl CANNOT accommodate low frequencies.
      > There is no reason why one cannot inscribe an 18 Hz note on a vinyl
      > record--and play it back if you have the right kind of playback.
      > In practice, this was not done much in the old days. But it could
      > have been. There is not much intrinsic limitation to the medium in
      > the bottom end until you get down to the warp deep- subsonic(10 Hz
      > or lower) region(and even that can be dealt with if one is clever
      > enough).
      > It is an historical accident that not much is on records way down in
      > olden times--plus the fact that most music has not got much below 40
      > Hz. (the bottom note of the piano is 27 Hz but one mostly hears it
      > via harmonics--there is not a lot of energy in the fundamental).
      >
      > There is no point in extension to daylight, but there is actually
      > some point in flat to DC.
      > Namely--and we have discussed this before but apparently without its
      > sticking--a high pass filter causes phase shift--audible phase shift.
      > In well-known experiments, KEF EQed a speaker to be flat to 5 Hz(as
      > I recall). Of course the speaker could not really produce much at 5
      > Hz but it was EQed flat in principle to that point albeit only for
      > very small signals in practice.
      >
      > Then KEF played orchestral music material that had almost nothing
      > below 40 Hz(orchestral music has no fundamentals below 40 Hz
      > usually). Sibelius, I think, among other things.
      > Everyone could hear easily the effect of switching the EQ in and out-
      > -even though the frequency response shifted essentially not at all
      > in the area of 40 Hz on up.
      >
      > What was happening? The EQ was fixing the phase further up and the
      > bass became cleaner and more defined. (People could hear pitches
      > more clearly etc.).
      >
      > This is real.
      >
      > It is important not to take measurement equipment and decide that it
      > proves something it does not prove. Even music that has very little
      > below 40 Hz still sounds more nearly correct with correct phase
      > behvior from 40 Hz on up.
      > And the only way to get that (in a minimum-phase system) is to have
      > the system flat to very low frequencies--in principle to DC.
      >
      > I know it sounds like a paradox but the mathematics is
      > incontrovertible as is the audible effect. Phase shifts from
      > nonflatness do not just happen right where the nonflatness is--they
      > spread all over. (This effect is smaller for low pass filters which
      > is why not much changes if you use a tweeter that goes out further,
      > not much compared to EQing the bottom end anyway).
      > The relationship between phase and frequency response in a minimum
      > phase system is an integral transform as it is called: localized
      > changes in response have phase effects all over the frequency range
      > at least to some extent. And in high pass filters these effects can
      > be important!
      >
      > REG
      >
      >
      > ------------------------------------
      >
      > Yahoo! Groups Links
      >
      >
      >
      >
      >
      > --
      > No virus found in this incoming message.
      > Checked by AVG.
      > Version: 7.5.524 / Virus Database: 269.24.4/1475 - Release Date: 5/30/2008 2:53 PM
      >


    • Roberto Ripio
      ... No room is a high pass filter... Regards, Roberto
      Message 2 of 28 , Jun 1, 2008
      • 0 Attachment
        Jose Antonio Almagro Pastor wrote:
        > Of course there's no limitation below 20Hz in vynil nor CD (I can
        > see my woofers moving) but why do you want a speaker with no phase
        > shifts if you have an oscillating system rotating phase (a room with
        > eigen-modes).

        No room is a high pass filter...

        Regards,

        Roberto
      • Ted Rook
        There s an implication here not just about vinyl where the thread began but LF from all sources. I m thinking of subwoofer integration about which I have some
        Message 3 of 28 , Jun 1, 2008
        • 0 Attachment
          There's an implication here not just about vinyl where the thread began but LF from all
          sources. I'm thinking of subwoofer integration about which I have some knowledge of the
          theory but no practical experience so someone correct me if I'm wrong.

          In the KEF experiment quoted it seems a single speaker was EQd and there were probably
          precautions to ensure the electronics were flat and didn't screw up the phase picture.

          Single speaker systems are not usually extended this way, the addition of a subwoofer has
          become a popular method.

          In a typical system the woofer output and the roll off in amplitude is it seems accompanied by
          relative phase shifts, presumably increasing in amount as the frequency moves away from
          the pass band. Now if we simply add a subwoofer the original woofer output persists, with its
          associated phase shift. It would seem desireable to cut off the woofer output where the sub
          woofer takes over. But if a filter is added to cut off the low end extension of the original
          woofer doesn't this introduce a phase shift itself?

          Looked at another way the issue seems to be how to extend the bandwidth of a system
          downwards towards DC while maintaining phase coherence with the main system. The
          question arises how to cut off the unwanted phase distorted LF output of the original woofer
          without introducing additional phase errors. It seems to me that these conditions must be
          satisfied if the desired goal of flat phase shift up into the audio band is to be realised.
          Thoughts on this are welcome.

          Ted
        • Ted Rook
          For those interested here s the AES article that discusses high pass filter phase requirements thoroughly (over my head in parts) but illuminating about the
          Message 4 of 28 , Jun 1, 2008
          • 0 Attachment
            For those interested here's the AES article that discusses high pass filter phase requirements thoroughly (over my head in parts) but illuminating about the differences between high and low pass bandwidth for the same phase linearity. Personally I suspect that we are all quite used to hearing high pass filter phase shift effects without realising it and without the option of doing anything about it.

            AES preprint #1415 Leach 1978

            ABSTRACT
            A rational basis is presented for the maximum bandwidth and maximum
            slew rate required in audio amplifiers. The bandwidth requirements
            are obtained by examining the phase and differential time delay distortions after the group or signal delay has been accounted for. The slew-rate requirements are obtained by examining the worst case signal conditions for a given bandwidth and power output. It is shown that a high-pass filter with a cutoff frequency in the signal bandwidth can reduce the amplifier overload margin by a factor of two.





            Next this is the AES KEF experiment Robert referred to, interesting reading, at the start of the CD era.


            AES preprint #2056 Fincham 1983

            Abstract
            Analog recordings always have high group delay at low
            frequencies due to the combined effects of all the components
            in the record / replay chain, and in particular the analog.
            recorder. Digital recorders now make it practical to
            remove much of this group delay. This paper will discuss
            whether it is worthwhile to produce a record / replay chain
            having uniform group delay down to the lowest audible
            frequencies and a demonstration will be given Of such a
            system.

            Ted


            On 1 Jun 2008 at 22:09, Ted Rook wrote:

            > There's an implication here not just about vinyl where the thread began but LF from all
            > sources. I'm thinking of subwoofer integration about which I have some knowledge of the
            > theory but no practical experience so someone correct me if I'm wrong.
            >
            > In the KEF experiment quoted it seems a single speaker was EQd and there were probably
            > precautions to ensure the electronics were flat and didn't screw up the phase picture.
            >
            > Single speaker systems are not usually extended this way, the addition of a subwoofer has
            > become a popular method.
            >
            > In a typical system the woofer output and the roll off in amplitude is it seems accompanied by
            > relative phase shifts, presumably increasing in amount as the frequency moves away from
            > the pass band. Now if we simply add a subwoofer the original woofer output persists, with its
            > associated phase shift. It would seem desireable to cut off the woofer output where the sub
            > woofer takes over. But if a filter is added to cut off the low end extension of the original
            > woofer doesn't this introduce a phase shift itself?
            >
            > Looked at another way the issue seems to be how to extend the bandwidth of a system
            > downwards towards DC while maintaining phase coherence with the main system. The
            > question arises how to cut off the unwanted phase distorted LF output of the original woofer
            > without introducing additional phase errors. It seems to me that these conditions must be
            > satisfied if the desired goal of flat phase shift up into the audio band is to be realised.
            > Thoughts on this are welcome.
            >
            > Ted
            >
            > ------------------------------------
            >
            > Yahoo! Groups Links
            >
            > <*> To visit your group on the web, go to:
            >     http://groups.yahoo.com/group/regsaudioforum/
            >
            > <*> Your email settings:
            >     Individual Email | Traditional
            >
            > <*> To change settings online go to:
            >     http://groups.yahoo.com/group/regsaudioforum/join
            >     (Yahoo! ID required)
            >
            > <*> To change settings via email:
            >     mailto:regsaudioforum-digest@yahoogroups.com
            >     mailto:regsaudioforum-fullfeatured@yahoogroups.com
            >
            > <*> To unsubscribe from this group, send an email to:
            >     regsaudioforum-unsubscribe@yahoogroups.com
            >
            > <*> Your use of Yahoo! Groups is subject to:
            >     http://docs.yahoo.com/info/terms/
            >
            >
            >
            > --
            > No virus found in this incoming message.
            > Checked by AVG.
            > Version: 7.5.524 / Virus Database: 269.24.4/1476 - Release Date: 5/31/2008 12:25 PM
            >

             
          • Fred
            Somewhat complex when considering left and right channels with convoluted phase components projected from one sub woofer. Have you tried good headphones for
            Message 5 of 28 , Jun 2, 2008
            • 0 Attachment
              Somewhat complex when considering left and right
              channels with convoluted phase components projected
              from one sub woofer. Have you tried good headphones
              for comparison?

              Fred.


              --- Ted Rook <rooknrol@...> wrote:

              > There's an implication here not just about vinyl
              > where the thread began but LF from all
              > sources. I'm thinking of subwoofer integration about
              > which I have some knowledge of the
              > theory but no practical experience so someone
              > correct me if I'm wrong.
              >
              > In the KEF experiment quoted it seems a single
              > speaker was EQd and there were probably
              > precautions to ensure the electronics were flat and
              > didn't screw up the phase picture.
              >
              > Single speaker systems are not usually extended this
              > way, the addition of a subwoofer has
              > become a popular method.
              >
              > In a typical system the woofer output and the roll
              > off in amplitude is it seems accompanied by
              > relative phase shifts, presumably increasing in
              > amount as the frequency moves away from
              > the pass band. Now if we simply add a subwoofer the
              > original woofer output persists, with its
              > associated phase shift. It would seem desireable to
              > cut off the woofer output where the sub
              > woofer takes over. But if a filter is added to cut
              > off the low end extension of the original
              > woofer doesn't this introduce a phase shift itself?
              >
              > Looked at another way the issue seems to be how to
              > extend the bandwidth of a system
              > downwards towards DC while maintaining phase
              > coherence with the main system. The
              > question arises how to cut off the unwanted phase
              > distorted LF output of the original woofer
              > without introducing additional phase errors. It
              > seems to me that these conditions must be
              > satisfied if the desired goal of flat phase shift up
              > into the audio band is to be realised.
              > Thoughts on this are welcome.
              >
              > Ted
              >



              __________________________________________________________
              Sent from Yahoo! Mail.
              A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html
            • Roberto Ripio
              Well, first a salutation to the list, that I have been reading with great interest. DSP can make make the phase linear while still having a LF cutoff that
              Message 6 of 28 , Jun 2, 2008
              • 0 Attachment
                Well, first a salutation to the list, that I have been reading with
                great interest.

