Saturday, 18 September 2021

MEASUREMENTS: USB Isolation with Nobsound ADuM4160 device and hi-res DAC. The world these days, and the audiophile microcosm.


After writing about very high resolution measurements with the E1DA Cosmos ADC last week and touching on basically "state of the art" fidelity like the Topping D90SE, for this week's post, I thought it would be good to talk about something more "fundamental". Often spoken of among audiophile tweakers but rarely have I seen evidence of actual value/effect for many of these devices.

Before I begin with this post or even show you anything, let me just say that I do not believe audiophiles need USB isolation for use with devices like DACs unless you know you have noise issues. I think we've all see indiscriminate suggestions that audiophiles need all kinds of noise filters and isolators (like this, or this). Sometimes these things cost quite a significant chunk of change compared to the downstream device (like DAC) itself!

Over the years, I have tried out a number of USB DACs with computers and simple streamer devices like the Raspberry Pi. With a normal set-up, I cannot say I have ever heard an issue with any of the various decent DACs I've tried. Likewise, measurements already suggest that in general there are no major noise issues with at least reasonable modern devices.

However, there are times when the system is complex enough where indeed you do see ground loops and noise like the 8kHz USB PHY noise pop through. A couple months back, I showed you some of this when we took the rather useless AudioQuest JitterBug FMJ for a spin in some of these situations! Today, let's look at the effect of that little inexpensive Nobsound ADuM4160 USB Isolator (typically can be purchased for <US$25) you see in the picture above.

This device uses the Analog Devices ADuM4160 "Full/Low Speed USB Digital Isolator", see that linked datasheet for full details. Notice that although this is USB2.0 compatible, speed is limited to 1.5Mbps or 12Mbps which means that if we connect a DAC to this, we are limited to 24/96 stereo playback (USB Audio Class 1 fallback for most modern DACs). Remember this limitation and only proceed if you can accept this for whatever purpose you might have.

As you can see from the images, this device allows you to connect the isolator with a microUSB cable on the right to your computer (USB-A to microUSB). And then on the left side, there's a female USB-A connector where you can plug devices that you want isolated from the computer/source into. There are a couple of small switches to change between Lo-speed (1.5Mbps) and 12Mbps Full-speed, plus one to select whether +5V should come from an external power supply (the side microUSB connector). I'm not sure what the power limit is without an external source, but I have been able to successfully run the Topping D10 Balanced as shown below which seems to need ~200mA.

That clear plastic hood protects from stray fingers while keeping the cost down.

You might be wondering: "So Arch, why do you have such a thing if you don't believe USB DACs need isolation?"

The answer is simple. It's because when I run measurements using a multipurpose computer, I typically have multiple USB devices attached to the computer, so in order to prevent noise from the computer, or ground loops affecting the ADC, or DAC, a device like this should do the job. Let's put that hypothesis quickly to the test with this post.

Let's start by showing the set-up:

Note the Linear Audio AutoRanger not being used.

As you can see, we have my Intel NUC 6i5SYH computer on the left. Here's how it's all connected up:

Obviously, we can take the Nobsound isolator in and out of that chain to see if it makes a difference. In the picture above, we have the Topping D10 Balanced (silver) with balanced cables. Let's see how a simple 1kHz 0dBFS tone measures using this DAC (at 24/96). Let's use ASIO4ALL to play the test tone to the DAC, and capture the data from the RME ADI-2 Pro FS, all through the same computer as shown. Note that I did not need to use a separate USB power supply; the D10 Balanced (and D10s) is able to run with just the current provided through the isolator, more power-hungry devices will need an external power source plugged in through the side microUSB connector:

As you can see, with balanced cabling (in this case I used TRS --> XLR), the isolator made no difference. Balanced devices with their inherently superior noise rejection is designed to function very well even in "noisy" set-ups.

I trust the result above is no surprise! So, let's try the single-ended setup. This time, we'll use the Topping D10s DAC instead:

Voilà! The effect of USB isolation in improving the noise level, but not so much the THD. Notice that the 60Hz mains noise with harmonics have been cleaned up. Also, notice the 8th harmonic is significantly reduced with the isolator - 8kHz is the dreaded USB PHY microframe packet noise and it's good to see this being dealt with. Although this is just a simple THD(+N) test, the improvement in noise can be seen with the other tests like RightMark. For those wondering, no this does not have an impact on jitter performance which is already excellent (see D10 Balanced, and D10s measurements).

Like I said at the start, if I were just playing audio on my sound system from a streamer to DAC, I have not found a use for a USB isolator even if I'm using RCA outs. No concern with XLR/balanced outputs. It's only when things get a bit complicated such as a measurement chain with both the DAC and ADC attached to the same computer or USB hub that these noise and ground loops seep into the system.

