Bits are bits… or are they?
A year ago, my intrepid editor John Reekie asked if I’d like to participate in a tag team review of USB cables. Audiophile USB cable manufacturers claim superior fidelity. At the time, my reaction was “Color me skeptical, this should be amusing…” Going into this review, I knew that, the longer the cable, the more opportunities for the physics of capacitance and inductance, in cahoots with EMI or electromagnetic interference, to degrade the signal. I erroneously assumed that I wouldn’t notice any differences between short USB cables, regardless of the construction or cost. I was… mistaken.
Let me preface the remainder of this discussion with a declaration: I am not a Bits Are Bits kinda guy. Just because I do not yet have a theory that explains the underlying mechanism of an effect, that doesn’t invalidate my sensory experiences. The behavior of digital data is not necessarily deterministic end-to-end, from analog through digitization and back to analog. In other words, what you put in at the beginning, you do not necessarily get out at the end. It is true that digital data, once captured, is or should be deterministic. The complications lie with that pesky analog domain we humans require.
I started my quest for truth in advertising by visiting my fellow ASM (Audiophile Society of Minnesota) board member Chuck, who was gracious enough to invite me over one crunchy cold weekend. We descended into his comfy, well lit man cave, where he has a modest but highly resolving system set up. At the time of testing, Chuck’s rig consisted of:
A generic Intel machine with Win7 running JRiver Media Center 19, DSP disabled, which feeds a generic USB card with an outboard linear PSU of Chuck’s design. USB data is isolated from USB power. A Bel Canto uLink USB–to–AES over fibre converter feeds, in turn, a Bel Canto DAC 3.5 MkII DAC. The DAC’s output feeds a DIY 25 Watt 845 push–pull, transformer–coupled zero feedback amp, also of Chuck’s design, with Lundahl transformers and Cardas internal wiring. His loudspeakers are each single (Lambda Acoustics) TAD compression drivers with beryllium diaphragms mated to a Peavey Quadratic Throat waveguide with a 15″ direct radiator bass driver. The bottom is provided by active stereo dipole subs, à la Siegfried Linkwitz, with DSP and IcePower amplification. The amp feeds the loudspeakers via Level 3 cables from AntiCable Audio.
We had several USB cables on hand, all 1 meter in length. Our cables under test were:
- A generic USB cable
- Chuck’s Light Harmonic LightSpeed
- My Wireworld Starlight 7
- A VUE Cables VU-1 loaner
As test material, we used 44.1/16 FLACs: Simon & Garfunkle’s “Overs” from Bookends, Willie Nelson’s “Stardust” from Stardust, and Copland’s “Fanfare for the Common Man” from Eiji Oue & the Minnesota Orchestra’s COPLAND 100. To train me, as Chuck had already been doing quite a bit of comparison with cables 1 through 3, we started with the commodity cable… Using as test material a recording of Elmer Bernstein’s score from The Magnificent Seven, my notes read: “smeared transient response, shallow narrow soundstage, veiled.” From there, we jumped directly to ultra–bling; the Lightspeed. This time I wrote, “…vastly improved transient response.” I was also hearing “… nice side (wall) reflections in the (acoustic) space” and “… very good localization.”
Next, we compared the Starlight 7 directly to the LightSpeed; the Wireworld yielded an “immediately shallower soundstage, softening of brass’ bite,” and a “slight veiling.” The LightSpeed offered “better (phantom image) placement, more air, (and) better transient response.” So, there really was something to this whole USB thing! The Wireworld was a no contest winner over the generic and, at ten times the price, the Light Harmonic trounced the Wireworld…Fascinating!
On to the VUE, which was the reason I dove into this madness in the first place. The VU-1 is about the same price as the Starlight 7, although they look quite different. The VUE is hand-made, with a round cross-section, while the Wireworld is machine-made, with a flat cross-section. The Wireworld has a softer “hand,” while the VUE is less flexible and a bit stiffer. Both are flexible enough to not cause your DAC to randomly relocate itself in space.
VUE Cable’s VU-1
The manufacturer of the VU-1, Kevin Best, has a background in electronics, working on computer hardware. He’s been in the audio industry since 1990. Early on, he was drawn to the pairing of computers and audio: “When I first heard a computer setup playing music files through a USB input on a DAC, I thought that it was okay but nothing special, and really did not think a lot about it’s potential. As time went by and… improvements were being made, it became apparent that this was the future in the form of another media playback. I however did not really think that the USB cable, especially 1 metre length, would make any difference in sound quality. After all, it was just sending digital information, wasn’t it?” That was my view as well, at least until a few months ago…
Best mentioned that the first time he heard an audiophile USB cable, he was amazed that it “really did make a difference.” He started experimenting to see what areas could be improved. “… The first and the most obvious was trying to eliminate noise. The amount of EM (electromagnetic) radiation that comes from processors in computers is quite a lot. Not only inside the computer, but around it also, so that cables would be picking this up as noise and then help make that “grunge” that covers up the blackness between the notes of the music. I do not feel that having the data lines and the DC voltage lines together without isolation from each other was ever going to eliminate the noise. So isolating the data lines really did help, however I have found that isolating the voltage lines improved (the sound) further.”
In all three VUE USB cables, the philosophy is to “make sure that data lines are isolated from the outside noise and the DC Voltage lines are also isolated from the outside noise and both are isolated from each other.” The Light Harmonic uses a similar design approach, though that product physically separates the two paths with quite some distance between them.
The VU-1 wraps the 5 volt DC line that computers use to power a USB peripheral in its own foil shield to isolated any noise that the power line may be carrying out of the box. The data lines are also isolated in their own foil shield to protect them from radiated EMI or electromagnetic interference. By the way, foil shields are just that; a gauge of aluminum foil a good bit thicker than what you’d wrap your turkey in.
The attention to detail doesn’t end with individual shields. The data lines are then separated from power by a plastic spline so the two foil shields are not touching each other. This eliminates any “short circuit” between the shields that may occur, which would, electrically speaking, negate the whole separate shield scheme. It also reduces the capacitive coupling of noise from one shield to the other, as the farther apart they are, the less coupling. The VU-1 then has a third foil plus braided shield – stranded wire knitted into a tubular shape – around the entire cable to further isolate the data from outside EMI and the massive radio frequency (RF) noise that is generated by switching power supplies and the digital circuitry being powered. As with all the VUE range, ROHS–compliant silver solder is used to terminate the wires at the USB plugs.
Chuck and I did another day of listening as he wanted to compare some prototype cables he had from a local manufacturer. Again, the same results: when evaluated against the far, far more expensive LightSpeed, the VU-1 retained low frequency solidarity, and the wider, deeper soundstage.
I took my Starlight 7 and the VU-1 home and continued to listen. This time, I had my standard rig, with one exception. I had a Violectric Audio V200 headphone amp for eval. After running it in, I sub’ed it out with my Phonitor for these tests with the same basic results, although one aspect we noticed at Chuck’s was also apparent on my system: the Wireworld sounded more rounded, more “analog,” for lack of a better word. We had noticed that as well with one of his prototypes. The Light Harmonic, and to a less extent the VUE, offered a more analytical rendition. Actually, the Light Harmonic was found to be a bit strident with some recordings. Kevin Best mentioned that his top of the line cable can, as with other high-end gear, expose any weaknesses in a recording. During my listening with Chuck, we found the VU-1 exhibited less glare and generally, a slightly deeper soundstage than the Light Harmonic.
Lest you think this is all esoterica and could not possibly apply to your system, consider this: I could hear the differences my three choices made even when using iTunes as my source. One day, my buddy Steve dropped by the house to take a listen. He’s a pro audio guy, through and through. Any whiff of audiophile voodoo sends him running. So, I sat him down in front of my speakers and did some swapping. After a bit, he shook his head and stated, “You have ruined me for life.” Bits are not always bits.
Best’s three models are all handmade by him and are tested at various times during assembly. The next model up in the line, the VU-2, uses Ohno† copper while the top VU-3 model has data lines made from Ohno silver. They are also cryogenically treated and have more elaborate shielding. He seems to discount his time quite a bit, considering the price. “I certainly could not sell them for the prices they are if they were manufactured by someone else.” Also, he sells direct to eliminate any additional markup. “The VU-1 is the entry line of the VUE range and is a great value for (the) money… and most likely is all you will need in most… systems.”
I agree. The VU-1 is an affordable tweak that really does improve any USB source. It was unfortunate that I couldn’t keep the VU-1. It has to go back out, this time to fellow editor Thad Aerts. Sucker that I am for good value coupled with excellent performance, I purchased a VU-3 as my soon-to-be new USB reference wire. It should be… amusing!
Music in heavy rotation during this review:
Kate Bush — The kick inside (1978 EMI re–release) [The young Ms. Bush at her best]
Doug Macleod — There’s A Time (2013 Reference Recordings) [straight up blues, recorded live at 176.4/24]
Tia Fuller — Angelic warrior (2012 Mack Avenue) [a definitive Body & Soul featuring Dianne Reeves]
Nicholas Payton — Into The Blue (2008 Nonesuch) [fine modern jazz with the occasional soul interlude]
Poliça — Give you the ghost (2012 Totally Gross National Product) [the best proof that delay is your friend]
Gregory Porter — Be Good (2012 Motéma) [Be Good (Lion’s Song) is a jazzy yet mercilessly invasive ear worm]
Additional gear used for this review:
Sources: Amarra Symphony w/iRC, iTunes
Cabling: Soundstring GEN II Beta 2-22S, Perfect Vision RG-6
Conversion: Mytek Stereo192-DSD Mastering Edition, Calyx DAC 24/192
Amplification: Bryston 3B, SPL Phonitor (factory mod’d), Violectric Audio V200
Cans: Audeze LCD-3, Etymotic ER•4PT
Speakers: Bowers & Wilkins 685, IsoAcoustics ISOL8R155, Sanus Foundations Steel
Power: Soundstring GEN II Digimax-18
† The Ohno Process —
Since ye olden days, the process of making wire involves cast ingots of metal being drawn or pulled through an unheated tapered die that “cold works” the metal, brutally deforming it into the desired cross–section. This process, still used today for commodity wire, results in an anamorphic bulk material, where the crystalline structure is stressed and disorganized.
The Ohno Continuous Casting Process®, as described in 1986 by Dr. Atsumi Ohno, a Japanese metallurgist, is a process akin to how ultrapure silicon is drawn from a molten pool to create the single crystal boules used to manufacture semiconductors . In the O.C.C. Process, wire is continuous drawn from a pool of molten metal though a heated mold,“… to hold its inner surface at a temperature above the solidifying temperature of the molten metal to prevent the formation of new crystals on the wall surface of the mold.” The result is a long, single crystal ingot which can be drawn into the desired cross–section.
Obviously the people at HIFiZine must know all this, but I’ll state it anyway:
A vital question is: does your DAC use asynchronous transfer or isochronous? If it’s asynchronous then, barring a catastrophic cock-up of a cable that causes actual errors, then bits really should be bits. If your DAC is isochronous then jitter *will* reach the output and *may* be marginally influenced by the cable. It may be negligible, but it must be there – maybe just a single minuscule rate correction every hour which of course would not have an audible effect, but you’d rather not have it at all, and you don’t have to.
If the DAC is asynchronous and electrical noise is being coupled from your PC to the DAC then the cable will not really have much influence on this – there are real engineering solutions to eliminate it rather than mysticism.
A DAC system with asynchronous transfer, electrical isolation between it and the source, and a cable capable of transferring bits without errors (no great feat) will then be independent of the cable quality, just as a word processing file transfer would be. This is as a result of straightforward design, eliminating (not just reducing) errors. No measurements and no mysticism are necessary. As a further check, measurements of various types can be used to verify the design.
To doubt this is to doubt the ability of each and every engineer to find solutions to fairly simple problems. Better stay in bed to be on the safe side!
Which DAC’s would you then recommend which have asynchronous transfer, and perfect electrical isolation between it and the source? I’ve spent a lot of effort and $$$ with cable solutions, and am interested in trying another approach.
Hello Brad
I’m not up on the latest in DACs etc. but something like these?
http://www.bladelius.com/products/bladelius/usb-dac/
“- Asynchronous USB transfer, re-clocking with a fixed low jitter crystal clock
– Galvanic isolation between USB and the analog circuits, which means no noise from the computer will be transferred to your amplifier”
http://www.jdslabs.com/products/84/c5d-amplifier-dac/
“ASYNCHRONOUS AND GALVANICALLY ISOLATED”
You could use a separate USB isolator with a non-isolated, but asynchronous, DAC such as the Arcam rDAC.
This is cheap
http://www.whiplashaudio.com/firegreenkey.html
And another
http://www.cesys.com/en/products/kategorie/signal-processing/produkt/usb-isolator-1/
The problem does have professional grade solutions:
http://www.analog.com/en/interface-isolation/digital-isolators/adum3160/products/product.html
with a ready made board
http://www.analog.com/en/evaluation/eval-adum4160/eb.html
What do you think?
We need a lot more ingsthis like this!
@Brad
I’m behind the times. The Arcam irDAC (not the obsolete rDAC) is described as follows:
“The Arcam engineering team gain a great deal of new insight while developing the class leading FMJ D33. Areas such as isolation of digital and analogue stages, ultra-low noise power supplies and direct coupled signal paths make a big difference to outright performance. The irDAC uses the outstanding Burr Brown 1796 DAC and 8 separately regulated power supplies to ensure class leading performance that are unmatched in the irDACs price category. Jitter reduction is an obsession within the Arcam engineering team. The circuits developed for the D33 anddeployed in the irDAC produce a signal that is almost entirely jitter free.
The irDAC is designed to be the heart of a digital system and can be connected to a host of different types of digital sources and connections. This includes asynchronous USB…”
Sounds pretty good to me!
(Not to mention their FMJ D33 DAC which also has isolated asynchronous USB)
@montraco whatever…. But the cable does not carry a stream of bits ….it does not carry zeros and ones, it carries voltage values which then must be interpreted as on or off values or bits if you will. The value of the voltage can vary more than it should and then be “misinterpreted ” as a zero instead of a one. Or vice versa. One may engineer solutions for transmitters and receivers , but if one doesnt control the variable in this case the transmission pathway , then best guesses or in engineering practice assumptions about the quality and quantity of the signal deal with expected, and in theory some unexpected values based on an ideal electrical transmission pathway . However ideal pathways only exist in simulations and rarely in practice . Therefore the practice that you stipulate to doubt each and every engineer to find a solution to a simple problem only exists in your head and no one else’s for people get out of bed each and every day despite the mistakes engineering practices make
I have to agree with Isis in that the real world is not as simple as we could wish it to be. Underneath all that “digital” is analog circuitry…Schmitt triggers and the like that act “digital” but operate on analog voltages, with power supply noise sensitivity and other non-ideal behaviors. — OMas
If I was a digital cable manufacturer every advert of mine would begin “The cable does not carry a stream of bits ….it does not carry zeros and ones, it carries voltage values”!
There are genuine mysteries in life, but the transmission of digital information over a cable isn’t one of them. As I said above, if the system is so bad that the bits are misinterpreted at the other end then all bets are off, but it’s easy to design so that doesn’t happen, and then easy to demonstrate that over the cable there are no errors. The digital cable that carries the bits from your PC to your HD monitor is doing a far harder job, but the bits are never wrong at the other end. Just as within the motherboard signals always get through their respective PCB tracks, and the same in miniature within the CPU chip itself.
It’s true that from one cable to the next, the actual voltage levels, shapes of the pulses, ringing etc. will vary by very small amounts, but the very essence of digital communication is that these differences are irrelevant. However, in the case of the isochronous DAC these variations may make it to the output in the form of almost infinitesimal differences in timing. In the case of the asynchronous DAC they are truly irrelevant. Think of it this way: if you download a file from the WWW with your PC, does it matter whether that file has come via a satellite, undersea cable, microwave link, piece of string? Clearly not. Your PC completely eliminates every trace of the analogue aspects of that communication. The asynchronous DAC is just the same system in miniature.
Yes, voltage values that get interpreted as a 1 or a 0. Now, say that somehow there is a single bit, or even a few, in a packet that would cause that voltage level to change enough that it would no longer be the correct level and would flip (this is a large variance, of course, and unlikely to ever happen). Well, that’s not a problem because every single packet of data that is coming also has a CRC error control value to make sure the data that is received is the same as what is sent. Not only would those bits in the packet of data have to change, but the CRC data would also have to change in the exact same method to cause it to pass. That won’t happen.