                DSP can make make the phase linear while still having a LF cutoff that
                preserves the woofers life. I have that setup made with a computer
                convolver.

                The setup you are describing is in fact a crossover between woofer and
                sub, and so, if properly made, it should combine in the listening
                position to an all pass filter: Unity gain (ahem...), and phase
                rotations. But this is how 99% of multiway loudspeaker aims to work. I
                tend to think that the low end high pass gives a different effect, where
                the group delay bends, and that bending propagates upwards. I can can't
                argue why this is worst, or if the crossover to another driver stops the
                group delay shift upwards propagation in a beneficial manner (probably so).

                But provided a linear phase crossover, and equalization of peaks in
                in-room response, what leaves in a more a less flat response with
                (hopefully) narrow valleys, that is, high Q notches. That seems not to
                be very intrusive, so maybe the violent phase rotations that goes with
                them, confined in narrow frequency bands, doesn't hurt a lot either.

                As I said, I have a loudspeaker that meets the above conditions, and
                have a corrected low end phase. The effect, as I see it, is more
                notorious with bass of impulsive nature (so to speak), where it gives a
                more realistic sense of impact, more "tactile", or less artificial.

                All of the above to say that DSP can easily answer your last question.

                Regards,

                Roberto

                Ted Rook wrote:
                > There's an implication here not just about vinyl where the thread began but LF from all
                > sources. I'm thinking of subwoofer integration about which I have some knowledge of the
                > theory but no practical experience so someone correct me if I'm wrong.
                >
                > In the KEF experiment quoted it seems a single speaker was EQd and there were probably
                > precautions to ensure the electronics were flat and didn't screw up the phase picture.
                >
                > Single speaker systems are not usually extended this way, the addition of a subwoofer has
                > become a popular method.
                >
                > In a typical system the woofer output and the roll off in amplitude is it seems accompanied by
                > relative phase shifts, presumably increasing in amount as the frequency moves away from
                > the pass band. Now if we simply add a subwoofer the original woofer output persists, with its
                > associated phase shift. It would seem desireable to cut off the woofer output where the sub
                > woofer takes over. But if a filter is added to cut off the low end extension of the original
                > woofer doesn't this introduce a phase shift itself?
                >
                > Looked at another way the issue seems to be how to extend the bandwidth of a system
                > downwards towards DC while maintaining phase coherence with the main system. The
                > question arises how to cut off the unwanted phase distorted LF output of the original woofer
                > without introducing additional phase errors. It seems to me that these conditions must be
                > satisfied if the desired goal of flat phase shift up into the audio band is to be realised.
                > Thoughts on this are welcome.
                >
                > Ted
                >
                > ------------------------------------
                >
                > Yahoo! Groups Links
                >
                >
                >
                >
              • Ted Rook
                When you see your woofers moving the motion frequency is many octaves below audio, this is not useful musical information it is caused by the limitations of
                Message 7 of 28 , Jun 2, 2008
                • 0 Attachment
                  When you see your woofers moving the motion frequency is many octaves below audio, this
                  is not useful musical information it is caused by the limitations of vinyl playback warp and
                  ripple on the disc surface are in the frequency range 0.5Hz to about 10Hz and these are
                  visible but not audible.

                  There are high pass filters everywhere in the audio chain, one of the major ones used to be
                  the analog tape used to master recordings and drive the vinyl disc cutting lathe, with analog
                  tape as the source there is no useful information at subsonic frequencies because the tape
                  medium does not go that low down, it has no response at DC and not much at 10Hz. So the
                  fact that you can see your woofer move proves that the stylus can generate subsonic signals
                  unfortunately those signals are not useful audio.

                  Ted


                  On 1 Jun 2008 at 13:55, Jose Antonio Almagro Pastor wrote:

                  >
                  > Of course there's no limitation below 20Hz in vynil nor CD (I can see my woofers moving) but
                  > why do you want a speaker with no phase shifts if you have an oscillating system rotating phase
                  > (a room with eigen-modes).
                  >
                  > Phase is very important to feel dry bass, but that important phase linearity it's not in LF but in
                  > the mid region. You can try using Smaartlive to get a flat phase in LF and spend a lot of hours
                  > trying to solve a natural problem that happens, and our ears-brain know well.
                  >
                  > Best regards.
                  >
                  >
                  > 2008/6/1, Ted Rook <rooknrol@...>:
                  > Thank you Robert, it would also be interesting to hear the views of a cutting engineer on
                  > extreme LF limitations of vinyl, Goran?
                  >
                  > Yes I agree about the phase linearity effect but how would you quantify it subjectively I
                  > wonder? I am inclined to believe it is a secondary effect, coming after the importance of the
                  > amplitude extension that allows the primary signal to be heard. It was interesting to discover
                  > the principal behind the effect of LF phase linearity being greater than HF phase linearity, I
                  > seem to recall coming across it in some AES papers on high pass filters in amplifiers. If I
                  > recall correctly the mathematics of the phase shift effect have opposite signs at the HF and
                  > LF ends of things, causing a much greater extension of the LF phase shift up the band than
                  > the HF shift does coming down the band, which is what you said earlier.
                  >
                  > If what we were discussing was a CD issue I would be concerned but the effect with vinyl of
                  > a
                  > subsonic filter on the system resonance is a measureable and audible improvement.
                  >
                  > Long term I have a project in preparation to make some measurements at extreme LF using
                  > my system as a test setup.
                  >
                  > Ted
                  >
                  >
                  > On 1 Jun 2008 at 0:13, Robert Greene wrote:
                  >
                  > >
                  > > It is quite true that many of the old recordings were bass shy--RCAs
                  > > in particular. I think everyone who has listened to them carefully
                  > > realizes this! They are very midrangey.
                  > >
                  > > But that does not mean that vinyl CANNOT accommodate low frequencies.
                  > > There is no reason why one cannot inscribe an 18 Hz note on a vinyl
                  > > record--and play it back if you have the right kind of playback.
                  > > In practice, this was not done much in the old days. But it could
                  > > have been. There is not much intrinsic limitation to the medium in
                  > > the bottom end until you get down to the warp deep- subsonic(10 Hz
                  > > or lower) region(and even that can be dealt with if one is clever
                  > > enough).
                  > > It is an historical accident that not much is on records way down in
                  > > olden times--plus the fact that most music has not got much below 40
                  > > Hz. (the bottom note of the piano is 27 Hz but one mostly hears it
                  > > via harmonics--there is not a lot of energy in the fundamental).
                  > >
                  > > There is no point in extension to daylight, but there is actually
                  > > some point in flat to DC.
                  > > Namely--and we have discussed this before but apparently without its
                  > > sticking--a high pass filter causes phase shift--audible phase shift.
                  > > In well-known experiments, KEF EQed a speaker to be flat to 5 Hz(as
                  > > I recall). Of course the speaker could not really produce much at 5
                  > > Hz but it was EQed flat in principle to that point albeit only for
                  > > very small signals in practice.
                  > >
                  > > Then KEF played orchestral music material that had almost nothing
                  > > below 40 Hz(orchestral music has no fundamentals below 40 Hz
                  > > usually). Sibelius, I think, among other things.
                  > > Everyone could hear easily the effect of switching the EQ in and out-
                  > > -even though the frequency response shifted essentially not at all
                  > > in the area of 40 Hz on up.
                  > >
                  > > What was happening? The EQ was fixing the phase further up and the
                  > > bass became cleaner and more defined. (People could hear pitches
                  > > more clearly etc.).
                  > >
                  > > This is real.
                  > >
                  > > It is important not to take measurement equipment and decide that it
                  > > proves something it does not prove. Even music that has very little
                  > > below 40 Hz still sounds more nearly correct with correct phase
                  > > behvior from 40 Hz on up.
                  > > And the only way to get that (in a minimum-phase system) is to have
                  > > the system flat to very low frequencies--in principle to DC.
                  > >
                  > > I know it sounds like a paradox but the mathematics is
                  > > incontrovertible as is the audible effect. Phase shifts from
                  > > nonflatness do not just happen right where the nonflatness is--they
                  > > spread all over. (This effect is smaller for low pass filters which
                  > > is why not much changes if you use a tweeter that goes out further,
                  > > not much compared to EQing the bottom end anyway).
                  > > The relationship between phase and frequency response in a minimum
                  > > phase system is an integral transform as it is called: localized
                  > > changes in response have phase effects all over the frequency range
                  > > at least to some extent. And in high pass filters these effects can
                  > > be important!
                  > >
                  > > REG
                  > >
                  > >
                  > > ------------------------------------
                  > >
                  > > Yahoo! Groups Links
                  > >
                  > >
                  > >
                  > >
                  > >
                  > > --
                  > > No virus found in this incoming message.
                  > > Checked by AVG.
                  > > Version: 7.5.524 / Virus Database: 269.24.4/1475 - Release Date: 5/30/2008 2:53 PM
                  > >
                  >
                  >
                  >
                  >
                  >
                  >
                • Robert Greene
                  While every thing said is right, it is still important to note that none of this is an intrinsic limitation of vinyl--more one of the tape machines! A 20 Hz
                  Message 8 of 28 , Jun 2, 2008
                  • 0 Attachment
                    While every thing said is right, it is still important to note
                    that none of this is an intrinsic limitation of vinyl--more one of
                    the tape machines!
                    A 20 Hz note can be cleanly cut on a vinyl record and cleanly played
                    back,too.