I think most of the time when audiophiles buy these kinds of products, they're really not improving anything. This is why it seems only to be the "subjective" guys possibly with devices sent to them by the manufacturers making claims that they "hear" lower noise or even "low jitter". As usual, this isn't helpful to consumers and one could even say it's dishonest since such "reviews" are basically just advertisements.

As you can see, the <US$25 Nobsound ADuM4160 USB Isolator works as advertised. The catch of course is the 12Mbps speed limit. For my measurement purposes where I might have a system like this set-up with a DAC as signal generator for amplifier tests, the 24/96 limit is just fine and the Topping D10 Balanced and D10s would provide clean enough signals. Maybe one of these days, a low priced high-speed USB2.0 or USB 3 device with a similar level of performance might come along and I'll upgrade then.

I think it's important when companies like AudioQuest, iFi, and Intona make their USB filtering/isolation products that they also show us tests demonstrating how the products can improve the function of devices like hi-res DACs if that is the intended use case. Over the years, there have been many of these products that claim to lower noise, but honestly, how often have you seen the reviewers/companies actually provide demonstrations to show the effect? In fact, I was hoping to see something even vaguely similar to the result above when testing the AudioQuest JitterBug FMJ given their advertising of the product. While I wasn't expecting any true isolation with the JitterBug, a little attenuation of noise, especially the 8kHz tone would have been nice!

As consumers, I think it's important that we insist on demonstrations of efficacy especially when things like noise improvements are easy to show! "Testimony" by audio magazines and review sources with no measurements IMO are also usually suspect. Be careful with companies and supposed engineers that use words only whether it's for tweaks, power supplies, or even streaming devices. It's probably best to ignore such products/companies and wait until there is data; honestly folks, there has to be data because these are engineered devices based on science - USB filters and isolators are not works of "art"!

Finally, be careful with supposed customer reviews on places like Amazon as well since many of these could be fake. As usual, plenty of questionable stuff in Cyberspace pertaining to audiophile products with no oversight to compel truth in advertising.


"Mechanized Microcosm" - saw this image while surfing and traced it to this artist, cool stuff!

In other news... Well, actually it seems pretty quiet in audiophile-land out there this week with really nothing all that exciting.

With the pandemic still problematic (although I'm optimistic things are on the mend), governments of the world propping up basically everything and stoking inflation, it looks like the "normal" mechanisms of the industrial supply chain and availability of goods to consumers from smaller companies remain distorted. Obviously, gatherings and reporting at trade shows and local dealer events have been severely limited. 

With inconsistent regional health advisories, worldwide disparities in vaccination rates, and the start of mandated "vaccine passports" whether to dine at the local restaurant or international travel, it seems like we are at the beginning of another phase in this pandemic (IMO, one of recovery and herd immunity in the countries with mass vaccination save for pockets of susceptibility). As we enter the fall and the traditional flu season ahead here in N. America, I'm sure there will still be challenges; hopefully nothing too unexpected like a more pathogenic variant.

Over the years, I have suggested at times that the audiophile hobby is like a microcosm of the real world, slightly ahead of the curve even. Of all the technology-based hobbies, audio "hi-fi" (specifically 2-channel) was the first to have reached maturity, achieving a state where sound quality became "good enough" for its intended entertainment goal - the enjoyment of music. Beyond "good enough" and the associated principle of "diminishing returns" is where opinions fracture between the polarities of those who target ultimate fidelity / accuracy / resolution as the goal, and those who seek purely subjective satisfaction with countless positions in between. "High fidelity" sound does not necessarily mean "better" subjective sound (or vice versa) depending on the system, album, and listener. I try to hit the middle ground on this although I do tend towards "high fidelity" which can be defined objectively across frequency, time, and amplitude domains; useful information for me to gauge my own perception and subjective impressions.

Subjective satisfaction with how music sounds through a piece of equipment or system by definition has to do with one's feelings/emotions. That's great so long as we know where to draw the line when it comes to trying to insist upon others the "truths" of one's opinions. Regardless of how we feel, not every emotion needs to be taken seriously by others. (Those of us who are parents for example, I trust that most of us know when it's time to engage and when to disengage from the tantrums of our kids, right?). The audiophile hardware hobby had been shifting towards excessive subjectivity I suspect since the days of thinking about some "Absolute Sound" in the early '70s and the navel-gazing that comes with such a vague target. Subjectivity is formed as a conjunction of biological sensory stimulation, cognitive awareness and creative imagination - can we with any confidence achieve consensus agreement on this "Absolute Sound" quality?

[Remember, one definition by Harry Pearson back in the day was "the sound of actual acoustic instruments played in a real space" which if you think about it is imprecise, perhaps meaningless, and impossible with 2 speakers. Which instruments? What space? What about very popular electronic genres? Isn't the recording/mastering itself just as much if not more important in achieving this than audiophile hardware? Then of course there's all that stuff about whether what's heard is "as the artist intended" or "true to the actual event" which are impossible to judge.]

Notice that despite the futility of talking about an "Absolute Sound", how it might be perceived or even defined, realize that there is such a thing as "Absolute Accuracy", at least when we think about achieving "perfect" analogue signal reproduction of digital data from our sources whether it be CD, SACD, hi-res PCM/DSD download, MP3 file, or stream. Defined mathematically, verifiable within the error limits of testing techniques.

If we zoom out of the audiophile microcosm and "our" battles, I've always had a sense that this "subjective" mode of audiophilia over the last number of decades is contemporaneous (maybe even a forerunner) to what we see these days with the various news networks and their opinionated pundits offering outrage among a polarized public. Every day for the last few years, it seems like facts, lies, "fake news", unsubstantiated claims, opinions, politics and feelings are colliding in extreme ways, each vying for our mindshare. Communications technologies have democratized whatever message anyone wants to project through social media unfiltered to the masses; deficiencies of evidence notwithstanding. Popularity trumps knowledge, or even common decency as the case may be.

My hope is that humanity does possess an adequate amount of intellect, common sense and nobility to work towards a good collective future (in all arenas, not just with the pandemic of course). I strongly believe that this "better tomorrow" would not be based primarily on opinions of the outspoken, but demands basic truths in what we consume (including in audiophilia of course!).

Not all opinions are equal. Notice how many people speak as if from authority, yet ultimately when pressed, have nothing to show (like this guy). At its core, I think education - especially the importance of learning to discern, through the use of critical thinking - is the key. In a complex world, it's impossible to imagine true wisdom to be bestowed by simple faith (whether in individuals, news agencies, institutions, or organizations) but does require one's own initiative to be educated, wrestling with the complexities when demanded, and weighing evidence. We also need to maintain discipline to find balance, to resist childish simplistic extremes. There is a level of maturity I think as adults to know when emotions are integral to achieving objectives, and when intellect is what's needed for a truly "good" outcome. Let's be honest as well, not everyone possesses equal intellect or even common sense in necessarily adequate quantities, so we need to apply discernment socially, and interpersonally as well.

Compared to many more important arenas, the audiophile microcosm isn't actually that complex. ;-) At least if we can start by finding truth, achieve understanding, and maybe even a sense of "order" in this tiny corner of the universe, then perhaps we can apply the same to other domains in our lives and how we interface with the World honorably.

Stay safe out there friends. I hope you're enjoying the music!

BTW: Chris Connaker is right, Lady Blackbird's Black Acid Soul (2021, DR9) is very good as he discussed recently, especially if you're a fan of the blues and soul.


  1. The 8 kHz noise obviously goes away with that isolator since the USB frame frequency is 1 kHz for full/low speed. A better way to see the effect on noise is to play silence or a low-level (-60 dB or so) tone at some unrelated frequency.

    See also, the USB Troll:

    1. Very nice Mans,
      Thanks for that and the link as well! The USB Troll looks great. Lemme know when I can get one ;-).

  2. I totally agree that USB isolators should only be used if there is an evidence of a ground loop, and there is no other connection where it could be broken. I suspect that the success of various "USB decrappifiers" and "JitterBugs" is partly because there are surprisingly many bad USB cables. I have tested the "collection" of USB cables I acquired over years and thrown a good half of them away because they either had no shield at all—these cables are known as "charging cables"—or had a high DC resistance of the shield. And visually you could never distinguish "good" USB cables from "bad" ones, you always need to measure with a DMM.

    And I have a question regarding your test setup, Archimago—did you use any clock synchronization between the DAC and ADC—either in software or using an SPDIF/TOSLink? Because without syncing, the clock on them can accumulate enough drift after running for a few hours to produce observable phase shifts in measurements.

    1. Mikhail,

      Your comment on bad USB cables was interesting - and a little bit scary! Could you perhaps elaborate a little on how to test this, for someone who isn't an electrical engineer? I must admit that currently I don't even own a DMM...

    2. Hi Freddie, usually the issue is with A-to-micro-B USB cables, however it's not limited to them.

      The first issue I have encountered is a complete lack of shielding on the cable. This is diagnosed by measuring continuity between the metal parts of the USB plugs on the cable—any cheapest DMM can do that. If there is no connection between them, that means the cable lacks shield or the shield is not connected to one of the plugs, or to both. This of course allows any EMF interference to couple to the cable.

      The second issue is that sometimes the shield is there, however it's done cheaply (aluminium foil and likes), and its DC resistance is high. Generally it should be fraction of an Ohm (and you'd want to have a good GMM that is able to measure this correctly). If the resistance is high, any current induced by nearby EMF fields will result in a noise voltage developed across the cable. Depending on the scenario of use this can affect signals that connected devices transmit.

      This is a good post on the issues with USB cables shields:

    3. Mikhail, thank you for responding so quickly, with a very good explanation!

      I like Archimago's approach of science and measurements, but it can be hard to understand the details when you don't have the skills or tools...

    4. Thanks for the note Mikhail,
      For this test, no did not use synchronization here. Just a "quick and dirty" look at the noise primarily between the devices with and without the ADuM in place!

      Nice link. Would be cool to have a running website with list of tested USB cables that are easily found and tested to be fully compliant. Of course things could change if the company cheepens out on costs and such!

    5. Hey Freddie,
      Yeah, it could be a bit overwhelming at first but I think if you stick with it you'll soon see the basic patterns around performance of devices and things like noise floor, distortion levels, etc. as they start to coalesce in your mind. I know when you read "mainstream" magazines and watch YouTube videos, there's often the concept that "measurements don't matter". I think this is an unfortunate reminder of the ongoing debate between those who seek "fidelity" and those who are aiming for "euphonia"; a bit of a false dichotomy of course because each of us is presented with the ability to choose and it's not necessarily "either/or"!

      The ears only have so much "resolution", and "subjectivists" typically don't want us to think beyond all that *they* are able to adjudicate since they don't take the time and effort to look at objective performance.

      I believe it is up to manufacturers to make a case for themselves when it comes to devices like the JitterBug and such. Likewise, magazine reviewers, "journalists" should have helped the audiophile be educated in matters of high-fidelity instead of churning out opinions and proselytizing products all these decades.

      Hopefully in the days ahead, more audiophiles can understand what works and what doesn't. Can use the technical lingo openly. Dealers and magazine reviewers likewise will need to start upping their game if they are to come across as knowledgeable and worthwhile. [Rather than obvious targets for ridicule and dirtying the "audiophile" hobby as if they're the crazy uncle who nobody wants to talk to over Thanksgiving dinner as he engages in mad conspiracies.] ;-)

    6. Archimago, thanks! Obviously I have no intention of becoming a measurement expert, but I'd like to understand a little bit more, rather than just reading the conclusions. I'm learning along the way, and sometimes the hard way - last year I bought two pieces of expensive hi-fi that I probably wouldn't have bought if I had read the recent ASR reviews of them...

    7. Hello, based on this thread I just measured some USB cables that I have at home and two of them have resistance of 15, where as the others 0,5-1,5 on my multimeter. Does it mean I should not use them?

  3. Not sure if this is an usb issue, but I had problems with electrical interference and huge pops sending my amp into overload with ASUS Sound XONAR U5. It was literally unusable, not sure if it was the usb, but that was the connection from pc to dac, then rac to amp. I replaced it with Marantz ND 8006, which is super quiet and problem free.

    1. Ouch Martin,
      Sounds like that's something more severe. Ground loops tend to be continuous events (eg. constant hum), and when you have noise from the computer, they tend to be soft bursts correlate to activity whether it's the mouse moving and connected to the same USB hub, or maybe hard drive transfers causing buzzing.

      Big pops (that can be damaging to speakers) speak to high-level signal anomalies like power supply-related issues. Even digital errors with USB cables should not be so awful:

      Best to replace the U5 I think.

    2. As I said I replaced it. After it shut down the amp, I stopped using it! You are probably right the U5 was the culprit, not the usb cable.

  4. AudioPoutine Odac was using the same usb isolator circuit built-in,
    just before the ESS DAC.


    Hi, do the above products work following the same principles? New to these stuff

    1. Hi PS

      No these products use audio transformers to break the ground loop.
      They will work OK in that respect, but it depends on the quality of the transformer how good they will sound.

  6. Hi Arch

    Having performed many tests on usb "noise" issues I can only concur with your observations.

    Proper galvanic isolators, as opposed to some of the useless inline reclocking noise reducers, have always solved any problems I have encountered.

    However, as you allude to, the real enemy is the use of single ended RCA connections that allow noise currents from computers or power supplies to flow in the signal low conductor (shield). It's a fundamental flaw that xlr doesn't suffer from. Unfortunately domestic hifi is still stubbornly based on RCA.

  7. I get incredible information in your blog.

  8. One of these solved a specific problem I had playing from a Mac Mini into my Audiolab M-DAC. I was getting intermittent clicking sounds, and watching the DAC's USB buffer display I saw the buffer level drop quickly every time it happened. Some data was being dropped for some reason - I have no idea why. With the isolator, the clicks and USB buffer errors have stopped.

  9. Thank you for penning this. I was a subjectivist when I was much younger. But a career in eng/tech showed me that proper application of analytical principles and processes is what allowed me to understand what I subjectively like. It makes it so much easier to communicate my subjective experiences via objective terms. Perhaps if the message is spread and comprehended by a broader audience, there would be fewer attacks based on vocabulary.