A USB Cable Test Designed to Satisfy Skeptics AND Believers—Feedback Welcome

[antcollinet]

the current evidence still leaves the debate open.

It really doesn't.

 

[antcollinet]

Edit: The problem with audio analysis of the DAC output using an audio analyzer is that the analyzers themselves are extremely shielded and resilient against RFI/EMI interference.

Doesn't matter. If there is noise resulting from the cable it will be present on the output of the DAC, and will be measured by the analyser.

 

[Guestuser]

I can't speak for anyone else, but I am not opposed to the OP carrying out these tests. The tests have no value to me, and won't persuade the true audiophoolery die-hards, of course, and I'd rather see all this time and effort (and money) put to study something that isn't already scientifically "settled," but if the OP wants to do this study its their time and effort and money wasted not mine.

I don't think that's what's happening here, though. Correct me if I'm wrong, but this looks like an experiment design with no provision (or budget) for actual execution.

The point of this isn’t just about USB cables, it’s also about promoting open research practices and addressing the issues of previous research. If those are wasted efforts, I’m at peace with that.

As for the conduct and budget pieces, the protocol is in a public comment phase. Once feedback is closed, the exact budget and timeline can be determined. I’m less worried about funding than getting people to actually read the document and provide specific suggestions with the experimental design.

Not saying you fall into this category, but so far on the two forums it seems most people who share their strong opinions aren’t willing to read other people’s forum posts much less an actual protocol.

 

[mcochris]

If you can't measure a difference, you certainly can't hear one either.

That sounds over-confident to me. I hear my tinnitus, but can be measured? There may be things we're not measuring that are perceived.

 

[solderdude]

The volume is matched exactly for each test.

Not needed at all when the only thing that is changed is the cable simply because the volume is NOT altered in any way (it is not possible).
Of course when the comparison is between 2 different DACs then they need to be matched withing 0.1% (0.1dB).

You can even do the same test with analog interlinks (and analog signals) b.t.w. and power cables, power conditioners etc.
That's the fun part of cable testing (except for speaker cables).
The volume does not change so no need for matching cables.

The person collecting the ratings doesn't know which cable is being used. Some tests include 'fake' switches to check for bias. Finally, the cables are swapped between the DACs and tested again to make sure the results are accurate.

Good idea ... Do it enough times (say ... 20+) and you'll score close to 50%.

Note: on SPDIF (especially optical) one has to depend on the optical receiver and locking of the DAC's internal receiver.
The latter is true for coax SPDIF as well.

 

[Svend P]

That sounds over-confident to me. I hear my tinnitus, but can be measured? There may be things we're not measuring that are perceived.

Master Thread: Are Measurements Everything or Nothing?

Seems like so many review threads get challenged with: 1. Measurements are not everything. 2. You all never listen. 3. I trust my ears, not graphs. 4. I don't listen to graphs. I listen to music. 5. You all must not listen to music at all. 6. Why don't you all buy the best SINAD gear? 7...

www.audiosciencereview.com

 

[wwenze]

Some people believe USB cables transmit only binary digital data (0s and 1s), and others assert that USB cables can still affect the output stage of a DAC (antenna effect, etc.). This test is designed to control for both scenarios by using shielded and unshielded cables, as well as two different DACs: one bus-powered DAC with no galvanic or chassis isolation, and another DAC with a separate power supply, galvanic isolation, and chassis isolation. If there is no effect with any combination of cables and DACs, the "digital only" part of the debate will now have data to support or disprove it.

Edit: The problem with audio analysis of the DAC output using an audio analyzer is that the analyzers themselves are extremely shielded and resilient against RFI/EMI interference. The objections I've seen are that not all DACs have this type of shielding, and therefore, the antenna effect could affect the DAC output stage regardless of whether the data stream is stable. This protocol directly addresses that objection by including a bus-powered DAC with no shielding.

With any experiment, you have to know what you are going to expect with the result and what that means

If I replaced the word "USB cables" with "amplifiers", in a properly executed test the expected result is still that, people cannot tell the difference between two amplifiers, even tho measurements can.

So what did that prove? Does it prove there is no difference, or people cannot hear the difference? Or in the words of the audiophile, "the system is not resolving enough to show the difference"?

Response to the Edit: Basically, this can be grouped under common-mode issues. (Wireless signals are referenced to earth and thus also common-mode by technicality)

We know that dongles are bad at it than big standalone DACs
We know that AP analyzer is very good at rejecting common-mode issues
We know that an SPDIF DAC on an unisolated signal has measurable noise worse than one on an isolated signal

The problem with common-mode is, even tho we know its effect exist, the amount of effect is different between different setups, different products, different houses. And there is no established standard for quantifying the performance. Let's say something has a common-mode rejection ratio of -100dB vs another product with -50dB, what does that mean? Will the -50dB be immediately failed? What if the user setup has no common-mode issue?

We have a similar situation with crosstalk - Some products measure better than others. But there is no established standard as to what is the minimum score required.

 

[RexrothPigeon]

Not saying you fall into this category, but so far on the two forums it seems most people who share their strong opinions aren’t willing to read forum posts much less an actual protocol.

This is fair, I think, for those who, like me, believe this is completely settled science

 

[amirm]

Complete Protocol

Man, you put incredible amount of work in this! Much appreciated.

Alas, there are some technical issues.

1. Maintaining sync. Due to the way async USB works, two identical DACs connected to the computer will drift apart as their internal clocks are not identical. You could use the rare DAC that has external clock sync but then this introduces another variable.

2. Relay clicks can be audible by themselves. But even if you hide this, then there may be glitches as you switch waveforms. If there is indeed differences in the two systems, this will create noise, providing a "tell" to the listener that a switch occured.

3. You would need special software to play two USB DACs at the same time and at least attempt to keep them in sync. Roon does this but many do not.

 

[amirm]

The point of this isn’t just about USB cables, it’s also about promoting open research practices and addressing the issues of previous research

The question is who needs the research. As you can imagine, our camp doesn't need it.

The camp that needs it doesn't believe in any controlled testing. You mention any protocol and the game is over.

Let me give you some evidence of how hard it is to convince the diehards in the other camp. A local audiophile has a system that by now is clocking close to a million dollars. He believes in every and all audio tweaks, no matter how remote. He once said not only did new outlets improve the sound of his systems, but actually changing the covers on the outlets helped remove another veil!

About 10 to 15 years ago, the above audiophile agreed to participate in a test of his MIT cables against another cheap cable, in his room and his system. The objectivist who conducted test performed it blind. After a few trials, the above audiophile agreed that he was not able to detect his expensive cables in blind tests even though he was so sure he could. This was all published in another forum and said audiophile said he had some reflections to do.

Well, he reflected alright. After a short while, he simply went back to his old ways, throwing everything at his system, including said outlet upgrades. If you as much as mention the word "measurement," he will throw you out of the room. "His ears tell the truth" and that is all there is to it.

Now, the above used to be limited to uber audiophiles. But in the last decade, it spread to new and everyday audiophiles. They read the forums and traditional rags and started to believe that you must spend hundreds of not thousands of dollars on audio tweaks. That is the bad news. The good news is that our work, through explanation of technology and measurements to back it, has made an incredible dent in this. Many were skeptical of audiophile claims but had read nothing to the contrary. Now they have.

You could argue that your testing will help this new audience and it probably can. But whatever the test is, it needs to be dead simple like what the above objectivist did with the cable swap. Any story more complicated than that will get lost. To wit, Ethan Winer did null hardware test of cables. Problem is that even in a good null, there is still noise. That was enough for the test to be dismissed.

So while I very much applaud your efforts, as a practical matter, I don't think it is worth the trouble.

 

[Sir Sanders Zingmore]

Could someone still dismiss results they don’t like? Sure. But they’d have to argue the blinding failed, the placebo controls don’t count, and the validation was a fluke. All of that while the analysis plan is open to public comment before data collection.

Have you read the countless arguments that happen on just about every hifi forum in the world. That’s exactly what they’ll argue. And then some.
If your aim is to develop a good protocol then I applaud you. If your aim is to change anyone’s mind then I very much doubt you’ll have any success.

 

[Geert]

Could someone still dismiss results they don’t like? Sure. But they’d have to argue the blinding failed,

Blind tests don’t work, that’s a fact in the audiophile community. Stress, unfamliar hifi system, need to listen for hours, alpha brain waves not kicking in, and so on.

I also notice that the equipment you listed in your document won’t get an audiophile stamp of approval. Genelec, Benchmark, Sennheiser ... all soulless unmusical lab equipment. To get the attention of audiophiles you at least need to use a $1M exotic Hifi system, assembled by a connoisseur who understands the synergy between components and who can get the most out of the system via appropriate cable lifters and crawling around over the floor. We don't have that expertise over here.

And lastly, the outcome of your test will only be valid for the DAC you tested with. ”With the DAC I have at home the difference is clear, I heard it”.

Unfortunately, I’m not being cynical. Please spend your time on your medical research.

 

[solderdude]

Still always good to do well performed (listening) tests.
If only for another datapoint.
When the goal is to do truth finding that is.


For the vast majority of audiophiles the truth is what they hear and 'measurements' are just for plot gazing nerds that never listen.
You can't convince people otherwise if they are already convinced. Regardless how facts stare them in the face.
Their personal experience, situation, gear, 'test methods' always differs from the performed test so they will ignore the results based on their findings and will have tons of 'reasons' why this is so.

 

[olieb]

Randomized controlled trials are considered the "gold standard" by doctors and regulatory agencies worldwide, and the level of rigor applied to human clinical trials is among the highest in any industry. When you want to put a new drug in a human for the first time, you don't have room for errors. This is the level of control that comes to my mind when designing a debate-ending USB cable test.

These kind of trials did neither end the discussion about globuli nor about horoscopes. Even the shape of the earth is in issue for quite some people and I could go on with this list. So I would not expect discussions over cables to end any time soon.
The core of the problem is that no scientific test/experiment can convince people who deliberately choose to ignore science.
Best idea is to just not participate and instead do something useful.

 

[somebodyelse]

3. You would need special software to play two USB DACs at the same time and at least attempt to keep them in sync. Roon does this but many do not.

Keeping them in sync without a common external clock would require ASRC which introduces another potentially audible variable. You could do that with a Mac, or with linux using a sufficiently recent version of PipeWire - I don't remember the version number.

If the test sections are short enough for drift not to be audible you can use a composite device that doesn't resample. How short will depend on how far apart the device clocks are, so would have to be determined for the hardware being used.
https://www.alsa-project.org/main/index.php/Asoundrc#Virtual_multi_channel_devices
I think it can also be done in PipeWire by setting the same clock.name property for both device nodes to disable the ASRC:
https://docs.pipewire.org/page_man_pipewire-props_7.html#props__scheduling_properties

 

[danadam]

Unlike other USB protocols, Asynchronous Isochronous USB (UAC2) does not have any error detection and correction built in.

It does have error detection. It doesn't have error correction.

 

[AudiOhm]

Wireshark anyone?
Nothing like being able to test a cable before it hits a DAC or any other audio device...

Ohms

 

[Guestuser]

Thanks for all the great feedback and suggestions. I genuinely appreciate your candid responses. I took some time to reflect on your comments and reassess my goals for this effort. Even though I agree that some folks will remain entrenched no matter what evidence is presented, I do still think there are opportunities to re-frame the discussion for those reasonable folks who have their opinions but are open to new information. Along those lines, I've been doing two follow-up activities since posting the protocol.

First, I am working on a redesigned protocol to make it easier to conduct. For example, Amir highlighted the challenges of synchronizing multiple DACs and eliminating relay clicks. Totally valid issues, and I think there are ways to reduce both of those by changing the study design. Still working that out...

Also, I spent some time researching the technical aspects of USB audio to inform my understanding further. I began by researching the potential issues from using a USB cable for audio playback. I then looked at which DAC design features and USB cable design features can be used to mitigate each of those issues. Finally, for each DAC design mitigation, I tried to find at least one example of DAC with that feature.

Important caveat: Just because something is technically measurable doesn't mean it's audible. And just because a DAC lacks a specific mitigation doesn't mean it will have problems - good overall design can compensate. This is meant as a technical reference, not a buying guide.

Just sharing my research notes in case it's interesting to anyone, and also just wanted to say thanks again for all your comments and feedback. I'm not an expert in these topics, so I'm happy to take feedback on missing or inaccurate information.

Issue from USB audio​	Mechanism​	DAC Design Mitigations​	Mitigated by Omitting USB Power Wire?​	Mitigated by USB Cable Shielding?​	Mitigation by USB wire thickness / composition?​	Mitigation by cable geometry?​
Ground loop noise	Multiple ground paths between computer and DAC create current flow and voltage differences	USB isolation barrier (galvanic or equivalent) Ex: Schiit Yggdrasil, Auralic VEGA G2.1 Floating ground design Ex: Naim DAC-V1 Single-point/star grounding topology Ex: Naim DAC-V1, Song DMP-Z1 Ground lift switches Ex: Cambridge DacMagic 200M	Partial - can reduce loop current associated with VBUS power draw (if any), but still depends on DAC grounding design	Minimal - may reduce capacitive coupling of AC mains noise onto the ground conductor, but doesn’t address fundamental multiple-path problem	Low - Thicker ground wire helps a little, material composition negligible impact	Minimal - geometry doesn’t break ground loop path, but tighter twist of VBUS/GND pair slightly reduces internal loop area (external system loop dominant)
Power supply noise injection	Computer's noisy 5V supply couples into DAC ground or power rails	Don't use USB power for audio circuits (separate PSU) Ex: any DAC with a power supply Heavy linear regulation if USB-powered Ex: iFi Audio micro iDAC2 Separate power domains for USB vs. analog Ex: TEAC UD-701N Low-dropout regulators with high PSRR Ex: Benchmark DAC3 Pi-filters and LC filtering stages Ex: Volumio Primo Plus, Benchmark DAC3	Partial - kills VBUS-borne coupling, but noise can still couple via GND/shield and data pair (common-mode)	Minimal - shielding doesn’t prevent conducted noise on VBUS, but can reduce external EMI adding to power line noise	Low - minor benefit if DAC is USB-powered and under-regulated; otherwise negligible	Low - Geometry helps somewhat by reducing coupling paths
Power wire radiating noise	5V wire acts as antenna transmitting computer PSU noise	Not using USB 5V at all Ex: any DAC with a power supply Heavy filtering if 5V is used Ex: iFi Audio micro iDAC2 Shielded power input Ex: Chord Hugo 2, RME ADI-2 DAC FS Isolation from sensitive circuits Ex: PS Audio DirectStream MK2, Holo Audio May (Level 3)	Partial - If the wire doesn't exist, it can't radiate	Moderate to High - grounded shielding (foil, braided, triple) can significantly reduce radiation from the power wire	Low - wire properties don’t significantly affect radiation	Moderate - twisted power pair significantly reduces radiation
