Jitters

January 17, 2023
 by Paul McGowan

Jitter, also known as timing deviation, is the variation in the time interval between two consecutive digital audio samples. This timing deviation can cause easily heard distortion in high-end audio systems including a loss of clarity, definition, and detail in the music, as well as a smearing of transients.

Jitter can be caused by a number of factors, including inadequate signal routing and EMI, though it is particularly obvious in systems that use asynchronous USB or S/PDIF interfaces (both coax and AES/EBU), as these interfaces are prone to jitter due to the lack of a common clock signal.

In the new DS MK2, reducing jitter through lowering noise and EMI through galvanic isolation and power supply improvements is important; the biggest trick in designer Ted Smith’s technology bag is his idea to ignore incoming data timing altogether.

Instead, Ted “simply” looks at the incoming data (DSD or PCM) as a source of data bits and ignores any timing issues or jitter-related impacts like edge transitions.

Once he has the data in hand, it is upsampled and sent on its way to be handled by clean and extremely low jitter clocks, unaware of what the original data signals and problems might have been.

The results are stunning and prompted PS Audio Community member Wally Hurst to write to me:

“I am 68 years old, have been listening to audio for over 55 years, and have put together nice systems in the past.

The complete PS Audio source with the DirectStream Memory Player and the new DirectStream Dac ll is the best I have ever heard an audio system in my entire life…

The DirectStream DACII is simply phenomenal sounding…unbelievable how the music just surrounds you like you are part of the band.

I want to thank PS Audio for providing these State of the Art Source components.”

Thanks, Wally, and thanks to all our beta testers who put their hearts and souls into helping us create what is likely one of the best sound audio components on the market today.

If you own an MK1 or want to upgrade your existing DAC, there’s no better time than right now.

Subscribe to Paul's Posts

41 comments on “Jitters”

  1. Didn’t Ted once explain that it requires extremely low jitter for getting his concept of a DS DAC running? Thus doesn’t this mean that it finally requires that the clock in the DS DAC has to generate exactly the same jitter pattern which was present during the digital recording? I guess this is an absolutely impossible goal. Thus the old question of soundmind comes to mind: which degree of jitter and jitter difference between recording and reproduction is finally audible?

    1. PS, You are right that there is recording chain jitter in everything that uses digital in the recording chain. The nature of jitter is that it is unpredictable. If it were predictable then one could create an anti jitter signal and combine it with the original to remove the jitter. What ever comes from the ADC recording process you are stuck with. ( Unless you subscribe to MQA’s claims of “time correcting” the signal. )

      What Ted is doing is trying to lower any additional jitter being added to the signal from the DAC process. And, IMO, doing a great job of it.

  2. I’ve streamed using either Ethernet or USB for some 13 years. The choice of Innuos for streaming was in part because one of the two output options is Ethernet as a data link. It is galvanically isolated, does not carry a clock and you can run long cables – I kept the unit in my office and used a 25m fibre cable. I think other popular brands like Lumin and Melco do the same, and all you need is a DAC or player with an Ethernet input. I am then informed that the sound quality is dependent upon the quality of the clock in your DAC/player and whatever upsampling or processing it chooses to do. When I bought my unit in 2016 it had a processor designed in 2010, but the manufacturer offered a free card upgrade in 2017, which was an improvement and allowed more processing like Roon Ready.

    External reclockers have become all the rage, including by Innuos for Ethernet and USB. If the receiving device reclocks the signal, I wonder the point of these devices at all. I’ve never tried one. The way I’m set up I can use either the Innuos streamer over USB or Roon Ready over wireless Ethernet. I can’t split them for quality. I presume that’s because they are both reclocked and upsampled to DXD by the player.

  3. From a retired Hi-Fi salesman’s viewpoint, can the PS Audio – ‘DS MkII’ DAC compete with those of say, dCS, MSB or Playback Designs…taking into consideration, of course, that the ‘DS MkII’ is close to a third of the price of those mentioned above?

    Law of Diminishing Returns.

    1. The dCS DACs are most commonly used with an external clock – a £10,000 optional extra – even if the internal clock is really rather good. Most of their DACs include streaming, one reason why they stopped making their streaming Bridge.

      Curiously, although dCS pioneered the use of DSD in consumer audio, they have never supported DSD256 and still don’t in their new Apex range. DSD256 is the freaky end of audio formats and DSD256 usually spent some time as something else, and it is hardly used anyway, so no great loss to the vast majority of people. They support up to DSD128 and 24/384 PCM and you can choose your level of upsampling.