                    REG


                    --- In regsaudioforum@yahoogroups.com, "Ted Rook" <rooknrol@...>
                    wrote:
                    >
                    > When you see your woofers moving the motion frequency is many
                    octaves below audio, this
                    > is not useful musical information it is caused by the limitations
                    of vinyl playback warp and
                    > ripple on the disc surface are in the frequency range 0.5Hz to
                    about 10Hz and these are
                    > visible but not audible.
                    >
                    > There are high pass filters everywhere in the audio chain, one of
                    the major ones used to be
                    > the analog tape used to master recordings and drive the vinyl disc
                    cutting lathe, with analog
                    > tape as the source there is no useful information at subsonic
                    frequencies because the tape
                    > medium does not go that low down, it has no response at DC and not
                    much at 10Hz. So the
                    > fact that you can see your woofer move proves that the stylus can
                    generate subsonic signals
                    > unfortunately those signals are not useful audio.
                    >
                    > Ted
                    >
                    >
                    > On 1 Jun 2008 at 13:55, Jose Antonio Almagro Pastor wrote:
                    >
                    > >
                    > > Of course there's no limitation below 20Hz in vynil nor CD (I
                    can see my woofers moving) but
                    > > why do you want a speaker with no phase shifts if you have an
                    oscillating system rotating phase
                    > > (a room with eigen-modes).
                    > >
                    > > Phase is very important to feel dry bass, but that important
                    phase linearity it's not in LF but in
                    > > the mid region. You can try using Smaartlive to get a flat phase
                    in LF and spend a lot of hours
                    > > trying to solve a natural problem that happens, and our ears-
                    brain know well.
                    > >
                    > > Best regards.
                    > >
                    > >
                    > > 2008/6/1, Ted Rook <rooknrol@...>:
                    > > Thank you Robert, it would also be interesting to hear the
                    views of a cutting engineer on
                    > > extreme LF limitations of vinyl, Goran?
                    > >
                    > > Yes I agree about the phase linearity effect but how would
                    you quantify it subjectively I
                    > > wonder? I am inclined to believe it is a secondary effect,
                    coming after the importance of the
                    > > amplitude extension that allows the primary signal to be
                    heard. It was interesting to discover
                    > > the principal behind the effect of LF phase linearity being
                    greater than HF phase linearity, I
                    > > seem to recall coming across it in some AES papers on high
                    pass filters in amplifiers. If I
                    > > recall correctly the mathematics of the phase shift effect
                    have opposite signs at the HF and
                    > > LF ends of things, causing a much greater extension of the LF
                    phase shift up the band than
                    > > the HF shift does coming down the band, which is what you
                    said earlier.
                    > >
                    > > If what we were discussing was a CD issue I would be
                    concerned but the effect with vinyl of
                    > > a
                    > > subsonic filter on the system resonance is a measureable and
                    audible improvement.
                    > >
                    > > Long term I have a project in preparation to make some
                    measurements at extreme LF using
                    > > my system as a test setup.
                    > >
                    > > Ted
                    > >
                    > >
                    > > On 1 Jun 2008 at 0:13, Robert Greene wrote:
                    > >
                    > > >
                    > > > It is quite true that many of the old recordings were bass shy--
                    RCAs
                    > > > in particular. I think everyone who has listened to them
                    carefully
                    > > > realizes this! They are very midrangey.
                    > > >
                    > > > But that does not mean that vinyl CANNOT accommodate low
                    frequencies.
                    > > > There is no reason why one cannot inscribe an 18 Hz note on a
                    vinyl
                    > > > record--and play it back if you have the right kind of playback.
                    > > > In practice, this was not done much in the old days. But it
                    could
                    > > > have been. There is not much intrinsic limitation to the medium
                    in
                    > > > the bottom end until you get down to the warp deep- subsonic(10
                    Hz
                    > > > or lower) region(and even that can be dealt with if one is
                    clever
                    > > > enough).
                    > > > It is an historical accident that not much is on records way
                    down in
                    > > > olden times--plus the fact that most music has not got much
                    below 40
                    > > > Hz. (the bottom note of the piano is 27 Hz but one mostly hears
                    it
                    > > > via harmonics--there is not a lot of energy in the fundamental).
                    > > >
                    > > > There is no point in extension to daylight, but there is
                    actually
                    > > > some point in flat to DC.
                    > > > Namely--and we have discussed this before but apparently
                    without its
                    > > > sticking--a high pass filter causes phase shift--audible phase
                    shift.
                    > > > In well-known experiments, KEF EQed a speaker to be flat to 5 Hz
                    (as
                    > > > I recall). Of course the speaker could not really produce much
                    at 5
                    > > > Hz but it was EQed flat in principle to that point albeit only
                    for
                    > > > very small signals in practice.
                    > > >
                    > > > Then KEF played orchestral music material that had almost
                    nothing
                    > > > below 40 Hz(orchestral music has no fundamentals below 40 Hz
                    > > > usually). Sibelius, I think, among other things.
                    > > > Everyone could hear easily the effect of switching the EQ in
                    and out-
                    > > > -even though the frequency response shifted essentially not at
                    all
                    > > > in the area of 40 Hz on up.
                    > > >
                    > > > What was happening? The EQ was fixing the phase further up and
                    the
                    > > > bass became cleaner and more defined. (People could hear pitches
                    > > > more clearly etc.).
                    > > >
                    > > > This is real.
                    > > >
                    > > > It is important not to take measurement equipment and decide
                    that it
                    > > > proves something it does not prove. Even music that has very
                    little
                    > > > below 40 Hz still sounds more nearly correct with correct phase
                    > > > behvior from 40 Hz on up.
                    > > > And the only way to get that (in a minimum-phase system) is to
                    have
                    > > > the system flat to very low frequencies--in principle to DC.
                    > > >
                    > > > I know it sounds like a paradox but the mathematics is
                    > > > incontrovertible as is the audible effect. Phase shifts from
                    > > > nonflatness do not just happen right where the nonflatness is--
                    they
                    > > > spread all over. (This effect is smaller for low pass filters
                    which
                    > > > is why not much changes if you use a tweeter that goes out
                    further,
                    > > > not much compared to EQing the bottom end anyway).
                    > > > The relationship between phase and frequency response in a
                    minimum
                    > > > phase system is an integral transform as it is called: localized
                    > > > changes in response have phase effects all over the frequency
                    range
                    > > > at least to some extent. And in high pass filters these effects
                    can
                    > > > be important!
                    > > >
                    > > > REG
                    > > >
                    > > >
                    > > > ------------------------------------
                    > > >
                    > > > Yahoo! Groups Links
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > > --
                    > > > No virus found in this incoming message.
                    > > > Checked by AVG.
                    > > > Version: 7.5.524 / Virus Database: 269.24.4/1475 - Release
                    Date: 5/30/2008 2:53 PM
                    > > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    >
                  • Robert Greene
                    Quite so about DSP and phase linearization. E.g., this is one of the thing UBs system does, as he has explained earlier (and would no doubt be willing to do
                    Message 9 of 28 , Jun 2, 2008
                    • 0 Attachment
                      Quite so about DSP and phase linearization. E.g., this is one of the
                      thing UBs system does, as he has explained earlier (and would no
                      doubt be willing to do again).

                      REG

                      --- In regsaudioforum@yahoogroups.com, Roberto Ripio <filarete@...>
                      wrote:
                      >
                      > Well, first a salutation to the list, that I have been reading
                      with
                      > great interest.
                      >
                      > DSP can make make the phase linear while still having a LF cutoff
                      that
                      > preserves the woofers life. I have that setup made with a computer
                      > convolver.
                      >
                      > The setup you are describing is in fact a crossover between woofer
                      and
                      > sub, and so, if properly made, it should combine in the listening
                      > position to an all pass filter: Unity gain (ahem...), and phase
                      > rotations. But this is how 99% of multiway loudspeaker aims to
                      work. I
                      > tend to think that the low end high pass gives a different effect,
                      where
                      > the group delay bends, and that bending propagates upwards. I can
                      can't
                      > argue why this is worst, or if the crossover to another driver
                      stops the
                      > group delay shift upwards propagation in a beneficial manner
                      (probably so).
                      >
                      > But provided a linear phase crossover, and equalization of peaks
                      in
                      > in-room response, what leaves in a more a less flat response with
                      > (hopefully) narrow valleys, that is, high Q notches. That seems
                      not to
                      > be very intrusive, so maybe the violent phase rotations that goes
                      with
                      > them, confined in narrow frequency bands, doesn't hurt a lot
                      either.
                      >
                      > As I said, I have a loudspeaker that meets the above conditions,
                      and
                      > have a corrected low end phase. The effect, as I see it, is more
                      > notorious with bass of impulsive nature (so to speak), where it
                      gives a
                      > more realistic sense of impact, more "tactile", or less artificial.
                      >
                      > All of the above to say that DSP can easily answer your last
                      question.
                      >
                      > Regards,
                      >
                      > Roberto
                      >
                      > Ted Rook wrote:
                      > > There's an implication here not just about vinyl where the
                      thread began but LF from all
                      > > sources. I'm thinking of subwoofer integration about which I
                      have some knowledge of the
                      > > theory but no practical experience so someone correct me if I'm
                      wrong.
                      > >
                      > > In the KEF experiment quoted it seems a single speaker was EQd
                      and there were probably
                      > > precautions to ensure the electronics were flat and didn't screw
                      up the phase picture.
                      > >
                      > > Single speaker systems are not usually extended this way, the
                      addition of a subwoofer has
                      > > become a popular method.
                      > >
                      > > In a typical system the woofer output and the roll off in
                      amplitude is it seems accompanied by
                      > > relative phase shifts, presumably increasing in amount as the
                      frequency moves away from
                      > > the pass band. Now if we simply add a subwoofer the original
                      woofer output persists, with its
                      > > associated phase shift. It would seem desireable to cut off the
                      woofer output where the sub
                      > > woofer takes over. But if a filter is added to cut off the low
                      end extension of the original
                      > > woofer doesn't this introduce a phase shift itself?
                      > >
                      > > Looked at another way the issue seems to be how to extend the
                      bandwidth of a system
                      > > downwards towards DC while maintaining phase coherence with the
                      main system. The
                      > > question arises how to cut off the unwanted phase distorted LF
                      output of the original woofer
                      > > without introducing additional phase errors. It seems to me that
                      these conditions must be
                      > > satisfied if the desired goal of flat phase shift up into the
                      audio band is to be realised.
                      > > Thoughts on this are welcome.
                      > >
                      > > Ted
                      > >
                      > > ------------------------------------
                      > >
                      > > Yahoo! Groups Links
                      > >
                      > >
                      > >
                      > >
                      >
                    • Ted Rook
                      It seems to me the requirement goes beyond linearization in the sense of putting the phase error back to zero at the crossover point. Isn t it the case, and
                      Message 10 of 28 , Jun 2, 2008
                      • 0 Attachment
                        It seems to me the requirement goes beyond linearization in the sense of putting the phase
                        error back to zero at the crossover point. Isn't it the case, and I'm trying to understand this
                        from the theory, that there are phase errors well into the woofer pass band that are
                        unwanted. Does a digital high pass filter, such as Tact perhaps, when put in the path to the
                        original woofer amp, also correct the phase errors further up the band? How? How is
                        information about the phase error due to the woofer being a high pass filter get into the
                        syetm? In a practical sense how is the information about the phase error for this particular
                        woofer in this particular cabinet obtained?

