Jud

If I were to look at potential causes (I don't have the facilities to measure, nor have I seen measurements), then greater common mode noise is certainly one possibility, dependent on the particular system. (That is, do the specific network components in the specific system configuration result in a common mode noise current flowing through the system such that there's an audible difference?) Another possibility that occurs to me is electrical noise from the final opto-electronic conversion, again dependent on the particular system.

Nope, nothing has to be. But being curious myself about possible causes of less than perfect sound and ways to improve, it just seems more efficient and less wasted time (to me as an old guy who's more conscious of time) to try to figure out what's going on.

It was less than 40, something like 32 or 36, for 10G optical as I recall. It was shown in UI, which I back calculated to ps. The slave jitter number of 20ps shown in your reference is quite close to my recollection of 16 or 18 for 10G copper.

What we want to do then in order to improve things is to determine what's causing any distortions, in order to be able to remove them if possible. For example, we've discussed jitter a fair amount here. But jitter is a *digital* distortion, and we don't hear digits, we hear music, analog. So what is the actual "sound" of jitter, that is, what sort of analog distortion does it cause? Well, the analog distortion caused by jitter is "FM sidebands." These are frequencies higher or lower than the true, or "carrier" frequency, modulated by that carrier frequency. In other words, the analog result is harmonic and intermodulation distortion. The amount and audibility of distortion is going to vary not only with the amount and type of jitter, but also with the frequencies, harmonics, and intermodulations going on in the piece you're listening to. I (and virtually everyone else who's tried it) has found that this sort of distortion is quite a subtle thing, when it becomes audible at all, in modern electronics. For example, I believe I can hear just a bit of distortion in a second or two of a particular test track when I add 800 picoseconds of jitter, more than twenty times the maximum allowed by the 10G optical spec, and more than forty times the maximum allowed by the 10G copper Ethernet spec (at least as I read those specs). (As @jabbr has often pointed out, this is taking place in the network itself, and is not a measure of what's going on in the DAC.) You're welcome to try this out on your own ears, as our friend @pkane2001 has provided the aptly named "Distort" software that allows the addition of various types and amounts of jitter and other distortions. It's available at https://distortaudio.org . Note that I and others experienced errors the first time we tried to run the software, but a restart of the app cured this. With the aid of such software and the various types of distortions it permits, you may be able learn what sorts of distortions might be causing any problems you're having, and thus have a better idea of how to cure them.

Also note that no distinction is made in the linked article between cause, effect, and correlation. That is, are the gross differences in noise the result of picosecond differences in jitter, or are both of these effects of another cause (for example, the different power supplies)?

Yep, just responding to remarks about what jitter caused distortion might sound like by saying you can actually listen and hear for yourself what it sounds like, and at what level you can detect its effects. This in turn should help you to determine whether any shortfall you hear in the sound of your system could possibly be caused by jitter or not.

There is software that can introduce various levels of phase noise into the signal, so that you'll be able to hear what jitter caused distortion sounds like, and the minimum phase noise level that's audible to you.

Yes, I've made a distinction between jitter (timing) and data corruption (bit flipping or loss of data). The former can cause slight, perhaps even inaudible distortion, depending on level. The latter you'll hear.

Don't allow this fact to obscure a more important one. In the analog world you're accustomed to, minor variations in data correspond to minor variations in sound. In the digital world, this intuition no longer applies. Flipping single bits will cause large, clearly audible changes in sound. So unless you are hearing dropouts and grossly distorted sound, your data isn't changing.

Yep. Yeah, the only part of this I'm at all interested in is "the last mile" - where any electrical noise hits the circuitry of the endpoint, and if that endpoint contains the DAC and its clocking. Still, my supposition is that at least with modern 10G spec, it's far more likely any electrical interference is coming from other sources.

Except that I am not talking about electrical noise disrupting the Ethernet signal, which would result in corrupted data and the type of grossly audible interference I was saying we *don't* hear. Rather, when I refer to electrical noise in a complex system, I'm talking about a large number of boxes plugged into power with lots of wires running between them, IOW a typical high end audio system in a typical modern home. I know you have taken measures to minimize electrical noise entering or circulating through the system, but not everyone has. And whether, for example, photoelectric conversion might introduce more noise than a copper connection in a given system, and whether this affects in any way the audible result, is something that may well vary from system to system. (Unless I'm reading it wrong, I note the optical 10G Ethernet spec allows for about twice the jitter of the copper 10G Ethernet spec.)

If you want to understand what you're hearing, you need to know a few things: (1) The data itself isn't changing unless you're hearing very grossly audible distortions and dropouts. Digital isn't analog - change or deletion of a bit or two will be clearly audible, not give you slight differences. (2) Any changes will almost certainly come from two sources: Either the timing of the data (jitter), or good old electrical noise and interference. The latter is more likely, especially in a complex system. (3) Changes in your perception of the sound can come from many causes other than changes in measured sonic characteristics. The type of music and whether it excites room nodes, your room, background noise, what's going on in the rest of your home's electrical system, your own mood, can all vary very quickly.

What is it about a specification that requires various optical Ethernet cables and terminations not to transmit jitter that you would like us to derive from a test of common mode noise reduction methods in copper Ethernet cables and terminations that aren't built to this specification?

Yes.

I'm almost certain it does. The only caveat is that although the Fitlet3 has an SFP+ input (and I have a 10G spec transceiver in it), internally it is set up to handle 1G, so I'm sending a maximum of 1G over a 10G spec connection.