      I’m not sure people at that end of the market are looking for a bargain. The level of transparency you can get with dCS just from a CD is so astonishing, if you don’t spend a chunk of change building a really good music room, I don’t see the point. I don’t have dCS, I have some Wilson speakers that weren’t cheap, and I still spent more rebuilding the music room than on the audio system.

        1. “My [fill in appropriate car engine, bodily appendage, sampling rate, etc.] is bigger than your [ditto]”.

          Of course bigger is better, didn’t you know that? Surely that was you go-to sales schtick?

          More seriously, the designers I’ve spoken will say that the biggest source of electrical noise is from inside a device rather than from outside, so designing devices that do a vast amount of processing and generate a lot of heat and noise is perhaps not such a great idea. My unit tells you the temperature of the processors and power supply, maxes out at 42dC, which is still pretty hot given it’s extremely efficient heat dissipation.

          p.s. To his credit, David Warner got out to a fabulous inswinging yorker the other day. Couldn’t have happened to a nicer guy.

        2. Yes, was a little surprised that this DAC only supports up to DSD256. Maybe a future firmware update will push the limit higher as with the final firmware update for the MKI.

          1. gorm.yoder,
            From what I’ve been told there is virtually no detectable sonic difference to the human ear between DSD256 & DSD512, so it’s not really a big deal.

      1. As far as I remember Ted’s tech-talks the clock must be positioned as close as possible to the FPGA chips for minimizing jitter! Thus obviously external clocks seem to be only the second best design idea!

        1. Perhaps you should listen to the impact of an external clock on dCS (which I have) and other brands (which I haven’t).

          1. The business of external clocks and their impact on the jitter of a DAC is tricking business even for such brands as dCS that use them. These external clocks often have ‘accuracy and precision on the order of femtoseconds ( a femtosecond is equal to 10 to the -15 power or 1⁄1 000 000 000 000 000 of a second; that is one millionth of one billionth, of a second ). Yet when these external clocks are added to the DAC the DAC still has jitter on the order of picoseconds ( a picosecond is 1000 times larger than a femtosecond ).

            This is because there are other sources of jitter. If ground plane variation is causing picoseconds of jitter and you then clock the signal with a femtosecond clock you will not be adding any jitter with the clock, but you still have a signal with picosecond jitter.

            Thus, it seems that Ted Smith’s approach of just using the raw unclocked digital data and putting a very good internal clock as close as possible to that actual DAC chip is a better approach to the problem of jitter.

          2. I am pretty sure that every manufacturer of audiophile highend cables will design the cable “properties” in a specific way that finally everybody will rank the most expensive cable as “best” in a shoot-out of the company’s cable series. Why should a manufacturer of most expensive external clocks design a DAC with clock input which sounds better without this external clock? My initial understanding of external clocks was that there is a requirement for studio equipment having every digital component in the chain supplied by the same “master” clock.

            1. I’ve seen manufacturers in addition to dCS that use world clocks connected to several devices in multi-box DACs and digital systems. There is a lot of hardware in these external clocks and I suppose having both a good internal clock and a better external option gives the customers a choice.

              I don’t agree with the cable thing. I have quite expensive speaker cables from a company nobody has probably ever heard of, but fibre ethernet costs pennies and have just sold my relatively expensive power cables and replaced them with much cheaper ones. The UK market doesn’t buy into the cable marketing thing anything like the USA. Some brands start affordable and make increasingly expensive products, like PS Audio, others make fiendishly expensive ones and make increasingly affordable ones, like dCS and Wilson.

              1. The funny thing is aside from the recent APEX upgrade, the biggest improvement in sound quality that could be made to a dCS Rossini with Rossini Clock was to replace the CLOCK CABLES with better ones.

                A jump from Cardas Clear to AudioQuest WEL was like upgrading to a better DAC – more pinpoint sound staging, more sense of space in the original recording. The best way to describe going back to the Cardas clock cables was like taking off your glasses if you need them.

                1. For what it is worth, I have found that the higher the quality of the gear that I have the more sensitive ( or perhaps i should say the more reveling ) it is to any change I make in the system.

                  I am a little surprised because most external word clocks use BNC coax cables.

      2. Steveen, I think the basic reason dCS does not support DSD256 is that like most brands that are based on a PCM DAC ( which is what the Ring DAC is, and is one of the best PCM DAC’s ) they handle DSD using DoP. As you can see from this link: https://dsd-guide.com/dop-open-standard#.Y8a8ExXMLy2, you can only support DSD128 if your highest PCM sampling rate is 352.8 kS.p.s. If you want to do DoP for DSD256 you need a PCM sampling rate of 705.6 kS.p.s.