So if that packet is bad, the asynchronous DAC will just request a new packet to replace the missing one. Since the DAC is controlling the clock, and has a buffer of data, it can request the packet and get it into place without causing any issues. A USB cable won’t cause jitter with an async DAC since the DAC is in control of the clock, not the cable or the PC. It also won’t cause packets to drop, because the USB specification already controls for that. A nicer cable will certainly look pretty, but that’s where it stops.
Of course you can’t disprove that someone believes they are hearing something that they aren’t. And I’m certain you can engineer a nicer USB cable that when hooked up to a testing device will show characteristics that you might think would provide a benefit. The way that USB is designed and DACs are designed now eliminates the cable as a source of interference or issue.
Hey Chris,
Yup, I used to subscribe to the Bits are Bits school as well but, as I mentioned in the piece. I heard distinct differences between cables. I didn’t have access to the test equipment necessary to eliminate the possibility of pathological misbehavior on the part of the cables under test. That said, I have heard the same effect on completely different systems.
Rather than going on blind faith, why not try a critical listening session or two yourself, on a highly resolving system…or two. If you don’t hear anything; fine. You can fall back on Ye Olde BaB. — OMas
“Rather than going on blind faith, why not try a critical listening session or two yourself, on a highly resolving system…or two. If you don’t hear anything; fine. You can fall back on Ye Olde BaB.”
What makes audiophiles so confident that they are superior to all other humans? i.e. that they are immune to false perceptions based on psychological factors? Any psychologist knows they can influence subjects to perceive false sensations just by telling them what to expect, and they would find it amusing that you believe that if you feel something it *must* be real. I know that if I tested myself in a sighted test I would hear the differences you describe – even if I mistakenly didn’t change the cable, and I’ve actually done that!
http://www.engineering.com/Library/ArticlesPage/tabid/85/ArticleID/5301/Five-Biggest-Engineering-Disasters.aspx
Maybe people should stay in bed
Since we are talking about digital transmission, it should be really easy to connect a logic analyzer to both ends of a given cable and compare the waveforms. we don’t have to rely on our ears to compare digital cables. I doubt any cable is perfect so then the question becomes how tolerant is the USB input of the DAC. I’m sure some are better than others and might explain why some hear a difference.
Hey Kevin,
Of course, that should be done at the early stages of the design process, to make sure that the cable meets spec.s and is actually doing its job well. One of the main points I tried to make is that designers, and consumers, _should_ rely on subjective as well as objective evaluation since a logic analyzer clearly is not telling the whole story. — OMas
“… designers, and consumers, _should_ rely on subjective as well as objective evaluation since a logic analyzer clearly is not telling the whole story”
Why “clearly”? What is the evidence for there being a flaw in the system? If the answer is that people think they can hear a difference, well what are your qualifications in psychology?
I will add that my best results with various DAC’s have been achieved when using split USB cable solutions (with attached battery or low noise power supplies), where the DAC does not see any power from the computer’s USB port. Not very elegant solutions, but the sound improvements have been worth the effort with my DAC’s.
That makes sense. But the Arcam DACs above, and similar, have their own power supplies I believe, and the USB cable is for data only.
Hey Brad,
As we found, isolating power from data yields improvements. — OMas
Hello Mr. Tautliner,
I do not pretent to be a psychologist, but I am an engineer and I am also paid to be a critical listener. I performed my test with two other engineers and, though they could not explain the result, nonetheless we are not so sure of ourselves that we can dismiss what we heard.
As I mentioned before, try this with yourself and an assistant, and then tell us what you heard or did not hear. – OMas
Hi OMas
I am not dismissing your particular test, but I challenge the conclusion that you draw from it i.e. that bits are not bits. Truly conclusive experiments are not easy.
In your mind, the only thing that changed was the cable, but in fact many things changed: where you or your assistant were standing, how long the equipment had been switched on for, the temperature of the room, the tiredness of your ears, the fact you were hearing the same piece of music several times in a row, the background noise etc. etc. And the test was, presumably not double blind – a requirement that scientists didn’t invent for fun. These factors were all real *and measurable*, but I’ll hazard a guess that the differences between cables would not be measurable. Yet the narrative you have chosen to construct is that any audible differences, real or imagined, were due to the cable alone.
