In this Applications Note we focus on how to run audio wires long distances. There are numerous challenges that need to be dealt with including signal amplitude loss, distortion, ringing, crosstalk, ground loops, and so forth.
There are a couple of things we don’t have to worry about. We are running line-level voltages, so conventional noise sources are not a significant problem. Also, at less than 1000 feet (300 meters), transmission line effects are negligible.
The MX3 differential input mode
The MX3 series of devices has an option to adjust an internal jumper so that it works in differential input mode where the Left and Right of the 3.5 mm input connector are differenced before being sent on the A/D converter. Thus common mode noise that raises both wires by the same voltage is rejected. The differential mode is very useful for the long wire challenge. As of this writing, we do not believe that it is available from any other Wi-Fi Audio vendor. (Note that by default, the systems are not shipped in this mode, but can be configured at the factory or in the field.)
One of the big challenges in running long distance wires is interference. The lines intersect all sorts of interference sources such as the areas near electric motors, florescent lights, other wires, computers, and so forth. The best way to deal with this is to send signals in what is called a balanced configuration where one uses two wires and drive them so as the voltage on one goes up, the voltage on the other goes down by the same amount. The signals that come out of satellite receivers, DVD players, TVs, and so forth are almost all single ended, not balanced. The device the converts a single-ended signal into a balanced one (or back) is called a “balun.” With this clever device, interfering signals affect both the + and the – wire the same way and the difference is zero. Thus common mode interference sources are rejected. This is a huge advantage when running a long distance in an electromagnetically hostile environment. You may have seen XLR connectors that are used for microphones—they use this balanced wire technique.
Baluns, first try
So we started by ordering a bunch of low-cost baluns off the Internet. I got three types, shown here, and tested them. Not one was actually a balun.
They were all just connectors. Argh. Real baluns generally use transforms and so are more expensive.
If we had real baluns then we could have connected the single-ended signal into the balun, run balanced over the long distance, and then converted back with a matching balun at the far end. With these non-baluns, that technique does not work. But one can still send the signal from one end to the other and count on noise sources influencing both wires the same, but one needs a way the subtract the two signals.
Happily, our latest generation of audio input devices has several jumper positions where one of those configurations enables a difference amplifier input. Thus by connecting the two wires at the far end of the line into our difference amplifier, common mode noise is removed.
Long transmission lines are part distributed capacitor and part distributed inductor.
In order for them not to ring, you need to terminate them.
If the line was kilometers long, one would need to do a matched termination, but for lines less than 300 meters such as we are using, that is not necessary. Most transmission lines have a characteristic impedance Z0 between 25 and 300 Ohms. CAT6, which is what we were using for this experiment, has a Z0 of 100 Ohms.
When we used a matched termination the waveform was beautiful, but we lost a lot of amplitude so it wasn’t practical. We had to make do with a non-matched termination.
For our experiment, we used 250 feet (76 meters) of CAT6. The wire was solid bare copper. It had a gauge of 23 AWG, had a measured series resistance of 11.6 Ohms and a measured capacitance of 3.8 nF.
From the capacitance, length, and characteristic impedance, one can calculate the inductance, but we don’t bother here because it is negligible in the audio frequency range. The square waves we used for testing were at 2 kHz. While we used CAT6, a good audio wire can be had here.
Using our non-balun connectors, we were careful not to ground the line at both ends so as avoid creating a ground loop.
Ground loops act like the secondary to naturally formed transformers and pick up voltages from any time-varying magnetic field that cuts through the ground loop. In this case v2(t) is not just a delayed version of v1(t) but has the addition of noise from all sorts of adulterous sources.
60 Hz noise was particularly bad. We used an old electric drill to generate extra noise, though there was plenty without it. In the real world, electric motors in the environment, florescent lights, and a thousand other sources conspire against the ground loop.
So what we did was keep the receiver end ungrounded and connected one wire to the + input on our differential amplifier and the other to the – input. For our MX3 system, the + and – come in on the R and L ports of the 3.5 mm connectors. The signal looked pretty good in difference mode, but horrible if one compared either wire to “ground.” We tested open, terminated with a 600-Ohm resistor, and letting the ground loop happen but still pulling the signal differentially. The 600-Ohm termination looked the best, but needed about 12 dB of gain to get the signal back to the right level. Happily, we have that option in the system. So this technique worked, but didn’t feel ideal. It is, however, less expensive than the balun option, below.
We went back to the Internet looking for baluns. We found two options that looked great for our application since they not only provided the balun function, but also summed the left and right signals at the source. One is the ASA 141 available from Extron Electronics . We didn’t try it but the specs looked good. The other was a DB11-STMB from decibel11, which is what we used. At the receiver end we used a three-terminal to stereo 3.5 mm connector, leaving the ground connection open (though you could connect it to a shield if the cable had one). We simply hooked it up as seen and put the output differentially into our system. The signal looked great and when we hooked it up to the system, with an extra 12 dB of gain, it sounded indistinguishable from the original, at least when used with moderate-quality ear buds. Note that we did not use a terminating resistor but we were careful not to ground either of the +/- signals on the receiver. So this is our recommended configuration.
Works great. Avoids ground loops. And is pretty inexpensive.
By the way, one can always use a balun on the end closest to the ExXtractor also. In this case one would not need to change the jumper in the box, but one must be very careful to attach the second balun correctly.