More than what’s obvious

April 26, 2021
 by Paul McGowan

In yesterday’s post, I made the point that regardless of the delivery method—transport or streaming—identical digital audio bits received at the DAC should sound the same.

They do not.

Why?

Let me first start with a little story. When we had the opportunity to listen to the PerfectWave SACD Transport (PST) for the first time, we had high expectations. Inside was a new way to handle bits: an extension of the work we had been pursuing for years, the Digital Lens.

A DL is a big buffer with a low jitter fixed output clock. Bits go in one end of the DL, gather together in a holding pen, and then when the jitter-free output clock has the “time” (pun intended), it pulls from the holding pen the next set of digital audio bits to send to the DAC.

The lowered jitter produced by the Digital Lens provided a revolution in sound quality.

What was different inside the PST—the new innovation we had been sitting on pins and needles to hear—was more than just a DL (we already knew what that sounded like). The PST’s internal DL had been galvanically isolated as if it were an entirely separate entity from the PST. We had built this new structure in the hopes of removing the last vestiges of sonic degradation: noise and jitter introduced by the power supplies and shared grounds inside the transport.

It worked. The sonic differences between the new PST and the older DMP it replaced were more than just better. They were extraordinarily better—a first-note-obvious better.

And therein lies what I believe to be the answer to yesterday’s question. Identical bits cannot sound different unless something else has changed. That something else is noise and induced jitter on shared grounds.

Which is why, in large measure, digital audio received from a computer via USB sounds remarkably different than the exact same bits as received from a transport.

It is not the veracity of the bits but all the baggage associated with the gathering and delivery of those bits.

As is almost always the case, it’s more than what’s obvious.

Subscribe to Paul's Posts

35 comments on “More than what’s obvious”

  1. Similarly it’s not about the DAC ‘chip’ but more about the design & the topology
    of the rest of the DAC processing unit that determines the sound quality.
    So, as a non-streamer, & born-again technophobe, it sounds to me that
    audiophiles are waiting for a galvanically isolated PS Audio DirectStreamer 😮 😉

    1. Even in the mid 1980s, it was shown that it wasn’t the chip, but implementation. A couple of companies used that original chip from the original Magnavox,used a different design around it, with better components, and got far better sound. They were out of my price range, but they were out there.

  2. If I’m right, this makes clear, that the mass of people using (probably generally sound-wise inferior) cloud streaming services, can’t improve with whatever boxes and measures what got lost in this context (not even with a better isolated streamer, as the information already got harmed), they can just improve following additional noise sources doing harm. The only way to care similarly to the PST transport from the beginning would be to stream from optimized local hardware.

    A further topic:

    So far I understood, noise influences during transport of digital data just harm in the moment of listening, they don’t harm the stored data as such. This would mean, a noisy burning process generates no worse data. Is this really true? I remember from past discussions about burning processes, that noise plays a role for the quality of the created disc.

  3. Misinformation abounds. I don’t know when the Digital Lens arrived (I had one), but low noise reclocking has been available in consumer products at least since 1999 (dCS 992). My first streamer/DAC (Linn Akurate DS) had galvanic isolation and an ultra low noise power supply, and that product was released in 2007. I have found performance can further be improved by mains ground plane noise reduction, high frequency noise reduction and cross-contamination isolation. I’ve been using fibre-optic for years. Even usb can now be used very effectively, it is the primary output on one of the world’s best streamers (the $27k Taiko SGM Extreme), also at more realistic products by Auralic, Innuos, Sonore etc.

    So comparing to “digital audio received from a computer via USB” simply ignores 20 years of streaming development, where much the same problems Paul describes (it’s the same digital data) were identified and solved sometimes decades ago, and it tends to be a lot cheaper because it does not require mechanical devices. High quality streaming can be achieved from around $1,000.

    Comparing a high-end transport to “computer via USB” is more equivalent to comparing a high-end transport to a CD spun from the disc drive in a laptop. If the high-end transport was not better I’d return it.

    Yesterday’s post talked about “transferring that same data from a hard drive on a Qobuz or Tidal server should then be identical to that of a Dropbox server.” Now Paul is talking about computers and usb. The goalposts have been moved.

    I suggest that SACD only still exists because of the Japanese market for physical media (where streaming is only 30% of the market compared to over 90% in Europe and the USA), so Marantz and TEAC/Esoteric continue to make SACD transports that western manufacturers like PSA rely on. They even made SACD products exclusively for the Japanese market. If you ask a bunch of mostly not-young people who compare to reel-to-reel what they think of streaming, you will get a 0.0% positive response, some of whom admit they’ve never even tried it. 0.0% also happens to be SACD’s USA market share of media sales (per the RIAA).

    I decided I wouldn’t post until PS Audio brought out a streamer, which will no doubt feature here with its various merits. I broke that because I suspect it will transpire that much of the technology PSA uses to optimise data from a disc will be applied in its streamer to data from online, because so far as I see it, the issues are largely the same.

    1. ———————
      „ Yesterday’s post talked about “transferring that same data from a hard drive on a Qobuz or Tidal server should then be identical to that of a Dropbox server.” Now Paul is talking about computers and usb. The goalposts have been moved.“
      ———————-

      That’s true and my first thought was, it tends not only partly but mainly towards marketing for the PST…but in fact it also explains the effect which also affects the cloud streaming matter Qobut/Tidal/local HW. The example is transferable.

    2. Dear SNTBCWS,
      Paul said to me that he doesn’t have the updated AP machine to measure jitter. The AP SOTA measures to 140dBs of jitter so I finds it fascinating that Paul can make an argument on something he cannot or does not measure. Most modern DACs already have basically immeasurable jitter, and I don’t believe that bits coming form a PC over USB are any different than those from a proper ($$$) server.
      To quote from a classical American movie, “Show me the data!”

  4. In a summary, I have just to put my DMP in the garbage, and buy a PST around $10250 (have to buy it from UK as there is none available from EEC distributors so $9000 retail plus 14% import fees). And of course, no trade in.

    Price in USA is $5000 with trade in. Add 40 % (maxi) import fees +VAT + processing file = $7000

    Do you really think it is fair ? I can understand you can sell directly because of your distributors but when a product in not available from them (from the French distributor, only the direcstrem memory player is available and cost is (from the price list) 6750 euros ( $8165).

    May be you no longer want any EEC customer :
    -PST not available in EEC
    – Np trade in
    => PST costs in EEC twice the price than in USA.

    “It is extraordinarily more—a first-note-obvious more”

    1. I sympathize with your economic situation ( and worry that with our new socialist regime we could end up in the same quagmire ), but I do not think that economics is what we are discussing today.

      1. May be I haven’t understand. Most of the post is about PST making DMP completely obsolete and the reason why. The streaming is just a way to introduce the PST. I don’t think the topic is between streamingand transport as it is obvious that LPs are still way ahead. I just wanted to draw Paul’s attention to the fact that PST is not avalable in EEC. Otherwise, I’ve read tons of pages about streaming vs transport, jitter (audiophile switches, sometimes one after the other, the Linn Organik DAC and it’s femtoseconde clock, what is the clock of the PDC and the one of the of the bridge II ?).
        My opinion is listen and choose or use both.
        Last but not least, my direcstream DAC and transport are already in the garbage. Where else ?

  5. One question that continues to be a source of confusion for me is this business of the clock being so crucial at the source – – as you describe, it apparently can be the difference maker. A DAC is going to ignore all clock information received and use its own clock. It’s a question I’ve posed to anyone that will listen, and mostly met with silence. I’m a believer that the source clock is critical and introduces jitter downstream to the DAC – – I’m just unclear how it assists the DAC’s conversion when everything suggests the DAC ignores the clock information.

    And I’m not convinced a transport by rule is better than streaming. Maybe YOUR transport is. A high-end streaming device (network audio adapter to be precise) has similar challenges and attempts to do the same thing your DL tries to do. These NAA’s, most notably by Sonore or SOtM, use expensive clocks, filters, and. isolated power. And if cost is the measuring stick,they every bit as expensive as your PerfectWave SACD. So I’m not sure anything was “proven” in today’s post other than you have a great transport that sounds better than most streaming alternatives but I think the jury might still be out when going up against the expensive NAA sources.

    1. Mike, I raised the same question over in the PS Audio Forums when I first heard the improvement in sound I got when I replaced my DMP with my new PST. The answer I got from Ted Smith and others is the PST improvement is mostly about noise improvement. There are those who argue that a well deigned DAC should be impervious to what transport is used to send it the digital data. That was probably true when the whole idea of “computer” audio started and the transport was a PC laptop. The sense I am getting is that in today’s high performance digital audio world the transport and DAC have to be viewed as a system in which if the transport can reduce the work load on the DAC then you will probably get better sound. I too wish that someone from the industry would really drill down on the transport / DAC interaction.

      1. Initially, all CD players were architected so they would provide the Master Clock. And those clocks were generally pretty bad with respect to jitter. (S/PDIF and Toslink still depend on a host clock.) The DAC has to phase-lock-link onto the host’s clock and hope its PLL doesn’t add too much additional jitter. But USB evolved a so-called “asynchonous mode” that allows the DAC to provide its own clock, “pulling” the data on demand rather than having to deal with it being stuffed in its face. And with Ethernet/WiFi architectures you can do the same in the sense of providing data independent of the host clock. Either asynch USB or Ethernet transports let the consumer decide on the quality of the clock because now it’s in the DAC. And in your DAC you can have reclocking buffers and galvanically isolate to your heart’s (and wallet’s) content. So presuming you are using either asynch USB or Ethernet/WiFi, I just fail to comprehend how reclocking buffers and galvanic isolation AT THE TRANSPORT matter. Might just be a lack of imagination on my part, I admit.

  6. As I mentioned yesterday, I do not stream audio so I have never heard the difference between what some of the streaming services supply. There is a wide spectrum of understanding digital audio within both the general population and the audio enthusiast community. I have no idea how many people actually understand how important it is to reduce jitter and to control process noise. I also think that reclocking ( Digital Lens ) and galvanic isolation have been in the digital audio industry for the last decade or two.

    There are people who honestly believe that cd quality sampling rates are all that is required for sufficient sound quality in digital audio. For those of us who believe that higher sampling rates improve sound quality I wonder if the difference in sound between streaming services is that some of them are simply upsampling cd data and passing it off as hi-rez data. HFN&RR does a monthly segment were they analyze the frequency content of digital downloads and points out to readers when a download that claims to be hi-rez has no more HF content than a cd has.

  7. So what I’m taking away from today’s post is that the issue of galvanic isolation has to do more with the usb side of things when it comes to streaming since galvanic isolation improved things significantly on the transport side.

    This leads to a few observations….

    To get the most out of a Directstream Dac one should be spinning and connected to the DAC via the I2 buss. So either buy SACD or burn from downloads?

    If using the Ethernet bridge in the DSDAC then that should eliminate the usb issue, but does it have galvanic isolation?

    If galvanic isolation is crucial to usb playback then shouldn’t that be integrated in to the DSDAC already?

    Is there a galvanic isolator to put in line with the usb you would recommend?

    Or should one stick with slightly less than optimal DSDAC performance for streaming via a usb input?

    1. ——————-
      “ So what I’m taking away from today’s post is that the issue of galvanic isolation has to do more with the usb side of things when it comes to streaming since galvanic isolation improved things significantly on the transport side.”
      ———————

      This is exactly what was so misleading as also Steven pointed out. IMO what you should take away is, that everything that adds noise before your DAC is reached, matters. May it be within the processing of cloud streaming providers, within your own storage/streaming environment or within a transport. PC/USB was just a correct but misleading example.

      The fact that within the PST transport care was taken for avoiding noise is also just one example (which admittedly promotes sales of those transports), but similar care could be taken inside a streamer, server or maybe even PC.

      The most interesting part and where we actually came from from yesterday was the difference of streaming services, which makes clear, that even there noise problems vary.

      1. The real confusion comes from how it was stated (back in the day) the DMP / earlier transport was Sonically superior to usb or streaming. Now that galvanic isolation has been introduced that seems to be the buzzword, or the at the very least the next step up for sound quality. It’s in the new transport and has been talked about in the new high end DAC. So the concept is not new, but the use of it in PSA products may be.

        This instead of just saying computer /microprocessor based audio is to some degree sonically inferior to disc spinning.

        I, for one, am not looking for perfection from home playback, but rather enjoyment and satisfaction. I enjoy the ability to shuffle and some playback features of computer audio. Usb connections are convenient and it’s not hard to keep external hard drives on separate busses. If galvanic isolation is the answer (or at least one of them) then there are many aftermarket choices. I’ll agree that noise is and always has been an issue. So where and how to eliminate all sources of it is the trick, I guess.

        1. I also find improvement by having every product that is connected to the production of my music plugged into my Power Plants. They power my DAC/Streamer, my Computer, my NASs, my router, EVERYTHING. Some of the units benefit more than others by doing so, but I’ll take every improvement I can get.

          1. Hey tarheelNeil,

            What you say makes perfect sense. I also know that PSA has at the very minimum all of their audio gear plugged into power plants. So the improvements being heard must be above and beyond the power plant(s). What I was trying to get to is why galvanic isolation is not a ‘requirement’ if one is using usb as an interface. I’m the using the i2s interface with an usb to i2s adapter and external power supply for that. But it’s not clear how ‘isolated’ that device is. (The sound is better though).

            I guess rather than wind myself up, I’ll just head to work, and then just enjoy the tunes when i happen to return home. It’s fun to play around though…. now that I think about it, nothing in my audio world will ever be ‘right’ and the only one I’m competing with is myself… the worst part is I keep loosing 😀

    1. You could buy the Marantz 30n for €2,998. It has the same transport as the PS Audio transport. It also is a full streamer with Amazon HD, Qobuz etc and DSD upsampling DAC. (No loss to PSA if they don’t sell in your territory.)

      It crossed my mind yesterday as the only valid way to compare SACD and streaming would seem to be through a device that does both and the Marantz 30n is the only box that I know of that does.

      Please keep buying books. Authors are starving. For something as uplifting as this discussion I would suggest Michel Houellebecq.

      1. Yes, buy books. And Records, CDs, and SACDs. Viva physical media.

        Despite my handle here on Paul’s Posts, I am not at all confused on this topic.

        Many (most?) of you have likely seen this YouTube vlog already, but in case you haven’t John Darko’s “10 reasons why I still buy CDs” (04/26/2021) is well worth 8:33 of your time.

      2. You’re comparing apples and oranges there hombre. @ 30lbs, the swiss army knife Marantz N30 player looks like an interesting piece of upper-mid pedestrian kit but in few way ‘s would compare to the sonic playing field of a separate PST/DSD combo.

        That said, i dig selectable digital filters and wish all Dac’s provided the listener selectable sampling rate/bit depth in an attempt to tune questionable sounding digital recording’s as compared to upsampled everything.

        While scanning the marketing propaganda product sheet at the Marantz website reminded of the humour when Glimmie used to belittle for simplifying technical concepts to differentiate feature, benefit and musical performance advantages of various components accusing moi of writing ad copy.

  8. I got rid of a $3000 CD player over 12 years ago because ripping to a hard drive sounded better, and many many people said the same thing. It’s still true today. What has always made a big difference in sq is how you are reading the files. iTunes always sucked, but audirvana/pure music and others provided better sq. As for using aws or another cloud service for providing the music, if your network is properly configured and especially if using fiber, it won’t matter. Music files are tiny compared to other streamed data. Ethernet does a very good job at sending and receiving small packets of data. If anybody thinks that packets off data can get mixed up, think about the millions of stock market transactions that occur every second and when was the last time somebody complained that their orders were out of sync?
    Network configurations make a huge difference. If your internet connection is copper, you already have a noisy line. If you have fiber, you are starting with a clean line with all the other network issues solved.
    USB is a flawed interface and there are many vendors selling products to try to make it better. You are better off using i2s or Ethernet for the best sq, no matter what your source is.

  9. Now I don’t need to post this!

    I can tell a funny story. The first generation of SPDIF/AES/EBU/Toslink transceiver chips generated so much jitter that they overwhelmed DAC chip buffers. This turned out to be great for the high-end cable business because there could be massive audible differences between different digital cables. Toslink was especially amusing because you could have someone else move the cable about and listen to the soundstage moving around!