        Since you are more familiar with dCS than I am, can you confirm that dCS does not support 705.6 kS.p.s. PCM?

        1. I have no idea. I’ve heard dCS quite a few times (most recently with Alexia V) because in many places it is considered a reference DAC so it is often used in demonstrations.

          All I know is that it apparently has a unique design, which goes back to the late 1980s, with a major upgrade in 2022. By all accounts it has extraordinarily low jitter and in my personal experience, the most transparent and holographic sound I’ve ever heard from digital audio.

          I have a lot of interest in low-noise digital and have been in a happy place for quite some time, although it came into its own with a dedicated room with care over acoustics built two years ago. Low-noise server and player, fibre cabling, good power conditioning, no switching.

          I lost interest in standalone DACs years ago because I see no reason for not integrating streaming, which is increasingly prevalent, including by dCS. I bought my first streaming DAC in 2009. I would only change my system if it involved getting a streaming DAC and an integrated amplifier. I was expecting PSA to make a streaming DAC, everything addressed in one box, but it has never happened. Not yet at least.

    2. FR, The DS MK2 DAC more than holds its own against dCS, MSB, Playback Designs, etc. It is a world class DAC. It is those names that you mentioned who are faced with the Law of Diminishing Returns far more than PS Audio is.

        1. The last time I heard those three DAC’s was 2018. At that time the DS MK1 DAC was in my own system. Since the pandemic I have not been to audio shows or NYC dealers.

    3. Value upgrading
      Staying one step behind the Joneses

      I’m wondering how much better is the Mkii version,
      alternatively how much worse is the half price (?) superceded version.

      1. Only your ears can answer that question…along with how much are you prepared to spend?
        What other greater financial priorities do you have, etc.
        Audition, audition, audition.

  4. “Instead, Ted “simply” looks at the incoming data (DSD or PCM) as a source of data bits and ignores any timing issues or jitter-related impacts like edge transitions.”
    It always seemed strange to me why the timing should be taken from the PCM data stream, when it is clear what clock is to be used and it can only be generated at the end in the DAC.

  5. Not one comment about creating music. Great sound quality is important but does not guarantee a great musical experience. I’m still trying to understand why a lowly preamp tube in the signal path imbues the reproduction of music with more life and emotion.

  6. Paul
    When I bought my initial Perfectwave Transport and DAC, I was under the mistaken impression that software would be continuously updated for years to come. It has become customary for Android manufacturers to be very clear about how many years of updates you can count on for a particular model, prior to your purchase. Are any commitments being made by PS Audio for the new dac and transport?

    1. We’re certainly behind the times as compared to Apple or Android (and a lot smaller). When the PWT and PWD came out the only software updates we ever performed were ones that were needed to keep the hardware running. Up until the DirectStream was launched, that’s all we have ever done of promised with hardware.

      With the introduction of DS we were able to move to a new model on that one piece of gear. Software updates that we supported for 7 years until we simply ran out of room in the processors.

      The MK2 should give us another similar run and we can expect yearly upgrades to the software.

  7. So if the new DAC disregards the source clock and then upsamples everything to DSD, disregards the files original formate and sample rate, where does that leave the the “all important source” in the equation?

  8. @Paul

    Paul wrote: “Jitter can be caused by a number of factors, including inadequate signal routing and EMI, though it is particularly obvious in systems that use asynchronous USB or S/PDIF interfaces (both coax and AES/EBU), as these interfaces are prone to jitter due to the lack of a common clock signal.”

    If I use a toslink optical connector between my CD transport and DAC, is jitter excluded from my digital playback chain?

    Related, does a toslink optical connector gain me galvanic isolation?

    Thanks, y’all, for advice/enlightenment here,
    Chaz

  9. I think Ted should have his own series of TED Talks.
    I’d certainly go…
    (Nodding my head in a distinguished and (con)fident manner whenever I heard the parts that went way over my brain’s pay grade…)
    I can listen to smart people talk all day.
    Thus I’ll shuddup now 😉

  10. There’s always the question of finding something a DAC doesn’t do well.

    I remember I tried a DS DAC a few years ago and found that it rendered solo piano very mechanically with a metallic leading edge.

    Luckily it was at RMAF and I was able to show Ted what I meant and he instantly thought of how he could fix it.

    Sadly I was never able to find if it did.

    The funny thing is he said “the last version of the firmware won’t have that issue”… and it didn’t but it got a lot of OTHER things wrong.

Leave a Reply

© 2023 PS Audio, Inc.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram