1. Join Now

    AVForums.com uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Why DD/DTS is no immune to jitter

Discussion in 'AV Receivers & Amplifiers' started by Halcion, Apr 29, 2002.

  1. Halcion

    Halcion
    Standard Member

    Joined:
    Oct 11, 2001
    Messages:
    78
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    9
    Ratings:
    +0
    We had a discussion on this forum earlier about how different dvd-players and/or dd/dts decoders sound different (discouting the analog stage differences).

    Some of us came to the conclusion that this was not possible from signal theory point of view and that DD/DTS streams are immune to jitter and all audible differences are imaginary.

    Well, guess what?

    DD / DTS stream is not immune to jitter - or to be more precise, the DA-stage after the DD/DTS decoding and decompression stage is NOT.

    Why?

    Because the clock for the bitstream to the DAC is derived from the incoming DD/DTS bitstream in order to keep synch correct.

    Supposedly some really high end receivers/decoders use their own high precision crystal to re-clock the data after DD/DTS decode/decompress stage, so that jitter becomes almost a non-issue for DAC. But most devices do NOT do this.

    So, the quality of DVD transport spewing out DD/DTS bistream can in theory affect sound quality.

    There is an ongoing thread about this in the AVSForum Home Theater PC section.

    Cheers,
    Halcion

    References: Julian Dunn, Jitter Theory 2, Audio Precision
    "There are many circumstances in
    which a sample clock must be derived
    from an external source. In a digital
    audio recorder or a digital surround
    processor, for example, the sample
    clock controlling the DAC is extracted
    from the input data stream."
     
  2. John Dawson

    John Dawson
    Member

    Joined:
    Jan 16, 2001
    Messages:
    852
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    43
    Location:
    Cambridge, UK
    Ratings:
    +112
    At Arcam we wondered how long it would take you to realise this :)

    You are correct BTW and anyone else telling you otherwise is mistaken.

    We have put a crystal based PPL clock recovery circuit in the upcoming AV8 preamp processor for this very reason. Very early development units didn't have it and you could certainly hear the difference.

    HTH.

    John Dawson (Arcam)
     
  3. Nic Rhodes

    Nic Rhodes
    Well-known Member

    Joined:
    Mar 23, 2001
    Messages:
    17,133
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    133
    Location:
    Cumbria
    Ratings:
    +1,277
    Isn’t this what I have always said?

    Clocks close to the DACs allows optimisation of both PCM and DD/DTS signals. I bought Tag because of this and it is the reason why I say they are in a handful of companies that understand digital. Arcam are obviously in this select band as well. In fact I always knew they were hence my promotion of ARCAM CD and DVD players here.

    See AV forum current thread. I still think they are.

    :D
     
  4. MarkB

    MarkB
    Active Member

    Joined:
    Oct 29, 2000
    Messages:
    737
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    31
    Location:
    Suffolk, England
    Ratings:
    +26
    It would be nice to know which chipsets and systems had this clock recovery unit.
    Mark
     
  5. John Dawson

    John Dawson
    Member

    Joined:
    Jan 16, 2001
    Messages:
    852
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    43
    Location:
    Cambridge, UK
    Ratings:
    +112
    AFAIK there are no commercial chipsets to do this. You have to do it yourself, remembering to cover 44.1, 48 and 96 kHz, and being able seamlessly to drop the circuit into the system after 2 - 3 seconds when the long time constants of the PLL have settled. It's not trivial :)

    I don't know of any integrated AV amps or receivers that do this, though I am happy to be corrected.

    John Dawson (Arcam)
     
  6. Plump

    Plump
    Standard Member

    Joined:
    Feb 22, 2002
    Messages:
    636
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    19
    Location:
    under the bridge
    Ratings:
    +0
    John, how did you do those audible tests - in stereo on in multichannel.
    From my testing I can say I almost never spot any difference in multichannel, not even with test DVDs. In stereo it is easier, but it still needs some ear - training!
    Any special recordings for this?

    cheers
     
  7. Halcion

    Halcion
    Standard Member

    Joined:
    Oct 11, 2001
    Messages:
    78
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    9
    Ratings:
    +0
    Quite so, but the technical reasons haven't been evident earlier - at least not on this board, AFAIK.

