Ethernet Cables and Sound

Well, well. this is good and bad news. The good news is that the asynchronous transmission really isolates your dac from the upstream clocks/jitter, and your dac does a great job ignoring those irregularities.

The bad news is that we cannot pinpoint the differences that we hear. Not in noise, not in jitter.


I guess you’re right! Still I couldn’t let it go yet. These pictures shows only 50ms. When we listen to music we are probably able to judge samples of longer periods. Let’s say 5sec to 20sec. For sure I can’t judge sample of 50ms by hearing :slight_smile:

What if it goes wrong a few times per second and not every millisecond. I’m sure it’s possible to hear the deviations. What we need is enough samples that correspond to a sample that we can judge by hearing. Maybe than we find some differences. Luckily I made far more samples than I showed and judged yesterday. Here’s a clear difference. I’ve soomed in on the samples and they really shows some clear deviations. I guess the digital stream was in trouble there!

First a sample with conditioning and second unconditioned

Doubleclick on the picture to swap the images

I have a dozen high end Ethernet cables from half a dozen high end manufacturers and I cannot hear a difference. I have decided to accept this and will no longer play. I now feed my gear fiber optic.

The end.


I have 4 budget ethernet cables from 2 manufacturers (3 x 1m BJC CAT6a and 1 x 25m AQ Pearl CAT6a) and a fibre optic cable because there is no difference. I don’t think I was ever in the game.

p.s. Am I the only one to flick through these charts as being totally meaningless? Normally these things are overlaid to show a difference, whether or not audible. These are just random squiggles?


They should not be meaningless :slight_smile: They should give us answers wheter or not the deviations in the digital signal could point out deviations in the analog signal. The first few were not proparly lined out (sorry for that!), but you can simply click and swap the images to see if you can detect some differences.

It first looked as if there was no influence at all, but if you look into the last two pictures. You can see a clear difference. I already found more differences that I will post later this week. My momentarily conclusion is that the better a digital signal is the lesser jitter will provoke deviations in the analog signal. For sure short samples like a millisecond or 50 milliseconds can be free of jitter, but on longer periodes like a second or 10 seconds several deviations will appear when a signal is bad. This actualy is logical because we hear clear deviations as well, but not in 50 milliseconds.

Oh my. this is very interesting. Congrats for your persistence.

Now, let’s dig a bit deeper. You would help to know the magnitude of the differences in dbFS. Could you tell us what the Y-axis scale is? Also, could you FFT one of those differences, so we can see them in the frequency domain?

Much appreciated all your effort to take the measurements and to share them.

Hey @aangen,

see that we are talking way more than cables. @Wijnand described the deep mods he has implemented in his streamer.

We are looking at the effects of linear power supplies, more precise clocks and dedicated ethernet card (@Wijnand, please correct me if I got this wrong). This is a chance for us to see what we hear.

1 Like

Thanks! I will for sure dig deeper, we get close to a breakthrough :slight_smile: The Y scale is the voltage of my dac output. The volume is at 0.0db maximum is +14db. At 0.0db you can see +/- 3.5V on the Y-axis. The measurements are done on the RCA output which is -6db in comparison with the XLR output. Yes I will FFT. The next images that I will shoot will be of a violin… so much higher frequency. It’s more likely to capture more deviations on which I can zoom in.

More later this week! It’s quite a job changing network and shoot images :stuck_out_tongue:

You are welcome!

Had to do a quick tidy-up, so here are some images of cheap and effective wired ethernet with no switches or crazy gadgets.

My modem has two outputs, one is a BJC CAT6a that goes to a TPLink media converter powered by an iFi 9v low noise power supply. The other goes to a standard Netgear switch that distributes around the house generally.

There are three cables going downstairs, all are 25m.

The first is fibre optic for audio, that goes to another media converter powered by a SoTM 9v battery supply (lucky to get this used very cheap, otherwise would have used another iFi). This has a BJC CAT6a going into the Innuos server above and that outputs by another CAT6a to Devialet.

That is the entire wired signal path before the speaker cables. Using iFi power supplies, the total cost for power, converters and cables would be about $250.

There are two other cheap 25m cables, CAT6a for an access point next to the audio for wifi and Spotify to the Devialet (which my kids prefer) and a CAT 5e for the TV. These both come off the Netgear switch.


I started with a lower cost Ethernet to Fiber converter then spilled for the more expensive Sonore device that I power with an LPS 1.2. So now the only connection to my audio rack is Fiber Optic to the Sonore Signature Rendu Optical. I moved my Roon server into another room and it is being fed by a EtherREGEN with a Keces power supply powering it. The Ethernet cables are Inakustik CAT 6 formally Inakustik Cat 7, formally Wireworld, formally BJC, formally oh the heck with it. :slight_smile:

What on earth do you have under your P8 power strip???!!!

This has been a really interesting thread.

Before I ended up returning to my DSS and I2S, I spent a month messing around with the network feeding a DCS Bartok and found that just about everything changed the sound in some way. Using an EtherRegen+LPS 1.2 and Zenith MKIII made the most difference. The ethernet cable did change the sound somewhat, but so did the DC cable to the EtherRegen.