                        Ted


                        On 3 Jun 2008 at 2:11, Robert Greene wrote:

                        > Quite so about DSP and phase linearization. E.g., this is one of the
                        > thing UBs system does, as he has explained earlier (and would no
                        > doubt be willing to do again).
                        >
                        > REG
                        >
                        > --- In regsaudioforum@yahoogroups.com, Roberto Ripio <filarete@...>
                        > wrote:
                        > >
                        > > Well, first a salutation to the list, that I have been reading
                        > with
                        > > great interest.
                        > >
                        > > DSP can make make the phase linear while still having a LF cutoff
                        > that
                        > > preserves the woofers life. I have that setup made with a computer
                        > > convolver.
                        > >
                        > > The setup you are describing is in fact a crossover between woofer
                        > and
                        > > sub, and so, if properly made, it should combine in the listening
                        > > position to an all pass filter: Unity gain (ahem...), and phase
                        > > rotations. But this is how 99% of multiway loudspeaker aims to
                        > work. I
                        > > tend to think that the low end high pass gives a different effect,
                        > where
                        > > the group delay bends, and that bending propagates upwards. I can
                        > can't
                        > > argue why this is worst, or if the crossover to another driver
                        > stops the
                        > > group delay shift upwards propagation in a beneficial manner
                        > (probably so).
                        > >
                        > > But provided a linear phase crossover, and equalization of peaks
                        > in
                        > > in-room response, what leaves in a more a less flat response with
                        > > (hopefully) narrow valleys, that is, high Q notches. That seems
                        > not to
                        > > be very intrusive, so maybe the violent phase rotations that goes
                        > with
                        > > them, confined in narrow frequency bands, doesn't hurt a lot
                        > either.
                        > >
                        > > As I said, I have a loudspeaker that meets the above conditions,
                        > and
                        > > have a corrected low end phase. The effect, as I see it, is more
                        > > notorious with bass of impulsive nature (so to speak), where it
                        > gives a
                        > > more realistic sense of impact, more "tactile", or less artificial.
                        > >
                        > > All of the above to say that DSP can easily answer your last
                        > question.
                        > >
                        > > Regards,
                        > >
                        > > Roberto
                        > >
                        > > Ted Rook wrote:
                        > > > There's an implication here not just about vinyl where the
                        > thread began but LF from all
                        > > > sources. I'm thinking of subwoofer integration about which I
                        > have some knowledge of the
                        > > > theory but no practical experience so someone correct me if I'm
                        > wrong.
                        > > >
                        > > > In the KEF experiment quoted it seems a single speaker was EQd
                        > and there were probably
                        > > > precautions to ensure the electronics were flat and didn't screw
                        > up the phase picture.
                        > > >
                        > > > Single speaker systems are not usually extended this way, the
                        > addition of a subwoofer has
                        > > > become a popular method.
                        > > >
                        > > > In a typical system the woofer output and the roll off in
                        > amplitude is it seems accompanied by
                        > > > relative phase shifts, presumably increasing in amount as the
                        > frequency moves away from
                        > > > the pass band. Now if we simply add a subwoofer the original
                        > woofer output persists, with its
                        > > > associated phase shift. It would seem desireable to cut off the
                        > woofer output where the sub
                        > > > woofer takes over. But if a filter is added to cut off the low
                        > end extension of the original
                        > > > woofer doesn't this introduce a phase shift itself?
                        > > >
                        > > > Looked at another way the issue seems to be how to extend the
                        > bandwidth of a system
                        > > > downwards towards DC while maintaining phase coherence with the
                        > main system. The
                        > > > question arises how to cut off the unwanted phase distorted LF
                        > output of the original woofer
                        > > > without introducing additional phase errors. It seems to me that
                        > these conditions must be
                        > > > satisfied if the desired goal of flat phase shift up into the
                        > audio band is to be realised.
                        > > > Thoughts on this are welcome.
                        > > >
                        > > > Ted
                        > > >
                        > > > ------------------------------------
                        > > >
                        > > > Yahoo! Groups Links
                        > > >
                        > > >
                        > > >
                        > > >
                        > >
                        >
                        >
                        >
                        > ------------------------------------
                        >
                        > Yahoo! Groups Links
                        >
                        >
                        >
                        >
                        >
                        > --
                        > No virus found in this incoming message.
                        > Checked by AVG.
                        > Version: 7.5.524 / Virus Database: 269.24.4/1478 - Release Date: 6/2/2008 7:12 AM
                        >
                      • Uli Brueggemann
                        Ted, typically passive crossover filters introduce phase changes. These filters introduce a minimum phase behaviour. In addition the sum of drivers and
                        Message 11 of 28 , Jun 2, 2008
                        • 0 Attachment
                          Ted,

                          typically passive crossover filters introduce phase changes. These filters introduce a minimum phase behaviour. In addition the sum of drivers and crossovers create excess phase. So all in all you get a mix. And the higher the filter order the easier it is to get aware.

                          TacT and others carry out a minimum phase correction which does not compensate al the phase deviations.

                          With linear phase crossover filters designed on subractive basis in addition with mixed phase correction (minimum phase and excess phase) you can overcome such limitations but this needs definitely more DSP than supplied by TacT.

                          Please see also my white paper at http://www.acourate.com/XOWhitePaper.doc for firther explanations.

                          Uli

                          On Tue, Jun 3, 2008 at 4:44 AM, Ted Rook <rooknrol@...> wrote:

                          It seems to me the requirement goes beyond linearization in the sense of putting the phase
                          error back to zero at the crossover point. Isn't it the case, and I'm trying to understand this
                          from the theory, that there are phase errors well into the woofer pass band that are
                          unwanted. Does a digital high pass filter, such as Tact perhaps, when put in the path to the
                          original woofer amp, also correct the phase errors further up the band? How? How is
                          information about the phase error due to the woofer being a high pass filter get into the
                          syetm? In a practical sense how is the information about the phase error for this particular
                          woofer in this particular cabinet obtained?

                          Ted




                        • Goran Finnberg
                          ... As I have just spent 5 days getting a Neumann VMS 66 lathe cutting equipment back to fully working for a client who moved it some 300 km to its new home I
                          Message 12 of 28 , Jun 3, 2008
                          • 0 Attachment
                            Ted Rook, 31 May 2008 20:48:33:

                            > Thank you Robert, it would also be interesting
                            > to hear the views of a cutting engineer on
                            > extreme LF limitations of vinyl, Goran?

                            As I have just spent 5 days getting a Neumann VMS 66 lathe cutting
                            equipment back to fully working for a client who moved it some 300 km to
                            its new home I have some recollecions to do.

                            The Neumann SX 68 cutting head of the very late 60´s, Neumann sales
                            leaflet SX 68-910-02-03 dated Nov.69 was specified thus:

                            Recording range: 30 - 16 000 Hz.

                            Frequency response:

                            100-10 000 Hz +- 0.5 dB.
                            40-16 000 Hz +- 1 dB.
                            30-16 000 Hz +- 2 dB.

                            Low frequencies eats land, the lower the frequency content the more land
                            is eaten. This will decrease the actual maximum lenght of each side of
                            an LP.

                            At low frequencies the distortion of the cutter drive amp together with
                            the cutting head rises due to much lower attainable feedback values due
                            to the always necessary avoidance of oscillation due to Nyquist caused
                            by low frequency phase rotation of the actual feedback from the feedback
                            windings on the SX 68 / SX74 cutter head.

                            As the actual feedback on the cutter head goes down then the control of
                            the cutting head gets worse and the actual limits below 50 Hz is shown
                            within an envelope of +- 3 dB at 27.5 Hz on the fourth side of the SX 68
                            datasheet. However the indicated normal response, shown as a black line,
                            shows the actual normal response to be down - 3 dB at 27.5 Hz going down
                            at extrapolated to be down some -17 dB at 15 Hz.

                            The very lates VMS 80 produced at the very end of this product is better
                            than the above but the day in day out operational choices made to the
                            audio being cut is no different due to physical constraints as follows:

                            Having wide stereo, ie having bass that shows time or amplitude
                            differences between channels also eats land decreasing the actual
                            playing time of the LP so is not a good thing.

                            Also wide stereo gives rise to cutter excursions eating cutter depth so
                            one must let the cutting head goes deeper in the laquer which also eats
                            land which is not a good thing.

                            So most disks are cut with a 40 Hz 12 dB/octave HP filter to protect the
                            cutter head from subsonics that will not be audible on the wast majority
                            of home stereo equipment that will use speakers that very rarely goes
                            below 50 Hz in practice.

                            And most musical intruments have nothing at all below 40 Hz too.

                            And in most cases the elliptical equalizer will be in use that will make
                            everything below 150 Hz mono, switchable in later Neumann lathes to 75,
                            150, 300 Hz, normally set to 300 Hz

                            The reason for this is not only to get longer playing length using a
                            louder cut to lessen the use of noisy vinyl but also to decrease
                            problems when pressing the actual disc - time is always money and the
                            cycle time is normally less than 10 seconds in the press and to get this
                            reliably one needs to avoid the non fill problem that happens if there
                            is to much near vertical, 45 degree, excursions that gives you a near
                            swiss alps cut where the vinyl has to fill up the actual cut in depth
                            too.

                            Failure to fill up in depth gives all sorts of strange sounding
                            distortions and to avoid it the pressing plant must increase the
                            pressing cycle time and this is not looked at with kind eyes so the
                            cutting person gets a kick in his butt to always use the elliptical EQ.

                            The LP was/is a mass produced cheap way for the masses and most of the
                            things done while cutting is to protect the cutting head from burning
                            out, achieving a louder cut while having decent playing time, and
                            lessening the chance to get thousands of disks back as unplayable on
                            normal cheap home equipment and or getting audible distortion due to non
                            fill.

