JohnK's latest

Collapse
This topic is closed.
X
X
 
  • Time
  • Show
Clear All
new posts
  • Dennis H
    Ultra Senior Member
    • Aug 2002
    • 3791

    JohnK's latest

    Ah, I see the board is back up and our resident EE poobah is back in residence. :LOL:

    JohnK just posted his latest effort in his search for the holy grail of transient perfect crossovers, i.e. pass a square wave that looks something like a square wave.



    This one is a spreadsheet for doing Lipshitz and Vanderkooy's subtractive filters with time delay. The advantage over his previous stuff is you can use steep filter slopes. Have you poobahs tried this?

    It would be trivial with a DSP box, somewhat harder with op amps, although doable at higher frequencies, and very difficult with passive components. The pic shows the basic DSP signal routing. I left off the EQ stuff. Separate from the crossover filters, you would need to EQ each driver as flat as possible (close mic'd) on each side of the crossover frequency and time align the two drivers' acoustic centers. Then you would probably do your baffle step and voicing on the input side.

  • JonMarsh
    Mad Max Moderator
    • Aug 2000
    • 15261

    #2
    JohnK sure has put a lot of work into this.

    Back in the 70's I built "transient perfect" speakers with both passive and active crossovers, including some experiments with the phase link driver approach pioneered at B&O. I also had the opportunity to compare these systems with some other fairly top flight designs of the time, such as B&W and Snell.

    One of the real drawbacks of the TPF designs for a simple WMT system was the inherent lobing in the vertical axis of the crossover. As a result, a system designed correctly for axial pressure and time response had a rather lumpy reverberant room power response. So, you needed to do a fair amount of room treatment.

    A way around that is what Dunlavy does, or rather, did, as they've cesased operations as of Nov. 7. That was to use what's described as a D'Apollito array, such as an WMTMW. This kills a lot of the vertical off axis lobing and alleviates to a degree the lumpiness in the power response.

    Note JohnK doesn't like MTM's, apparently, and doesn't use them.

    Now, here's the crux of the matter, in my opinion. Some of the finiest speakers in the world are transient perfect, but the reason they sound as good as they do (IMO) is not because they are transient perfect in a limited axial angle (such as is possible with multi-way dynamic speakers), but because of the inherent low coloration of the driver system. In this case, I'm talking about the Quad ESL's in their various incarnations.

    In my own experience, if you can get everything else right, transient perfect is nice if you can do it. But, for an idea of it's importance and where it fits in with the overall scheme of things, compare a "transient perfect" B&O speaker with an Avalon Eclipse, Ascent, Arcus, Opus, or Eidelon, for example.

    Last, in considering all the evils which occur in the reproduction chain, I think I'd rather have a well designed conventional passive crossover that isn't transient perfect, as opposed to having a lot of additional active electronics in the signal path. (just think of all the all-pass stages's you'd need to use to get adequate delay for a midrange crossover using pure analog techniques- no DSP!). But that's just a function of my curmudgeonly irritation with conventional opamps, DSP, and the like. Why go to the trouble of building a hotshot DAC with no conventional analog signal path (opamps, electrolytic caps in the signal path, etc), and then hook it up to a ADC+DSP+DAC + op amps crossover? :?

    But, as John Stuart Mills said (he was a favorite of mine in college, long before I'd heard of Audio Amateur and Ed Dell's affinity for him),

    The peculiar evil of silencing the expression of an opinion is, that it is robbing the human race; posterity as well as the existing generation; those who dissent from the opinion, still more than those who hold it.

    So it will be interesting to see what comes of these efforts.


    Best regards,

    Jon




    Earth First!
    _______________________________
    We'll screw up the other planets later....
    the AudioWorx
    Natalie P
    M8ta
    Modula Neo DCC
    Modula MT XE
    Modula Xtreme
    Isiris
    Wavecor Ardent

    SMJ
    Minerva Monitor
    Calliope
    Ardent D

    In Development...
    Isiris Mk II updates- in final test stage!
    Obi-Wan
    Saint-Saëns Symphonique/AKA SMJ-40
    Modula PWB
    Calliope CC Supreme
    Natalie P Ultra
    Natalie P Supreme
    Janus BP1 Sub


    Resistance is not futile, it is Volts divided by Amperes...
    Just ask Mr. Ohm....

    Comment

    • Dennis H
      Ultra Senior Member
      • Aug 2002
      • 3791

      #3
      Hi Jon,

      Yep, I know where you're coming from with the "do no harm" philosophy. The thing I have a hard time getting my mind around is we put all that effort into "pure" upstream electronics and then run the signal through a passive crossover that does this to the signal. YIKES!! I know we're not "supposed" to be able to hear that but would we accept an amp or a DAC that butchered the signal like that?

      About the lobing, JohnK's favored setup gives vertical response very similar to an LR4. He uses an LR4 lowpass and adds an extra 3.5% of delay. That shifts the lobe slightly and gives an HP response that's 4th order for the first 1.7 octaves and then flattens out to 1st order below that. Should work pretty well with drivers that can handle a conventional LR4.

      Ultimately, the TacT way is probably the way to go for signal purists, at least with digital sources. The signal stays digital all the way from the transport, through the preamp and through the power amp (power DAC). Crossovers are steep linear phase. The only analog components are a choke and a cap on the speaker terminals of the power DAC to lowpass below 60K. Zero feedback anywhere in the system and the volume is controlled by adjusting the power supply voltage in the power DAC so you maintain full digital resolution at any volume. Spendy though. :LOL:

      Comment

      • JonMarsh
        Mad Max Moderator
        • Aug 2000
        • 15261

        #4
        I'd love to hear a speaker like this that would make a believer out of me. Hopefully (but unlikely) it would even exhbit this time perfect behavior over some reasonable front listening area.

        Next question, what drivers will be pick that will be pistonic in operation over a wide enough range to avoid problems with these still relatively gentle slopes? Theres a clear tradeoff between LP slope and the resulting HP, plus the summing error. 4th order Bessel LP produces a roughly symmetric response, but, for example, with a 2.2 kHz crossover doesn't protect the tweeter much at 1 kHz, and doesn't attenuate the the upper pass band of the woofer as much as I would like. This still requires a tweeter that's pretty happy down to 1 kHz, (remember, the drivers have to have a net flat acoustic transfer function for these "ideal" active filteres to work), and a woofer pretty clean to 5 kHz plus. Doesn't hardly exist in the real world. So we may have to scratch a two way.

        Three way? Let's say you combine an Accuton C44 mid with any number of good tweeters- you could make that fly with a 3500 Hz crossover, I think, using a 4th order Bessel LP, with about 220 usec HP delay. (that's a fair amount, if we have to do it analog) Then the mid to woofer crossover might be managed at about 800 Hz with some care. That would get us crossed over to a good 7" woofer, or maybe a very good 8" woofer with some careful EQ applied. Time delay is about 600 usec... ouch! Forget an allpass chain, this is definitely a DSP problem.


        From an overall engineering perspective, it always comes down to "which approach buys me the best sound for a given investment in engineering dollars in the hardware"-

        (Note I didn't say, which approach gives me the best specs for the dollar!

        My own experience and the research I've seen done, as well as anecdotal reports by guys like Linkwitz suggest to me that the ear is not particularly sensitive to the phase shifts which occur to the timing of harmonics in waveforms. This is why we don't worry so much about what a 4th order L-R does to the square wave output. The harmonics are all still their (if the crossover sums flat), there just changed slightly in their relative phase.

        How audible is that compared with cabinet and driver resonance? Or driver distortion (IM and THD)? How much does it cost to fix that, versus the audible improvement, as compared with how much it costs to fix the other problems, and the resulting subjective improvement.

        Joeseph Audio sells speakers with crossover designs by Modaferri which use what they call "Infinite slope" crossovers- certainly not transient perfect. Yet they pay a fair amount of attention to all these other details, and their speakers have won "Best of Show" awards, even for relatively unlimited price. Now, they aren't perfect, and they have some issues which I would trace to driver integration in the case of their three way, the RM33si, but they hold their own quite nicely against some speakers like Vandersteen's and Theils and Dunlavy's which do place a great emphasis on transient perfect behavior.

        It's just a typical audio quandry- you'd really like to have a full helping of column A and column B, and not pay a price penalty for it- but in the real world, that's rarely the case.


        Best regards,

        Jon




        Earth First!
        _______________________________
        We'll screw up the other planets later....
        the AudioWorx
        Natalie P
        M8ta
        Modula Neo DCC
        Modula MT XE
        Modula Xtreme
        Isiris
        Wavecor Ardent

        SMJ
        Minerva Monitor
        Calliope
        Ardent D

        In Development...
        Isiris Mk II updates- in final test stage!
        Obi-Wan
        Saint-Saëns Symphonique/AKA SMJ-40
        Modula PWB
        Calliope CC Supreme
        Natalie P Ultra
        Natalie P Supreme
        Janus BP1 Sub


        Resistance is not futile, it is Volts divided by Amperes...
        Just ask Mr. Ohm....

        Comment

        • JonMarsh
          Mad Max Moderator
          • Aug 2000
          • 15261

          #5
          Here's another idea for a speaker system that you could test and compare the audibility of the crossover characteristics somewhat indepent of the limitations of more conventional drivers: (had this idea on the train home tonight)

          A two way - RD50 ribbon tweeter on "top end", and line source of 8 6-1/2" or 7" Excel woofers on the bottom end. Nominal crossover with the "TPF"s at 800 Hz; this gives close to two octaves of response for each driver above and below the crossover, the driver performance and overlap for filter design shouldn't be an issue. In this case, the limitations of the drivers shouldn't affect the overall performance significantly, and various orders of TPF (Transient Perfect Filters) can be tried and evaluated, and compared with the simpler and less expensive conventional crossover. I don't see practical way to do the crossover without a DSP system- the required delay is too high- and I'm not sure if there's an off the shelf reasonably priced unit that could be readily adapted to this kind of evaluation. There maybe, though.

          -Jon




          Earth First!
          _______________________________
          We'll screw up the other planets later....
          the AudioWorx
          Natalie P
          M8ta
          Modula Neo DCC
          Modula MT XE
          Modula Xtreme
          Isiris
          Wavecor Ardent

          SMJ
          Minerva Monitor
          Calliope
          Ardent D

          In Development...
          Isiris Mk II updates- in final test stage!
          Obi-Wan
          Saint-Saëns Symphonique/AKA SMJ-40
          Modula PWB
          Calliope CC Supreme
          Natalie P Ultra
          Natalie P Supreme
          Janus BP1 Sub


          Resistance is not futile, it is Volts divided by Amperes...
          Just ask Mr. Ohm....

          Comment

          • Dennis H
            Ultra Senior Member
            • Aug 2002
            • 3791

            #6
            Good idea, Jon. One of the CAD-style DSP units like the QSC would probably be the easiest way to experiment. Any damage to the sound, caused by the DSP, should be equal for all the crossover topologies. Of course, that damage might be so severe that it masked any difference. :LOL:

            Linkwitz drew a circuit you could experiment with using headphones but, as I recall, nobody took him up on the challenge. That said, I've heard some people claim headphones aren't a valid test because real hearing of phase stuff depends on the crosstalk between the ears.

            For that matter, I suspect the TPF concept would make less difference with a line array as the array is already smearing the first arrival by some large part of a msec. Who knows? :LOL:

            Ah, found the Linkwitz link. http://linkwitzlab.com/phs-dist.htm He's testing at 100 Hz and the hard-core phasaholics seem to indicate that higher frequency XOs bother them more.

            Comment

            • Dennis H
              Ultra Senior Member
              • Aug 2002
              • 3791

              #7
              Now that I think about it some more, headphones and a DSP might be a pretty good way to test. You could just mix the highpass and the lowpass back together and listen to the summed result. Your reference would be straight through the DSP, bypassing all the filters. You could program up a whole bunch of topologies (B1, LR4, subtractive, whatever) and page through them as you listen. As well, you could put a scope on the output with a square wave into the input and fine tune the delay of the subtractive filters to give the best looking waveform.

              Comment

              • JonMarsh
                Mad Max Moderator
                • Aug 2000
                • 15261

                #8
                In an absolute sense, I think that's a good suggestion.

                As regards mimicing what actually happens when you try to build a multi-driver TPF crossover with non-coinciedent drivers, it would miss the rather imperfect realization that is possible. The idea of ad RD50 and Line array was attractive to me because it would still be a "real" system, with a crossover and radiation interaction, but occuring at a low enough frequency relative to the driver spacing that the polar issues would be minimal.

                -Jon




                Earth First!
                _______________________________
                We'll screw up the other planets later....
                the AudioWorx
                Natalie P
                M8ta
                Modula Neo DCC
                Modula MT XE
                Modula Xtreme
                Isiris
                Wavecor Ardent

                SMJ
                Minerva Monitor
                Calliope
                Ardent D

                In Development...
                Isiris Mk II updates- in final test stage!
                Obi-Wan
                Saint-Saëns Symphonique/AKA SMJ-40
                Modula PWB
                Calliope CC Supreme
                Natalie P Ultra
                Natalie P Supreme
                Janus BP1 Sub


                Resistance is not futile, it is Volts divided by Amperes...
                Just ask Mr. Ohm....

                Comment

                Working...
                😀
                😂
                🥰
                😘
                🤢
                😎
                😞
                😡
                👍
                👎
                Searching...Please wait.
                An unexpected error was returned: 'Your submission could not be processed because you have logged in since the previous page was loaded.

                Please push the back button and reload the previous window.'
                An unexpected error was returned: 'Your submission could not be processed because the token has expired.

                Please push the back button and reload the previous window.'
                An internal error has occurred and the module cannot be displayed.
                There are no results that meet this criteria.
                Search Result for "|||"