    While conversion from DD/DTS data format to PCM before DA-conversion can INDEED cause jitter on itself EVEN when using proper circuits and low jitter crystal clock - this is NOT the main source of jitter.

    I doubt anybody could hear that jitter - it'd be so low (order of 10-100 ps peak RMS) when properly implemented.

    The feasible technical explanation for me at least is that decoders use the input stream clock (which can be HUGELY jittery due to bad dvd clock, ground/voltage swings, interface jitter, cable induced jitter, intersymbol interference and quite a few other factors) to clock the PCM.

    This is the real source of jitter and can only be cured (eliminated to inaudible levels) through the use of a very high quality new master clock (+ associated implementation) inside the decoder or using an extremely low jitter dvd transport to begin with.

    I wouldn't have dreamed that anybody would draw the PCM clock from the input DD/DTS stream, but apparently that is the standard in the industry and the main reason for audibility differences among dvd transports in DD/DTS use.

    I hope this finally resolves this issue and gives a feasible technical explanation to the possible audible differences among DVD transport when used with DD/DTS bitstreams.

    cheers,
    Halcion
     
  8. Stuart M. Robinson

    Stuart M. Robinson
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Guys,

    Packed bitstream formats really can be immune to jitter… but only if implemented correctly. Dolby Digital, for example, was designed to be incredibly robust, able to be transmitted not just three feet from a DVD player to a receiver, but down cable TV lines, over the air and even thousands of miles via the Internet in real-time. It really doesn’t matter how poor the transmission method is, provided that when all the data blocks get to the other end of it they’re buffered, unpacked and then re-clocked before being passed to the system DACs.

    What it turns out is happening in the mainstream, is that manufacturers are not producing independent circuitry in their decoders and are relying on the S/P-DIF clock to control the clock in their DACs. It’s disappointing to say the least.

    Unfortunately, what is now going to happen is manufacturers are going to play up the importance of de-jittering bitstream formats and use it as another marketing tool, so I’d advise on the side of caution before throwing away that trusty Denon.

    A key point about S/P-DIF is that no matter what jitter it may carry when the data enters a receiver or processor, it will be filtered by the PLL action within. PLL is not an infinite filter and of course designs will vary, but will there be any audible impact?

    At one level, it’s necessary for the output clock rate to be equal to the input clock rate and it could be argued that the acquisition of the clock from the input IEC 61937 signal is in itself a jitter-filtering process (jitter is caused when two clock rates do not match or there are timing errors). Whether the clock is acquired using a separate PLL, or something else, any high-frequency components of the jitter will be attenuated.

    I’d be very interested in hearing more details about the listening tests conducted at Arcam. I asked Eric Benjamin, digital engineering guru at Dolby about the impact of jitter on Dolby Digital and his response was along the lines of…

    “Most estimates of jitter audibility are based on the assumption of a full-scale 24kHz sine wave reproduced at 120dB SPL, but in reality the amount of jitter-induced distortion decreases proportionately according to frequency and amplitude. That distortion is also likely to be masked by the audio itself, for example if you have a sound with some strong fundamental component, it also contains a lot of energy at higher frequencies.”

    Benjamin’s estimate of the audibility of jitter, based on actual listening tests, is about three orders of magnitude less sensitive than most published estimates, but actual listening tests have only actually been conducted twice, once in1974 by the BBC, and once by Benjamin himself. Both works reached the same conclusion.

    Folks should read ‘Theoretical and Audible Effects of Jitter on Digital Audio Quality’, AES Preprint 4826, by Eric Benjamin and Ben Gannon.

    http://www.aes.org/publications/preprints/search.html

    For the record, aside from those manufacturers already mentioned, both Meridian and EAD double buffer the incoming bitstream data so each stage of their decode chains has isolated jitter filtering. Lexicon processors also have de-jittering circuitry associated with their S/P-DIF inputs.