I don’t know whether your DAC was operating asynchronously, or that the analogue sections are isolated from the digital nasties, or indeed that the DAC is truly immune to the bits’ arrival times. Only if these (perfectly practical) conditions were met would the headline be in any way justified, but even then by Occam’s Razor, I would rule out the cable being the culprit in favour of all the other variables. In fact I would put a large sum of money on it!
MT, I’m going to interject and suggest that perhaps you might take a step back and consider what it is that you are doing here. There is no war here, never mind a battle, to win. If you would like to step up to the plate and demonstrate the courage to actually write and publish an article on this matter, please use the contact form. Thank you.
Sorry…. I know it seems churlish to have posted what might be construed as negative comments. I apologise. But simultaneously the comments are also positive in the sense that they support my fellow engineers who, in the audio world at least, are somewhat under siege these days! Our job is to solve problems and the only tools we have are science, reason and logic. We are put in a very awkward position when the received wisdom says that, in fact, there are clearly-audible imperfections that we literally cannot explain, never mind cure. Is it a total exaggeration to suggest that the livelihoods of certain types of engineers are at stake, or that the future of the audio industry itself will be determined by what is said in articles like this one?
I’ll leave it there!
“Is it a total exaggeration to suggest that the livelihoods of certain types of engineers are at stake, or that the future of the audio industry itself will be determined by what is said in articles like this one?” Yes – I’d say that would be more than a little far-fetched.
P.S. Please feel free to remove my posts – I hate to think I might have upset anyone.
Interesting comparison to add to others that have been done – It seems pretty well a given at this stage of the game that USB cables do, in fact, sound different. It has been some years since Wavelength Audio’s asynchronous protocol emerged, and started the buzz about the different sound of various cable. I currently run a USB adapter from my laptop to the DAC. Both lay on the floor. That sounds the best from my experience. No ‘cable’ at all. Im sure YMMV…
No double-blind testing — results are inaccurate.
Agree with Dik, double blind testing is the elephant in the room.
I have known since around 1980 when 7 out of 8 “golden ears”
(respected hifi journalists) were unable to pick out the “superior”
Linn Sondek to it’s rivals (guessed wrong) in a double blind test.
Ever since DBT’s have been trashed by the hifi press (wonder why). Without DBT the psychological components of the human brain are able to run rampant. The problem then is our egos overide any objectiveness(it’s way more expensive it has to be better is one). The brain will fool you every time. Don’t believe me, try this: after several months of using your expensive new component, put back the old component (change nothing else) & see who is kidding who. After countless (long term) experiments this is what I have found.
1) In a digitally sourced system the biggest influence on sound is
the chip. Delta Sigma chips are sonically inferior to the earlier R2R
ladder chips. Audiophiles were given the shaft way back then as the criteria for the new chips were: less costly, easier to implement, end of story.
My Magnavox (I hear the chortles) CD player with TDA 1541 chip outperforms any SD based system I have tried (way more expensive), by a country mile.
Can anyone tell me why it is so many of the top of the line DAC’s
use R2R chips that have been out of production for close to 25 years?
2) discussions about cables are really side issues compared to the
fundamental issue of sonic inferiority (read digititus) since the introduction of SD chips.
Try this, buy a $100 CD player off ebay with a 1541 chip (or similarily respected R2R chip & do a double blind test with your current set up & publish the results.
Hi Mr. Loomis,
I will say that my findings are incomplete, more work is needed. Now that I own sufficient test examples, I will be arranging some double blind testing. I am working to borrow a protocol analyzer or ’scope that can help to track down the cause of what we heard. In the meantime, I have opinions, as do you, that are not verified by double blind tests.
I will also say that one should not lump all listeners together in one, clueless mass. Some audio engineers are trained listeners, having to make very subtle quality judgements every day. Mastering engineers come to mind as an example. Same goes for sommeliers; they rely heavily on their senses. These folks are rigorous yet also rely on their subjective judgement, without which they could not do their job or create products that we all either love or hate.
OMas
Hi folks,
Re-reading this discourse after some passage of time has brought not a new perspective but a renewed sense that our ear-brain mechanism is more complex that some of you folks seem to assume…
I wrote a piece for another editor friend of mine, over at audioXpress, that you may find thought provoking or will simply dismiss as more touchy-feely nonsense. Take a look:
http://audioxpress.com/article/The-High-Resolution-Audibility-Test.html
Thanks for listening,
OMas