                            One should NOT confuse the above what can be done given say two-three
                            days to optimize everything, doing test cuts, which costs you the laquer
                            and money for it for each test cut, taking risks that might burn up the
                            cutting head and thus a stiff price getting it operational again, and
                            say 40 seconds cycle time in the press to provide a state of the art
                            audiophile product with a sound that only a few can hear the difference
                            compared to the usual mass produced item.

                            And to most executives it is clear which product gives the best profit
                            versus quality since most people do not hear any difference or wrongs in
                            a normal disk.

                            BTW, I always use a high pass filter at 30 Hz 18 dB/ octave when playing
                            vinyl disks to get rid of the sub sonics that robs the amplifier of
                            needless power and users of reflex speakers should note that the bass
                            driver flapping induces severe distortion due to the bass driver being
                            driven in a region of no loading at all from the box resonant tube and
                            here a closed box speaker wins hands down as the enclosed air works like
                            a restoring spring that will not have the driver go into severe non
                            linearity when fed subsonics.

                            This filter is set sufficiently low that it will pass the desired music
                            content, 99.9999 percent, while getting rid of the non music subsonics
                            that only causes sonic problems.

                            I don´t agree with Robert Greene about any negative impact from this,
                            phase - delay effects etc.

                            The audible effect of such filters have been studied for a long time and
                            there is even a limit EBU specification to what is truly AUDIBLE and
                            this is not so from such a filter in most cases.




                            --
                            Best,

                            Goran Finnberg
                            The Mastering Room AB
                            Goteborg
                            Sweden

                            E-mail: mastering@...

                            Learn from the mistakes of others, you can never live long enough to
                            make them all yourself. - John Luther
                          • Goran Finnberg
                            ... http://www.bostonaudiosociety.org/pdf/bass/BASS-05-07-7704.pdf Page 9 - 10 Why You Need a High-Pass Filter by Tomlinson Holman Broadcast and standards
                            Message 13 of 28 , Jun 3, 2008
                            • 0 Attachment
                              I wrote regarding HP filters:

                              > I don´t agree with Robert Greene about any negative
                              > impact from this, phase - delay effects etc.
                              >
                              > The audible effect of such filters have been studied
                              > for a long time and there is even a limit EBU specification
                              > to what is truly AUDIBLE and this is not so from such
                              > a filter in most cases.

                              http://www.bostonaudiosociety.org/pdf/bass/BASS-05-07-7704.pdf

                              Page 9 - 10

                              Why You Need a High-Pass Filter by Tomlinson Holman


                              "Broadcast and standards organizations have perceptible group delay
                              standards, since long telephone lines are subject to phase effects."

                              "The German Post Office and Broadcasting Organization has made 70 mS at
                              50 Hz the acceptable limit, and the CCIF has made 80 mS at 50 Hz the
                              limit for imperceptibility on program material, while Bell Labs
                              concludes 70 to 90 mS at low frequencies is inaudible."





                              --
                              Best,

                              Goran Finnberg
                              The Mastering Room AB
                              Goteborg
                              Sweden

                              E-mail: mastering@...

                              Learn from the mistakes of others, you can never live long enough to
                              make them all yourself. - John Luther
                            • Robert Greene
                              I did not mean to suggest that the negative effect of high pass were spectacular in practice. But they are definitely real. On the other hand, anyone who is
                              Message 14 of 28 , Jun 3, 2008
                              • 0 Attachment
                                I did not mean to suggest that the negative effect of high pass
                                were spectacular in practice. But they are definitely real.

                                On the other hand, anyone who is looking for technically exact sound
                                reproduction is not going to be listening to records for it.
                                I listen to records a lot, but one has to be rational about what to
                                expect, especially in the context that records were made, as GF says,
                                very much as a compromised commercial product, with a few exceptions.
                                One of the nice things about CD is the good bass--this is reflected
                                in the fact that subwoofers really took off when digital sources
                                appeared.

                                REG



                                --- In regsaudioforum@yahoogroups.com, Goran Finnberg
                                <mastering@...> wrote:
                                >
                                > Ted Rook, 31 May 2008 20:48:33:
                                >
                                > > Thank you Robert, it would also be interesting
                                > > to hear the views of a cutting engineer on
                                > > extreme LF limitations of vinyl, Goran?
                                >
                                > As I have just spent 5 days getting a Neumann VMS 66 lathe cutting
                                > equipment back to fully working for a client who moved it some 300
                                km to
                                > its new home I have some recollecions to do.
                                >
                                > The Neumann SX 68 cutting head of the very late 60´s, Neumann sales
                                > leaflet SX 68-910-02-03 dated Nov.69 was specified thus:
                                >
                                > Recording range: 30 - 16 000 Hz.
                                >
                                > Frequency response:
                                >
                                > 100-10 000 Hz +- 0.5 dB.
                                > 40-16 000 Hz +- 1 dB.
                                > 30-16 000 Hz +- 2 dB.
                                >
                                > Low frequencies eats land, the lower the frequency content the
                                more land
                                > is eaten. This will decrease the actual maximum lenght of each
                                side of
                                > an LP.
                                >
                                > At low frequencies the distortion of the cutter drive amp together
                                with
                                > the cutting head rises due to much lower attainable feedback
                                values due
                                > to the always necessary avoidance of oscillation due to Nyquist
                                caused
                                > by low frequency phase rotation of the actual feedback from the
                                feedback
                                > windings on the SX 68 / SX74 cutter head.
                                >
                                > As the actual feedback on the cutter head goes down then the
                                control of
                                > the cutting head gets worse and the actual limits below 50 Hz is
                                shown
                                > within an envelope of +- 3 dB at 27.5 Hz on the fourth side of the
                                SX 68
                                > datasheet. However the indicated normal response, shown as a black
                                line,
                                > shows the actual normal response to be down - 3 dB at 27.5 Hz
                                going down
                                > at extrapolated to be down some -17 dB at 15 Hz.
                                >
                                > The very lates VMS 80 produced at the very end of this product is
                                better
                                > than the above but the day in day out operational choices made to
                                the
                                > audio being cut is no different due to physical constraints as
                                follows:
                                >
                                > Having wide stereo, ie having bass that shows time or amplitude
                                > differences between channels also eats land decreasing the actual
                                > playing time of the LP so is not a good thing.
                                >
                                > Also wide stereo gives rise to cutter excursions eating cutter
                                depth so
                                > one must let the cutting head goes deeper in the laquer which also
                                eats
                                > land which is not a good thing.
                                >
                                > So most disks are cut with a 40 Hz 12 dB/octave HP filter to
                                protect the
                                > cutter head from subsonics that will not be audible on the wast
                                majority
                                > of home stereo equipment that will use speakers that very rarely
                                goes
                                > below 50 Hz in practice.
                                >
                                > And most musical intruments have nothing at all below 40 Hz too.
                                >
                                > And in most cases the elliptical equalizer will be in use that
                                will make
                                > everything below 150 Hz mono, switchable in later Neumann lathes
                                to 75,
                                > 150, 300 Hz, normally set to 300 Hz
                                >
                                > The reason for this is not only to get longer playing length using
                                a
                                > louder cut to lessen the use of noisy vinyl but also to decrease
                                > problems when pressing the actual disc - time is always money and
                                the
                                > cycle time is normally less than 10 seconds in the press and to
                                get this
                                > reliably one needs to avoid the non fill problem that happens if
                                there
                                > is to much near vertical, 45 degree, excursions that gives you a
                                near
                                > swiss alps cut where the vinyl has to fill up the actual cut in
                                depth
                                > too.
                                >
                                > Failure to fill up in depth gives all sorts of strange sounding
                                > distortions and to avoid it the pressing plant must increase the
                                > pressing cycle time and this is not looked at with kind eyes so the
                                > cutting person gets a kick in his butt to always use the
                                elliptical EQ.
                                >
                                > The LP was/is a mass produced cheap way for the masses and most of
                                the
                                > things done while cutting is to protect the cutting head from
                                burning
                                > out, achieving a louder cut while having decent playing time, and
                                > lessening the chance to get thousands of disks back as unplayable
                                on
                                > normal cheap home equipment and or getting audible distortion due
                                to non
                                > fill.
                                >
                                > One should NOT confuse the above what can be done given say two-
                                three
                                > days to optimize everything, doing test cuts, which costs you the
                                laquer
                                > and money for it for each test cut, taking risks that might burn
                                up the
                                > cutting head and thus a stiff price getting it operational again,
                                and
                                > say 40 seconds cycle time in the press to provide a state of the
                                art
                                > audiophile product with a sound that only a few can hear the
                                difference
                                > compared to the usual mass produced item.
                                >
                                > And to most executives it is clear which product gives the best
                                profit
                                > versus quality since most people do not hear any difference or
                                wrongs in
                                > a normal disk.
                                >
                                > BTW, I always use a high pass filter at 30 Hz 18 dB/ octave when
                                playing
                                > vinyl disks to get rid of the sub sonics that robs the amplifier of
                                > needless power and users of reflex speakers should note that the
                                bass
                                > driver flapping induces severe distortion due to the bass driver
                                being
                                > driven in a region of no loading at all from the box resonant tube
                                and
                                > here a closed box speaker wins hands down as the enclosed air
                                works like
                                > a restoring spring that will not have the driver go into severe non
                                > linearity when fed subsonics.
                                >
                                > This filter is set sufficiently low that it will pass the desired
                                music
                                > content, 99.9999 percent, while getting rid of the non music
                                subsonics
                                > that only causes sonic problems.
                                >
                                > I don´t agree with Robert Greene about any negative impact from
                                this,
                                > phase - delay effects etc.
                                >
                                > The audible effect of such filters have been studied for a long
                                time and
                                > there is even a limit EBU specification to what is truly AUDIBLE
                                and
                                > this is not so from such a filter in most cases.
                                >
                                >
                                >
                                >
                                > --
                                > Best,
                                >
                                > Goran Finnberg
                                > The Mastering Room AB
                                > Goteborg
                                > Sweden
                                >
                                > E-mail: mastering@...
                                >
                                > Learn from the mistakes of others, you can never live long enough
                                to
                                > make them all yourself. - John Luther
                                >
                              • retired_old_jj
                                ... that ... What sort of length does your FIR filter have? Are you doing multiband filtering (to shorten the calculation filter length)? I wonder somewhat
                                Message 15 of 28 , Jun 3, 2008
                                • 0 Attachment
                                  --- In regsaudioforum@yahoogroups.com, Roberto Ripio <filarete@...>
                                  wrote:
                                  > DSP can make make the phase linear while still having a LF cutoff
                                  that
                                  > preserves the woofers life. I have that setup made with a computer
                                  > convolver.


