Well, I wanted to find out if it is possible to run ethernet on two pairs of a Cat.7 cable, while running audio (line-level) on the other two pairs.
As you probably know, ethernet TP cable consists of four pairs of conductors. For 10Base-T and 100Base-TX, only two of those pairs are required (pins 1+2 and 3+6 on the connector), leaving the other two unused (pins 4+5 and 7+8).
Note: There is also a standard called 100Base-T4, intended for fast-ethernet over non-Cat.5 cable. This one uses all four pairs. It is not very common and not supported by nowadays "standard" hardware. HP invented yet another thing called 100Base-VG "AnyLan", which also uses all four pairs. It's not very common either.
In a Cat.7 cable, each of the conductor pairs has its own shielding, in addition to the outer shielding. Therefore it was my idea to use the two "spare" pairs for audio, with a good chance that the ethernet link would not disturb the audio quality.
I started by taking a 40m piece (about 130ft) of high-quality Cat.7 cable (Daetwyler Uninet 7002 4P), certified for up to 800MHz and suitable for gigabit ethernet. This is the ring of orange cable on the left picture. Actually, I'd need no more than 20m for connections within my flat, but for the sake of this experiment, I decided to try 40m in order to have a worst-case scenario. If it worked fine with 40m, then 20m would be even less of a problem.
Next I cut an old, short Cat.5 patch cable in the middle. I connected two conductor pairs of each half of it to the ends of the Cat.7 cable, using luster terminals and a screwdriver. I connected the RJ45 ends to two different fast-ethernet switches. You can see the switches in the middle of the Cat.7 cable ring on the left picture, stacked on top of each other.
Finally, I had to connect the other two pairs for the audio line. I used crocodile clamps and a 3.5mm plug for the side where the amplifier was located, and another pair of luster terminals on the other side of the Cat.7 cable, where it was connected to the line-out socket of a soundcard (a Creative Labs AWE64) using 3.5mm-to-cinch cable.
You can see the amplifier-side of the cable on the right picture.
ASCII graphics rule. :-)
MPEG audio player Computer 1 | | | | soundcard line-out fast-ethernet switch | | | | 0.5m audio 1m Cat.5 \ / \ / \ / +-----------+ | 40m Cat.7 | +-----------+ / \ / \ / \ 0.5m audio 1m Cat.5 | | | | amplifier line-in fast-ethernet switch | | | | headphones Computer 2
1. Direct connection
For reference purposes, I started by connecting the amplifier directly to the line-out socket of the soundcard, without the whole Cat.7 stuff.
When playing music, the loudest setting I'd normally use is at about 25% of the volume knob. The music is quite loud at this point. I stopped the music, and then I didn't hear anything. I turned the volume up slowly, with the music off.
At about 70% I started hearing a soft noise. At this point I didn't dare to start playing music again, as it would probably have destroyed my headphones, not to mention my eardrums.
I was pretty satisfied with the result of this part of the experiment. The signal-to-noise ratio of the analogue audio parts involved was more than enough for my needs.
2. Without ethernet
Next I inserted the whole Cat.7 mess into the chain. I started using the audio connection only, i.e. the RJ45 cables were not connected to the switches; there was no link.
The audio signal was a tiny bit softer now, so I had to turn up the volume knob slightly to achieve the same loudness (or increase the mixer setting of the soundcard). At first I wasn't sure if it really was the case or just my imagination, but another direct comparison with a direct connection confirmed it. The difference was barely noticeable. I could not determine any impact on the signal-to-noise ratio.
Bottom line: The 40m Cat.7 cable attenuated the audio signal just slightly, which could easily be compensated by turning up the volume a little.
3. With ethernet
Finally I connected the RJ45 ends to the switches. The LEDs indicated that a 100base-TX full-duplex link was established.
Now a soft, low-pitched noise added to the noise that I could hear starting at about 65% of the volume knob. I guess that this is caused by the carrier frequency spectrum of the link.
Interestingly, the traffic on the fast-ethernet connection had no influence on the audio signal whatsoever. It did not make any difference if the connection was idle, if a pingflood was running, a large download, or random NFS traffic. My guess is that the actual traffic is modulated upon the carrier at such high rates that it is way beyond the audible range.
Bottom line: The ethernet link added some noise and decreased the volume level somewhat at which the noise started to be noticeable (from about 70% to 65%). Network traffic is not audible and does not change the appearance of the noise at all.
I regard this experiment a success.
The additional noise introduced by the fast-ethernet link on the Cat.7 segment is completely insignificant. It is merely noticeable at volume levels way beyond what is normally used to play music.
Furthermore, the experiment was performed using a 40m Cat.7 segment and very questionable ways of connecting (crocodile clamps with non-shielded cables, luster terminals intended to be used for 220V lamps). The final cabling in my flat will be no more than 20m, and the connections and wirings will be much better, using fixed wall sockets. Therefore I expect the audio quality to be even better in the final installation.
At a later time I have also measured the frequency range using a noise generator and a 10-band analyser / equalizer. The result was that the influence of the cable was not measurable with that equipment. I'll add a picture and more detailed description of this at a later date.