They may well be. However in an interview in the German magazine Stereo some years ago , Alan Ainslie of Melco Audio specifically points out that the improvements that the S100 switch may bring have nothing at all to do with replacement clocks or power supplies. He does not reveal what the precise reason for any difference may be but gives a vague explanation along the lines of it being the way that the data are handled. Nothing further I am afraid so it remains their secret sauce.
2 Likes
Pete, unfortunately T-elmi is correct; there has been enough “forensic” work done on the Melco S100 Switch to know for a fact that other than the pretty chassis and beefed-up power supply, it has an identical motherboard to the parent Company’s BS-GS2016 switch (which sells for under US$300 on Amazon currently).
That Buffalo Switch is actually a highly configurable Managed Switch with a Web interface. The Melco S100 on the other hand seems to be touted as a “plug & play” unmanaged Switch. It’s possible Melco “dumbed-down” the Buffalo Software, but that cannot have any impact on Ethernet packet integrity, let alone any “sound quality” difference.
1 Like
Thanks but my point is not regarding the unit’s origin but rather that according to Ainslie it does not have ( or require) a fancy clock or power supply. In regard to the latter it comes with an unremarkable SMPS which the user can replace of course. I did which made no difference IMO.
2 Likes
A fancy case and/or a high quality power supply are critical to the sonic performance of a network switch. By fancy, I of course mean a case which performs a function like blocking RFI: ideally made from a good conductor as the best conductors make the best shields, with no LEDs and no ventilation holes. By power supply, I mean definitely low noise and ideally linear. Clocks are a red herring in the ethernet space.
(Declares an interest as a manufacturer of award-winning audio-optimised network switches)
3 Likes
Come now Nigel, we’ve covered this before; any such supposed RFI on Ethernet connections are quite easily mitigated with a simple fibre connection. Problem solved. And PSU issues are also easily solved very cheaply.
Absolutely no need to spend thousands on an Ethernet Switch for home audio 
Come now Anup, we have indeed but you’re still not listening!
The RFI is not supposed, and is the very reason many audiophiles use a switch (whether expensive or not) a short ethernet cable before their streamer. And a “fancy” case can play a very important part in RFI mitigation, way beyond any cosmetic considerations.
A simple fibre/fibre connection is a fine alternative, and thanks to your help amongst others some months ago this is what I now use. I also use a switch between the downstream FMC and my Innuos streamer which I’m sure you’ll assert can do nothing at all for sound quality… all devices powered by independent linear PSUs.
Spending thousands is great!
Quite right, RFI is not just “supposed”, it’s always there at some level. The “supposed” part is whether it actually has any real effect on a Streaming DAC past proper Ethernet magnetics.
Here’s an idea, since you’re an Ethernet Switch manufacturer and you own a dCS, why not share some objective data to show us the before/after effects of Ethernet Switch RFI being “fixed” with your connected dCS? 
Great idea @anupc, do you have any ideas of a suitable test? Since any induced distortion can only be measured on the analog outputs I would imagine the problem is not so much the measuring as the proof of both correlation and causality. Maybe if one can induce RFI at a known frequency under controlled conditions the resulting effects will be easier to identify?
I would be interested to hear your thoughts on Paul Miller’s lab test of the Melco S100 switch where he found minuscule but repeatable reductions in jitter and uncorrelated noise from inserting the switch before three different DACs (unfortunately no dCS), despite the presence of proper Ethernet magnetics.
Now you’re just being silly, and perhaps rather rude. None of my customers use measurements, they use their ears. Are you implying, through your insistence on measurements and inverted commas, that either I’m kidding my customers, or they’re kidding themselves, or it’s some other form of psychoacoustic manipulation? This is nonsense if so.
RFI absolutely has an audible effect on a DAC, including such an excellent DAC as a dCS.
As the RFI is not supposed, the RFI is fixed (no inverted commas) by the proper use of a switch.
Peace and love and, please, understanding. 
Calm down Nigel if you don’t have any measurement data, thats your business.
A bit much though calling the ask for measurements “silly” or “rude”. It’s standard practice to have objective measurements for HiFi (and Networking) equipment. Whether vendors chose to share those publicly is a different question (which is why Stereophile and others who do measure are doing us consumers a service).
2 Likes
All calm here. I am happy to stand by my description of this as silly or rude:
“Here’s an idea, since you’re an Ethernet Switch manufacturer and you own a dCS, why not share some objective data to show us the before/after effects of Ethernet Switch RFI being “fixed” with your connected dCS? ” Perhaps it’s the inverted commas on “fixed” suggesting that nothing is really being fixed at all.
My point is that any measurements provided are unlikely to correlate to heard experience. Sure, they may give an indication of direction of travel, but ultimately they are only useful WITH that heard experience. If you tell me you’re a potential consumer of my products then I’ll see what I can do… though my pen is not poised over my order book!
Seconded! @Anupc , if you can confirm (a) precisely what one would measure and (b) what “good” would look like, that would be super-helpful. Thanks.
Exactly. Keep your pen comfortably capped. It’s never gonna happen. 
Oops, sorry I missed your post (in my haste to try and “calm” Nigel down 
)
We wouldn’t even need to look for correlation/causality to start with, just for any consistent measurable difference. I did read the review when it was first published.
My best guess based on the very brief outline, is that the Ethernet cable shielding on the Melco CA1E CAT7 cable used caused the S100’s and Lumin’s Ground planes to be interconnected ← that’s the most probable cause for the very slight measured difference.
A couple of additional thoughts maybe worth noting;
-
The fact that Paul didn’t publish any other graphs suggests there were no other measurable impact from the S100.
-
One would have to be pretty super human to hear 5psec of analog jitter difference and a 0.4dB noise reduction at the -146dB noise floor! It certainly doesn’t line-up with what Andrew Everard claims to hear; “Pulling out the Netgear and inserting the Melco S100 put things back on the right path, the music gaining body, detail and insight, and not in a subtle way.” (Yeah… sure…
) -
The differences are so tiny that I don’t expect anyone else can duplicate those exact results as the setup won’t match Paul’s precisely, including minor variations in Lumin/S100 unit production. Pity Paul didn’t immediately test with a UTP cable as well. Plus the fact that he hasn’t investigated this further suggests it hasn’t been repeatable.
3 Likes
The Melco C1AE is indeed shielded and the shield is connected at both ends. The newer C100 is admirably explicit in having the shield grounded at only one, identified, end. Bravo!
2 Likes