                                  What sort of length does your FIR filter have? Are you doing multiband
                                  filtering (to shorten the calculation filter length)?

                                  I wonder somewhat about audible effects of such systems due to the very
                                  long impulse response, and the nonlinear nature of the human ear.
                                • Uli Brueggemann
                                  ... JJ, e.g. I use linear phase FIR filters of length 65536 taps or even 131072 taps with double precision floationg point filters and convolution. No problem
                                  Message 16 of 28 , Jun 3, 2008
                                  • 0 Attachment
                                    On Tue, Jun 3, 2008 at 8:29 PM, retired_old_jj <jj@...> wrote:

                                    What sort of length does your FIR filter have? Are you doing multiband
                                    filtering (to shorten the calculation filter length)?
                                     

                                    I wonder somewhat about audible effects of such systems due to the very
                                    long impulse response, and the nonlinear nature of the human ear.







                                    JJ,

                                    e.g. I use linear phase FIR filters of length 65536 taps or even 131072 taps with double precision floationg point filters and convolution.
                                    No problem with the tail and any audible effects.

                                    Why do you expect problems? And how is the nonlinear nature of the ear connectecto this?

                                    Uli



                                  • Roberto Ripio
                                    ... 64k samples. Yes, less would surely be enough, but it s a lot of fun to do a brickwall high pass at 18 hz... ... No, this is pretty brute force. Do you
                                    Message 17 of 28 , Jun 3, 2008
                                    • 0 Attachment
                                      retired_old_jj wrote:
                                      >
                                      > What sort of length does your FIR filter have?

                                      64k samples. Yes, less would surely be enough, but it's a lot of fun to
                                      do a brickwall high pass at 18 hz...
                                      > Are you doing multiband
                                      > filtering (to shorten the calculation filter length)?
                                      >
                                      No, this is pretty brute force. Do you mean multirate filtering? I don't
                                      have software to do that.

                                      CPU load is about 23%. That with 3 ways crossover and global
                                      equalization (8 convolutions in total). It´s a lot for general use
                                      computer but fine for a dedicated one, I think.
                                      > I wonder somewhat about audible effects of such systems due to the very
                                      > long impulse response, and the nonlinear nature of the human ear.
                                      >

                                      Please correct me if I'm wrong, but I think and interesting property of
                                      brickwall linear phase filters is that the long tail have almost only
                                      the cutoff frequency, so the oscillation gets excited only in a very
                                      narrow band. Of course, impulsive sounds in usual program material will
                                      have content in that frequency, but the tiny energy amount that gets on
                                      it are likely inaudible. I (and others) fail to notice an adverse
                                      effect, but I haven´t run a formal comparation test. So, maybe it
                                      follows the same rule as high Q resonances, and can be disregarded.

                                      Add to the above that we are talking about subsonics...

                                      Off axis imperfect summation of crossovers is another story, though.

                                      When you refer to the non linear nature of human ear, what kind of
                                      effect are you thinking on?
                                    • Roberto Ripio
                                      Hi Ted, I wasn´t clear enough. I was referring, aside for having linear phase crossovers, also to linearization of the high pass filter phase in the low end.
                                      Message 18 of 28 , Jun 3, 2008
                                      • 0 Attachment
                                        Hi Ted,

                                        I wasn´t clear enough. I was referring, aside for having linear phase
                                        crossovers, also to linearization of the high pass filter phase in the
                                        low end. When you see a group delay graph of that kind of response, it
                                        becomes clear that it has influence well into the passband (say, a
                                        decade or so). That group delay shift means dispersion of energy.

                                        I can say that I don´t notice nothing special *except*, as I said, with
                                        percussive bass (pizzicato double bass, ketteldrums...).

                                        For correction of the phase shift you have to go for it, it doesn't go
                                        with the linear phase filter per se. I do the usual near field measure
                                        to have an idea of the low end woofer roll off, then get the impulse
                                        response, and work from it.

                                        Roberto

                                        Ted Rook wrote:
                                        > It seems to me the requirement goes beyond linearization in the sense of putting the phase
                                        > error back to zero at the crossover point. Isn't it the case, and I'm trying to understand this
                                        > from the theory, that there are phase errors well into the woofer pass band that are
                                        > unwanted. Does a digital high pass filter, such as Tact perhaps, when put in the path to the
                                        > original woofer amp, also correct the phase errors further up the band? How? How is
                                        > information about the phase error due to the woofer being a high pass filter get into the
                                        > syetm? In a practical sense how is the information about the phase error for this particular
                                        > woofer in this particular cabinet obtained?
                                        >
                                        > Ted
                                        >
                                      • Ted Rook
                                        Thank you Roberto, although I m comfortable around analog audio (I used to design analog mixing consoles) but I m a beginner still learning with digital audio.
                                        Message 19 of 28 , Jun 3, 2008
                                        • 0 Attachment
                                          Thank you Roberto, although I'm comfortable around analog audio (I used to design analog
                                          mixing consoles) but I'm a beginner still learning with digital audio. When you say "you have
                                          to go for it" I interpret that as the construction of a custom filter that changes the phase with
                                          frequency but doesn't change the amplitude with frequency, by using some filter construction
                                          software, is that right? Maybe that is what Uli's software does also?

                                          Ted

                                          On 3 Jun 2008 at 22:06, Roberto Ripio wrote:

                                          > Hi Ted,
                                          >
                                          > I wasn´t clear enough. I was referring, aside for having linear phase
                                          > crossovers, also to linearization of the high pass filter phase in the
                                          > low end. When you see a group delay graph of that kind of response, it
                                          > becomes clear that it has influence well into the passband (say, a
                                          > decade or so). That group delay shift means dispersion of energy.
                                          >
                                          > I can say that I don´t notice nothing special *except*, as I said, with
                                          > percussive bass (pizzicato double bass, ketteldrums...).
                                          >
                                          > For correction of the phase shift you have to go for it, it doesn't go
                                          > with the linear phase filter per se. I do the usual near field measure
                                          > to have an idea of the low end woofer roll off, then get the impulse
                                          > response, and work from it.
                                          >
                                          > Roberto
                                        • Roberto Ripio
                                          ... Thanks! A lateral reference for that experiment, Michael Gerzon, was involved in a similar one at B&W: http://www.stereophile.com/reference/706deep/
                                          Message 20 of 28 , Jun 3, 2008
                                          • 0 Attachment
                                            Ted Rook wrote:
                                            >
                                            > Next this is the AES KEF experiment Robert referred to, interesting
                                            > reading, at the start of the CD era.
                                            >
                                            >
                                            > AES preprint #2056 Fincham 1983

                                            Thanks!
                                            A lateral reference for that experiment, Michael Gerzon, was involved in
                                            a similar one at B&W:
                                            http://www.stereophile.com/reference/706deep/
                                          • Roberto Ripio
                                            ... Exactly. Although it may be desirable to combine both functions, usually to lift the rolloff within the woofer capabilities. I´m using octave/matlab
                                            Message 21 of 28 , Jun 3, 2008
                                            • 0 Attachment
                                              Ted Rook wrote:
                                              > When you say "you have
                                              > to go for it" I interpret that as the construction of a custom filter that changes the phase with
                                              > frequency but doesn't change the amplitude with frequency, by using some filter construction
                                              > software, is that right?
                                              Exactly. Although it may be desirable to combine both functions, usually
                                              to lift the rolloff within the woofer capabilities. I´m using
                                              octave/matlab scripts.

                                              > Maybe that is what Uli's software does also?
                                              >

                                              Good question, I tried the demo but didn´t look for that.

                                              Roberto
                                            • Ted Rook
                                              Fred, I put another example in the photo folder, the most recent image, last in the folder
                                              Message 22 of 28 , Jun 3, 2008
                                              • 0 Attachment
                                                Fred, I put another example in the photo folder, the most recent image, last in the folder

                                                http://ph.groups.yahoo.com/group/regsaudioforum/photos/view/f470?b=12&m=f&o=0

                                                The horizontal scale is different but the vertical range is the same. This example is from a
                                                different piece and from CD not vinyl, it is an orchestral tutti with tympani. This is what I
                                                expect to see, plenty at 50Hz, some at 40Hz but by 30Hz almost nothing.

                                                As I said before, it doesn't bother me a bit, the content down to 30Hz has all the bass
                                                extension I'm looking for. But then, I only got my hifi set up properly in the past month after an
                                                eight year gap, so who knows where things will be in a little while, I could be looking for subs!

                                                Ted
                                              • Ted Rook
                                                Goran thank you for the reply and the links. If I understand you correctly then the engineer who has to cut test discs and put on there a resonance test band
                                                Message 23 of 28 , Jun 3, 2008
                                                • 0 Attachment
                                                  Goran thank you for the reply and the links. If I understand you correctly then the engineer
                                                  who has to cut test discs and put on there a resonance test band sweeping down to 5Hz then
                                                  the engineer is on his own I think, outside normal limits. That this can be done at all I find
                                                  surprising, I think the engineer knew his machine well.

                                                  I also note your use of a subsonic filter, I use one also 30Hz but 12dB/octave, it is this subject
                                                  that began this thread. My speakers are reflex, there was large amplitude woofer motion with
                                                  a vertical resonance from the cartridge, at about 20Hz, and I hear the difference, with the
                                                  filter it is better, the benefits outweigh any drawbacks I find. I have been encouraged to
                                                  experiment with different tone arms which I would like to do but funds do not permit it.

