Ethernet switch

Is anyone using an expensive ethernet switch with their DCS equipment ?

1 Like

Yes, I use a Melco S100 and there are others here with the same.

I have to say that I can discover no reason at all why it should make any difference.There has been a lot of discussion here and I have had long conversations with a friend who before retiring ran the network for a national airline. Switches should make no difference ( so long as correctly configured for the purpose). However , irrespective of any logical conclusion, they do. So the question for me is now not do they make a difference but why they do? Incidentally everyone who has reported hands on use here has reported audible differences. It just that one or two found it a negative change compared to what they already used. All the rest have reported a gain in SQ.

I link to my most recent comment ( final paragraph):

Indeed they should not. I believe they actually don’t make a difference with respect to the bits being transmitted. A change in bits would never cause a change in soundstage, frequency response etc. The only possible consequence would be snap crackle and pop.

However my hypothesis is that there is analog (high frequency) noise that couples through to the analogue stages of the connected audio device. That hypothesis is consistent with the fact that power supplies (of the switch) with less noise have an impact and optical links have an impact

I am hoping the next generation of dCS products will have an SFP input that will put this issue to bed.

Ethernet connections are galvanically isolated on both ends so nothing adverse should carry over. I have no idea on the other hand about the frequency range of those network isolators. What is their band pass range? Key also is avoid use of a shielded Ethernet connection, which will connect the chassis grounds of equipment on both sides.

I am using an Innuos Phoenix net and although the difference is not night and day it does improve SQ. It just sounds more natural and voices have more of sense of 3D.

Switch does make a difference and it is not illogical.

Any switch gets the zeros and ones perfect, so for storage purposes, no difference at all, but in music what makes the difference is Phase Jitter. The pace they are presented to the clock.

Sound quality is about pace, so clocks for music (dCS …) are designed to be precise in that aspect, one second in one year is irrelevant.

Noise is poison, so a good switch must have a very good power supply (silent) and a good clock…

And, why noise affects zeros and ones?

Well, digital cables, Switches… are actually analogue. They transmit electrical pulses, short ones are read as zeros and long ones are read as ones.

But if we have noise, short and long become unclear and pace is ruined = Phase Jitter = poor analogue conversion.

I use now an Innuos PhenixNet, that replaces my former Melco S100. The Innuos plays in another league.

I hope it helps :grinning:

On the other hand, here is what dCS have to say about Ethernet switches:

Technically speaking, I see no reason not to believe them.


Nor do I. Unfortunately hands on experience of real switches tells me other than theory predicts or, rather, that limiting the theory to data transmission may not hold the answer. As the old ( Buffalo Springfield?) song goes “there’s something going on here but we don’t know what it is”.


What I mean is that data packet are data packet, so jitter=0.

Maybe there are different reasons why noise gets to the output via a poor cabling. A possible explanation is that using some esoteric shielded cable, noise gets from a power unit into the DAC, and that can influence the sound. But using unshielded cable that’s not a possible vehicle for noise injection.

My personal take of the whole thing is: I buy an expensive ethernet cable, which does not conform to the standards and is shielded. I hear a difference w.r.t. an ordinary 15$ cable. My audiophile brain says: “Hear that! It’s different” (maybe it’s worse :slight_smile: ). But the audiophile brain immediately translates “different” into “better”. Then I read that there are some wonderful High End Switches, and I say “why not”. So I try the expensive switch and find that everything sounds better, because I am injecting less noise through the wrong cable. So I caused a problem in the first place and solved it later.

If Ethernet cables, switches and all that made a difference, dCS would know. And they would tell us, because it adds value to their products.

Just my opinion of course.


1 Like

Hi Manel, and thank you for your attempt to help but unfortunately what you have said is not correct:

DCS DAC’s buffer the incoming data stream in internal memory before passing it on to the digital to analogue conversion stage so data will be passed from internal memory to the Ring DAC accurately clocked by either the DCS internal or external clock source as long as it arrives at a rate greater than the sample rate of the file being played. An ethernet switch will have no impact on the clocking of data in a DCS DAC.

Noise might be a problem but not in the way that you suggest. A chip set reading a digital data stream only needs to be able to determine ones from zeros which have specified voltage values for each. The actual voltage values do not have to be exact in order for the chip set to determine between ones and zeroes so any variance in the voltages caused by the addition of a noise component will not change the interpretation of the digital data stream. If the noise were so extreme as to change a value then the checksum check would fail and a retransmission requested.

It’s also worth noting that the ethernet signal is carried on four sets of twisted pair cables and that the twisted pair design dates back to the early days of telephony and is valued for its inherent external noise rejection.

Noise added to the mains by the (usually cheap and cheerful) power supply of an ethernet switch might be a problem for hifi components on the same mains circuit as might electrical emissions from an unshielded switch in the vicinity. The same issues could be generated by a NAS.

An ethernet switch or NAS designed specifically for audio presumably will address these potential issues and may offer a better solution than a run of the mill ethernet switch or NAS if it used in close proximity to a hifi system.

In my opinion an even better solution, if feasible, is to not have an ethernet switch or NAS anywhere near to a hifi system but to run ethernet cables from a router or switch located in another part of the house and preferably on a different mains circuit.

1 Like

I thought that at first and tried it out.