    Then came digital TV. I went to a SMPTE meeting at the first all-digital TV station studio in San Francisco where they were discussing what it took to make their conversion to digital. The first buildout had a problem with little artifacts in the picture. This turned out to be in the cabling network and finally got traced to the transceiver chips which just happened to have been designed by the same folks who designed the audio chips. The thing is that they could no longer get away with just demanding ABX tests because producers were pointing at monitor screens and demanding to know what the f!ck is THAT?

    The next generation of chips got it fixed but still many manufacturers were too cheap to upgrade their products.

  10. Can we please have a Digital Lens in the DAC? (It sounds like galvanic isolation would be a good idea, too.) Then let’s talk about whether digital sources matter and why.

    I see no reason why the music data bits laid down in the recording process by the original A-to-D converter cannot be reliably reconstructed in the “holding pen” used by the Digital Lens. If the Digital Lens were to be placed in the DAC, why would we need a fancy streamer or a fancy transport, provided those devices don’t mess with the music data itself? True, any *lossy* transformation of the music data by the source devices would add jitter and maybe a lot of it. THOSE clocks would matter, but if lossy transformations are avoided entirely, the problem goes away.

    With a DL in the DAC capable of receiving the original music data, “induced jitter” wouldn’t matter. It would be no more of an issue than the fact that computer networking software breaks up data into small packets that it can disperse over multiple carriers as the data makes its way across the country, only for the original stream of data bits to be reconstructed at the receiver – perfectly, every time.

    Induced noise is a different problem. Maybe galvanic isolation solves it, I don’t know.

    If this were to work well, it would simplify the process of setting up an audio system. Source devices that faithfully pass on the digital music data should all sound the same, regardless of cost. And digital cables to the DAC shouldn’t matter either. The reality might well be more complicated, but it would be clarifying to get the question of reliable data transmission between the A-to-D converter that recorded the music to the D-to-A converter that plays it back off the table.

  11. Bits are bits…

    And, vinyl grooves are vinyl grooves.

    Yet, everything in the audio chain from the bits to the speakers has an effect on the sound. Everything effects, including the room its heard in..

    The way some talk here, and elsewhere in the audiophile community? Is like no one has to yet heard music on a system that pleases them. Satisfies them.

    How much does that depend on the state of the one doing the listening? I remember being in bliss when I used to listen to records on a portable Magnovox record player with attached speakers. Some try drinking or smoking grass to recapture that ability… but? What’s the real problem?

    1. I remember being in bliss listening to my own record purchases on my quadrophonic Radio Shack system bought with my own summer job earnings. Had my dad listen to Neil Young’s then current album with Four Strong Winds and he merely said his voice is awful and that’s not his song ( Ian and Sylvia, apparently). Nicolette Larson is sadly gone, but I saw Neil with Promise of the Real in 2019. Neil’s still fine and Willie’s kids will represent well, too.
      As for affordable digital today, I enjoy Tidal streamed from a Cambridge CXN and CDs off its companion transport CXC. Is it PS Audio grade, no, but it’s all sufficient for blissful evenings with all of the above.

  12. “The way some talk here, and elsewhere in the audiophile community? Is like no one has to yet heard music on a system that pleases them. Satisfies them.”

    I think you may have something there Genez. We could be back to FOMO again. That’s the thing with this audio lark, you have a great system, you’re really happy with it and you enjoy listening to it. Then, a new product comes along and the doubts start to creep in. Or you read a great review. The other day I saw this one, ‘Taking the Denafrips Teminator to 11’. Now the Denafrips Terminator is already a respected DAC (with such a name it has to be) so taking it to 11, wow, that must make it really fantastic, what must I be missing by not having one in my system? Then a deep breath, a pause, a step back and think. Apparently, like many products, the Denafrips has a distinct sound of its own which I may or may not like, even taken to 11. Then it is not necessarily better than my DAC and/or I might not prefer it. That’s okay then, no more need to worry.

