Let’s talk why USB cable makes a difference

Status
Not open for further replies.
It’s a controversial subject and even more so because it’s not some analog speaker cables that you could intuitively see why, but it’s a USB cable that carries purely digital 1 or 0. To make thing worse, I have not heard a clear and logical explanation from anybody using actual electrical engineering principles. I’ve once read an Hi-Fi + article in which they interviewed a bunch of highly reputable … ahem… cable engineers to explain why a USB cable would make a difference. Although most of them had first hand experience about the differences, I don’t think any of them gave a really satisfactory answer.

Before going into any detail, let’s approach it from a perspective that at least makes sense from an engineering point of view that most sane people will agree. Whenever you hook up two electronic components together, it’s indisputable that there will be an exchange of fluid, in this case electronic fluid so get your head out of the gutter. And the conduit that allows that fluid to flow in this case is the cable. And how the cable is constructed will affect the way the fluid will flow between the two components – and this is also indisputable. So at least in theory, if you follow the entire chain, you have component A does something to component B through a cable and component B ultimately produces an electrical sound wave, then by purely logics, you can conclude that component A ultimately affects the electrical sound wave coming out of component B, and since A affects B using the cable, therefore I have established at least in theory, a cable can affect the sound of component B. But that is a Big Gap between A and B. Since I don’t work in the marketing department, I thought I’d give my own explanation and fill in the details between … ahem … the gap.

But let’s stop fooling around and be serious. I think the idea that a USB cable makes a difference is offensive to some people. And nowadays, most USB DAC supposedly work in asynchronous mode which means the input USB digital stream and the output DAC stage are completely separate in logics. The digital stream from the PC are buffered using the USB clock. The clock used to clock the data into the DAC stage is a completely separate clock as well so there is no possibility of any interaction between the two. The beauty of USB is that even if there is jitter on the incoming USB data stream, logically it does not have any affect whatsoever on the DAC stage. So the DAC should theoretically have ZERO jitter. But why then some very reputable people claim that you can hear differences with different usb cables?
Notice that in the above paragraph, I kept using the word “logically” and intentionally avoid the word “electrically”. Even though in principle, the USB and the DAC stage are logically isolated, I am not too sure that they are “Electrically” isolated and it has to do with the “fluid” I mentioned above. And because of this fluid, once it flows, theoretically if something happens at one corner of the pc board, can affect something at a different corner of the pc board.

There are three possible specific areas in which A can affect B:
1. Power rail
2. Ground
3. Jitter induced noise : that is usb jitter of the incoming data induces noise.
4. Data rise time.

And those are three specific areas are what filled the “gap” between A and B. Let’s talk specifically about each 1, 2, and 3. Actually I can lump 1&2 together since they sort of go together. 3 can be a bit trickier since as jitter is never an easy thing and can be expensive to tame.

1&2: Power and ground coupling are easy to understand. You can design your USB DAC in such a way that can minimize the noise from the computer but it’s never going to be completely zero. There will be some amount of coupling. Theoretically Electrical field falls off exponentially and although it’s approaching zero, it won’t be zero. For example, Exp-33 is very small but not zero. It will be small and hopefully small enough that it will be below your DAC intrinsic noise level that it may not matter. Different usb cable construction will couple the power and noise differently so at least in this respect, the cable will have an affect on the DAC power and ground ultimately the electrical sound wave of the DAC. There is one way in which you can eliminate power noise, at least for the most part. Ayre QB-9 DSD has it’s own usb power so that the computer usb power rail will not affect the DAC power. But still the ground has to be shared so you still have the ground to worry about. But then you ask, why not optically isolate the usb data stream so neither the power and the ground will be shared. Yes, optical isolation will greatly minimize power and ground noise coupling but there is still a back door where the computer power and ground noise can sneak into your DAC. Basically you will need an IC to convert the electrical to optical, and if this IC is located on your DAC, it will need power and ground from your DAC and this is the backdoor where noise could be couple. A better solution is to place the IC on the computer side but I don’t know of any optical USB, and it may not possible since the USB format may not lend itself very well to optical signal so you can see at the end of the day, it may be better just to deal with power and ground noise directly instead of beating around the bush.

Number 3: is a little difficult to explain as it should be. How can jitter on the usb data can induce noise on the DAC? It comes down to the implementation. In order to accept the usb data, you need something called a “CDR” – clock data recovery. When you have more jitter, the cdr has to work harder. When it works harder, it takes more current from the power grail and put more stress on the ground as well and when the power/ground is stressed, you have noise. And although I am sure the DAC designer did his job, he should not be naive to think he has completely isolate the power/ground of the CDR to the power/ground of the DAC. And if the DAC power/ground has noise, equivalently it will leads to jitter of the DAC stage as well. So you can see, there is really no “complete” isolation. Logically yes, but electrically no.
But it potentially can get worse. In electrical engineer, there is something called “data dependent jitter”. That is the jitter signature depends on the nature of the usb data which means the jitter is deterministic and not random and I won’t go into detail but this type of jitter since it is not random, it has a certain periodic frequency. As a DAC designer, if you have noise, you want the noise to be as random as possible because if the noise is periodic, it will ultimately produce high frequency distortion which result in treble harshness, sounding digital, non-analog and so on which are the things audiophiles hate the most. I’ve read somewhere that some DAC are designed with an intentional small amount of random noise in it. This is also the motivation of the famous dCS ring-dac design. I don’t go into detail but periodic noise is very BAD for audio. But let’s get back to usb data. The thing is usb or any data format will have very specific convention and rules such as pre-ample, post-ample, error correction, packet size, and a lot of these things are periodic, predictable and not random (otherwise usb wouldn’t work). They are periodic so the coupling noise will be periodic as well. So you can see how this can be very bad.

Number 4: rise time. This is related to Number 3. When you design a cable, you can affect the rise time of the signal. A fast rise time can generate more noise and more coupling. And I have explained about how noise on the usb data can affect the DAC stage jitter.

Anyway, I thought I started a discussion.
 
The only real question here is what difference do those "experts" claim to hear? Is it noise or are they going on about how this cable is more open sounding or that cable has tighter bass? Cable induced noise is a real possibility with any copper interconnect but I'd wager any cable that does add noise is either substandard or defective. You won't find a scientific explanation for how a cable alters the sound character of a digital data stream because data streams don't have a sound character, that doesn't even exist until the data is converted back into an analog signal.
 
Here is the question asked by magazine HiFi+ to some of the leading industrial experts in cable design. I posted here so as FYI. Although I do admire their cables, I thought they could have done a better job articulating thing. Some of the response are down right condescending.

How do you answer those who argue that specialised digital cables don’t and can’t make a difference since digital audio is ‘all ones and zeros’ anyway?

“(Laughs) No argument there—you would
not be talking to an audiophile. The fact is
that digital audio is ‘all ones zeros’ and time.
In digital cables, clocking and time arrival
integrity is crucial to musicality. When the
timing aspects of digital cables are right,
the sonic differences should be (and are)
obvious, so if you can’t hear a difference, don’t buy the cable.” George Cardas of Cardas Audio

“First, there is no music until something
vibrates, electrically or acoustically. Second,
in terms of cables, a digital cable “pulses” off
and on quickly. If the cable does not vibrate
in sync with the music signal, there will be errors. “ Bruce Brisson of MIT


“I don’t argue this point. All cables affect
the final sound of a system so if they don’t
understand this, it’s obvious they have not
listened or can’t overcome their negative
expectation bias. Amusingly there is no
such thing as a “one” or a “zero” where
digital transfer through a cable is concerned.
There is only an analogue signal, which
is “interpreted” as a one or a zero, so
there really is no difference between why
analogue cables effect sound vs digital
cables effecting sound. All cables affect sound in a system.” Ted Denney
of Synergistic Research

“Sit down and shut up and have a listen! Any digital cable
(with the exception of optical cables) in a
hi-fi system is carrying an electrical signal.
That signal is as subject to interference and
mechanical noise as any analogue cable is.
We’ve already mentioned how critical we
think the relationship between conductor
and dielectric material is and if anything, this
is more so when it comes to digital cables.
For the same reason, the quality of shielding
and the ability of that shielding to work
at high frequencies are also critical. One
of the areas of musical performance that
digital cables seem to have a profound effect
on is the coherence with which music is
presented. Producing cables that are able to
carry a musical signal coherently is, we think,
fundamental to every cable design.” Nigel Finn of
The Chord Company
“It might start out as “all ones and zeros”, but
the signal from the output must change from
one to the other and the input device has
to determine at which point a one becomes
a zero and vice-versa. If a cable causes
the voltage rise time to lag or linger it will cause problems.” Ray Kimber of Kimber Kable