                                                  Ted

                                                  On 3 Jun 2008 at 18:21, Goran Finnberg wrote:

                                                  > I wrote regarding HP filters:
                                                  >
                                                  > > I don´t agree with Robert Greene about any negative
                                                  > > impact from this, phase - delay effects etc.
                                                  > >
                                                  > > The audible effect of such filters have been studied
                                                  > > for a long time and there is even a limit EBU specification
                                                  > > to what is truly AUDIBLE and this is not so from such
                                                  > > a filter in most cases.
                                                  >
                                                  > http://www.bostonaudiosociety.org/pdf/bass/BASS-05-07-7704.pdf
                                                  >
                                                  > Page 9 - 10
                                                  >
                                                  > Why You Need a High-Pass Filter by Tomlinson Holman
                                                  >
                                                  >
                                                  > "Broadcast and standards organizations have perceptible group delay
                                                  > standards, since long telephone lines are subject to phase effects."
                                                  >
                                                  > "The German Post Office and Broadcasting Organization has made 70 mS at
                                                  > 50 Hz the acceptable limit, and the CCIF has made 80 mS at 50 Hz the
                                                  > limit for imperceptibility on program material, while Bell Labs
                                                  > concludes 70 to 90 mS at low frequencies is inaudible."
                                                  >
                                                  >
                                                  >
                                                  >
                                                  >
                                                  > --
                                                  > Best,
                                                  >
                                                  > Goran Finnberg
                                                  > The Mastering Room AB
                                                  > Goteborg
                                                  > Sweden
                                                  >
                                                  > E-mail: mastering@...
                                                  >
                                                  > Learn from the mistakes of others, you can never live long enough to
                                                  > make them all yourself. - John Luther
                                                  >
                                                  >
                                                  > ------------------------------------
                                                  >
                                                  > Yahoo! Groups Links
                                                  >
                                                  >
                                                  >
                                                  >
                                                  >
                                                  > --
                                                  > No virus found in this incoming message.
                                                  > Checked by AVG.
                                                  > Version: 7.5.524 / Virus Database: 269.24.6/1480 - Release Date: 6/3/2008 7:00 AM
                                                  >
                                                • Ted Rook
                                                  Goran and Uli thank you, I ve downloaded the white paper and will study it. I wrote what follows to satisfy my own curiosity about audibility of high pass
                                                  Message 24 of 28 , Jun 3, 2008
                                                  • 0 Attachment

                                                    Goran and Uli thank you, I've downloaded the white paper and will study it.

                                                    I wrote what follows to satisfy my own curiosity about audibility of high pass (subsonic) filters, it is not original but may be of some interest. I'm sure this is familiar territory to you from your Acourate work. It is bit long-winded!

                                                    This is about analog audio filters and should not be assumed true for digital audio filters.

                                                    There are now available two pieces of hard data about LF phase shift. The first in Leach's AES paper quoted yesterday the required bandwidth is calculated for 5 degrees phase error at 20Hz and 20kHz relative to 1kHz. The precise figures vary with filter configuration but the typical number for high frequencies is a -3dB response at about 40kHz, an octave higher than the 20kHz requirement. My experience designing mixing console systems was that we used to be -3dB at 100kHz or more in order to be flat at 20kHz within less than 1dB.

                                                    At low frequencies the situation is different, according to Leach for 5 degrees error at 20Hz the LF filter corner frequency (-3dB) must be about 1Hz and this is more than 4 octaves away from 20Hz.

                                                    Bandwidth extension to 1Hz is not typical of hifi audio hardware, 10Hz yes sometimes, 1Hz rare, most equipment is many dB down by 1Hz, deliberately, there's nothing there to reproduce, and not only is there no audio down there but much audio test gear won't generate or read 1Hz, 10Hz yes. I have a DC coupled preamp but my CD player and power amplifier are AC coupled and don't reproduce DC. DC coupled amplifiers are rare in hifi, there is a significant risk of output offsets being troublesome to the woofer and crossover, which is prevented (partly) by the use of a simple input capacitor. Similarly few CD players have DC coupled outputs. So even without using a preamp at all, the minimum signal path possible, CD player direct to amp, includes two simple high pass filters in series. This leads me to the conclusion that nearly all of us is hearing recordings played back with more, probably much more, than 5 degrees phase error at 20Hz because we have numerous high pass filters in our systems.

                                                    Goran's numbers for frequency dependant delay in broadcast sound lines is the second hard data. It is reported that delay at 50Hz of more than 50milliseconds relative to the midband is required to reach a threshold of audibility. A single cycle of 50Hz audio occupies 20 milliseconds, so it seems that a delay equal to this amount is below the threshold of audibility, put another way, the signal at low frequencies can be delayed relative to the midband by more than a full cycle without the effect being audible. The figure of 50 milliseconds at 50Hz is equivalent to two and a half cycles required for the threshold of audibility. I think it safe to asume that at higher frequencies the threshold of audibility changes in the direction of the same delay being more easily noticed, or put another way, one wants less delay towards the midband and high frequency areas.

                                                    These two numbers can be compared. The first is based on the theoretical consideration that errors should be small and 5 degrees is a small proportion of one cycle 360 degrees. The second is based on audibility and states 50 milliseconds delay of 50Hz relative to midband to be inaudible, this is more than 800 degrees of phase error.

                                                    It seems likely that what we are used to hearing from our systems lies somewhere between these two extremes, more than the 5 degree figure but less than the 800 degree figure.

                                                    From theoretical considerations the introduction of a high pass filter in an otherwise linear phase system would introduce an amount of phase error at the filter corner frequency of about 45 degrees, by definition. 45 degrees represents one eighth of a cycle and less than one sixteenth of the delay considered audible in broadcast lines, it seems reasonable to expect no significant audible effect. The phase shift rates for 2nd (12dB/oct) and 3rd order (18dB/octave) filters are higher but still small compared with the 800 degree audibility criterion.

                                                    The effect of a filter at frequencies above the corner becomes proportionally less than original phase shift at the corner frequency.

                                                    Taken together these conclusions are good evidence for there being no significant audible phase effect from introducing a (lowish order) high pass filter. There may be amplitude ripple effects however. Someone care to explore that topic?

                                                    Hope there's not so many mistakes ;-)

                                                    Ted


                                                    On 3 Jun 2008 at 8:24, Uli Brueggemann wrote:

                                                    >
                                                    > Ted,
                                                    >
                                                    > typically passive crossover filters introduce phase changes. These filters introduce a minimum
                                                    > phase behaviour. In addition the sum of drivers and crossovers create excess phase. So all in all
                                                    > you get a mix. And the higher the filter order the easier it is to get aware.
                                                    >
                                                    > TacT and others carry out a minimum phase correction which does not compensate al the phase
                                                    > deviations.
                                                    >
                                                    > With linear phase crossover filters designed on subractive basis in addition with mixed phase
                                                    > correction (minimum phase and excess phase) you can overcome such limitations but this needs
                                                    > definitely more DSP than supplied by TacT.
                                                    >
                                                    > Please see also my white paper at http://www.acourate.com/XOWhitePaper.doc for firther
                                                    > explanations.
                                                    >
                                                    > Uligraphic
                                                  • Uli Brueggemann
                                                    Ted, as you say there are two data: - 5 degrees at 20 Hz corresponding to a group delay of 0.7 ms - 900 degrees at 50 Hz corresponding to a group delay of 50
                                                    Message 25 of 28 , Jun 4, 2008
                                                    • 0 Attachment
                                                      Ted,

                                                      as you say there are two data:
                                                      - 5 degrees at 20 Hz corresponding to a group delay of 0.7 ms
                                                      - 900 degrees at 50 Hz corresponding to a group delay of 50 ms (for 20 Hz this would mean 125 ms!)

                                                      According to http://en.wikipedia.org/wiki/Delay_distortion
                                                      the best source is

                                                      Blauert, J. and Laws, P "Group Delay Distortions in Electroacoustical Systems", Journal of the Acoustical Society of America, Volume 63, Number 5, pp. 1478–1483 (May 1978):

                                                      FrequencyThreshold
                                                      500 Hz3.2 ms
                                                      1 kHz2 ms
                                                      2 kHz1 ms
                                                      4 kHz1.5 ms
                                                      8 kHz2 ms

                                                      and there seem to be no data for low frequencies.

                                                      Actually I'm playing with allpass filters created by well-defined group delays and I can hear group delay differences of one cycle length (360 degrees).
                                                      But please note: I can only hear with music with transient signals (there are artifacts which I can identify blindly) AND with stereo playback (the two channels are treated with different group delays, this means I hear the interchannel behaviour). With 1 channel playback or with equally treated stereo channels I cannot hear anything.

                                                      As your intended highpass filter will be identical for both channels I believe that the 5 degrees requirement is nonsense. Even 360 degrees are safe.

                                                      Frequency ripple:
                                                      The ripple typically is quite small depending on the filter but the filter choice is not a big problem. Please also be aware that the ear is not very sensitive at low frequencies. I guess: no chance to notice it.

                                                      BTW I use a 20 Hz subsonic filter of type Nevile-Thiele 2nd order. I have never noticed any negative effect.

                                                      Uli





                                                      On Wed, Jun 4, 2008 at 3:37 AM, Ted Rook <rooknrol@...> wrote:

                                                      Goran and Uli thank you, I've downloaded the white paper and will study it.

                                                      I wrote what follows to satisfy my own curiosity about audibility of high pass (subsonic) filters, it is not original but may be of some interest. I'm sure this is familiar territory to you from your Acourate work. It is bit long-winded!

                                                      This is about analog audio filters and should not be assumed true for digital audio filters.

                                                      There are now available two pieces of hard data about LF phase shift. The first in Leach's AES paper quoted yesterday the required bandwidth is calculated for 5 degrees phase error at 20Hz and 20kHz relative to 1kHz. The precise figures vary with filter configuration but the typical number for high frequencies is a -3dB response at about 40kHz, an octave higher than the 20kHz requirement. My experience designing mixing console systems was that we used to be -3dB at 100kHz or more in order to be flat at 20kHz within less than 1dB.

                                                      At low frequencies the situation is different, according to Leach for 5 degrees error at 20Hz the LF filter corner frequency (-3dB) must be about 1Hz and this is more than 4 octaves away from 20Hz.

                                                      Bandwidth extension to 1Hz is not typical of hifi audio hardware, 10Hz yes sometimes, 1Hz rare, most equipment is many dB down by 1Hz, deliberately, there's nothing there to reproduce, and not only is there no audio down there but much audio test gear won't generate or read 1Hz, 10Hz yes. I have a DC coupled preamp but my CD player and power amplifier are AC coupled and don't reproduce DC. DC coupled amplifiers are rare in hifi, there is a significant risk of output offsets being troublesome to the woofer and crossover, which is prevented (partly) by the use of a simple input capacitor. Similarly few CD players have DC coupled outputs. So even without using a preamp at all, the minimum signal path possible, CD player direct to amp, includes two simple high pass filters in series. This leads me to the conclusion that nearly all of us is hearing recordings played back with more, probably much more, than 5 degrees phase error at 20Hz because we have numerous high pass filters in our systems.

