Here’s a video from Belden with an analyser hooked up in real-time showing deviations based on movement and coiling, and a bit about return loss and reflections:
Interesting read these CA ones but very confusing as to what’s good and what’s not.
Yeah, his posts got me thinking about two things.
Cables - I wonder how close the Belden bonded cables are to his $700 testing cable. He doesn’t end up mentioning what it actually is either.
PHY - To get the best performance, aside from the cable, you’d need decent PHYs on both the audio streamer and the switch? You’d hope that the audio manufacturer got that right on their streamer end. Even if they didn’t, I don’t think there’s much an average user can do on that end. However, the user I guess could look at the PHY’s design on the switch/router.
Hmmm, I wonder if that was maybe part of the rationale for suggesting direct connection between the Macs and the endpoint directly. I remember reading that Macs in general have decent PHYs from Broadcom.
I’m just thinking out load. It’d be good to better understand with what’s actually happening and what the measurements tell us.
Yup as John has said, ethernet input DACs (or other endpoints) suffer similar issues to USB physical interfaces in this regard (extra work done by the receiver when fed ‘poor’ signal integrity). But just like USB DAC (and endpoint) physical interfaces (and designs to implement them) have improved greatly over the last few years, we should expect the same improvements as ethernet input DACs become more common over the next few years.
Obviously Uptone have hinted at a product of their own solving this issue (as you’d expect) but John had great success with his REGEN’s and Rendu’s that focused on signal integrity, with positive feedback by both end customers and DAC manufacturers.
Couple of other things that I’ve thought about (BTW is John active here?):
- With regards to return loss and reflections, for a constant given cable like the Belden bonded, what would be the recommendation regarding length?
I do know that with some switches, like my D-Link DGS-1005 and DGS-108, these are green switches and use full power if the cable is longer than 20m. Under the 20m, the power consumption is less. However, I don’t know how the overall behaviour of these switches under both settings differ, or in comparison to an enterprise switch? Is one more preferred or stable (aside from the power supply considerations of course)?
- Fiber ethernet - With light, we know that there is no electrical connection, but is signal integrity and variance in power consumption also a thing with fiber? Is it more stable than the copper systems, in general?
I wonder if anyone has experience with Bel Canto’s “The System” with the fiber inputs. It looks serious at $50k.
Minor side note: I do have a separate story to share about the wireless adapter I used directly off the BDP-1’s bus power that I posted a picture of above awhile back. It’s variance in power consumption and subtle impact on SQ could be heard depending on what frequency (2.4 vs. 5 Ghz) the dongle was on and/or what frequency the other devices nearby on the network were also on. The less harder the dongle had to work, the better. Of course, this is highly dependant on everyone’s own environment physical (walls/structure) and network saturation. Simply moving the Wifi adapter to an external powered USB hub could offload all that, provided the power supply for the hub is decent enough.
Anyways, back to ethernet!
He doesn’t use Roon, the last time I checked with him - maybe he’s been converted since lol.
You can just signup to CA Forum and hit reply to any of his posts and he should reply directly back to you.
The fiber is just used between the controller and the two monoblocks. The D/A converter is moved from the controller to the amplifier chassis and ST fiber is used to connect them. So this fiber has nothing to do with the LAN connection that is brought in on a regular RJ45 connection.
I use Ghent ethernet cables that are made from Belden 1303E CAT6A cables (screen is not connected at the ends but use a grounding technuiqe John Swenson came up with on Computer Audiophile called JSSG loop)
The cables and contacts are really well made and fit and finish is really good. They are tested with a Fluke testinstrument and PASS CAT6A without any problems
He uses guy from Blue Jeans fo this test.
John has stated that he believes the audio companies that make Ethernet cabling don’t know what they are doing.
Sorry but you are spreading misinformation with that.
3 ways STP cable is done:
Floated shield. It isn’t tied at either end. Generally used used in horizontal runs with other cables. Prevents inter cable loading.
Chassis tied shield. When your premise equipment doesn’t have a grounded Ethernet port
Fully tied shield. When you need shield integrity to extend from Chassis to Premise equipment. Potential for ground loop. Often solved in data centers by running balanced power to a DC supply fanout for the rack equipment.
I concur considering they claim to not know why it sounds better either makes me rub my chin a lot. If they cant back up claims with imperial evidence I call snake oil and assuming audiophiles will buy any old thing if we tell them it makes it sound better.
Funny thing is, that audio seems to be the only digital data that is affected by the big cable differences. Hm, as I use the same ethernet cables for my computer while reading this forum, maybe some of the comments here could also be affected by ‘noise’ and are meant in a total other way
When you consider how audio data is a trickle in comparison to what modern eth ports and cables were design to robustly handle, it’s yet another example of subjectivist audiophoolery.
If ethernet cables were really that sensitive, storage area networks would collapse under the weight of continuous packet loss and re-transmission. Even at 10GigE, it’s just not a problem that requires cables hand-wrapped by Tibetan monks (although in the 10Gig case, it’s often an optical SFP for better performance at longer lengths).
A minor point but I don’t believe anyone is discussing/assuming dropped bits… Hopefully this entire discussion is about bit perfect playback. This was (hopefully) established way up in the thread.
The points I was trying to make, with links to John S’s posts, was that an audiophile cable (typically shielded) may potentially actually make things worse, if it provides a path for ground / leakage current loops to form - depends on the rest of the system.
So to be safe, stick with unshielded - or a floating shield design if you really needed shielding, like Belden 10GX series Cat 6A cable (sold by BJC).
The Meicord seem to get consistently favorable reviews and their cost isn’t prohibitive.
They don’t make outlandish claims , I use them throughout my system and don’t think or worrry about it any more.
Thats like saying a 5 oz bag of potato chips that costs $25 isn’t ‘cost prohibitive’.
Any way you slice it the price is absurd. But if there are foodies willing to pay $25 for a 5 oz bag of standard ol’ taters. Then there is someone smart enough to capitalize on them.
No they don’t. IMO they pay someone else to… When someone doesn’t believe something about what our software does (and it does indeed) I tell them don’t take my word for it, take my customers word for it and I give them a few names/numbers.
Can you share more about these unexpected phenomena?
Is it mainly the issue/s around leakage currents via USB ground that John Swenson has shed bright light on? And the extensive (and expensive) efforts required to block these leakage current over USB? And even then, they are still not entirely blocked in most cases, whether done inside the DAC or with a USB isolator.
Any other gremlins?
Cheers in advance
Mark, I have Meicord cables, short, 1m and long - 5 m. The are “sounding” definitely better than standard cheap no-name Ethernet cables. I can tell you even more - if you’ll power ethernet switch from battery it will improve sound again.