EMI/RFI pickup by cable	Cable acts as antenna, picking up radio frequency interference	Shielded USB receiver section on PCB Ex: RME ADI-2 DAC FS, Auralic Vega G2.1 EMI filters at USB input Ex: Audio Amari, Chord DAVE Ferrite beads on USB lines Ex: Grace Design m900, Focusrite Scarlett Series Metal enclosure/Faraday cage Ex: Schitt Yggdrasil, Auralic VEGA G2.1 PCB ground plane design Ex: Benchmark DAC3, Schitt Audio Hel 2	Partial - Fewer conductors means less antenna surface area, but data lines can still pick up RFI	High - primary benefit of shielding is blocking external RF and preventing cross-talk between pairs	Minimal - skin effect is negligible, shielding vast more important	High - twisted pair geometry is critical for EMI rejection
Capacitive coupling between wires	Power/ground wires couple noise into data lines within cable	Input filtering to block high-frequency coupled noise Ex: Chord Hugo TT2 Common-mode chokes Ex: Benchmark DAC2, Luxman DA-07X Differential input protection Ex: Benchmark DAC3 B, RME ADI-2 DAC FS Good CMRR/differential PHY + proper EMI components/ layout near the connector Ex: Benchmark DAC3 B, RME ADI-2 DAC FS	Partial - Eliminates coupling from power wire to data lines, but ground wire remains and data lines can still couple to each other	Moderate - individual pair shielding can reduce cross-talk between pairs; outer cable shielding has minimal effect between wires	Minimal - wire guage/material is negligible except at extremes	High - Geometry (twisted pairs, coaxial) highly effective at reducing capacitive coupling
Common-mode noise on data lines	Noise appears equally on both D+ and D- lines	High common-mode rejection ratio (CMRR) in USB receiver Ex: Benchmark DAC3 B, RME ADI-2 DAC FS Differential line receivers Ex: RME ADI-2 DAC FS, Aesthetix Pandora Common-mode chokes/filters Ex: Benchmark DAC2, Luxman DA-07X Balanced input stages Ex: Benchmark DAC3 B, RME ADI-2 DAC FS	NO - Common-mode noise on data lines unaffected by power wire presence	Moderate to High - shielding helps significantly if properly grounded	Minimal - no significant effect	High - twisted pair geometry is a primary defense against common-mode noise
Differential-mode noise	Noise appears differently on D+ vs D- (actual signal corruption)	Signal integrity by spec-compliant PHY+layout/ESD/EMI components Ex: Benchmark DAC3 B, RME ADI-2 DAC FS CRC detects errors Ex: Benchmark DAC3 B, RME ADI-2 DAC FS Recovery differs by transfer type (bulk can retry; isochronous generally cannot)	NO - Data line signal integrity independent of power wire	Moderate - shielding helps somewhat by reducing external sources, but pair quality matters more	Minimal - pair symmetry matters but otherwise guage/material are non-issues	High - geometry and pair symmetry critical for maintaining pair balance and impedance
Clock jitter from computer	Computer's USB clock has phase noise/timing variations	Asynchronous USB mode (DAC controls timing) Ex: Benchmark DAC3 B, RME ADI-2 DAC FS High-quality local oscillator/clock Ex: Benchmark DAC3 B, RME ADI-2 DAC FS, Chord Hugo 2 Phase-locked loops (PLLs) with jitter reduction Ex: Gustard DAC-X26, Benchmark DAC3 B, RME ADI-2 DAC FS Large FIFO buffers Ex: Auralic Vega G2.1, Chord DAVE, SMSL VMV D2R Multiple stages of re-clocking Ex: Ideon Audio Absolute Epsilon, RME ADI-2 DAC FS, iFi NEO iDSD2 Low-jitter crystal oscillators (TCXO/OCXO) Ex: SMSL VMV D2R, Gustard DAC-X26 Pro	NO - Power wire irrelevant; host packet timing jitter is largely decoupled in async designs; synchronous/adaptive modes can be more hosting-timing dependent	None - shielding is irrelevant to host clock jitter	None - no effect	None - geometry doesn’t affect source jitter, though extreme geometry problems can degrade signal integrity and indirect affect receiver timing
Impedance mismatches	Cable impedance variations cause reflections and signal integrity issues	Adaptive equalization (USB 3.x but not USB 2.x) Ex: iFi NEO iDSD2 Wide receiver input tolerance Ex: Benchmark DAC3 B, RME ADI-2 DAC FS, iFi Zen DAC 3 Termination resistors Ex: RME ADI-2 DAC FS, Benchmark DAC3 B, iFi NEO iDSD2 On-chip impedance matching Ex: RME ADI-2 DAC FS, Benchmark DAC3 B, iFi Zen DAC V2	NO - Impedance of data line differential pair unaffected by power wire presence	Minimal - impedance control and shielding are separate features, although better cables often have both	Low - thicker wires have slightly better inductance but wires that meet USB spec are equally good	High - Geometry is dominant factor in impedance control
Data timing errors	Signal edge degradation, inter-symbol interference	Error detection and recovery Ex: Auralic Vega G2.1, Benchmark DAC3 B, RME ADI-2 DAC FS Adaptive sampling Ex: Benchmark DAC3 B, RME ADI-2 DAC FS Eye pattern monitoring (in sophisticated designs) Ex: RME ADI-2 DAC FS, Eversolo DAC-Z10	NO - Data timing determined by data lines only	Low - shielding helps prevent edge degradation but cable quality and length matter more	Low - within USB 2.0 length limits, negligible effect	Moderate - geometry may affect timing through propagation and skew, but well-twisted pairs maintain equal path lengths
Ground bounce	Rapid current changes in ground conductor cause voltage spikes	Separate analog and digital grounds Ex: RME ADI-2 DAC FS, Benchmark DAC3 B Star grounding Ex: PS Audio Stellar, DirectStream, Benchmark DAC3 B Decoupling capacitors Ex: RME ADI-2 DAC FS, Chord Mojo 2, Benchmark DAC3 B Low-impedance ground planes Ex: Topping D90 III, RME ADI-2 DAC FS, Benchmark DAC3 B	NO - Ground wire typically still present even when power wire omitted	None - shielding doesn’t reduce the inductance of the ground wire	Moderate - thicker wire has lower inductance with less voltage spikes; material choice negligible	Low - Twisted pairs measurable reduce ground bounce, but may not be audible
Crosstalk between USB and analog	Digital noise couples into sensitive analog audio stages	Physical separation on PCB Ex: SMSL VMV D2R, RME ADI-2 DAC FS, Benchmark DAC3 B Separate shielded compartments Ex: MSB Technology Discrete DAC, Pro-Ject DAC Box E Digital and analog ground isolation Ex: MSB Technology Discrete DAC, RME ADI-2 DAC FS, Benchmark DAC3 B Careful PCB routing and layer stackup Ex: WiiM Ultra, Benchmark DAC3 B, RME ADI-2 DAC FS	NO - Internal DAC design issue, not affected by external cable configuration	None - Not applicable	None - not applicable	None - internal DAC issue, cable geometry irrelevant

 

[antcollinet]

That sounds over-confident to me. I hear my tinnitus, but can be measured? There may be things we're not measuring that are perceived.

Your comment sort of proves the point. I'm sure you agree that the sound you "hear" coming from your tinnitus is not coming from any sound reaching your ears.

Similarly even when there is no difference in the sound waves coming from two different devices - it is still possible for our brain to cause us to perceive a difference.


So the statement could be clarified as "if you can't measure a difference, then you certainly can't hear one (coming from the sound reaching your ears) either.

If you know what has changed though (This cable or that one, this amp or that one) - it is quite possible for your perceptive biases to create a change in your perception. If that is happening though it has nothing to do with any difference between (in this case) USB cables.

 

[Timcognito]

The point of this isn’t just about USB cables, it’s also about promoting open research practices and addressing the issues of previous research. If those are wasted efforts, I’m at peace with that.

Pick something that is not an established fact and go now in peace on this. There are many tests here at ASR and other websites that show, that copper with all the tweaks, have no audible difference. If it is audible a microphone, far more sensitive than anyone's ears, will pick it up and the resulting data will verify any change. It has been demonstrated that braiding numerous stands help prevent a cable from becoming an antenna.