    I think this happens a lot in audio, new hot product, must have, until the next one. We just need to sit back, relax and enjoy what we’ve got. As a youngster my wife heard the oft repeated line “be grateful for what you’ve got”, which isn’t such a bad thing but I admit if taken too literally we wouldn’t make any progress.

    Up until yesterday’s post and comments I’d got the impression there was great love for streaming on this site but now it seems that’s not as universal as I thought. All very interesting. Like all mediums it has its problems and the deeper we go I expect the more we will find. Heck, these issues even brought Steven out of ‘retirement’. I don’t stream but see it’s major benefit as convenience, not necessarily the absolute best sound, even though that’s inferred. Reminds me very much of when CD was introduced all those years ago. Whatever happens, I believe to get the BEST sound with streaming you’re going to have to pay, and likely a lot.

  13. Richtea….

    Not long ago I made an audio streaming discovery that immensely improved streaming, and without seeking to. AT&T recently mandated that all DSL users in my area switch over to their high speed fiber. The combination of long runs of fiber and a new improved DAC took me by surprise. Apparently, long runs of copper going to the house, and then to the computer, does cause noise added to the sound. And, I turned off the WiFi option on the modem and running an audiophile Ethernet cable to my PC… Sometimes it sounds more like records, Not, digital. Sometimes. It all depends on the web provider.

  14. There’s something rather interesting in there, and that is the fact (because I hear it clearly myself) that the same tracks sound usually different on TIDAL and Qobuz, with Qobuz almost inevitably being preferable.

    Unfortunately, in my case at any rate, I am constrained to listening to both via their dedicated Mac Apps. As an an audio software developer, I am aware that different playback Apps have the capacity to sound different. However, the differences I hear when comparing TIDAL and Qobuz typically go beyond what I’m used to hearing when comparing different playback Apps using the same source material. Which leaves me wondering whether the source material is in fact identical. Or, if they are, whether the actual streamed bitstreams arriving at the local computer are identical.

    Unfortunately, my TIDAL subscription has now lapsed, otherwise I think it might have been an interesting idea to set about capturing both bitstreams and digitally compare them.

    1. I really think you’re onto something here Richard. I have a background in IT systems, applications and networking and I believe (based on my experience) that the processes involved in getting the analog to digital output generated at the recording session to the digital to analog converter at the ‘listening’ end are largely unimportant. As you say, provided that the bit stream that’s fed into the DAC is ‘bit-perfect’ then the sonic quality of what you hear is all down to the what the DAC designer has created (and of course to the recording and mixing engineers 😉 ). Let’s just reflect on the ‘bit-perfect’ terminology. I understand ‘bit-perfect’ to mean that the bit stream as ‘seen’ at the ‘listening’ end is identical to that created by the analog to digital process at the recording session. That is, no single bit has been ‘flipped’ from its original ‘on’ or ‘off’ state. So, IMO, it doesn’t matter how you ‘transport and deliver’ the bit stream to your DAC, be that via CD, SACD, PWT, PST, Quobuz, Tidal or whatever, so long as it’s ‘bit-perfect’ on input to the DAC.

      This is why I have great expectations for the upcoming Ted Smith Signature DAC. His idea to galvanically isolate the digital to analog processing side from the analog side using two ‘boxes’ seems such an elegant concept. I don’t normally lust for the ‘next best thing’ to come along but I think in this case I do.

  15. Wow. Tons of comments on this baby. Paul. You must be loving this! 🙂
    To follow up on the main topic of Paul’s post.

    I use a Cyrus CD i and it employs what is known as their patient SERVOS technology for ascertaining bits of information from a cd. What SERVOS specifically does is gather up all the information at once to avoid ever using error correction. People wonder why Apple Music sucks. They use error correcting and that creates a whole host of different sound anomalies and of course timing/transient errors/misinformation.
    I have to admit. Once knowing about that tech that Cyrus uses it really put me over the edge to buy the CD i. I have not regretted since because compared to my OPPO BDP 83 running through the same dac…yup. You guessed it. The Cyrus sounds better because of how that player takes in the bits of information.

    Honestly I love to compare the PS Perfect wave to my CD i someday….maybe someday. 😉

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301
1-800-PSAUDIO

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

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