Interesting questions and comments.
I must admit my interest in something like the Fibrebox lies not in the claims it is making regarding clocking etc., but rather in the fact it may simply be an elegant single box solution (as the original poster claimed).
My current fiber solution consists of two separate Cisco 2960 switches. One receives ethernet - it then connects via an optical cable to a second Cisco 2960 - and finally an ethernet cable from the second 2960 to my Rossini.
The Fibrebox may simply be a single box solution performing the same duties as my dual Cisco switches (perhaps with the benefit of one less power supply in the mix?)
Then again, in my case, it may be a solution chasing a problem.
Yeah, similar to the GigaFoil that I’m using:
Have a Keces P8 supplying power to it, my Nucleus+ and a basic unmanaged Cisco switch (CBS110-8T-D).
Works a treat. Sold everything else after doing my best to make something more complicated sound any different.
The thinking goes: the more (frequently) the frequency of the transmitter and receiver change relative to each other (for example due to phase or thermal noise of an oscillator), the more frequently the Phase Locked Loop (PLL) has to adjust its output to compensate.
It would seem to follow that a PLL constantly making adjustments would output its own noise, but how much that influences other nearby circuits would be dependent on the design. I think dCS have really thought through noise as an issue, however this isn’t true of all manufacturers. In the case of the Vivaldi, the whole ethernet circuit is eliminated from the DAC.
The Fibrebox (and others) use (what we’re told is) a tightly controlled oven-controlled crystal oscillator, which should minimise this effect as it shouldn’t suffer so much from frequency drift.
For the reasons Zapp gave, a typical ethernet card isn’t going to be too concerned about phase noise, but in an audiophile environment where noise matters much more, having a more accurate clock could help reduce the noise from the PLL being injected into the rest of the system.
PS I should add that, while I have heard others talk about this topic, this is merely a suggestion from my side as to why reclocking an ethernet signal might make a difference
I’m no electrical engineer, so I’m not qualified to opine on this argument, but I confess I am circumspect. This excerpt from the Fibrebox website in particular sounds like all-too-common audiophoolishness:
We invented the synchronous clock to optimize the 2 fiber modules to reduce jitter to the minimum! Because the two modules completely use synchronized clock signals, the optical fiber has exactly the same action requirements when transmitting and receiving signals! So as to achieve the effect of reducing jitter. In addition, a high-quality signal transformer completely isolates the clock and fiber module, so it is absolutely very low phase noise.
Like Ben, I use a GigaFoil powered by a Keces PSU in front of my networked audio devices. I don’t know that my SQ is better for it, but once I established it wasn’t worse, I quit worrying.
I’m inclined to agree, my primary reasons for buying the Fibrebox were to optically isolate the music server and its simplicity doing so; the fact it was built for audiophiles meant that it has a decent PSU; and it sounds better than the TP-Link fibre converter I tried out.
The fact it seems to have better clocking is a bonus that I’ll take. I believe it does make a difference on one of my Dacs, but not so much on my Rossini.
It pains me to say that this is often a result of someone in marketing not doing their job. If the writer doesn’t understand what they’re writing about or the audience they are writing for, then gobbledygook is typically what comes out.
I hasten to add, this may not be the writer’s fault, but of those who are meant to brief them.
No argument there Jeremy. I want to stress I am not saying this is BS. Just that I am looking for an explanation that makes sense.
Jeremy, many Ethernet Transceiver chips do not use PLL for the Physical Medium Attachmemt (PMA); the Atheros AR8030/8035 transceiver chips used in dCS platforms are among those. Instead of PLLs, they have high speed internal ADC/DACs to detect/transmit bit streams, hence bit stream jitter is inconsequential to it’s operating characteristics (to a wide tolerance that you wouldn’t ever see in a home environment).
Plus, whatever noise from the incoming Ethernet line or generated on-chip, are shunted to ground when interfaced to the MAC. In the case of dCS, that’s the Texas Instruments Sitara ARM SoC, and the resulting I²S signal out of that Compute module are totally decoupled from the incoming Ethernet.
Without going into further excruciating technical detail, let’s just say categorically that having an OCXO on a Ethernet Media converter (or Switch) serves absolutely no useful function (other than to convince Audiophiles to spend more $$ )
@Anupc Interesting, thanks. I’m a bit confused that the notes for these chips appear to say that the PLL can only be switched off for 100Mbps connections, but haven’t had sufficient time to dig into this further.
As I say, I haven’t heard as much improvement by making changes on the ethernet side of things as with other manufacturers devices, so I’m happy to believe that dCS have largely designed/engineered many of these issues out.
Has always thank you for the technical explanation.
@glevethan Why are you using 2 switches? Where is the benefit of that?
Has been released in Japan: Telegärtner’s Japan new optical isolator “OPT BRIDGE 1000M” - as far as I know isn’t available outside Japan as of yet. Price is 349,800 Japanese Yen.
I don’t have any more information about the product.
And still my “audiophile” switch has a positive influence when streaming Qobuz to my dCS Rossini.
As the originator of this topic, I thought I would update that I did purchase ADOT but the single mode version with an 8m length of fibre between convertors. It definitely makes a subtle but useful improvement to the extent that Qobuz streaming now approaches CD replay with my Rossini APEX which is what I was hoping for. Could I assemble cheaper components from Amazon, yes obviously! Would they sound as good, no clue and can’t be bothered to find out. The outstanding support I have received from the ADOT supplier has been of great value to me.
Yup, whatever may be happening inside, I use an Innuos PhoenixNet and it does its magic. Yup, totally spooky if we want to believe technical explanation. All I can say to you all who have spent real good amount of moolah on dCS stacks, there is no harm in trying additional $3-4k and I am sure whoever dealer you go to, should offer a return.
+1 for the ediscreation Fiberbox.
It was a fantastic addition for me combined to edison’s switch.