Home of the Squeezebox™ & Transporter® network music players.
Page 2 of 4 FirstFirst 1234 LastLast
Results 11 to 20 of 38
  1. #11
    Senior Member
    Join Date
    Apr 2005
    Location
    Buckinghamshire, England
    Posts
    9,983
    Quote Originally Posted by darrenyeats View Post
    I was thinking of these (figure 5 in each case):
    http://www.stereophile.com/content/b...r-measurements
    http://www.stereophile.com/content/s...r-measurements

    Though, I imagine the dithered versions would appear smooth.
    Thanks... If you are referring to figs 5 & 6, they look pretty good to me. I note they are 24-bit data, so you can ignore my 1-bit remark!
    The waveforms aren't really "staircases"... They have some HF distortion which looks like low level non-linearity to me, but really as I said these signals are very very low-level and in audible in practice. It's not clear to what Stereophile think these measurements prove in terms of sound quality? - my guess is nothing, so they are only meaningful as an engineering benchmark in comparison with other DACs.
    Regards
    Phil
    You want to see the signal path BEFORE it gets onto a CD/vinyl...it ain't what you'd call minimal...
    Touch(wired/W7)+Teddy Pardo PSU - Audiolense 3.3/2.0+INGUZ DRC - MF M1 DAC - Linn 5103 - full Aktiv 5.1 system (6x LK140's, ESPEK/TRIKAN/KATAN/SEIZMIK 10.5), Pekin Tuner, Townsend Supertweeters,VdH Toslink,Kimber 8TC Speaker & Chord Signature Plus Interconnect cables
    Stax4070+SRM7/II phones
    Kitchen Boom, Outdoors: SB Radio, Harmony One remote for everything.

  2. #12
    Senior Member
    Join Date
    Mar 2007
    Location
    UK
    Posts
    1,170
    Quote Originally Posted by Phil Leigh View Post
    Thanks... If you are referring to figs 5 & 6, they look pretty good to me. I note they are 24-bit data, so you can ignore my 1-bit remark!
    The waveforms aren't really "staircases"... They have some HF distortion which looks like low level non-linearity to me, but really as I said these signals are very very low-level and in audible in practice. It's not clear to what Stereophile think these measurements prove in terms of sound quality? - my guess is nothing, so they are only meaningful as an engineering benchmark in comparison with other DACs.
    Regards
    Phil
    Figure 6 is 24 bit data, but figure 5 is 16 bit. John Atkinson wrote: "....reproduction of an undithered tone at exactly ľ90.31dBFS (fig.5) was perfect, with excellent waveform symmetry, and clear delineation of the three DC voltage levels described by this signal." This is what I was referring to.
    Darren

  3. #13
    Senior Member Mnyb's Avatar
    Join Date
    Feb 2006
    Location
    Vństerňs Sweden
    Posts
    13,549
    Time for another bad analogy ( in good forum tradition ) .

    "out of context" is my thought ?

    If you walk up to your HD tv some millimeters away with an magnifying glass, you see nothing but dots , but you don't get any sleepless nights over that in the visual case it is accepted that the dots blur together at the correct viewing distance and becomes invisible to us . Pictures with different resolution is more or less sharp and natural to us not " dotty " or anything some such.

    Correct viewing distance would be analogues to the correct dynamic range for the music .

    But maybe it is bad form to use picture analogy it is not the same and brain behaves different .
    And picture quality is still crude compared to sound quality .


    This test is probably for testing linearity at extreme low levels.
    --------------------------------------------------------------------
    Main hifi: Touch + CIA PS +MeridianG68J MeridianHD621 MeridianG98DH 2 x MeridianDSP5200 MeridianDSP5200HC 2 xMeridianDSP3100 +Rel Stadium 3 sub.
    Bedroom/Office: Boom
    Kitchen: Touch + powered Fostex PM0.4
    Misc use: Radio (with battery)
    iPad1 with iPengHD & SqueezePad
    (in storage SB3, reciever ,controller )
    server HP proliant micro server N36L with ClearOS Linux

    http://people.xiph.org/~xiphmont/demo/neil-young.html

  4. #14
    Senior Member
    Join Date
    Mar 2007
    Location
    UK
    Posts
    1,170
    Quote Originally Posted by Mnyb View Post
    Time for another bad analogy ( in good forum tradition ) .

    "out of context" is my thought ?

    If you walk up to your HD tv some millimeters away with an magnifying glass, you see nothing but dots , but you don't get any sleepless nights over that in the visual case it is accepted that the dots blur together at the correct viewing distance and becomes invisible to us . Pictures with different resolution is more or less sharp and natural to us not " dotty " or anything some such.

    Correct viewing distance would be analogues to the correct dynamic range for the music .

    But maybe it is bad form to use picture analogy it is not the same and brain behaves different .
    And picture quality is still crude compared to sound quality .


    This test is probably for testing linearity at extreme low levels.
    That makes sense to me. But this seems to go against the idea the analogue waveform is re-created as a continuous curve which intersects the data points. Instead, it seems the values are just output as a voltage during each sample interval.

    Even if the above is true, I would assume that all the nastiness of the jaggedness is already inherent to the PCM signal i.e. quantisation error and that this is normally changed from distortion into noise by dithering, making it inaudible. Am I way off? What I want to know is, where does this sinc function thingy come into it, if the DAC is just outputting the level determined by the samples?

    Note I am not stating anything with certainty - really I am asking how does this work?!
    Darren
    Last edited by darrenyeats; 2012-03-10 at 10:49.

  5. #15
    Senior Member
    Join Date
    Apr 2005
    Location
    Buckinghamshire, England
    Posts
    9,983

    Thumbs up

    Quote Originally Posted by darrenyeats View Post
    Figure 6 is 24 bit data, but figure 5 is 16 bit. John Atkinson wrote: "....reproduction of an undithered tone at exactly ľ90.31dBFS (fig.5) was perfect, with excellent waveform symmetry, and clear delineation of the three DC voltage levels described by this signal." This is what I was referring to.
    Darren
    OK ... Will read the whole article...

    ...
    In 16-bit, a -90dB sine wave will only have values of 0 (zero-crossing point) +full signal and -full signal. That's not a staircase, more a fact of life :-) essentially it's behaving more like a square wave... It's either full on or full off.
    No matter how hard we try, there's nothing to be done with such a simple representation of a waveform. The filter will still do what it always does which is to smear those voltage changes over time...


    The fact remains that what comes out of a DAC (assuming a more realistic number of bits!) is not a jagged staircase, but something that is largely indistinguishable from what went into the ADC in the first place.
    You want to see the signal path BEFORE it gets onto a CD/vinyl...it ain't what you'd call minimal...
    Touch(wired/W7)+Teddy Pardo PSU - Audiolense 3.3/2.0+INGUZ DRC - MF M1 DAC - Linn 5103 - full Aktiv 5.1 system (6x LK140's, ESPEK/TRIKAN/KATAN/SEIZMIK 10.5), Pekin Tuner, Townsend Supertweeters,VdH Toslink,Kimber 8TC Speaker & Chord Signature Plus Interconnect cables
    Stax4070+SRM7/II phones
    Kitchen Boom, Outdoors: SB Radio, Harmony One remote for everything.

  6. #16
    Senior Member
    Join Date
    Apr 2005
    Location
    Buckinghamshire, England
    Posts
    9,983

    Question

    Quote Originally Posted by darrenyeats View Post
    That makes sense to me. But this seems to go against the idea the analogue waveform is re-created as a continuous curve which intersects the data points. Instead, it seems the values are just output as a voltage during each sample interval.

    Even if the above is true, I would assume that all the nastiness of the jaggedness is already inherent to the PCM signal i.e. quantisation error and that this is normally changed from distortion into noise by dithering, making it inaudible. Am I way off? What I want to know is, where does this sinc function thingy come into it, if the DAC is just outputting the level determined by the samples?

    Note I am not stating anything with certainty - really I am asking how does this work?!
    Darren
    Dithering is entirely optional and affects what happens in the bottom half bit (I'm sure we've been over this before?).

    Basic electronics 101... What happens when you put a changing DC voltage into a capacitor?... Capacitor acts as an integrator... What comes out reflects the integral over time of what went in...
    You want to see the signal path BEFORE it gets onto a CD/vinyl...it ain't what you'd call minimal...
    Touch(wired/W7)+Teddy Pardo PSU - Audiolense 3.3/2.0+INGUZ DRC - MF M1 DAC - Linn 5103 - full Aktiv 5.1 system (6x LK140's, ESPEK/TRIKAN/KATAN/SEIZMIK 10.5), Pekin Tuner, Townsend Supertweeters,VdH Toslink,Kimber 8TC Speaker & Chord Signature Plus Interconnect cables
    Stax4070+SRM7/II phones
    Kitchen Boom, Outdoors: SB Radio, Harmony One remote for everything.

  7. #17
    Senior Member Henry66's Avatar
    Join Date
    Jun 2010
    Location
    New Hampshire, USA
    Posts
    278
    Thanks Phil and everyone, I am a software guy, so I really appreciate all this awesome hardware/electronics information!

  8. #18
    Senior Member
    Join Date
    Mar 2007
    Location
    UK
    Posts
    1,170
    Quote Originally Posted by Phil Leigh View Post
    Dithering is entirely optional and affects what happens in the bottom half bit (I'm sure we've been over this before?).

    Basic electronics 101... What happens when you put a changing DC voltage into a capacitor?... Capacitor acts as an integrator... What comes out reflects the integral over time of what went in...
    What I said about dither is correct AFAIK.

    The "bottom half bit", as you put it, is present in every sample whether the waveform is big or small. So if I am right, the jaggedness is changed from signal correlated jaggedness to random jaggedness by dither. This occurs regardless of amplitude.