“Digital signal is ones and zeros, a lot of them!
However, those who argue against the use
of specialized digital cables overlook one
exceedingly important factor which is the
“timely conveyance” of all of those ones
and zeros. In some digital components the
lack of timely conveyance of digital signal
is referred to as jitter. The electric current
in all cables, by which digital information
is conveyed, is impacted by conductivity,
dielectric absorption, impedance, eddy
current resistance and correspondingly
the integrity of the timely conveyance of
the massive amount of digital information
is also altered. At the end of the day, I use
the most sophisticated and precise test
instrumentation in the world, the human
hearing system. And, by that standard alone
it is clear that different digital cables have a big impact on the sound.” John McDonald of Audience


“The digital signals that are used in our audio
equipment are not dissimilar to an analogue
signal, only that they are constant amplitude
but still constitute numerous frequencies.
Exact representation of these frequencies in
relation to triggering and lack of unwanted
interferences influences the end result—the
music that we hear. It is really the DAC that
interprets the signal as ‘zeros & ones’. The
signal itself is just made up of waves, albeit
square waves.” Touraj Moghaddam
of Vertere Acoustics


“We are confident that we can demonstrate
the positive differences that come through
using our digital cables in music systems,
making for a far more enjoyable listening experience.” Graham Nalty of Black Rhodium

“We the only specialized manufacturer
of HMDI and 4K UHD cables in North
America, Nordost has put a lot of R&D into
digital transmission. We have found that
when digital cables are being used, it isn’t
numerical 1s and 0s, but analog square
waves that represent 1s and 0s that are
actually being transferred. The “less square”
a wave is, the harder it is for a DAC to make
a distinction between the so-called 1s and
0s, which results in timing errors. In order to
achieve sharp analog square waves, the cable
must have a large bandwidth and be capable
of high transmission speeds. Therefore,
quality cable design makes a tremendous
audible impact in digital cables.” Joe Reynolds of Nordost

“I explain that those ones and zeros pass
through cables as streamed waveforms with
jitter that creates measurable data errors
that change the sound we hear. I also offer
to send them white papers published by test
instrument manufacturers on that very issue.” David Salz of Wireworld

“They’re correct in that the ones and zeros in
the data remain ones and zeros through the
connecting cable. Where it all goes to hell is
when it’s converted to analog at the other
end, and that is where the connecting cable
and equipment interactions wreak havoc
with the clarity of the sound.
A digital source outputs data, and noise.
The quality of the cable determines both
the integrity of the data and how noise
travels. Today’s digital gear should be looked
upon as an RF system and while designed
to play through faults, the quality of the
analog output, particularly the very low-level
signal portions that brings about a sense of
space and other life-like qualities, is highly
dependent on the care taken in the transport
of the digital signal. Resolution is not black
and white, but rather shades of grey.” Joe Skubinski of JPS Labs

“The zeros and ones are not the problem. It
is about jitter and noise causing a different
sound. Jitter causes timing errors, easily
heard as an unpleasant sonic effect.” Edwin and Gabi Rijnveld
of Siltech and Crystal Cable
 
Can we pretty please not have a copypasta manufactroversy? It's pretty easy to see from a neatly EE mindset that a cable that affects the transmission of the bits OR changes the power delivery of a load is either:

1.) A defective cable. That would include it behaving like an antenna.
2.) A cable that is asked to behave outside of specifications due to the equipment surrounding the cable. Fix the equipment.
 
Just recently someone from another forum encountered common-mode noise caused by a poor-quality USB cable which was fixed by a better quality one.

But of course I would take this kind of anecdotes as 50% trustable at most due to how many such users (of other forums) eventually coming back to seek solutions for the same problem that became unsolved despite previously being thought of as solved because bad @ controlled testing.
 
IME different 'off the shelf' USB cables most certainly can alter system sound, subtly but evident nevertheless.
'Theoretically', fitting a ferrite filter around USB cable should cause zero change in system sound, in practice I find this is not true.
Fitting a non magnetic filter (propriety) around USB cable should cause zero change in system sound, in practice I also find this is not true.

Dan.
 
I´m using USB for military custom development.
I´ve used COTS cable at the beginning with very bad results although cable marking High Speed USB (cheap or expensive ones, don´t care).

After a lot of expertising with a lot of measurements (LeCroy with fixture according USB standard, eye diagram...) we state that this is like russian roulette.
So we decide to use only custom made and qualified cable from Draka.
We don´t need any ferrite on cable, but the interface is ESD protected and CMC EMI filtered on PCB.
This is a technical issue.

My 2 cents.
JP
 
IME different 'off the shelf' USB cables most certainly can alter system sound, subtly but evident nevertheless.
'Theoretically', fitting a ferrite filter around USB cable should cause zero change in system sound, in practice I find this is not true.
Fitting a non magnetic filter (propriety) around USB cable should cause zero change in system sound, in practice I also find this is not true.

Dan.



If there is a difference then the USB cable - which transmits a digital signal - is broken, because it's not transferring the correct bits.

Either it transfers all bits correctly, or it doesn't thus it's broken.
It's really not polishing up the ones while they're being transferred.

Even if you think it must be jitter, no really, it doesn't matter.
Archimago's Musings: MEASUREMENTS: USB Cables for Audio DACs. [2013-06-18 UPDATE]
 
USB is an asynchronous transmitting method, where clock is retrived from the data stream by using a PLL at the receiving end. The data stream can be characterized by an eye pattern, similar to the CD transport. The USB cable clearly affects the shape of the signal at the receive end, there are methods for physical layer measurements, like this:

Understanding and Performing USB 2.0 Physical Layer Testing

The USB cable affects signal shape, that has influence on jitter. In digital domain, jitter is bad for audio :mad: No wonder there is difference between sonic characteristics of USB cables (and Toslink, and SPDIF and all similar async digital audio tranfer).
 
I´m using USB for military custom development.
I´ve used COTS cable at the beginning with very bad results although cable marking High Speed USB (cheap or expensive ones, don´t care).
After a lot of expertising with a lot of measurements (LeCroy with fixture according USB standard, eye diagram...) we state that this is like russian roulette.
Like I said.

So we decide to use only custom made and qualified cable from Draka.
We don´t need any ferrite on cable, but the interface is ESD protected and CMC EMI filtered on PCB.
This is a technical issue.
Thanks JP.
Can you give details of the Draka cable type that you are now using ?.

Dan.
 
USB is an asynchronous transmitting method, where clock is retrived from the data stream by using a PLL at the receiving end. The data stream can be characterized by an eye pattern, similar to the CD transport. The USB cable clearly affects the shape of the signal at the receive end, there are methods for physical layer measurements, like this:

Understanding and Performing USB 2.0 Physical Layer Testing

The USB cable affects signal shape, that has influence on jitter. In digital domain, jitter is bad for audio :mad: No wonder there is difference between sonic characteristics of USB cables (and Toslink, and SPDIF and all similar async digital audio tranfer).
I think you're confusing adaptive and asynchronous protocols. What you describe is true for adaptive endpoints (the pcm2*** serie from TI for example). Definitely not for the more modern interfaces like xmos, which work as asynchronous endpoints.

There are 3 modes for audio transfer under USB audio. All 3 are falling in the isochronous category (real time with no error correction).

From the USB Device Class Definition for Audio Devices:

3.3 Audio Synchronization Types

Each isochronous audio endpoint used in an AudioStreaming interface belongs to a synchronization type as defined in Section 5 of the USB Specification. The following sections briefly describe the possible synchronization types.

3.3.1 Asynchronous
Asynchronous isochronous audio endpoints produce or consume data at a rate that is locked either to a clock external to the USB or to a free-running internal clock. These endpoints cannot be synchronized to a start of frame (SOF) or to any other clock in the USB domain.

3.3.2 Synchronous
The clock system of synchronous isochronous audio endpoints can be controlled externally through SOF synchronization. Such an endpoint must do one of the following:
• Slave its sample clock to the 1ms SOF tick.
• Control the rate of USB SOF generation so that its data rate becomes automatically locked to SOF.

3.3.3 Adaptive
Adaptive isochronous audio endpoints are able to source or sink data at any rate within their operating range. This implies that these endpoints must run an internal process that allows them to match their natural data rate to the data rate that is imposed at their interface

For more details on the synchronization methods, see section "5.10.4.1
Synchronization Type" from the Universal Serial Bus Specification Revision 1.1.
 
Copper quality could make a difference to the signal. A piece of wire isn't just a piece of wire, the resistance varies a lot between different wires.

Also excessive length of cable will make a difference.
You should be using the shortest length of cable as possible.

The number of items on the same bus will affect data quality too as each loads the bus a little more.
 
Status
Not open for further replies.