Honestly, I thought the whole situation was extremely analogous to the USB to I2S I had before (which I’ve since returned to), with the EtherRegen taking a place comparable with the Matrix X-SPDIF2. The main difference was that that ethernet cables don’t make as big a difference as USB and HDMI.

It is strange that this is so hard to measure, though.

I hope in the future we all go to something like I2S over fiber and be done with it.

1 Like

Nice the EtherRegen and the LPS.

Maybe it isn’t that strange that it’s so hard to measure. The clocks in our dac’s probably have a higher precision than the clocks in most scopes. The clocks that I use in my dac and in my network have a stability of 60 femto seconds, which scope could measure that? Even when one would measure with a GigaHz scope. Then it’s a million times short to measure a femto second. Due to the near perfect timing the waveform is build a tiny bit better and we can hear that, but still we talk about a precision that is very hard to measure.

Besides this I think the clocks in an average scope comes nowhere close to the precision of an audio clock. So the tiny differences that we see could be due to the deviations of the scope.

What are we looking at on a scope. We can see the frequency of a measured waveform, but we can’t measure the sound of the sound. I mean several voices can reach 3 to 4Khz, but they will al sound different. Maybe we approach this sound a bit better, by upgrading our networks. On the scope we alway’s see tiny differences even when we measure the same sound twice. How can we lay a finger on the differences without having the original soundwave?

Still, maybe if we search and look hard enough we can find something!

Simple = good. Complex = :man_facepalming:t2:

1 Like

I wish it were that simple with audio quality, of course most of us don’t want to unnecessarily complicate our setups. Hopefully someday we get a universal digital transfer format that doesn’t need a slew of little boxes to sound good. Unfortunately, we aren’t there yet and there are way too many options.

1 Like

I suppose I could replace the receiving TP-Link MC220L with an EtherRegen. i.e. a component that that costs £17 with one that costs £665.

I find it difficult to get my head around that given the excellent sound quality that I have.
Plus those and the Sonore stuff have product names that make no sense to humans, I assume were designed for computer audio which I don’t use (no computers in my system ever and usb only briefly), are not exactly mainstream products. You need a degree in engineering to understand a word of their blurb about what the thing actually does (see for example To me, flip-flops are beach footwear.

The P8 is actually a Shunyata Venom thing and it sits on a pair of these:

The Innuos sits on my other pair.
I bought them for my speakers, a bit of an experiment that was pointless, but they work a treat for isolating components and I have my turntable on a similar Townshend product.
If you know the right people in the UK you can get them at dramatically reduced prices, the retail list prices are a bit silly.They are extremely popular products, especially the speaker podiums.

I consider the Innuos to be a computer. I am not surprised that you don’t.
I don’t use an EtherREGEN to connect up with my Signature Rendu Optical, no sir.
I could, but why. I use it for other things, like my Roon server, which is a computer.
I was talking about stepping up from your TP-Link to this:

It’s a little bit better IMHO. It costs $289. I power mine with an LPS 1.2 that I stopped using for other things.
I have that same Shunyata Venom thing but I just went with the fancy metal feet.
It’s a power strip don’t ya know.

The Fancy metal feet were £250 and I had the bars, so that was that. There is also a few scraps of acoustic foam in there.

My definition of a computer is something that can run Excel. Can’t do that on my Innuos, it’s a music server. By your definition I’m a computer as given time I’ll compute that 2+2=5.

The Venom is a power strip, but a whizz-bang one that is quiet, provides a cleaner sound, a load cheaper than the regenerator it replaced, has 6 rather than 4 sockets, the plug sockets go up rather than down (a design fault on UK regenerators) and leaves space for other important things.

I might try one of those optical things if I can get one on trial, but a lot of money for a few extra parts. Almost $400 in the UK.

I understand when @aagen considers innuos as a computer. it runs on a supermicro motherboard custom built from the X1SAA model (modified, but yet a PC motherboard). It uses a pentium n4200. it operates as per a modified linux OS. Its storage is a regular SSD. It surely has enough horse power to run excel. And should you tinker enough, I believe you could install a standard linux/windows distro on it.

AFAIK, the advantages reside on the simplification of the chipset (elimination of some subsystems to reduce EMI), clocking (using high quality OCXO), and specially on the power supply. But it never ceases to be a computer inside.


Let’s not pooh-pooh putting anti-vibration products under power conditioners, power strips, regenerators. I attended a demonstration of such products by the CEO of Harmonic Resolution Systems, a PhD with much defense and NASA experience with vibrations in missiles and the like. The products were “pucks”-footers and plates to be placed in the tops of electronic components. First, playing a demo on a very high end systems Then, with pucks under each of the components and plates on top. A very nice improvement. Then, just use the pucks and plates on the Transparent power conditioner only. Result-a very noticeable improvement. I use these products and enjoy the improvements.

Mark H.

I never offered a definition. Its a computer. What you choose to believe has little bearing on reality. You are a special person.
I have a Venom Power strip. I know what it is. it is sitting two feet away from me. It’s a power strip.

1 Like