    Stuart M. Robinson
    SMR Group – http://www.smr-group.co.uk/
     
  9. garmtz

    garmtz
    Standard Member

    Joined:
    Mar 27, 2002
    Messages:
    305
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    21
    Location:
    Netherlands
    Ratings:
    +1
    This is very interesting stuff. It had already been clear to me that different DVD-players sounded very dissimilar with DD/DTS bitstreams. The best I have heard up till now have been the Arcam DV88 and Meridian 596.

    I also wonder how digital coaxial cables can sound so mysteriously different. I guess it's not all that clear cut.

    Stuart, could you elaborate on what Lexicon do exactly in their processors to dejitter the incoming data stream?

    Thanks!
     
  10. Nic Rhodes

    Nic Rhodes
    Well-known Member

    Joined:
    Mar 23, 2001
    Messages:
    17,133
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    133
    Location:
    Cumbria
    Ratings:
    +1,277
    Halcion

    I am sorry if you think the relevant points weren’t got over in the thread. These things take time to write and with the new Twins and a house move, time is the one thing I am short of.

    I still think the relevant bit is in my original text is right at the beginning and a later comment. The packed signal are transmitted through a SPDIF from which a PCM clock signal can and often is recovered. I assumed that people realised clocks were extracted from the SPDIF for bitstream. DD/DTS don’t have a separate interface standard. See DTS CDs.

    Again I hold by my comment that it is not the ‘spec’ that matters but how digital is done. DD / DTS can be largely resistant to jitter if done correctly, though not completely immune. It is just VERY few do it correctly. You need to seek out the few digital specialists. Tag, Meridian, Audio Synthesis etc…EAD? The differences are audible.

    The situation with modern (cheap?) kit is often much worse. Clocks from the SPDIF are bad enough, however it is now common practice to get the clock from the video clock! It doesn’t get much worse than this. This is getting to be the norm…. Doooh. This is why I promote the master clock right next to the DAC.

    Now I don’t detect mainstream manufacturers promoting low jitter DD/DTS. It is done for PCM and I believe done rightly as the differences are significant and audible. I have always argued the difference for DD/DTS are there but are MUCH smaller. However to some of us that small difference is very important.

    PLL are an interesting subject. PLL can cause as many problems as they solve. Even the best implementations of PLL are only partial solutions and are not ‘total’ filters, leaving substantial parts of the jitter spectrum un-touched. I for one believe this is audible and therefore prefer a more tightly engineered solution. However for the mass market, they are a decent enough solution and will satisfy many. Again it down to the quality of the engineering / how PLL are design implemented. There is more on this in my original text on what I think is good, audible and needs to be done for good PLL operation. Others will have opinions here as well. High frequency components of jitter are easy to get rid of, it is the low stuff that causes the problems (mushy bass on audition). I would also be very interested in the Arcam listening test.

    I believe that a decent number of people now understand jitter who design DVD / Processors. There are far more fundamental changes they need to do with designs than squeezing the last once from the jitter issue. Distortion and S/N are my big bug bears. Theoretical claims for S/N ration make me cry when compared with the amp you sling this stuff through.

    Is it so hard to do this stuff properly or are people just not willing to pay for quality engineering? I fear the latter knowing the TV market.
     
  11. Stuart M. Robinson

    Stuart M. Robinson
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Garmt,

    “…could you elaborate on what Lexicon do exactly in their processors to dejitter the incoming data stream?”

    As far as I’m aware, the old Lexicon platforms (the DC-1, DC-2 and MC-1) used a clock recovered from the S/P-DIF receiver passed through an extra PLL. It was based around ceramic resonators, and did a fairly good job of rejecting jittery clocks (frequency and/or duty cycle shifts). So rather than buffer data and then re-clock it, Lexicon were trying to create a pure clock from the incoming data.

    The MC-12 uses a similar system, but the PLL is a new design based on an LC-tank circuit with much higher Q and therefore superior performance. Looking at the design, it also seems that there is a can around the tank circuit to reject any stray fields that could influence its operation.