    My point is there appear to be voltage levels in the output from these DACs, so in fact they appear to be maintaining a voltage level during a sample interval (you can see the plateau shape in the waveforms). I agree this approximates to a smooth waveform at higher amplitudes but this is increasing the S/N ratio rather than reducing distortion due to this step effect, if the signal is dithered! In an undithered signal, it would indeed reduce distortion. (But if I'm wrong then please explain.)
    Darren

    Sent from my HTC Sensation Z710e using Tapatalk
    Last edited by darrenyeats; 2012-03-10 at 14:01.

  9. #19
    Senior Member
    Join Date
    Apr 2005
    Location
    Buckinghamshire, England
    Posts
    9,983
    Quote Originally Posted by darrenyeats View Post
    What I said about dither is correct AFAIK.

    The "bottom half bit", as you put it, is present in every sample whether the waveform is big or small. So if I am right, the jaggedness is changed from signal correlated jaggedness to random jaggedness by dither. This occurs regardless of amplitude.

    My point is there appear to be voltage levels in the output from these DACs, so in fact they appear to be maintaining a voltage level during a sample interval (you can see the plateau shape in the waveforms). I agree this approximates to a smooth waveform at higher amplitudes but this is increasing the S/N ratio rather than reducing distortion due to this step effect, if the signal is dithered! In an undithered signal, it would indeed reduce distortion. (But if I'm wrong then please explain.)
    Darren

    Sent from my HTC Sensation Z710e using Tapatalk
    Dither is only (these days) relevant when downsampling from 24 to 16 bits. It determines what happens in what I call the lowest half-bit. It pseudo-randomises it, which helps avoid quantisation distortion becoming apparent on very quite things like in fade outs, reverb tails etc.


    What you are seeing in that Stereophile trace are exactly the distortions caused by the fact that the16-bit signal was NOT dithered when it was created from a 24-bit starting point. That's why they have to make the point of saying it is "undithered".
    It also neatly illustrates why dithering is a good idea when downsampling :-)

    Those jagged edges Simply do not exist as you increase the number of bits in play. You will notice that Stereophile do not show you what a sine wave looks like at higher levels... That's because it looks exactly like a very nice sine wave!!!!!


    I can see this clearly on my oscilloscope... The sine wave is fine all the Way down until you run out of meaningful bits...
    You want to see the signal path BEFORE it gets onto a CD/vinyl...it ain't what you'd call minimal...
    Touch(wired/W7)+Teddy Pardo PSU - Audiolense 3.3/2.0+INGUZ DRC - MF M1 DAC - Linn 5103 - full Aktiv 5.1 system (6x LK140's, ESPEK/TRIKAN/KATAN/SEIZMIK 10.5), Pekin Tuner, Townsend Supertweeters,VdH Toslink,Kimber 8TC Speaker & Chord Signature Plus Interconnect cables
    Stax4070+SRM7/II phones
    Kitchen Boom, Outdoors: SB Radio, Harmony One remote for everything.

  10. #20
    Senior Member
    Join Date
    Nov 2005
    Posts
    656
    Quote Originally Posted by Phil Leigh View Post
    Dither is only (these days) relevant when downsampling from 24 to 16 bits. It determines what happens in what I call the lowest half-bit. It pseudo-randomises it, which helps avoid quantisation distortion becoming apparent on very quite things like in fade outs, reverb tails etc.


    What you are seeing in that Stereophile trace are exactly the distortions caused by the fact that the16-bit signal was NOT dithered when it was created from a 24-bit starting point. That's why they have to make the point of saying it is "undithered".
    It also neatly illustrates why dithering is a good idea when downsampling :-)

    Those jagged edges Simply do not exist as you increase the number of bits in play. You will notice that Stereophile do not show you what a sine wave looks like at higher levels... That's because it looks exactly like a very nice sine wave!!!!!


    I can see this clearly on my oscilloscope... The sine wave is fine all the Way down until you run out of meaningful bits...
    True. That 16-bit -90dB tracing at 1kHz is just to show the DAC's linearity down to the last bit. It's undithered and EXTREME to show how cleanly the DAC can flip that least significant bit in a 16-bit signal. It's nice to show off good DAC design:
    Transporter Fig. 5 - nice and clean like Benchmark DAC1:
    http://www.stereophile.com/content/s...r-measurements

    Touch Fig. 5 - clearly less clean but I'm not sure how audible this would be:
    http://www.stereophile.com/content/l...r-measurements

    Folks, remember just how extreme this test is and not get hung up about it since it is at -90dB and inaudible. I can see many non-technical folks will point to this and say just how bad 16-bit music is compared to 24-bits; hogwash! For actual audible material, the waveform will look nothing like this as Phil noted.

    Edit:
    Another beautiful DAC design - Weiss DAC202:
    http://www.stereophile.com/content/l...r-measurements

    Epic FAIL for >$15,000 - Zanden 5000 Mk. IV (LOL @ pretentious name), check out Fig. 4 among other issues, must have been 'designed by ear':
    http://www.stereophile.com/content/z...t-measurements
    Last edited by Archimago; 2012-03-10 at 17:25.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •