Installed a new switch today

The noise isn’t affecting data, it’s affecting other parts of the system.

If you use Toslink to transfer data via S/PDIF over a cheap optical cable bent into a figure 8, you’ll see no difference in the extracted data, but the clock jitter will be absolutely off the charts bad.

Would they allow in their design? No.

But might it have an effect they didn’t expect? Of course.

dCS has stated several times they see no need for expensive clock cables, yet it’s obvious to any who actually listen that there are significant audible differences between say the Apogee Wide Eye and Cardas Clear, Audioquest WEL and more recently, Transparent clock cables.

No one disagrees that the Ethernet data transfer is bit perfect, but we’re still struggling to understand why bit-perfect interfaces like Ethernet switches do in fact sound different from one another, and it’s not due to “bias.”

3 Likes

Indeed, with that I am struggling now. I hear a difference. Will do more (blind) testing.

1 Like

It’s funny how it’s always companies that struggle to understand how stuff works that sell you the “upgrade” products.

2 Likes

I accept what you say you hear, but how do you know this? What have you done to eliminate the variable of bias from your evaluations, in a way to provide a third party confidence in the assessment? Insisting in bold type there is no bias is a meaningless assertion without demonstrating its removal from the evaluation process.

Just as you say that without the Clock, your Rossini does not to your ear reproduce your test tracks as well as your former Wadia DAC, I can accept that assessment as a subjective evaluation. But unless you’ve taken steps to remove cognitive bias from the evaluation, please don’t ask anyone else to accept it for more or less than it is.

I don’t expect you to accept anything I have to say any more than I expect to ever agree with Anupc’s objectivist leanings.

I don’t double-blind, I do all my evaluation completely sighted, and I believe I control for bias given how often I don’t choose to purchase products that are very highly reviewed by magazines and others. A great example is I recently demoed a Simaudio MOON 810LP phono preamp, which is very highly regarded throughout the industry, and it took me all of thirty seconds to hear that it paled considerably compared to my Audio Research Ref 2 Phono SE. (Likewise it took about the same time to hear how much better the Ref 3 Phono SE sounds when I recently had an opportunity to trade up, and I did.)

You can choose to accept what I have to say or not, but I certainly would never, ever want anyone to make a purchasing decision based upon what I say.

Rather, as with magazine reviews, I would hope at best my words would help others to narrow down what they want to audition, using whatever methodology they choose. (I also only read reviews to determine what I want to audition, and don’t think I’ve ever spent more than a few seconds looking at Stereophile’s published measurements except to check whether amplifiers meet their rated power output.)

If others measure or listen and find no differences and consider me under the influence of snake oil peddlers, so be it; I know what I hear, I do my own testing and select products that make improvements in my system.

Likewise I find myself most often purchasing products from companies that use measurements to determine why what they’ve experimentally determined sounds better sounds better and those that listen to each product to make sure it sounds right before it leaves the factory.

Wadias actually almost always measured poorly in some way, but they were the DAC that previous to my Rossini + Rossini Clock sounded the most like… music, and were the first DAC I heard that ever did, something I never realized listening to earlier CD players until I heard it.

I’m just relating my experiences, feel free to place me in whatever category you choose.

This is just down to electrical noise, which can be bypassed by using an optical connection. There is nothing mystical about it. Clocking Ethernet, and jitter affecting the dac clock is just bs. And the sad part people keep quoting some “measurements” behind a paywall, but no sight of them, and no interpretation.

2 Likes

Well it’s easy enough for anyone interested in an audio-rated switch to test using a pair of Ethernet to fiber media converters; I recently installed a pair for lightning protection on the Ethernet output from my cable modem.

I can’t say whether they affected sound quality or not as I don’t ever stream audio except from a local file server, and even that’s rare which is why I’ve never tried an audiophile switch for my LAN.

Update:

iFi just released the LAN iSilencer which also galvanically isolates Ethernet but does so in a much. more compact form factor for just US $89, about the same price as a pair of media converters, a pair of SFPs and a fiber optic cable.

And where are your measurements ?

Extraordinary claims require extraordinary evidence. Or…anything at all really. I’m not the one claiming the entire planet functions perfectly using Ethernet protocols, except streaming dacs, where just like in most areas of this old audio tech “NEW AND EXCITING TECHNOLOGIES ARE DISCOVERED EVERY DAY” . I mean just using common sense, how does your Netflix work with the old router and not your overbuilt dac?

That’s fine, but why such fierce replies when other people say they can hear a difference in their systems, for me the alpha audio conclusion makes sence.

Jeez Bill, I said I accepted your listening assessment (twice), and simply asked how it was you could definitively say that cognitive bias was not a factor. I did not categorize or label you, suggest you were promoting snake oil, or anything of the sort. Simple question.

False equivalence; TOSLink is a synchronous physical layer interface, while streaming over Ethernet is asynchronous and involves layers of abstraction and error-correction specifically designed from inception to operate in an electrically noisy environment - the Postman delivering your mail dirtying the envelope is not somehow going to cause a typo on the letter contained within. That would be absurd, no? :rofl:

It’s by no means a “fact”. The only fact is that there are millions of devices operating perfectly well with Ethernet without any special Ethernet Switch required. Audio is not somehow special.

1 Like

Sorry, it was meant as a general case commentary, not a direct commentary on you or your post.

My apologies.

Oh that’s right, you don’t believe in power cords or power conditioners either, otherwise surely the fact that my computer works with any old power cord means my stereo couldn’t be affected by using a different one as “audio is not somehow special.” :roll_eyes:

Hi Bill, I don’t think @Anupc has ever written that power cords don’t matter, and @Ermos started this thread to discuss switches.

Hopefully it is not a controversial statement that everyone has natural biases, and repeated double-blind testing is an effective way to mitigate/eliminate them.

With respect to the matter at hand, I believe everyone is in agreement that ethernet delivers bit perfect data transfer. The interesting question, which you raised, is whether switches impact anything that could alter the analog signal coming from the DAC. I think that is a good question, and one I’m not qualified to answer. @Anupc clearly has some expertise in this area and his views are interesting to me, and I suspect others.

Of note, even though MSB offers an ethernet connection (“Network renderer module”), they recommend using a fiber/USB connection (“ProISL”)

(again, I own dCS but respect MSB’s architectural approach)

Cheers!
R

And Linn are now on record with their new Klimax Organik stipulating that fibre provides a superior result to Ethernet.

3 Likes

Apples and Oranges Bill, Apples & Oranges :laughing:

Also, try not to generalise please, context is everything. That line about Audio not being special was specifically with respect to Ethernet traffic; Audio streaming is not special at all.

Spot on.

I suspect previous hold-outs now agrees Ethernet delivers bit-perfect data transfer because it’s so easy to prove objectively, so they’ve abandoned that bandwagon.

These days “Ethernet noise”, and “Jitter” are the new catch-phrases, just look at the marketing blurb of Ansuz, Innuos, English Electric, Nordost, Uptone, Silent Angel, etc etc.

But suggesting Jitter as a problem for an asynchronous packet-based transport - especially one involving TCP buffers/windowing at the DAC end-point - is absolutely bonkers. Clear indication these folks don’t know what they’re talking about. :laughing:

Vendors who claim ultra-precise Clocks (e.g. Uptone, Innuos) or external clocking (e.g. Uptone, Silent Angel) to address jitter on Ethernet Switches are foolishly ascribing traditional Clocking problems onto a packet transport - “solutions” looking for a problem that doesn’t exist. Clocking precision affects data integrity, but that is already at 100% with the cheapest Switches out there.

So, that leaves just Ethernet Noise as the only remaining straw that some love to clutch at. I readily admit it’s slightly harder to debunk, as it could possibly affects lesser DACs. For example, the Metrum Amber that Alpha Audio used has an el-cheapo Ethernet port with integrated magnetics (circled in red); maybe that’s reasonable for a $1,200 streamer, I couldn’t say.

Whereas the dCS DACs have a separate dedicated Pulse Transformer that are pretty darn good at noise immunity. Here’s a blurb from the manufacture;

Pulse Electronics High Isolation Ethernet Magnetic Transformers

Pulse Electronics High Isolation Ethernet Magnetic Transformers meet IEEE 802.3 specifications and feature a compact footprint for use in space-restrictive appliances. These transformers offer high isolation for robust applications that require more isolation than traditional LAN applications. They provide safety isolation of up to 6kV and are ideal for both industrial and medical applications.

The Audio Alpha folks could have just measured the analog output of the DAC. Just stream a pure-tone into the DAC, any audible differences (if they exists) caused by Ethernet Switch Noise will show-up as visible difference in distortion spikes on that pure-tone FFT when compared to a less noisy Ethernet Switch. Can’t imagine why they wouldn’t do such a simple test (unless differences don’t exists? :man_shrugging:t2:)

3 Likes

Without wanting to defend them beforehand, we still did not see their measurements and their interpretation.

1 Like

That’s true though. But such a test wouldn’t require a “Yes: 100%. We have now proven that, we hope.:wink: