Sunday 5 February 2023

MEASUREMENTS: Intona 7055-C USB 3.0 SuperSpeed Isolator. And Darko & Lavorgna hear no difference with the Silent Angel ethernet switch. [Importance of open discussions.]

Over the last couple years, I've shown in my PC measurement system the benefit of using USB isolation to break ground loops when test devices are connected to the ADC. There was the ADuM4160 device which was limited to "Full Speed" (12Mbps). Then I posted on the Topping HS01 which operates at a fixed USB 2.0 "High Speed" (480Mbps).

Over the Christmas holidays, I got a hold of the Intona 7055-C USB Isolator which can run at USB 3.0 "SuperSpeed" (5Gbps), basically a pass-through device which provides galvanic isolation of the USB signal and ground lines, with the flexibility of being downward-compatible with 480Mbps (most important for USB audio), 12Mbps, and even Low Speed 1.5Mbps devices.

Although audiophiles have been talking about isolators like this for years used in home set-ups, they're really meant for industrial applications where devices with USB ports might be subjected to high-voltage surges and spikes. This "C" model is designed for protection up to a modest 1kV over 60 seconds. The more expensive 7055-D can handle up to 5kVrms over 60 seconds.

Compared to the ADuM4160 (cheap, often <US$20) and Topping HS01 (currently US$70), the Intona is significantly more expensive at a steeper US$350-380 currently. I purchased this myself through usual retail channels.

The box measures a small 120x30x70mm. It's made of aluminum and feels solid in the hand (weighing 300g). In the image above, we see the rear of the device with the ports, from left to right, the USB-A connector where you plug in the device you wish to isolate (say a DAC to be tested), there's a micro-USB AUX 5V power input if needed; for clean power, a battery is recommended, USB-C would have been nicer these days but this is fine. Then to the right is the USB-B 3.0 (USB 3.1 Gen 1) port which would be connected to your computer. To the far right (harder to see in picture) is a small square LED that blinks and can change color depending on connection speed.

Here's a view of the bottom of the enclosure:

Notice that I have "v 2.0.3", and it's Made in Germany.

The USB-A port can provide up to 2A with microUSB auxiliary power connected. Otherwise, the model "C" here can provide up to 1A which should be enough for most USB-powered audio devices like DACs. Compared to other 7055 models (like the 7055-D with 5kV isolation), this "C" version operates with lowest output noise level up to 80kHz measured as 0.9/1.6/8.6ΞV (bandwidth 20/80/500kHz based on datasheet).

I can confirm that it is able to switch speeds between Full Speed, High Speed, and SuperSpeed devices. Unlike the Topping HS01, I can actually plug in something like my Logitech Unifying keyboard/mouse receiver and it'll still work at the slow USB1 speed. Old UAC1 DACs like the AudioEngine D3 (2014) worked fine and likewise the UAC2 Light Harmonic Geek Out V2 (2015) dongle DAC also ran without issue. Modern XMOS based, USB-powered, "desktop" DACs like the Topping D10s and D10 Balanced ran without external power applied.

As with the previous articles on USB isolators, let's see what the isolator can do in reducing ground loops and noise in general. Here's a block diagram of the set-up I'll use:

If one were to use the Intona in an audio system, it would be like this. Basically as a USB isolator situated between the computer/streamer and DAC.

And here's what it looks like on my testbench:

As you can see, I've annotated the pieces of gear in the image for clarity. Notice I'm using the USB-powered Topping D10 series of DACs that when looped back to the E1DA Cosmos ADC and the measurement computer, can result in ground noise and potentially exacerbate signal line noise as well.

In the back (blurred), we can see the 10-port USB3.0 Hub I'm using along with the Intona isolator running at USB2.0 480Mbps speed denoted by the green LED with D10s DAC connected. Both the Topping D10s (at 192kHz as you can see) and the E1DA Cosmos ADC are connected to this hub which is subsequently plugged into the Beelink SER4 MiniPC for signal generation and measurement using REW.

Let me run an accurate 1kHz THD+N measurement of the D10B/s to show the effect isolation makes with the E1DA Cosmos "Stack" including the ADC, prototype Scaler and APU notch filter. I used 2 lithium batteries to separately power the APU and Scaler; doing this reduces the noise level as opposed to powering the devices with the same battery.

Starting with the Topping D10 Balanced (XLR) with and without the Intona isolator; to strain the system a little more, let's run the test at 192kHz (96kHz bandwidth):


While the THD+N number doesn't show a huge difference between -117.7 vs. -118.0dB, obviously we can see from the FFT that the isolated USB port helped reduce noise in the output of the Topping D10 Balanced in such a configuration.

We know from previous testing that XLR balanced connections are typically highly resilient to noise but as we can see in this example, it's not always perfect especially with a USB-powered device.

Let's switch over to the Topping D10s (RCA) which is the single-ended version of essentially the same DAC and ESS chip (ES9038Q2M) internally:


The USB isolation in this case made a huge difference to the analogue output quality of the single-ended DAC. A nice example of the benefit that isolation can provide in a system where ground loops and USB signal noise might be exacerbated with all kinds of devices plugged into and powered by the same USB hub.

As a test of compatibility with the Intona, here are a few measurements from the RME ADI-2 Pro FS R Black EditionSMSL DO100, and Sabaj A20d 2022 DACs, XLR outputs. They're mains-powered rather than USB-powered so will not be as susceptible to USB ground noise like the Topping D10-series above. Measurements done at 96kHz, which is what I've standardized on over the years on this blog for hi-res performance.

Good luck blind testing between DACs like these given their flat frequency response, very very low distortion, high linearity and ultra-low jitter as discussed in my reviews! Focus on features like EQ, headphone amp, output level adjustments. Room acoustics, speakers, headphones, music mastering, amps are way more important these days. [Hard enough blind testing 16-bit playback!]

Very clean performance for the DACs and measurement system as a whole. (Notice the RME's AKM DAC shows a more predictable "cascade" pattern with reduced higher-order harmonics compared to the lower level fluctuating ESS patterns.)

So, what about the dreaded "jitter" in the face of severe USB ground noise? Here's the J-Test from the Topping D10s RCA output comparing the un-isolated vs. isolated USB port:


Yup, we can see the high noise level without isolation in this configuration. It's tempting to wonder if the +/-4kHz noise at 8kHz and 16kHz might be jitter-related sidebands on either side of the 12kHz primary signal. In fact these are not jitter sidebands. The 8kHz noise is USB2 packet frame noise which we've discussed before, and the 16kHz tone is the 2nd harmonic. Clearly the isolated Intona port cleans this all up and we see the expected, extremely low jitter performance of the Topping D10s with ESS ES9038Q2M DAC chip. No need for special "femtoclocks" or OCXO stuff these days for precise DAC performance, folks.

I must say that I am quite impressed at how well the Intona cleaned up that signal considering that this 10-port USB 3 hub I'm connecting both DAC and ADC to is probably the noisiest USB set-up I've tried among tests of isolator devices.

As a quick test, notice that the Intona's LED is now a light bluish, confirming that indeed it's detecting and running at USB 3.0 (5Gbps) speed with the USB3 Kingston DT50 memory stick.


Conclusions:

Yes, the Intona 7055-C USB 3.0 SuperSpeed Isolator works as advertised. It's a much more versatile isolator than the Topping HS01 previously discussed with greater compatibility with USB devices from "slow" 1.5Mbps all the way to 5Gbps USB 3.0 (3.1 Gen 1).

While compatibility with USB devices seems good, and it's working well with all kinds of stuff I've plugged into it from the USB3 memory stick above to Logitech's mouse dongle to various DACs, conspicuously, the E1DA Cosmos ADC and #9038D6K DAC were not able to connect stably. Typically when I plug these in to the isolator, the computer will "see" the device for a few seconds then disconnect. I wondered if the ComTrue USB controller chip in these E1DA devices might be the issue. Curious if anyone out there has been able to get the E1DA devices to work reliably on an isolator. I've tried a number of cables including OTG varieties and still no-go.

For a peek, let's open up the device to see what's in there:

Notice that there's a lower main board split in two halves for galvanic isolation and then there's a daughterboard. First let's focus on the mainboard:

Straddling the two halves is a Coilcraft DA2303-AL SMT transformer for voltage isolation. That reddish box is a Schrack PE014005 power relay to switch over to the auxiliary voltage input when connected.

Data line isolation looks to be the job of the daughterboard. We see a couple of Xilinx Spartan FPGAs presumably doing the USB data-rate negotiation and transmission across the input/output divide. Underneath the "v 2.0.3" sticker are most likely a couple of high-speed digital isolators for the bidirectional communications. In previous USB2 generation Intona models, I see that they used Si86XX-series devices. For 5Gbps speed they must be using something faster; maybe the dual-channel 2.5Gbps Analog Devices ADN4620. I didn't bother to peel off the sticker to confirm.

As you can see, it's not a simple low-component-count design and there will be custom FPGA programming needed, making this device significantly more expensive. Plus I'm not sure how much demand there is for USB3 isolation, so there's probably lower economy of scale.

Let me be clear... Although I can show the effects of USB isolation using my measurement setup with the E1DA Cosmos "Stack", I do not believe as an audiophile, one would typically need to buy something like this to use between your computer/streamer and DAC.

More likely than not, the playback system from computer/streamer to your USB DAC is direct without any loopback to another USB-powered device or common USB hub like what I'm doing with my measurement system, thus causing those ground loops. However, if for whatever reason with your particular computer/streamer, you are hearing electrical noise seeping into your DAC and audible through the speakers - like humming, buzzing, intermittent noises correlated with CPU/HD/GPU activity, then by all means, a device like this or the Topping HS01 (USB2 only) probably can help. The Intona is a useful tool when there's a need, and like all tools, to be used judiciously. Whether spending around US$375 represents a valuable investment for you will therefore depend strongly on specific needs.

The ASR measurements of this device back in 2020 showed no effect (with a shrugging Pink Panther). This is not surprising because the Audio Precision gear used for measurements already has isolated ports so he wasn't going to see the kind of anomalies I'm showing here.

In the days ahead, you'll see me use this device in the measurement chain when needed.

--------------------

Let's end off with even more "audiophile media criticism". ðŸ˜ą

It's kind of fun addressing this stuff because there's just so much mis/dis-information out there including claims from longtime reviewers and "mature" folks who should actually know better! For some of these guys to perpetuate clearly uninformed positions (such as the discussion last week on MQA) after all this time suggests gross unwillingness to change and I think reflects a form of intellectual dishonesty with themselves (do they truly believe what they say?) and towards the readership/viewership. As audiophiles, with all the information we have now at our fingertips, I trust it doesn't take too much searching and critical thinking to be able to quickly point out "naked disinformation"; the kind of stuff that often seems to be designed to perpetuate all kinds of "Fear, Uncertainty, and Doubt" when in fact contemporary technology has already solved many problems.

Over the years, John Darko and his material has been interesting to consider (see here and here for examples, among others). I see in the first podcast of 2023, he again has a running discussion with Michael Lavorgna - "Podcast #43 – Do ‘audiophile’ network switches REALLY sound better?". For those who haven't followed along, Lavorgna left his site AudioStream back in 2018 (asked to leave the Stereophile network?) and the site contents have completely disappeared since.

The answer to that podcast question is obviously - "Of course not!" for all kinds of reasons as spelled out here and here. Zero evidence in support of "audiophile switches" over the years other than questionable anecdotes and speculative beliefs without evidence.

Having said this, I guess it's OK that both Darko and Lavorgna apparently did not hear a difference with the Silent Angel Bonn N8, an 8-port gigabit switch with a US$549 MSRP. Discussion of this starting at 42:00 into the podcast.

[Before that, there was talk about the US$3700 (!) Silent Angel Genesis GX Word Clock which would be odd to have in a home system and likely would worsen DAC jitter anyways. Master clocks are useful in studios with multiple digital devices like AD converters, effects units, or video synchronization needs. See here for serious discussion and testing.]

I find it fascinating how these guys still have a hang-up about "jitter" as some awful boogeyman attributable to ethernet switches. I don't see why Darko seems to have a problem with repeating a claim from a DAC manufacturer that their product's "architecture ... makes this DAC resistant to jitter" in a press release when he has no problem with quoting all kinds of other nonsense. Want an example? Just look at the silliness he repeated from the English Electric 8Switch press release back in 2020! With an asking price of £450, does he actually believe this ethernet switch has special "high and low electrical noise isolation, reducing troublesome data issues when streaming music over a network" (as opposed to just accurate data transmission from any decent switch) yet not believe DAC manufacturers have the capability to render jitter inaudible? So has he learned after all these years to differentiate facts from fiction, or is this some kind of cherry-picking of claims through his idiosyncratic filter based not on knowledge or verified perceptions?

I also find it funny that Darko would use the example of the Audiophilleo as justification of the importance of jitter in 2023. This is an old USB-to-S/PDIF interface first sold around 2010. It was released at the advent of asynchronous USB. I don't doubt that it sounded good (in fact a friend had one in his system back in those days), but realize that this level of performance has become a standard feature of USB DACs for a decade now. Are we as audiophiles supposed to be stuck with the same level of awe about jitter reduction as we might have been in 2010? Or do we in the 2020's basically have an expectation that all modern DACs from reputable brands should perform with minimum jitter, and we can measure anomalies still a problem in the minority of sub-par devices? To hang on to the same kind of fear/uncertainty/doubt Darko speaks of would suggest that devices have not advanced in a decade! This is patently false but I guess some companies want to perpetuate this belief.

At 52:00, Lavorgna chimed in that not hearing a difference with the Silent Angel switch "isn't the same as me saying that audiophile-grade ethernet switches are snake oil...". Uh-huh, so he's not willing to make a committed statement. But where is there evidence that any of these special ethernet switch claims are not snake oil given that there are examples of such disgraceful serpentine lipids being sold? If the Silent Angel doesn't sound different, and the AQVOX switch is overtly ridiculous, care to show us an "audiophile ethernet switch" that's verifiably worth the price? Since audiophile companies making ethernet switches seem to be making extraordinary claims, perhaps it's time to see extraordinary evidence first and ignore whatever "faith" reviewers like Lavorgna seem to be possessed with.

Of course, every time these guys talk about "jitter", soon they'll talk about scary "electrical noise" (57:00). It is strange that Lavorgna would suggest to audiophiles to do their own research on "mixed signal systems", hinting at the idea that one would find some deep truth that will somehow clarify things. As a writer and reviewer, isn't it his job to explain this to readers rather than sending audiophiles on some wild goose chase? Care to write something a little more detailed, maybe more technical to show us exactly what you think we'd find? I guess the answer to that is suggested around 58:00; he's just referring to what he was told in interviews with audiophile designers over the years - not that he has seen the data or specific research. Let's just say some of the names he dropped like the late Charles Hansen (of Ayre Acoustics) had choice words about the intellect of mainstream audio reviewers when he strongly disapproved of MQA shortly before he passed in some of our private conversations.

I see that recently Lavorgna appears to be reviewing the Mola Mola Tambaqui DAC; maybe he should take the opportunity to interview Bruno Putzeys about whether he uses an "audiophile" switch, which one, and whether Putzeys thinks they represent good value.

Without even any real depth to that conversation, Darko then announces that "that's where my knowledge ends" by ~59:00. If his knowledge ends so quickly, what makes him think he should even speak to this at all publicly? Perhaps it's time for Darko and Lavorgna to start doing their own research, their own testing, maybe even their own measurements. Maybe it's time to interview some more reputable audio engineers who aren't catering to fringe, purely subjective (and unreliable), audiophiles who accept merely a few words and run with it as if this is enough to then build their Industry-friendly narrative around.

Indeed, by this point, multiple horses have been flogged and sacrificed (1:00:45) when it comes to discussions of "audiophile switches" or even "audiophile cables".

I wish John Darko a speedy and full recovery from his eye surgery. When it comes to audiophile discussions though, I really hope he and Lavorgna can communicate like educated journalists and perhaps try their hand at applying critical thinking and mental flexibility when faced with evidence contrary to their beliefs especially while interviewing individuals with a sales agenda.

Seeking truth demands vigilance; it doesn't serve anyone (including themselves) to be in a state of willful blindness, passing along misinformation, or even worse, disinformation in the form of implied FUD yet have zero ability to demonstrate their claims about stuff like jitter and noise for example.

I would submit to Darko and Lavorgna that in my experience, being open with the audience, which means opening up the comments section (notice that SoundCloud comments were disabled), and seeing what kinds of responses one gets is a good feedback mechanism. Without open feedback that allows healthy debate, we end up getting stuck in small echo-chambers of our own creation, walled off to foster an "us vs. them" paranoia, or comforting delusions and sensitive egos by avoiding normal ways of verifying realities. After all, we're simply talking about computer hardware like ethernet switches as in this podcast! Trust that even if there are trollish comments, serious hobbyists can sort those out from honest feedback seeking to correct inaccuracies in an article/video/podcast.

If we honestly, logically, and methodically speak truth, what's there to be afraid of in open discussions with other audiophiles? IMO, if one is seeing a lot of resistance to what is being expressed based on technical arguments to the point where one is considering strong comment moderation and shutting down discussions, make sure to consider the possibility that you are wrong. Perhaps it's not you, but the claims of the company being discussed are strongly suspect - which is a bad thing only if your job is in advertising and the public can see through the false claims.

--------------------

To end, I want to send kudos to the Roon team for the recent releases of Roon 2.0.10 (build 1202) and ARC 1.0.10 that came out earlier this week. I'm seeing somewhat better performance without as much issues like >5 second latency when playing tracks. I see this excessive delay switching tracks after 30 seconds of playing a song when RoonServer is up for many days. Stopping and restarting RoonServer will fix the issue for awhile. Why does it do this Roon? Memory leak? Cache performance issue? Still some work to be done obviously since I don't think this should ever happen.

Remote playback on ARC is more stable even on Android emulation with Windows 10/11 BlueStacks and the Android Auto feature works pretty well. Good to see improvements given some of my concerns expressed recently. Still seeing some pauses and errors when jumping between FLAC and MP3-encoded music across ARC in "Balanced" (Opus 256kbps transcode) quality mode possibly due to timeouts from the latency issue above, but at least it's less frequent. One suggestion is that when the network gets disrupted, ARC seems to have difficulty reconnecting. So after a few "Poor connection" errors, it would be nice to have the option for the app to exit and restart which typically fixes the issue for me.

With that, dear audiophiles, I wish you joy in your music listening. This past week, I've been catching up on some of Hilary Hahn's recordings including her recent Paris (DR15, 2021) and her 2008 release Schoenberg, Sibelius: Violin Concertos (DR14). Very nice recordings for audiophiles to appreciate both the artistry and recording quality. Alas due to work responsibilities, I missed her performance here in Vancouver a few weeks back with the VSO, but glad my daughter and wife were able to attend.

I wish you well in negotiating the treacherous minefield of misinformation around audiophile topics out there. Until next time...

7 comments:

  1. Informative as usual, and of course I agree that it's disappointing to have reviewers who so often spread manufacturers' talking points without investigating them, looking at existing analyses, or applying much logic.

    Regarding Roon, I use a Roon server (Core) that is ROCK running on a Roon-approved NUC, and I don't see the problems with latency that you report. I don't know if that narrows down the problem to your version of Roon Server, or if there are other network, configuration, or library factors at play.

    ReplyDelete
    Replies
    1. Thanks for the note Roth,
      The issue I see appears to be the same/similar to this one on my Windows install:
      https://community.roonlabs.com/t/lagging-when-starting-playback-800k-track-library/157468/5

      I don't have 800k tracks! (More like 200k.) But that comment about latency worsening after 30 seconds of playback is also what I see. Not horrific but a little annoying and seems to be more severe if I have the server running longer...

      Hopefully Roon can look into this odd behaviour! An issue since 2021?

      Delete
    2. Here's another thread of the issue I'm seeing:
      https://community.roonlabs.com/t/roon-slows-down-over-time-core-server-reboot-temporarily-fixes-problem/133306

      No solution from Roon I see...

      Delete
  2. I went and bought an Intona 7054 USB2 in 2016 solely on the advice of others raves. It did subtly change the sound and I thought for the better but after listening for some three months and removing it, I found it had a slight glazing effect on the sound not dissimilar to what I have heard trying many power supply devices/filters. These devices do some good things but underlying also take something away. Intona has been put back in my system a year later and removed straight away. Since my experience Intona have done more work to improve things and as you show it does remove some spikes/noise. Intona did motivate DAC manufacturers to include galvanic isolation in there designs. Did you listen?

    ReplyDelete
    Replies
    1. Hey there Robo,
      Great to hear you and friends have had experience.

      Yeah, I had a listen with the Intona placed between the low-power MeLE Quieter2Q computer running Linux and HQPlayer to Sabaj A20d 2022 using RCA out the other night. Subjectively maybe a sense of better noise floor but not something I would want to blind test ;-).

      Certainly nothing wrong with getting manufacturers to consider ways to achieve better performance like adding galvanic isolation!

      Delete
  3. I do currently have the original industrial version 7054 and the same 7055-c and been using them in both of my systems. Because of the portability I had also brought the 7054 to some of my friends’ place to try in their systems. Even for some of the dacs that claim to have built-in galvanic isolation, the intona still makes a subtle but noticeable improvement to the sound. Now amongst my group of audiophile buddies, we all have a couple of these intona units in our systems

    ReplyDelete
    Replies
    1. Hey there Deric,
      Cool. And yes, I agree if I can hear a difference it would be subtle as noted above ;-).

      The important thing I think is to recognize that the Intona isn't "snake oil". It does perform a job which might have a significant effect in certain circumstances. It's up to each audiophile to figure out when something like this is needed and when it's not.

      Delete