|Interesting... But in physical reality impossible of course! Perhaps like many audiophile opinions? [Info on art here...]|
"Your desires and true beliefs have a way of playing blind man's bluff. You must corner the inner facts."
--- David Seabury (1885-1960)A few weeks ago, I read this "Quick Take" review of The Linear Solution DS-1 Network Streamer device published on Computer Audiophile (now Audiophile Style) by austinpop. Wow... I was impressed by how many computer audiophile "beliefs" were strung together all in one article! A good place to start and think about the "evolution" of computer audiophilia in 2019 perhaps.
So let's enumerate some of the claims mentioned as supposedly important elements in a good sounding streamer / computer in that article.
1. A low power device with low electrical noise production results in better sound.What a list! Yet which of the items above can be proven to be true? Are most of those beliefs of things that can be done for the sake of making a system "sound better" more likely to be "inner facts" or are they likely one's "desires and true beliefs" as per David Seabury? Remember, there is a big difference and it's not an issue of sincerity or absolute faith! A farmer can be totally convinced in his "true belief" that praying to
2. "Everything matters", especially PSU quality and clock quality.
3. Running the OS (Audiolinux for this device) completely in RAM "enhances sound quality".
4. Intel's 7th Gen NUC (like the NUC7PJYH at the heart of this system) + fanless case + OS + good PSU sounds "shockingly good" as a network streamer.
5. RAM latency (DDR4 in this case) is "customized" in the device and that presumably makes a difference to the sound.
6. OCXO (Oven-Controlled Crystal Oscillator), in this case of military spec, applied to USB, ethernet, and system clocks makes a difference. The claim is that this feature "is the reason for the bulk of" the sound improvement.
7. Network streaming computers like this still need to "burn-in". In this case, sound improved "dramatically" over 24 hours.
8. The CPU used makes a difference. An Intel i7-based NUC7i7DNBE apparently sounds "better" than the Celeron-based NUC7CJYH.
9. Sonic improvements include airiness, big/deep soundstage, micro-details, "refined" treble, and of course "better" bass definition/clarity.
10. Sonic improvements from better power supplies are similar to #9. Linear power supplies are better than switching power supplies.
11. BIOS tweaks make a difference to sound quality - turn off functions like SpeedStep, wireless (Bluetooth & WiFi), interfaces like SATA if not needed, and onboard audio.
12. Crippling the ethernet down to 100Mbps presumably improves sound quality? At least Linear Solutions seem to think so for their product since they apparently purposely did that.
In this article, the Intel NUC7CJYH being referenced is part of the 7th generation of Intel NUC devices. Through all these generations, I'm pretty sure Intel has learned a thing or two about creating low power devices that simply, reliably, work. Do we honestly think there's a significant difference between a NUC7 and a NUC6, or maybe a current NUC8?
While I'm sure the DS-1 streamer looks great in person, the fanless customized enclosure costs some money to produce, plus the manpower to create the customized Linux OS might be lightweight and responsive, let's not forget what this device is. Anyone of us can get most of the way there by buying a <US$200 Intel NUC7PJYH which has a Pentium J5005-based quad-core 10W processor with approximately the same computing power as an old Core 2 Q6600 processor from 2008 (in the last 10 years, power demands and architectural efficiencies have really improved!). Grab some high quality 8GB SODIMM RAM (perhaps a good Crucial DDR4 2400 for ~US$60? Tweak the timings a little in the BIOS if we want.). Replace the case with the fanless Akasa Newton JC for ~US$100. Grab a 64GB Samsung USB 3.0 stick (~US$25) as shown in the picture in the article. And buy a license for Audiolinux (US$50).
Adding up the price list above, we're conservatively looking at ~US$435; let's call it US$500 with whatever tax and shipping.
The only difference comparing the parts above to the device under review is the OCXO clock, customized case, and whatever other "optimizations" in software was done to the base Audiolinux. All this with a price differential of +US$1100 with claims of better sound and no factual evidence provided by the company.
For sure, I believe it is good to build an audio streaming device meant for the sound room that is reliable and looks good! But let's not over-reach with dramatic claims of sonic improvements when neither logically nor empirically does there appear to be reason or evidence of benefits from a device like this.
Beyond these basic principles:
- A fanless case is a good thing because fans make noise and we don't want that in our listening rooms.
- A quality power supply is good to have - more reliable, hopefully no audible transformer hum.
- No spinning hard drives or other active mechanical components that could be audible.
- The computer is fast enough and capable enough for the features it needs to provide. Running the appropriate OS/software.
- Able to achieve bit-perfect output to the DAC as a basic requirement for transparency. This is both a function of the hardware and software.What else can we really "hang our hat on" for sound quality?
The role of the streamer from the perspective of sound quality is to make sure the data being delivered to the DAC is available when the DAC needs it... And this can be satisfied quite easily and we have data buffers in place which will take care of tiny latencies. Even on the DAC, as I've expressed before, precise femtoclocks are not particularly beneficial if jitter is already minimal and of questionable audibility as shown a few months back.
As per John Swenson's post from June 2017, remember that OCXO clocks (which is what is being attributed as the major source of sonic improvements here) is simply about keeping the temperature of the "oven" optimal for the rated crystal frequency. Suppose the clock speed of a computer is super duper accurate and doesn't change with temperature, so what? Again, remember that the computer clock has essentially no effect by definition on the asynchronous USB DAC. It's not like a computer running at 3.0000GHz "sounds different" from 3.0154GHz, right? Clock anomalies of the DAC itself is where jitter and other temporal anomalies would originate and potentially change the sound. If say the computer is not fast enough and at times keeping the buffer populated is marginal and we run into underrun conditions, will a super duper stable clock speed improve things? Of course not, what is needed would actually be a faster clockspeed to process quicker, perhaps by overclocking the computer or replacing the CPU with something that's faster. No amount of tweaking a modern multi-core CPU machine, changing what already are reasonable OS parameters, fooling with the nanoseconds of RAM timings would make an iota of difference to how this works.
It's actually rather interesting what items are being "picked on" for these optimizations. For example, why was the ethernet speed downgraded to a mere 100Mbps by The Linear Solution? Assuming high-resolution PCM/DSD streaming is desired, wouldn't keeping gigabit ethernet (1000Mbps) be a good thing from the perspective of lowering data transfer latencies and if you want even lower ethernet transfer jitter, go with 10GbE as I showed last year! (I hope we don't see "audiophile" 10GbE in a few years. :-) If "everything matters", shouldn't these timing differences measured in milli- and microseconds be more significant than nanoseconds for RAM timing? Funny that the author makes a point to state that the ethernet clock is derived from the OCXO as if that would have any significance in the data transfer! [I see that the company sells TCXO and OCXO "audiophile routers", taking a page from the just-as-questionable JCAT people selling expensive computer hardware in Europe. Absurd.]
Looking at the Audiophile Style forum, we see that a discussion thread about the topic of optimizing an audio computer needs to have "guiding principles" to maintain "reasonably high S/N ratio". As plainly said, the guiding principles demand that the participants in the thread foregoes proof, requires no specific methodology, and measurements are unnecessary to this kind of "experimentation". In other words, reasonable, intellectual discussions that seek answers based on principles that we would apply to essentially any other endeavor where we seek knowledge and facts need not apply here! This is amazing considering that computers are the fruits of science and engineering.
One apparently cannot question the accuracy of "direct listening experiences". Despite the thread OP's insistence, does it not imply that "this is
By the way, the Audiolinux site does show a few measurements like CPU latency, and even a 24-bit J-Test. Of course, for the J-Test, there's no "before and after" comparison to show jitter was in any way diminished when the computer used Audiolinux compared to say a generic Linux install. They would of course find no difference yet I'm sure some will claim that the system "sounds better"; typically attributed to improved jitter and noise.
Humans are remarkably creative creatures. We want to control things - or at least believe that we can control things especially when we're feeling passionate about whatever it may be. But in the attempt to tweak the parameters and try to make things better (in this case sound quality), how do we make sure that we're not going down the path of blind faith or delusion (this could be self delusion or some kind of delusional community)? How does a discussion thread stay "truthful" or "factual" if utilizing our intellectual abilities to do things like produce measurements to falsify a theory is dissuaded? Without a desire for evidence, are we not then potentially (even likely) perpetuating what could simply be wishful thinking?
For technological devices, I think as modern consumers we understand that there are bound to be some intangible values we can ascribe to our purchases. Whether it's a fancy brand name, appreciation of build quality, recognition of rare products created in small volumes, or even being okay with the "hype" at times which can improve desirability. My belief is that when there is a "headline" purported benefit to differentiate an engineered product from competitors, this should really be verifiable. The obvious feature for this The Linear Solution Streamer is the OCXO clock which the subjective reviewer also attributes as the main factor that "improved" sound quality. If I were a prospective buyer ready to shell out US$1600, I think I'd first make sure to contact the manufacturer to understand how they confirmed that this "headline" feature actually makes any difference to the sound of a USB DAC.
To end, I suppose when it comes to audiophilia, there's no great harm in just having "fun" (so long as it's legal of course). But "fun" can be very different from "right", and something "fun" certainly might not be "good". I'm personally of the belief that it's better to accept reality as it is rather than seek magical "cures" no matter how much "fun" it might be to imagine (eg. consider the "snake oil" of various sorts). In 2019, figuring out if a computer sends data to a USB DAC properly as in being bit-perfect and in a timely fashion has become far from "rocket science" and really does not need all kinds of fanciful myths.
In the grand picture, I think as audiophiles, it's important to consider the hobby itself and whether the things we say and how we portray ourselves result in engagement by the general public. IMO, writers, reporters, and reviewers should introspect and make sure what is expressed approaches the ideals of what is "true" and "good" in a way that promotes a healthy vision of the hobby. I believe it's hard to have a strong, vibrant hobby if it's not based on good foundations. Seek the "inner facts".
BTW, for more on the topic of differences that streamers and computers may/may not make, refer to this recent article from late 2018.
As usual, remember to stay rational.
The "Do digital audio players sound different?" Blind Test is still "live" as of today's post. Please listen and report back. Let's gather some actual data! How else does one dispel myths and the need to appease
Remember to enjoy the music first and foremost of course!