                                                      Goran's numbers for frequency dependant delay in broadcast sound lines is the second hard data. It is reported that delay at 50Hz of more than 50milliseconds relative to the midband is required to reach a threshold of audibility. A single cycle of 50Hz audio occupies 20 milliseconds, so it seems that a delay equal to this amount is below the threshold of audibility, put another way, the signal at low frequencies can be delayed relative to the midband by more than a full cycle without the effect being audible. The figure of 50 milliseconds at 50Hz is equivalent to two and a half cycles required for the threshold of audibility. I think it safe to asume that at higher frequencies the threshold of audibility changes in the direction of the same delay being more easily noticed, or put another way, one wants less delay towards the midband and high frequency areas.

                                                      These two numbers can be compared. The first is based on the theoretical consideration that errors should be small and 5 degrees is a small proportion of one cycle 360 degrees. The second is based on audibility and states 50 milliseconds delay of 50Hz relative to midband to be inaudible, this is more than 800 degrees of phase error.

                                                      It seems likely that what we are used to hearing from our systems lies somewhere between these two extremes, more than the 5 degree figure but less than the 800 degree figure.

                                                      From theoretical considerations the introduction of a high pass filter in an otherwise linear phase system would introduce an amount of phase error at the filter corner frequency of about 45 degrees, by definition. 45 degrees represents one eighth of a cycle and less than one sixteenth of the delay considered audible in broadcast lines, it seems reasonable to expect no significant audible effect. The phase shift rates for 2nd (12dB/oct) and 3rd order (18dB/octave) filters are higher but still small compared with the 800 degree audibility criterion.

                                                      The effect of a filter at frequencies above the corner becomes proportionally less than original phase shift at the corner frequency.

                                                      Taken together these conclusions are good evidence for there being no significant audible phase effect from introducing a (lowish order) high pass filter. There may be amplitude ripple effects however. Someone care to explore that topic?

                                                      Hope there's not so many mistakes ;-)

                                                      Ted



                                                    • Ted Rook
                                                      Uli thank you for the comments and the link. The high pass filter I am currently using is done in the digital domain in the Rane DEQ60L that I use for room EQ,
                                                      Message 26 of 28 , Jun 4, 2008
                                                      • 0 Attachment
                                                        Uli thank you for the comments and the link. The high pass filter I am currently using is done
                                                        in the digital domain in the Rane DEQ60L that I use for room EQ, it is a 1/3rd octave graphic
                                                        that I am using for cut only, there is separate HP and LP with adjustable frequency. I no
                                                        longer have any concern about HP effects, they are positive sonically for LP playback.

                                                        My interest in ripple came because I wondered if the same thing happens with ripple in digital
                                                        audio low pass filters that created pre and post echoes until it was figured out how to get
                                                        really low ripple filters. People sometimes comment about changes in sound quality when
                                                        using different CD players and storage media, changes that don't fit well with the theory of
                                                        digital audio, bits are bits you know? I have a small untested theory that proposes each DA
                                                        converter digital filter IC has a different high pass filter, probably the corner frequency, the
                                                        slope and the ripple are different every time a machine is made. Then if high pass ripple
                                                        creates larger echoes then it might explain sonic differences that sometimes don't have an
                                                        explanation. This is something I will probably never get to investigate it would take maths I
                                                        don't have and I don't have the brains to learn it anymore ;-) I know you have made
                                                        experiments with CD media, your thoughts would be interesting to read.

                                                        Ted

                                                        On 4 Jun 2008 at 9:31, Uli Brueggemann wrote:

                                                        >
                                                        > Ted,
                                                        >
                                                        > as you say there are two data:
                                                        > - 5 degrees at 20 Hz corresponding to a group delay of 0.7 ms
                                                        > - 900 degrees at 50 Hz corresponding to a group delay of 50 ms (for 20 Hz this would mean 125
                                                        > ms!)
                                                        >
                                                        > According to http://en.wikipedia.org/wiki/Delay_distortion
                                                        > the best source is
                                                        >
                                                        > Blauert, J. and Laws, P "Group Delay Distortions in Electroacoustical Systems", Journal of the
                                                        > Acoustical Society of America, Volume 63, Number 5, pp. 1478-1483 (May 1978):
                                                        >
                                                        > Frequency
                                                        > Threshold
                                                        >
                                                        > 500 Hz
                                                        > 3.2 ms
                                                        >
                                                        > 1 kHz
                                                        > 2 ms
                                                        >
                                                        > 2 kHz
                                                        > 1 ms
                                                        >
                                                        > 4 kHz
                                                        > 1.5 ms
                                                        >
                                                        > 8 kHz
                                                        > 2 ms
                                                        >
                                                        >
                                                        > and there seem to be no data for low frequencies.
                                                        >
                                                        > Actually I'm playing with allpass filters created by well-defined group delays and I can hear group
                                                        > delay differences of one cycle length (360 degrees).
                                                        > But please note: I can only hear with music with transient signals (there are artifacts which I can
                                                        > identify blindly) AND with stereo playback (the two channels are treated with different group
                                                        > delays, this means I hear the interchannel behaviour). With 1 channel playback or with equally
                                                        > treated stereo channels I cannot hear anything.
                                                        >
                                                        > As your intended highpass filter will be identical for both channels I believe that the 5 degrees
                                                        > requirement is nonsense. Even 360 degrees are safe.
                                                        >
                                                        > Frequency ripple:
                                                        > The ripple typically is quite small depending on the filter but the filter choice is not a big problem.
                                                        > Please also be aware that the ear is not very sensitive at low frequencies. I guess: no chance to
                                                        > notice it.
                                                        >
                                                        > BTW I use a 20 Hz subsonic filter of type Nevile-Thiele 2nd order. I have never noticed any
                                                        > negative effect.
                                                        >
                                                        > Uli
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                        > On Wed, Jun 4, 2008 at 3:37 AM, Ted Rook <rooknrol@...> wrote:
                                                        > Goran and Uli thank you, I've downloaded the white paper and will study it.
                                                        > I wrote what follows to satisfy my own curiosity about audibility of high pass (subsonic) filters, it is
                                                        > not original but may be of some interest. I'm sure this is familiar territory to you from your Acourate
                                                        > work. It is bit long-winded!
                                                        > This is about analog audio filters and should not be assumed true for digital audio filters.
                                                        > There are now available two pieces of hard data about LF phase shift. The first in Leach's AES
                                                        > paper quoted yesterday the required bandwidth is calculated for 5 degrees phase error at 20Hz
                                                        > and 20kHz relative to 1kHz. The precise figures vary with filter configuration but the typical
                                                        > number for high frequencies is a -3dB response at about 40kHz, an octave higher than the 20kHz
                                                        > requirement. My experience designing mixing console systems was that we used to be -3dB at
                                                        > 100kHz or more in order to be flat at 20kHz within less than 1dB.
                                                        > At low frequencies the situation is different, according to Leach for 5 degrees error at 20Hz the LF
                                                        > filter corner frequency (-3dB) must be about 1Hz and this is more than 4 octaves away from 20Hz.
                                                        > Bandwidth extension to 1Hz is not typical of hifi audio hardware, 10Hz yes sometimes, 1Hz rare,
                                                        > most equipment is many dB down by 1Hz, deliberately, there's nothing there to reproduce, and not
                                                        > only is there no audio down there but much audio test gear won't generate or read 1Hz, 10Hz yes.
                                                        > I have a DC coupled preamp but my CD player and power amplifier are AC coupled and don't
                                                        > reproduce DC. DC coupled amplifiers are rare in hifi, there is a significant risk of output offsets
                                                        > being troublesome to the woofer and crossover, which is prevented (partly) by the use of a simple
                                                        > input capacitor. Similarly few CD players have DC coupled outputs. So even without using a
                                                        > preamp at all, the minimum signal path possible, CD player direct to amp, includes two simple
                                                        > high pass filters in series. This leads me to the conclusion that nearly all of us is hearing
                                                        > recordings played back with more, probably much more, than 5 degrees phase error at 20Hz
                                                        > because we have numerous high pass filters in our systems.
                                                        > Goran's numbers for frequency dependant delay in broadcast sound lines is the second hard
                                                        > data. It is reported that delay at 50Hz of more than 50milliseconds relative to the midband is
                                                        > required to reach a threshold of audibility. A single cycle of 50Hz audio occupies 20 milliseconds,
                                                        > so it seems that a delay equal to this amount is below the threshold of audibility, put another way,
                                                        > the signal at low frequencies can be delayed relative to the midband by more than a full cycle
                                                        > without the effect being audible. The figure of 50 milliseconds at 50Hz is equivalent to two and a
                                                        > half cycles required for the threshold of audibility. I think it safe to asume that at higher
                                                        > frequencies the threshold of audibility changes in the direction of the same delay being more
                                                        > easily noticed, or put another way, one wants less delay towards the midband and high frequency
                                                        > areas.
                                                        > These two numbers can be compared. The first is based on the theoretical consideration that
                                                        > errors should be small and 5 degrees is a small proportion of one cycle 360 degrees. The second
                                                        > is based on audibility and states 50 milliseconds delay of 50Hz relative to midband to be inaudible,
                                                        > this is more than 800 degrees of phase error.
                                                        > It seems likely that what we are used to hearing from our systems lies somewhere between these
                                                        > two extremes, more than the 5 degree figure but less than the 800 degree figure.
                                                        > From theoretical considerations the introduction of a high pass filter in an otherwise linear phase
                                                        > system would introduce an amount of phase error at the filter corner frequency of about 45
                                                        > degrees, by definition. 45 degrees represents one eighth of a cycle and less than one sixteenth of
                                                        > the delay considered audible in broadcast lines, it seems reasonable to expect no significant
                                                        > audible effect. The phase shift rates for 2nd (12dB/oct) and 3rd order (18dB/octave) filters are
                                                        > higher but still small compared with the 800 degree audibility criterion.
                                                        > The effect of a filter at frequencies above the corner becomes proportionally less than original
                                                        > phase shift at the corner frequency.
                                                        > Taken together these conclusions are good evidence for there being no significant audible phase
                                                        > effect from introducing a (lowish order) high pass filter. There may be amplitude ripple effects
                                                        > however. Someone care to explore that topic?
                                                        > Hope there's not so many mistakes ;-)
                                                        > Ted
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                        >
                                                      Your message has been successfully submitted and would be delivered to recipients shortly.