    I’d like to know more about Arcam’s solution, but it looks like Mr. Dawson is intent on being coy…

    Nic is right about not getting carried away with jitter – there are so many other factors that have a far greater bearing on the performance of digital circuitry. We must also remember that jitter is not as audible as some folks would make out, see the aforementioned AES paper for more.


    Stuart M. Robinson
    SMR Group – http://www.smr-group.co.uk/
     
  12. garmtz

    garmtz
    Standard Member

    Joined:
    Mar 27, 2002
    Messages:
    305
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    21
    Location:
    Netherlands
    Ratings:
    +1
    Thanks for your reply Stuart!

    What I think is strange is that although the Lexicon processors have good jitter-reduction, differences between transports are still perceptible with the MC-12, as well as the difference between digital cables.
     
  13. Stuart M. Robinson

    Stuart M. Robinson
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Garmt,

    “…as well as the difference between digital cables.”

    Let’s not go there! Digital cables make no difference whatsoever; in any type of controlled test nobody has ever been able to distinguish one from another. Way back when I actually sold digital cables, this was something we were aware of at the time – we just hoped consumers were gullible and would buy into anything, which they were, most of them anyway.

    If you get hold of a Dolby Digital bitstream checker (from Dolby) and test the transmission of signals down a coat hanger – I kid you not – it turns out to be just as effective as any other digital link for getting signals from A to B. There was a detailed write-up of this test on-line once, but the applicable link is currently unavailable. Anyway, see the following URL for more:

    http://2eyespy.tripod.com/myaudioandhometheaterhomepage/id3.html


    Stuart M. Robinson
    SMR Group – http://www.smr-group.co.uk/
     
  14. garmtz

    garmtz
    Standard Member

    Joined:
    Mar 27, 2002
    Messages:
    305
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    21
    Location:
    Netherlands
    Ratings:
    +1
    Sorry Stuart, I simply do not agree with you here... I do hear differences between digital cables and a lot of people agree with me here I think. But let's not discuss this here too deeply, as discussions about it have been known to last till eternity... :)
     
  15. Stuart M. Robinson

    Stuart M. Robinson
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Garmt,

    “…I simply do not agree with you here... I do hear differences between digital cables and a lot of people agree with me here I think.”

    Well, you know the party line… sure, you can hear differences, as can many others, but under what conditions? Run an ABX, which relies solely on the audibility factors rather than any other bias and you might be surprised. If you could still hear differences under those conditions, then you'd be a millionaire overnight, as every cable company in the world would want your services.

    “But let's not discuss this here too deeply, as discussions about it have been known to last till eternity...”

    Indeed, although they really should start and end with the question above. But then again, it could be said that an imagined difference in sound is real, at least in the space between our ears. ;-)


    Stuart M. Robinson
    SMR Group – http://www.smr-group.co.uk/
     
  16. stranger

    stranger
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    stuart, if there is no difference in cables then why does almost any other cable sound better than the free ones in the box? and they do.
     
  17. Nic Rhodes

    Nic Rhodes
    Well-known Member

    Joined:
    Mar 23, 2001
    Messages:
    17,133
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    133
    Location:
    Cumbria
    Ratings:
    +1,277
    This one will be fun.:devil:
     
  18. garmtz

    garmtz
    Standard Member

    Joined:
    Mar 27, 2002
    Messages:
    305
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    21
    Location:
    Netherlands
    Ratings:
    +1
    Ah well, at least we can (hopefully) both laugh about it. Let's just call it a difference of opinion. I am well aware of your very thorough ABX testing methods and they are well admired (there simply are not many reviewers willing to do it/that dare to do it this way).

    At least I know I am spending lots of money and THINK it sounds better. That's also worth something, don't you think..? ;)

    Just out of curiosity... Is it possible to do an ABX test between digital interconnects and have you ever done such an experiment?

    P.S.: I think the MIT Digital Reference is the best sounding digital coax around today... :)
     

Share This Page

Loading...