I have 5 dedicated lines for the music system, 7 different lines in total for the house with their own breakers, …. Also, a dedicated grounding in the garden (yes, a bit overkill). Router and NAS are 16m away fro my system and in another power line.

The empirical findings are that with a switch next to the dCS sound is noticeably better than running directly from the router.

At the end of the day, I trust my ears better than my theories :blush:

It has taken 40 years to find out that zeros and ones “perfect sound forever” is not what it seemed to be and there are more variables to be considered.

Game is not over and we are still learning, but ear is always the final judge.

tested Melco S100, Cisco 2960 and Ansuz switches- they all delivered different sound characters on dCS. I bought the Ansuz. Plan to test the Innuos soon, heard very good news about it.
PS: doesn’t matter what the physic /specs may say- test it in your set up:-)

I am using a modified Melco S-100 e-switch, sent to Jord at Pink Faun in the Netherlands…new ultra OCXO clock module and uses 2 external lps, a 5v lps for clock, and the 12v lps or the motherboard.
Night and day difference, background is beyond blackness, and using a Nordost Valhalla 2 ethernet cable to my dcs Bartok.
One of the best upgrades I have done, just priceless sonics.
Highly recommend Jord, his team is fast professional and masters of electronics at Pink Faun.

The upgrade for the Melco S100 includes:

  • The Pink Faun ULTRA OCXO clock.
    -Vibration dampers for the ULTRA OCXO clock board.
    -Cable upgrade, all cables inside the S100 are replaced by Pink Faun proprietary cable and all wires are soldered to the board instead of using connectors.
  • DC-input 5v for the clock. The side of the DC connector is 5.5/2.1mm center positive.
  • Labor and small parts.
1 Like


I do with aqvox solution (highend switch and ethernet cable).
i am only a dcs network user with AMR dac following

Visited their site. Intriguing. I would be tempted to invest 179 euros for the galvanic insulator, except the ethernet card in the Rossini is already galvanically insulated, and I think the guys at dCS know their job. Anyway, never say never.

But I am really puzzled about the clock. Why do I need a clock when the ethernet transmission is not synchronous and a lot of things happen after data packets reach the DAC, before they are converted into a clocked bitstream thanks to the external Rossini clock?

They might as well say that if I put an alarm clock in my car it will have a better acceleration, but I just won’t believe it.


1 Like


i have no answer to fully explain.
but in my set i had a great bonus to use those products
although it was before i buy the network bridge (i was with allo solution)
don’t ask me why however, but i won’t let them go back
i have even upgraded one of my tywo aqvox ethernet cable to best level.

note you can have a trial if you are not sure.
But Norman won’t give you more internal details on his stuff.


I am wondering how your analogy applies? A alarm clock in a car?
Please try to keep questions relevant to audio.
I would recommend you contact Pink Faun and ask them about clocks, as they have been developing this technology many years…and are true audiophiles.
Are you an expert in OCXO?

1 Like

Betsy, reading your message I assumed for a moment that you must be a Forum moderator telling me what I am allowed to write. But you have just two posts, so it seems you are not, and I will disregard the second line in your message.
I will also assume that you are a private enthusiast like me with no commercial affiliation.

So, we have dCS saying that “Audiophile Ethernet switches are a solution looking for a problem.” (Andrew Papanikolas July 2019)

And then in your first post on the dCS forum you write: “Night and day difference, background is beyond blackness”; “One of the best upgrades I have done, just priceless sonics.”; “Highly recommend Jord, his team is fast professional and masters of electronics at Pink Faun.” and add the website, then a well formatted upgrade list, including labor and small parts.

I checked the website, saw the galvanic insulator, and got intrigued: I think I will give it a try.

But the clock? What improvement may come from an improved clock when “The Network interface runs in asynchronous mode” (taken from the dCS Rossini manual)?

About the reference to the alarm clock: many years ago some audiophiles were sure that putting one thing that looked suspiciously like a toy alarm clock (and was in fact a toy alarm clock) anywhere in the listening room would improve sound. And if you purchased two alarm clocks, it was even better. And so on. In principle, you might improve SQ indefinitely for the cost of an adequate number of alarm clocks.

So with my audiophile friends we said “yes if you put the AC in your car it will go faster, if you study close to the AC you will get higher marks,” and so on. Was kind of a recurrent joke.

No offense meant, but every time I hear the words “let your ears be the judge” with no technical rationale offered, I think of the magic Alarm Clock. Ready to change my mind if a scientific explanation is offered.

For the record: many Alarm Clocks were sold, and according to some sources, someone bought three of them.

Regarding your final question: my research field is not close to OCXO; it’s closer to polynomials over Galois fields, the theory behind Reed-Solomon codes, which in turn are related to crc codes used in data transmission. I think there is no use here showing off titles, academy, scientific papers and so on: this forum is for friendly people who share a common passion.




Melco S100. I must confess I bought this primarily because I didn’t want the ethernet switch to be a “weak link”. I won’t even pretend to understand the science (or lack of it). The sound quality of my system is fantastic ( for me).
Whether it is worth $2000 is a contentious and different debate ( similar to whether a $5000 power cable makes a difference).
Over the last year or so I have listened to- and slowly upgraded- dCS components. They sound so good that I am “afraid” to pair them with anything that is not close to the top of the line ( if I can). Am I an audiophile or an audio-fool? I have often thought about this- I guess I am somewhere in the middle.
Thanks for letting me share my 2 cents.

1 Like